Philosophy of Science

Lecture 7 Kuhn normal and revolutionary science 1 Last Week We have now looked at two major aims of science – law and explanation. We have a richer picture of how science works. This week we are going back to theories of how science proceeds - Kuhn. This week, focus on understanding Kuhn’s theory itself. Next week, worry about implications – the ones Kuhn claims, and whether his account is plausible. 2 Introduction to Kuhn Thomas Kuhn was an American, and lived from 1922 to 1996. Originally a scientist, getting a PhD in physics from Harvard. Then he became interested in the history of science, and later in the philosophy of science. He published three major works: 1957 The Copernican Revolution A work in the history of science - the period covering the change in world view from the Ptolemaic system to the Copernican. Mainly historical but contains a lot of philosophical comment. Informed: 1962 The Structure of Scientific Revolutions very much on the philosophy of science, giving an account of the process of science. 1977 The Essential Tension a collection of essays modifying and defending his views in Structure. We've seen that the accounts of scientific method given by Bacon and Popper are opposed. But they share some crucial core assumptions which Kuhn opposes. Run through: 1 Observation and theory Both Popper and Bacon think there is a strict distinction between observation and theory. What we observe is independent of the theory we use to account for that observation. Observation is privileged – it is more Phyllis McKay Illari: Material provided for Philosophy of Science Unit important. You can compare two theories to see which fits the independent data better. Kuhn challenges this. 2 Logical relation between theory and data Both Bacon and Popper looking for a logical relation between a theory and the independent data. Bacon thinks theories are supported inductively by data, where Popper is concerned with a deductive relation. We accept those theories that are unfalsified. Relation is logical. The theory is justified by this logical relation to the independent data. Social and historical factors are irrelevant to whether a theory is justified. (Popper – historical and social factors might be relevant to finding a theory, the conjecture, the context of discovery, but not to the context of justification.) We have already seen that Duhem and especially Quine challenge this view. Kuhn does so even more. 2 Knowledge accumulated Transition from one theory to another is cumulative. Science makes progress, gradually accumulating more knowledge. When a new theory replaces an old one, the good logical and empirical consequences of the old one can be preserved. Newton’s theory shows Galileo’s laws of motion and Kepler’s laws true, because both are subsumed by Newton's overarching theory. Popper not so committed to this point as Bacon. Nevertheless, Kuhn’s views are much more radical. These three points are the core of the views Kuhn attacked: science does not accumulate knowledge, observations are not theory-neutral, and there is no logical relation between theory and observational data. So what does Kuhn say? QUESTIONS Phyllis McKay Illari: Material provided for Philosophy of Science Unit 3 Normal science Kuhn says that there are two sorts of episodes in science – normal science and revolutionary science. They are importantly different. As you may expect – revolutionary science involves the fun stuff. But look at normal science first. It is what most scientists spend most of their careers doing. Revolutionary science is much rarer. The two are distinguished using what Kuhn calls a paradigm. Normal science is ruled by a paradigm. In normal science the paradigm is unquestioned. The paradigm governs the theories, objectives and values of the community, and the models they use. It tells them what direction of inquiry looks fruitful and determines what counts as an experiment providing relevant observational data. Paradigm 1st discussion OHP Very important, and very complex. Best way to approach by thinking about the examples Kuhn mentions: Astronomy Ptolemaic geocentric, using cumbersome maths. Copernican heliocentric – eventually ellipses, then Newton. Physics Aristotelian function-based explanation, Newtonian notable because it joined separate areas eg encompassed astronomy and Relativistic paradigms notable space and time not absolute – relative. Biological sciences Darwinian evolutionary paradigm revolutionized how we saw our place in the universe. A paradigm is not just an important theory. An important theory couldn’t do all the things I said above that paradigms do. Paradigms do more. Eg Copernicus/Darwin – the paradigms are still around but we don't use theory, at least not anything like in its original form. A paradigm usually does involve an important theory, but it also includes a group of beliefs, values and techniques shared by a particular scientific community at a particular time. Paradigms also involve objectives, and the models that guide normal science. Phyllis McKay Illari: Material provided for Philosophy of Science Unit What a paradigm is is difficult to codify. This is part of Kuhn’s point – the relations within a paradigm are not logical. Science is a human pursuit happening in a particular place and time. Different aspects of different paradigms might be more important. So paradigms will standardly all have the aspects I have mentioned, but different aspects will be crucial to different paradigms. Take an example of normal science. Newtonian mechanics was the ruling paradigm in physics for about 2 centuries. Holding that theory didn’t just mean accepting that the claims of the theory are true. It also involved buying into the model of the universe as a deterministic system kind of like clockwork, and holding that the basic properties of things were their position and mass. Newton's understanding of gravity was the cause of lots of people refusing to accept it until later. It also involved using the standard apparatus governed by Newton’s laws, and adhering to the methodology. Tells you what approaches are fruitful and what major scientific problems are. Changes your world-view, and what you do as a scientist. Kuhn is not denigrating normal science. He thinks it is very important. We understood a lot more at the end of this long period of normal science than at the start. Much of the work of normal science develops the paradigm to its full potential. Newton’s theory was expanded, its implications explored, and its conceptual commitments taken on board by many. Especially the problems about gravity. Few find the lack of explanation for masses attracting each other a problem now. Exploring the full potential of the paradigm is very useful. It is what we’re still doing with Quantum Mechanics and Relativity. But however important all this work is, it is still all governed by the paradigm. Puzzles investigated were set by the paradigm, like the anomalous orbits of Uranus and Mercury, and the methods and instruments used were governed by the paradigm. What counted as success or failure was set by the paradigm. Kuhn calls the work of normal science ‘puzzlesolving’. The paradigm itself is utterly unquestioned during normal science. But this is very useful. Putting aside those sorts of debates allows normal science to Phyllis McKay Illari: Material provided for Philosophy of Science Unit be fruitful, by focusing research on problems selected by the paradigm as fruitful. Core three assumptions OHP. In normal science, the core three assumptions Kuhn opposes largely still hold. Observation is distinct enough from theory, relation logical enough, and the process is cumulative. The process is successful. Some people miss this point about Kuhn – his understanding of normal science explains why science is like this, and why it is so successful when it is. QUESTIONS 4 Revolutionary science Revolutionary science is not governed by a paradigm. It is what happens when a paradigm itself is under serious attack. Ultimately, the ruling paradigm is overthrown – replaced by a competing paradigm. Never just overthrown – always replaced. This relates to issues we've talked about regarding Bacon and Popper – you always replace one set of presuppositions with another. It is not possible to have none. So revolutionary science is what happens during paradigm shift. We've said that a paradigm is more than a theory – some are whole worldviews. During normal science the paradigm is not challenged. So why shift? Kuhn’s point – what happens is not a strictly logical process, something like the paradigm being falsified. Kuhn is interested in the social and historical context. Normal science runs into problems. Either the whole of science or a large section of it. Anomalies accumulate, and expected phenomena fail to appear etc. Eventually normal science is unable to proceed effectively because these puzzles are intractable – they remain even after many ingenious attempts to get round them. Scientists start to lose faith in the old paradigm. They stop believing that it will one day be able to solve all these problems. Phyllis McKay Illari: Material provided for Philosophy of Science Unit A bright spark or two start developing a new approach. If successful enough, they gradually win supporters, who tentatively start working in the new tradition. Some scientists from the old paradigm will convert, and greater numbers of scientists first entering the scientific community will be attracted to the new paradigm - attracted because they have no faith in the old paradigm and think the new one looks more fruitful. Some scientists refuse to abandon the old paradigm. Kuhn points out that they will eventually retire or die off. Then the shift is complete and the new paradigm reigns until it reaches crisis. The new paradigm does NOT win because it is rigorously compared with the old and shown to be better. That’s Bacon/Popper-style view of theorycompetition. On Kuhn's view, this sort of competition actually works socially and historically within a particular scientific community at a particular time. A paradigm attracts individual scientists. Especially in the early stages, these scientists convert for very personal reasons. Kuhn calls them ‘personal and inarticulate aesthetic considerations’. One of the important reasons for believing this might be true is that in the early stages, a paradigm is very poorly developed and can look very unpromising. It is the work of normal science to develop the paradigm. Paradigm 2nd discussion Understand better what Kuhn means by a paradigm when you understand what he means by paradigm shift. You can hopefully see better now that a paradigm is not just a theory. Even a new important theory coming along might not change the paradigm. It only changes the paradigm if the whole way of working and thinking in the science changes. New theories, values, techniques, objectives, standards and models are crucial. Example of paradigm shift: Ptolemy to Copernicus HANDOUT NewtonEinstein and more detail of Ptolemy-Copernicus. Phyllis McKay Illari: Material provided for Philosophy of Science Unit This shift part of huge impact on everyone’s daily thinking. Change Aristotelian-Ptolemaic worldview to now. We are not at the centre of the universe, not special in that way. I'll focus on the change in astronomy from the geocentric to heliocentric system. Not just Ptolemy and Copernicus themselves – the paradigm shift concerns the whole tradition. The theory advocated by Copernicus himself is long gone, but the Copernican paradigm is still here - it is still the ruling paradigm. The Ptolemaic paradigm was around for a long time. When it was first developed, it was very successful, much better than any other system at the time. Lots of work ensued, astronomers found problems but kept trying to solve them. The theory never quite conformed with the best observations for both the positions of the planets and the precession of the equinoxes. But for a long time confidence remained, there was no crisis. As astronomers worked on it, the theory became astonishingly complex for little gain in accuracy. Accumulated errors meant calendars were wrong. Some lost faith. Copernicus was the bright spark this time. But at first, Copernicus' theory was just as complicated as Ptolemy's and no more accurate. There were still converts. Galileo used the newly invented telescope to make important observations such as of the moons of Jupiter. And Kepler – we know about. With Kepler came the first really serious improvement in empirical adequacy and simplicity over the Ptolemaic model. No accident that there were huge numbers of converts then. The shift was complete when the last person still using the Ptolemaic model died off/stopped working. Kuhn points out that many new observations were made after the shift. New stars supernovas, sunspots etc were observed and recorded. But they hadn't been seen before the shift. In China, where there was no belief in the immutable unchanging heavens, astronomers recorded sunspots and new stars centuries before Galileo. The paradigm affected what astronomers looked for – they may have observed a new star, but disregarded. OHP Come back to core assumptions at start and explain. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 3 knowledge not accumulated. Shifts are too radical to talk of cumulative knowledge. The new paradigms completely knocks out and replaces the old paradigm. 2 relation theory and data not logical. Approximates in normal science – paradigm sets standards for comparison etc. But even in normal science this view is misleading since it only works in relation to the paradigm, and a paradigm is just a paradigm. It is merely a human construct, not an independent set of rules. 1 Observation not independent of the paradigm under which it is collected – because of models and assumptions and instruments etc used to collect it which are all part of the paradigm. We will focus on this issue next week the claim that two paradigms can’t be compared in a straightforward way with an independent body of data. QUESTIONS 5 Next Week Implications of Kuhn's views especially those about the relation between theory and observation for rationality and objectivity of science. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

71G5 The Philosophy of Science Unit Outline Autumn 2003 Phyllis McKay Room A82, Pathfoot Building Office Hours: 12.30-1.30 Tuesdays and Fridays 01786 467560 p.k.mckay@stir.ac.uk Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Course Information The course will introduce you to central issues in the philosophy of science. We will look at theories of how science progresses and achieves the success it does, whether science really does get at the truth about the world, philosophical problems with confirmation of scientific hypotheses, and what – if anything – distinguishes science from other disciplines. We will sometimes use discussion of scientific discoveries to illuminate philosophical theories about science, but the course will not presuppose any prior scientific knowledge. Outcomes (1) Knowledge and understanding of some of the central topics in current philosophy of science. (2) Ability to think critically and analytically about the topics covered, and display that ability in discussion and in written work. Practical details Prerequisites Any semester 4 Philosophy unit Teaching Autumn 2003, 2005 Honours course: semester 5 or 7. Assessment Essay 1 due 17 October 25% Essay 2 due 21 November 25% Exam 50% The Student Handbook explains penalties for late submission of essays. Reading Core text: Curd and Cover (eds.) Philosophy of Science: the central issues. (Norton and Company, New York, 1998) This is a very good anthology, which covers many of the central topics in the course thoroughly, giving carefully chosen extracts from the major figures in the field. It is also a very good resource if you are struggling, or if you get interested in a subject. It contains alternative papers on all the subjects it covers, and well-written commentaries at the end of every chapter which help to explain all the papers in the chapter. We will use this book extensively during the course. You should probably buy it. Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Some introductions we will be using: Alexander Bird Philosophy of Science (UCL Press, London, 1998) AF Chalmers What is this thing called Science? (Open University Press, Buckingham, Third Edition 1999) Donald Gillies Philosophy of Science in the Twentieth Century (Blackwell, Oxford, 1993) These are clear general introductions to the study of philosophy of science, and each covers several of the topics we will study. It would be helpful to get a copy of at least one of these books. There are several copies of all three in the library. John Losee A Historical Introduction to the Philosophy of Science (OUP, Oxford, 2001) James Ladyman Understanding Philosophy of Science (Routledge, London, 2002) These two books are also useful, but less widely available. The Losee takes a historical approach for those who like to see how the ideas developed, while the Ladyman covers most of the topics we will study. I will sometimes set readings from both of these. Samir Okasha Philosophy of Science: A very short introduction (OUP, 2002) A fun read, which might help set you on the right track if you’re struggling. And it really is very short. Reference works you might find useful: Curd and Cover, Losee and Bird all have useful glossaries. If they don’t help: Anthony Flew (ed): A Dictionary of Philosophy (Pan Books, London, 1979) Concise Routledge Encyclopedia of Philosophy (Routledge, London, 2000) This is expensive but very good, and it is available online. Go from library home page to Recommended resources by subject. Choose Philosophy and then Routledge Encyclopedia. Newton-Smith (ed): Blackwell Companion to the Philosophy of Science (Blackwell, Oxford, 2000) Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Course Timetable We will cover ten topics over the term, each topic involving a lecture and a seminar. In the lecture I will introduce the basic material, and we will use the seminar to get into that material in depth. Seminars will require much more participation by you, including careful reading and thinking about issues arising from the reading, short presentations on prepared answers to questions set, and discussion of those questions. Both lectures and seminars are prescribed for this course. The rationale for this is in the student handbook, section F. September Friday 19th 11-12am Lecture topic 1: What is science and what is not science? What is philosophy of science? Thursday 25th 12-2pm Seminar topic 1 Friday 26th 11-12am Lecture topic 2: Does science proceed by gathering evidence that supports a natural law? Problems for proving laws true. October Thursday 2nd 212-2pm Seminar topic 2 Friday 3rd 11-12am Lecture topic 3: Does science proceed instead by suggesting laws and testing to see if they are false? Thursday 9th 12-2pm Seminar topic 3 Friday 10th 11-12am Lecture topic 4: Can we really refuse to accept that a theory – any theory – has been proven false? Duhem and Quine’s problems for proving theories false. Thursday 16th 12-2pm Seminar topic 4 Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Friday 17th 11-12am Lecture topic 5: What are these laws of nature science aims to find? 17th October Essay 1 due Thursday 23rd 12-2pm Seminar topic 5 Friday 24th 11-12am Lecture topic 6: What is the nature of a scientific explanation? Is law or explanation the more crucial aim of science? Does all science have the same aim? (There will be no teaching for this course in the week beginning Monday 27 October.) November Thursday 6th 12-2pm Seminar topic 6 Friday 7th 11-12am Lecture topic 7: Is the progress of science more complex than either Popper or Bacon allow for? Kuhn’s description of progress in normal science, and his startlingly different claims about revolutionary science. Thursday 13th 12-2pm Seminar topic 7 Friday 14th 11-12am Lecture topic 8: Is a scientific observation really distinct from the scientific theory that describes and/or explains it? Implications for rationality in science. Thursday 20th 12-2pm Seminar topic 8 Friday 21st 11-12am Lecture topic 9: Does science aim at truth? Are the unobservable entities (things like electrons or genes) of science real? 21st November Essay 2 due Thursday 27th 12-2pm Seminar topic 9 Friday 28th 11-12am Lecture topic 10: Return to What is science? Should creationscience be taught in biology classrooms? Implications of the Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. theories of science and scientific progress we have studied. December Thursday 4th 12-2pm Seminar topic 10 Friday 5th 11-12am Revision class Wednesday 10th Exams start. Exact date for this course not yet decided. Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 1 What is science and what is not science? What is philosophy of science? Issues Many philosophers and scientists have been concerned to distinguish genuine science from any field of study that might purport to be a science. Popper, who is probably the philosopher most concerned with this issue, first focused on Marxist theories of history and early psychoanalysis. In his time these theories were just developing and were influential. Popper wanted to argue that whatever they were, they were not sciences. The issue is still topical with a big movement to force biology teachers to teach creationscience alongside evolutionary theory remaining influential in the US. We will come on to this particular debate in detail in topic 10. In this first topic we will start to consider what it might be that distinguishes science from non-scientific studies. What is surprising is that it is harder to do this neatly than most people first assume it will be. We will also consider what the philosophy of science is and what we can get out of studying it. Reading Thagard: ‘Why Astrology is a Pseudoscience’ in Curd and Cover 25-37 WH Newton-Smith: ‘Introduction’ to his Companion to the Philosophy of Science 1-8 Further Reading: Any of the articles in Curd and Cover Ch1, or the commentary. Questions 1) What is a pseudo-science? 2) How is science different from pseudo-sciences? 3) Is Astrology a science? 4) What is the philosophy of science? 5) Why should we study the philosophy of science? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 2: Does science proceed by gathering evidence that supports a natural law? Problems for proving laws true. Issues It is very intuitive to think that science proceeds by observing the world and merely generalising from that observation. General laws are proved true by means of this very straightforward generalisation from observation. This was at one time a common view in the philosophy of science, and we will look at Bacon’s theory as an example. But since Bacon died in 1626 problems for his view have been discovered. One of the major ones is the problem of induction, first introduced by David Hume. The problem of induction is a central problem in the philosophy of science. We will look at what induction and the problem of induction is, and how it affects Bacon’s method. We will also consider what might still be good about Bacon’s method in spite of the problem of induction. Reading Losee: Ch 7 Section 2 54-62 on Bacon Popper: ‘The Problem of Induction’ in Curd and Cover 426-427 Ladyman: Ch 1 11-30 Further Reading Russell: The Problems of Philosophy Ch 6 ‘On Induction’ 33-38 Bird: Introduction 10-17, Ch 5 182-186 Chalmers: Ch 4 41-58 Seminar Questions 1) What is Bacon’s theory of scientific progress? 2) What is induction? 3) What is the problem of induction? 4) Is there an adequate response to the problem of induction? 5) Does the problem of induction undermine Bacon’s method? How? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 3: Does science proceed instead by suggesting laws and testing to see if they are false? Issues In response to problems with theories of scientific progress like Bacon’s, Popper has an original alternative theory of scientific progress. It is that science does not in fact proceed by positive advances. Instead, science proceeds in a series of bold creative conjectures, followed by rigorous attempts to refute the conjectured theory – that is, attempts to prove that theory false. We will look at Popper’s theory and compare it to Bacon’s, identifying problems with both theories, and also pulling out their good points. To help in this, I will take you through a comparison of Bacon and Popper’s theory as they do and do not apply to a successful episode in the history of science – Kepler’s discovery of the laws of planetary motion. This discovery by Kepler was successful in persuading many that the planets do in fact orbit the sun, not the earth. I will not suppose that you have any prior knowledge of Kepler’s discovery. Reading Popper: Selections from The Logic of Scientific Discovery’ in Boyd et al (eds) The Philosophy of Science 99-112 Bird: Ch 5 177-182 Further Reading Popper: ‘Conjectures and Refutations’ in his Conjectures and Refutations Ch 1 33-59 5th Edition. Skip Appendix. Gillies: Ch 2 on Popper 26-53, Ch 6 124-131 Chalmers: Chs 5, 6, 7; 59-73, 74-86, 103 Ladyman: Ch 3 62-92 (You can skip Section 3.4 on Duhem if you like since we will cover that problem next week.) Questions 1) What is Popper’s theory of how science progresses? 2) Why is Popper drawn to a falsification based theory? 3) Does Popper’s theory solve the problem of induction? 4) Is Popper’s theory of scientific progress better than Bacon’s? 5) Does Popper’s theory have any new problems? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 4: Can we really refuse to accept that a theory – any theory – has been proven false? Duhem and Quine’s problems for proving theories false. Issues Popper’s theory of scientific progress generated a great deal of interest in the sciences, and also in subjects like the social sciences keen to be recognised as ‘real’ sciences. But it does have problems. Duhem attacks the view that choices between competing scientific theories can be made by means of a ‘crucial experiment’. On the way, he argues that a scientific theory, T, is never tested on its own. Since it can only be tested in conjunction with other theories, U, V, W... , Duhem argues that a single experiment can never definitely falsify theory T. We will see why this is true, and examine its implications. Quine makes stronger claims, saying that a scientific theory can only be tested in conjunction with everything else. This includes not just other scientific theories but also, for example, maths and logic. We will be interested in whether this is true, and what it might imply. Reading Duhem: ‘Physical Theory and Experiment’ Ch 3 S2+3 in Curd and Cover 260-66 Quine ‘Two dogmas of empiricism’ in his From a Logical Point of View. Also in Curd and Cover 280-301. Concentrate on thoroughly understanding Section 6 ‘Empiricism without the dogmas’. Klee: Introduction to the Philosophy of Science 64-67, 73-79 Extra Reading Curd and Cover rest of Ch 3 Ladyman: Ch 3 77-81, Ch 6 162-195 Gillies: Ch 10 205-230 Questions 1) What does Duhem think? Is he right? 2) What does Quine think? How does he differ from Duhem? 3) Is it really true that science might make us doubt 2+2=4? 4) Is it really true that I can continue to believe ‘there is a pink elephant on the desk dancing the Macarena’ in the light of any empirical observations? 5) What are the implications for science and scientific knowledge of the underdetermination of theory by data? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 5: What are these laws of nature science aims to find? Issues It’s natural to assume that the role of science is to find the laws that describe the world. But what are these laws? Are they mere descriptions of natural relations, saying that when you get one thing, like a brick thrown at a window, another follows, like the smashing of the window? And if they are such descriptions, are they exceptionless? In other words, is there no law governing bricks thrown at windows just because sometimes a brick is thrown but the window doesn’t break? We will also look at what else might be demanded of a genuine law of nature if laws are supposed to do something more than merely describe nature. Reading Mellor: ‘Necessities and Universals in Natural Laws’ in Curd and Cover 846-863 Bird: Ch 1 ‘Laws of Nature’ 25-41 Extra Reading Curd and Cover rest of Ch 7 Chalmers Ch 14 213-225 Questions 1) What is the regularity theory of a law of nature? What’s wrong with the regularity theory? What’s good about it? 2) What is a counterfactual claim and why is it relevant to what a law of nature is? 3) How might we adequately describe a law of nature to deal with this problem? 4) Can a law of nature be probabilistic? 5) What else might we expect of a law of nature? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 6: What is the nature of a scientific explanation? Is description or explanation the more crucial aim of science? Does all science have the same aim? Issues Many people think that science should not only describe nature but also seek to explain nature. But it is not clear exactly what it is that science should explain, nor is it clear exactly what it is that constitutes a scientific explanation. The most famous attempts to characterise scientific explanation are by Carl Hempel, and we will look at two of his attempts and discover the problems for them. We will then consider more widely what might be required of a scientific explanation, in particular whether it must cite a cause of the thing being explained. Reading Bird Ch 1 ‘Laws of Nature’ 42-60 Hempel ‘Two Basic Types of Scientific Explanation’ in Curd and Cover Ch 6 685-694 Extra Reading Curd and Cover rest of Ch 6 especially: Railton ‘A Deductive-Nomological Model of Probabilistic Explanation’ in Curd and Cover Ch 6 746-765 Klee: Introduction to the Philosophy of Science, Ch 6 104-127 Hempel: ‘Laws and their role in scientific explanation’ in Boyd et al. (eds) The Philosophy of Science (MIT Press, USA, 1992) 299-315 Questions 1) What sorts of things are explained in science? 2) What is Hempel’s Deductive-Nomological model of explanation? 3) What is Hempel’s model of probabilistic explanation? Can events which have an element of chance be explained? 4) Does an explanation need to cite a cause? 5) Must a scientific law offer an explanation? Does this add anything to our requirements for a law? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 7: Is the progress of science more complex than either Popper or Bacon allow for? Kuhn’s description of progress in normal science, and his startlingly different claims about revolutionary science. Issues Kuhn is probably the most revolutionary figure in the history of the philosophy of science, with the truth/falsity and possible implications of his views still hotly debated despite the fact that his major work The Structure of Scientific Revolutions was published in 1962. Much of this debate is over whether Kuhn has in fact shown that there is no rationality or objectivity at all in science. Kuhn’s views are complex, so we will spend two weeks on him. This week, we will focus on how Kuhn says science proceeds, looking at what he means by ‘normal science’ and by ‘revolutionary science’. To understand Kuhn, this difference, and the relation between the two sorts of episodes in science, must be mastered. One word of warning: Kuhn is subject to misinterpretation both by those who love his views and those who hate them. Both camps overreact. Take your time and try to understand what Kuhn is really trying to say. Reading Kuhn: ‘The Nature and Necessity of Scientific Revolutions’ in Curd and Cover Ch 2 86101 Ladyman: Ch 4 96-105 Losee: Ch 14 197-202 Extra Reading Losee: Ch 14 202-206 Hoyningen-Huene: Reconstructing Scientific Revolutions 4.4 159-62; 5.1, 5.2 167-179 Questions 1) What is Kuhn’s conception of normal science? 2) What is Kuhn’s conception of revolutionary science? 3) Why are paradigms so useful to science? 4) What initiates paradigm shift? 5) How is Kuhn’s account different from Bacon’s and Popper’s, and how is it better/worse? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 8: Is a scientific observation really distinct from the scientific theory that describes and/or explains it? Implications for rationality in science. Issues Now that we have understood the structure of Kuhn’s theory of scientific progress, we can investigate the implications of Kuhn’s views on revolutionary science. We will mainly focus on the possibility that Kuhn’s views imply that science has no more rationality or objectivity than any other non-empirical field of study. To understand this, it is crucial to get straight what Kuhn means by ‘incommensurability’. In doing this, we will discuss whether it is true that it is impossible to approach data without presuppositions in the way that Bacon requires. This is the view that all data is ‘theoryladen’. Note: When worrying about the implications of revolutionary science, don’t completely forget the implications of Kuhn’s views on normal science that we discussed last week. It would also be worth thinking about the relevance of Quine’s views on holism. Reading Kuhn: ‘Objectivity, Value Judgement and Theory Choice’ in Curd and Cover Ch 2 102118 McMullin: ‘Rationality and Paradigm Change in Science’ in Curd and Cover Ch 2 119137 Extra Reading Curd and Cover rest of Ch 2. If you feel lost you might find the commentary helpful. Ladyman: Ch 4 93-123 or 105-123 if you don’t want to re-read earlier section Questions 1) What does Kuhn mean by two theories being ‘incommensurable’? 2) Why does Kuhn think competing paradigms cannot be compared with a common body of empirical data? 3) If theories are incommensurable, why do scientists change paradigm? 4) What are the implications of incommensurability for science? 5) Is Kuhn right about these claims? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 9: Does science aim at truth? Are the unobservable entities (things like electrons or genes) of science real? Issues There is a vast debate on scientific realism, so we will only try to touch on central themes of it. We will focus on the question of whether scientific theories aim at truth, and compare this to the opposing view that these theories are merely instrumental tools for generating correct predictions in the future, with no claims to truth at all. We will also think about the relevance to realism of Kuhn’s views on the relation between theory and observation, and also Quine’s closely related holism about scientific knowledge. Reading Van Fraassen: ‘Arguments Concerning Scientific Realism’ in Curd and Cover Ch 9 106487 Maxwell: ‘The Ontological Status of Scientific Entities’ in Curd and Cover 1052-1063 Extra Reading Bird Ch 4 ‘Realism’ 121-161 Fine: ‘The Natural Ontological Attitude’ in Curd and Cover Ch 9 1186-1208 Losee: Ch 18 252-263 Chalmers: Ch 15 226-246 Ladyman: Ch 8 230-263 Questions 1) What does it mean to be a realist or an anti-realist in science? 2) Why might you not be a scientific realist? 3) Why might you not believe electrons are real? Could you consistently still believe mountains are real? 4) Does science aim at truth, or is it inescapably instrumental? 5) Does it really matter if you don’t think science can aim at truth? Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Topic 10: Return to What is science? Should creation-science be taught in biology classrooms? Implications of the theories of science and scientific progress we have studied. Issues We can now look at the relation between science and non-science in much more detail given our understanding of the theories of Popper, Kuhn and Quine. A much more thorough answer to the questions can now be given. Discussion of the issues will also help to develop and consolidate understanding of several of the theories we have studied. At the same time, the debate over creation-science in the US is interesting, and illustrates that the sorts of philosophical issues we have been discussing do sometimes matter intensely to both scientists and non-scientists – although I admit that political interest on the scale seen in the US is rare. Reading Popper: ‘Science: Conjectures and Refutations’ in Curd and Cover Ch 1 3-10 Ruse: ‘Creation-Science is not Science’ in Curd and Cover Ch 1 38-47 Laudan: ‘Commentary: Science at the Bar – Causes for Concern’ in Curd and Cover Ch 1 48-53 Extra Reading Curd and Cover Ch 1 Gillies: Ch 8 153-188, Ch 9 189-204 Questions 1) What does Popper say makes science different from the pseudo-sciences? 2) Can we produce a simple demarcation criterion to separate science and pseudoscience? 3) Does it matter if we can’t produce a simple demarcation criterion? (Think about the views of Quine and Kuhn.) 4) Is creation-science a science? Why? Is astrology? 5) Why is maths not a science? If this is different from astrology/creation-science, why? (Don’t forget Quine’s views.) Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Essay 1 17 October 1 What is the problem of induction and what do you think is the best way to deal with it? Revision of reading for Topic 2 Extra reading for Topic 2: Russell: The Problems of Philosophy Ch 6 ‘On Induction’ 33-38 Bird: Introduction 10-17, Ch 5 182-186 Chalmers: Ch 4 41-58 More advanced papers: B Skyrms: Choice and Chance (Wadsworth Publishing Company, 1986) Ch 2 D Papineau: ‘Reliabilism, Induction and Scepticism’ in Philosophical Quarterly 1992 120 PF Strawson: An Introduction to Logical Theory (Methuen, London, 1952) Ch 9 2 ‘Neither Popper’s nor Bacon’s theories of scientific progress describes accurately how real science happens.’ Discuss with reference to a real successful episode from the history of science. Revision of reading for Topic 2 and Topic 3 Extra reading for Topic 3: Popper: ‘Conjectures and Refutations’ in his Conjectures and Refutations Ch 1 33-59 5th Edition. Skip Appendix. Gillies: Ch 2 on Popper 26-53, Ch 6 124-131 Chalmers: Chs 5, 6, 7; 59-73, 74-86, 103 Ladyman: Ch 3 62-92 On Kepler: A Koestler: The Sleepwalkers Part four: The Watershed pp 227-427 You will be able to dip into the relevant sections, although you may find the whole story of Kepler utterly fascinating. Toulmin and Goodfield: The Fabric of the Heavens Ch 7 ‘Preparing the Ground’ 200229. You may wish to concentrate on the section on Kepler 218-229 Marie Boas: The Scientific Renaissance 1450-1630 (Fontana, London, 1962) Ch 10 ‘Circles vanish from Astronomy’ 269-292 Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. Essay 2 21 November 1 Does Quine show that there is no most rational way to alter your beliefs in the light of an observation inconsistent with them? Revision of reading for Topic 4 Extra reading for Topic 4: Curd and Cover rest of Ch 3 Ladyman: Ch 3 77-81, Ch 6 162-195 Gillies: Ch 10 205-230 More advanced papers: Newton-Smith: ‘The underdetermination of theory by data’ in The proceedings of the Aristotelian Society Supplementary Volume 52, 1978, 71-91 Laudan and Leplin: ‘Empirical equivalence and Underdetermination’ in Journal of Philosophy, 1991, 449-72 2 How should we characterise a law of nature? Should a law of nature help to explain something? Revision of reading for Topics 5 and 6 Extra reading for Topics 5 and 6: Curd and Cover rest of Ch 7 Chalmers Ch 14 213-225 Hempel: ‘Laws and their role in scientific explanation’ in Boyd et al. (eds) The Philosophy of Science (MIT Press, USA, 1992) 299-315 More advanced papers: M Christie: ‘Philosophers versus Chemists concerning laws of nature’ Studies in History and Philosophy of Science, 1994, 613-29 M Tooley: ‘The nature of laws’ in Canadian Journal of Philosophy, 1977, 667-98 Phyllis McKay Illari: Course taught twice, at King's College London, and Stirling University. 

71G5 Revision Notes Topic 1 Demarcation About whether we can characterise the difference between science and non-science and how. Mostly concerned with demarcation between science and pseudo-sciences. May be a different story for the difference between science and something like maths, or history. Popper’s classic account: 1 Claim scientific if it is falsifiable by comparison with empirical data. Theory scientific if it makes predictions that a falsifiable by comparison with empirical data. 2 For a theory/study to be scientific practitioners must hold theory tentatively. Must be prepared to compare it to evidence and reject it if evidence not favourable. Gets some intuitively right answers, (and point 2 is crucial in this) and is clear and snappy. But study of Quine and Kuhn makes us very aware of its inadequacies. Theories are never falsified in the straightforward way Popper envisages, and it can sometimes be the rational thing to do to keep trying to solve anomalies using the paradigm you still believe will work. Problem is that study of Kuhn and Quine suggests there is no distinctive scientific method. Is this true? If it is, is there no demarcation? Kuhn’s comments on demarcation: says astrology not a science, it’s a pseudo-science. This is because it has no central theory – no paradigm – and no puzzle-solving tradition like normal science. So idea is not to identify the difference between science and non-science by using a distinctive scientific method, but to use instead characteristics of the activity of science. Can fill in story about normal science with what Kuhn has also told us about revolutionary science. Question is – can this work? Examples of pseudo-sciences looked at: Astrology: The stars and planets affect your current emotions and actions. Study of them can help predict your future actions and emotions. The positions of the planets at your birth have an ongoing influence on your character and prospects of success. We thought: Its predictions are inaccurate but not significantly worse than some studies with more claim to being sciences. There are statistical methods for testing such predictions. There is no progress in a central body of theory all practitioners agree on. They won’t give up their theories, and seem unconcerned to compare them to data at all. We have better competing theories in the same domain eg cognitive science, psychology, evolutionary theory. Creation-science: The theory that the earth was created as the biblical book of Genesis claims. So the world was created in much its current condition in six days not very long Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. ago – ie complete with a fossil record etc suggesting it’s older. Man has separate ancestry from the apes, all species created by God and limited in variation since then. We thought: Does have paradigm, but no puzzle-solving tradition. Main job to try to pull down work of evolutionary theorists. Is falsifiable – we take it to have been falsified. Inconsistent with current theories, which seem to show us, eg that the world is much older than they think, and was never completely covered with water. We’ve got better theories about same stuff, that accord with other current theories. Eg physical theories – origin of the universe, evolutionary theories – origin of life and how there came to be so much diversity in living creatures. These do show a puzzle-solving tradition and a strong concern to compare their predictions with evidence. Not fruitful. Has object of remaining forever as it always was. Competing theories are fruitful, coming up with novel predictions and ideas. Note: it might be relevant, for both astrology and CS, to examine the thought that even buying Kuhn’s views, you don’t have to reject the idea that scientific choices are rational. So you can still argue that even if science is not objective, you can look at an old paradigm and claim that holding on to it is still irrational. If you want to reject Kuhn’s views and identify some paradigm-independent parts of scientific method, your case is even better. Implications: 1 What studies do/do not count as sciences might change over time. Perhaps astrology was once a science (when it was bound up with astronomy) but it isn’t any more. 2 The boundary between science and non-science is blurred – studies are more or less securely sciences or non-sciences. 3 There is no simple demarcation criterion. Nevertheless, an argument can be made against each pseudo-science on a case by case basis. These might be acceptable implications. Topic 2 Bacon 1561-1626. Major propagandist for organised and collaborative scientific inquiry, and emphasised science’s practical applications. Method. Thinks key to scientific method is the elimination of human subjectivity and guesswork. 3 stages. Doesn’t emphasise Stage 3 much because he is reacting to Aristotle and it is just the same as a stage in Aristotle’s method. Don’t forget it’s there. Bacon gives heat example – see handout. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Stage 1: Clear your mind of presuppositions. (Bacon calls them ‘Idols’.) Do not speculate or theorise before you collect the data. Collect data about the world in a systematic way. Stage 2: Put that systematically collected data in tables and study. Generalise from that data. This is the inductive stage. Bacon thinks it’s very simple. Remember the grapes analogy on the handout. Stage 3: You have your generalisation. Use it to make a prediction about something not studied in Stage 1. Go see if your prediction is true. This is the deductive stage. Note how like Popper’s it is. Criticisms: 1 Many scientific successes not mechanical – highly imaginative. Did Einstein’s special relativity, or Darwin’s theory come out of a simple generalisation? Also, many laws go beyond what the data could give by simple generalisation (eg Kepler’s second saying the planet moves faster at some points on the orbit). Need to assume the world is arranged along principles of mathematical harmony. 2 Science has big jumps – not simple steady progression. Eg Copernicus to Kepler, Kepler to Newton, Newton to Einstein. 3 Bacon’s method might only get general descriptions, such as a descriptive law. Science might want more, such as explanation. Think about the difference between Kepler and Newton. Kepler wants to explain why the planets move; Newton manages it. Can generalisation do this? 4 Presuppositions can be crucial to science. Think about theory conditioning observations, and our discussion of why paradigms so useful in normal science. Kepler eg we discussed – willingness to ditch presuppositions crucial to his success. a) Kepler was a convinced Copernican, so related the orbit of Mars to the Sun, not the Earth. This was because he wanted a dynamic explanation for the movement. b) He rejected the Aristotelian idea that the heavenly bodies move at constant speed in circles, or movements composed of circles, due to their perfection. But not clear Kepler became without presuppositions, as much as replaced them with something else – different presuppositions that worked better. Topic 3 Popper Two main motivations: a) Very impressed with problem of induction – thinks no solution. b) Very against the verification method claimed by early psychoanalysts and Marxists who saw confirmation everywhere and never seriously tested their theories. Method: (sometimes called method of conjectures and refutations) Separates context of discovery from context of justification. 2-stage method. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Context of discovery: Scientist produces hypothesis in a bold conjecture. Doesn’t matter where scientist gets hypothesis. At this stage scientist should be creative, daring, and needs presuppositions. This stage is the route to the hypothesis. What happens at this stage is a matter for psychology, NOT for science. How scientist gets hypothesis in the first place is irrelevant to whether we are justified in accepting it or not. Context of justification: Scientist tests and attempts to falsify the hypothesis by deducing predictions from it and seeing whether they hold. (refutation stage) This stage is what justifies us in accepting that hypothesis, which is quite distinct from how scientists get to the hypothesis. This stage is deductive. If the hypothesis withstands rigorous testing, it is corroborated. You are justified in accepting it tentatively. It is always subject to future falsification. Criticisms: 1) We put scientific results to practical use. Can Popper derive any future expectations from science, that is, from a ‘corroborated’ hypothesis? Since he explicitly eliminates induction from the justification of accepting a hypothesis, it looks like he can’t. Without inductive support, there doesn’t seem to be any reason to expect (even to a high degree of belief) that a hypothesis that has merely withstood testing and proven not to be false will continue to hold. Note that Popper sometimes talk as if corroboration can allow future expectations, but if he is consistent about his elimination of induction from the justification of hypotheses, he is not entitled to this. 2) See Duhem and Quine. Kepler example for Bacon and Popper: See handout with Kepler’s four hypotheses illustrated. Kepler discovered that the planets move in ellipses with the sun at one focus, and the radius vector from the sun to a planet sweeps over equal areas in equal times. Worked from data collected by Tycho Brahe before Kepler started work. Careful accurate and continuous (daily – which was not done at the time) observations of the movements of the planets and carefully recorded them. Far best data available at the time. Included no presuppositions of Kepler’s, and fewer than contemporary data. This crucial to Kepler’s success. But see Kepler’s presuppositions under Topic 2 Bacon. Brahe allocated Kepler Mars to study – happy accident since orbit of mars is the most ‘eccentric’ i.e. the most obviously NOT a circle. Kepler attempted to fit a succession of mathematical descriptions of possible orbits to the data, using detailed calculations. Generated hypothesis, tested it against the data, and rejected it if it didn’t work. Kepler clearly was creative as Popper says, but hypotheses do seem to show a succession towards the final ellipse, suggesting that how you get to hypotheses is important, perhaps even to their justification, as Bacon says. Status of laws: Kepler thinks he has reason to believe them true. Bacon agrees, Popper disagrees. What do you think? (But see Topics 5 and 9, laws and realism.) Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Topic 4 Duhem and Quine Duhem: You cannot test a hypothesis alone. You need at least auxiliary hypotheses and initial conditions to generate a prediction. So if the expected prediction doesn’t hold, you only know that one of the set is false, not which one is false. Means a small number of anomalous results mean nothing to a scientific theory. (Consider how many pupils in physics classrooms have ‘falsified’ Newton’s theory of gravitation.) Certainly doesn’t mean we should reject the theory. Note: Popper is aware of this problem and responds. Says we should take a theory as falsified ‘only if we discover a reproducible effect which refutes the theory. In other words, we only accept the falsification if a low-level empirical hypothesis which describes such an effect is proposed and corroborated.’ p110 in Popper reading set. Think about whether this is adequate. Duhem explicitly limits scope of the thesis to physics, saying it doesn’t apply to physiology or chemistry. Thinks maths and logic have a quite different status, saying that we take the axioms of Euclid as established true by common sense knowledge. Quine: Quine doesn’t limit the scope of the thesis. He thinks that it applies to any statement whatever – and this includes maths and logic. Whenever you compare a prediction with observational data, you have derived this prediction from the entirety of your knowledge (the web). If the prediction doesn’t hold, all you know is that something, somewhere in the web is wrong. You don’t know what. This is why he says ‘the unit of empirical significance is the whole of science’. Quine does give us pragmatic standards for theory revision. He thinks we are likely to make alterations that disturb the web as little as possible so that everything is consistent again. This means that we are more likely to alter statements near the ‘outside’ of the web – ie those most directly related to experience. So you might decide that your voltmeter is faulty, for example. You are therefore unlikely to alter maths or logic which lie at the heart of the web, but the difference is not a difference of kind – it is merely a difference of degree. It is also a matter of human decision, not forced on us by the world and logical relations between it and our beliefs. Examples Quine gives: Logic: As a response to problems arising from quantum mechanics some suggested ditching the law of the excluded middle (p or ¬p) and constructing an alternative logic. Some do still work on this project. Maths: Euclid’s fifth postulate is also no longer used in the geometry we use to describe curved spacetime. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Underdetermination of theory by data: Quine also points out that there are potentially indefinitely many entire webs all consistent with the same set of empirical data. There is no reason (forced on us by logic, anyway, but think about the pragmatic considerations) for choosing one over the other. Criticisms: we thought much of what Quine says is true, but questioned whether we always test everything in the entire web, as opposed to being able to identify sections tested in particular cases. We also thought there was an alternative way of understanding what was happening in the maths and logic cases. This idea is that the maths and logic are still correct, but we discovered, empirically, that it would be better to use a different mathematical or logical system to describe the world. That that discovery should be empirical is not so surprising. You need to decide what you think. Topic 5 Law Most people agree that laws are an important product of science. But they don’t agree how best to characterise them. Simple regularity theory: It’s a law that all Fs are Gs if and only if all Fs are Gs. Ie laws are nothing over and above collections of their instances. Most minimal account of a law, involving no worrying metaphysical assumptions. Problems: Regularities that aren’t laws, such as accidental regularities like Laws that aren’t regularities, such as laws governing a single instance like the Big Bang, or laws governing stuff that hasn’t actually happened (as a matter of accident or even by law). For example, functional laws like f=ma govern masses larger than any that actually exist. But not clear a simple regularity will support counterfactual claims – claims about something that hasn’t actually happened, like ‘If a large mass had entered our solar system at x velocity, then...’ Yet we do think Newton’s laws will let us work out what would have happened – for example, would it have missed the Earth? May also allow laws to have no instantiations at all, allowing both ‘zoogons are living creatures’ and ‘zoogons are not living creatures’ to count as inconsistent laws of nature. Doesn’t allow laws to explain their instances. Something cannot explain itself, so if a law is nothing over and above the collection of its instances, it cannot explain those instances. Full-blooded view: Put aside worries about metaphysical claims, and allow them. Hold something like: a law like ‘all Fs are Gs’ asserts a relation of necessity between having property F and having property G. If you have property F, you must have property G. It could not be otherwise. Good: Solve the problems above. Note it does sensibly allow laws to exist that have no instantiations. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Allows laws to support counterfactual claims. Allows laws to explain their instances. Problems: Giving an account of the necessity involved. It is not logical necessity (we think the laws are contingent – could have been otherwise so are not themselves logically necessary). So we need an account of something like natural necessity to work, and such an account is not easy to give. Might make trouble for how we establish that such a law exists. We can establish the existence of regularities, but how do we establish the existence of a relation of natural necessity? Topic 6 Explanation Everyday informal notion of explanation as being for clearing up a temporary confusion for a particular purpose. It is partial and contextual, depending on the contrast class implicit in the question asked and the surrounding context. Some think there is no more to scientific explanation. Some don’t, like Hempel. Hempel’s Covering Law Model: The key to explanation is that it shows that the think to be explained was to be expected. Close link between explanation and prediction. Hempel’s models are just models. Real scientists use informal shorthand. But the idea of the models are that they give the full story of what is going on that lies behind the informal explanations. There are two versions: 1 Deductive-Nomological (D-N) An explanation is an argument, with at least one natural law as a premise. The conclusion is the thing to be explained. An explanation for why the ball falls at 9.8ms-2: Natural law: Unsupported objects fall at 9.8ms-2 Conditions: The ball is an unsupported object Outcome: The ball falls at 9.8ms-2 Problems: Law plus initial conditions can entail an outcome without explaining it, for example if the outcome is preempted by some other thing. Jones and arsenic example. Also explanation appears to be asymmetric, whereas in a deductive argument, the law plus the outcome will entail (and thus on Hempel’s model also explain) the initial condition. But this is clearly not an explanation for the ball being an unsupported object. 2 Probabilistic-Statistical Model The explanation is still an argument, but the law used as a premise is now a statistical one. The argument no longer entails the conclusion, it merely shows that it is likely. An explanation for why the radium nucleus decayed within an hour: Natural law: It is very likely that radium nuclei decay within an hour of being made. Conditions: This is a radium nucleus just made. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Outcome: It is very likely that it will decay in the next hour. Problems: Doesn’t let you explain unlikely events. So if you are trying a new drug, and you know you will have 1/3 chance of cure, 1/3 chance of getting a bit better, and 1/3 chance of no change, you still can’t explain any of these outcomes, since each one is unlikely. Maybe explanation isn’t so closely tied to prediction and showing that an event was to be expected. Note link to law: Hempel’s quite minimal about laws, and there are lots of them. But if you adhere to the full-blooded view there will be fewer natural laws. You might then not want to insist every explanation involves a law. Think about whether explanation always involves a cause. Topic 7 Kuhn Science involves two distinct (and repeating) stages – normal science, where science is ruled by a paradigm, and revolutionary science, when the paradigm is shifting and no single paradigm is ruling. In normal science, the paradigm is unquestioned. It helps focus work on fruitful issues, helps scientists discuss problems and work together, and generally hugely aids scientific progress. Kuhn calls scientific activity in this stage ‘puzzle-solving’. But eventually the paradigm becomes less successful. Anomalies accumulate, expected predictions fail to appear, and fewer and fewer puzzles are solved. Scientists begin to lose faith in the old paradigm. The science is in crisis. One day, a bright spark thinks up a new idea. Some scientists begin investigating. It is not a new approach within the old paradigm but the start of a new one. Gradually, more and more scientists convert to the new paradigm, convinced of its potential fruitfulness. They do not choose in a strictly logical way, but according to ‘personal and inarticulate aesthetic considerations’. Eventually, scientists clinging to the old paradigm retire or die off, and all new entrants to the science are trained in the new paradigm. Paradigm shift is complete. The process starts again. Paradigms: Group of beliefs, values and techniques shared by a particular scientific community at a particular time. Scientists working in the same paradigm share theories, objectives and more. They share the models and standards that guide normal science. They use, for example, not just the same theory but the same models of the atom and the behaviour and categorisation of sub-atomic particles, and use the same experimental methods and instruments for investigating them. They direct their attention to the problems the paradigm suggests will be fruitful, and share standards for deciding when that problem is solved. See handout giving examples of paradigm shift as the best way to get an intuitive feel for what Kuhn means. See also ‘Incommensurability’ below. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Examples of paradigms Kuhn mentions: Astronomy Ptolemaic geocentric. Cumbersome maths. Copernican heliocentric – eventually Kepler’s ellipses, then Newton. Physics Aristotelian function-based explanation, Newtonian notable because joined separate areas eg encompassed astronomy and earthly motion, relativistic paradigms notable space and time not absolute – relative. Biological sciences Darwinian evolutionary paradigm revolutionized how we saw our place in the universe. Kuhn also mentions some more dubious examples, like the discovery of X-rays. But it is questionable whether this really is paradigm shift. Incommensurability: We can’t compare competing paradigms in any straightforward way with an independent body of data and decide which is better. 1 Observations conditioned by theory. So no way two scientists working under different comparisons can compare the paradigms with an independent body of data. The data is not independent because it is conditioned by the paradigm of the scientist collecting it. 2 Meaning change between paradigms. Kuhn talks about different language, but it is better to think of as different conceptual schemes. 2 reasons: a) clusters of interdefined concepts so that you need to understand the cluster to understand any of the concepts b) conceptual disparity between different theories. You can’t compare paradigms with each other in a simple way, and you have to take on the paradigm to understand it, so you can’t stand in a neutral position to compare paradigms. 3 Shifting standards. Scientific method comes along as part of the paradigm. If the paradigm tells you how to compare theory with data, you won’t compare competing paradigms with an independent body of data in the same way. Criticisms: We thought 1 was probably true to some extent but worried how far it went. We wondered whether 2 was a problem at all – it makes it tricky to compare paradigms, yes, but surely not impossible. 3 is a big worry. We agreed that sometimes paradigm shift has (historically) altered what we take to be good scientific method, but questioned whether that means there is no scientific method that is genuinely paradigm-independent. Question: do the plausible elements of all three add up to enough of a problem to claim that competing paradigms are incommensurable? If not, might science be cumulative after all? And see later Kuhn on being guided by values below. Objectivity and rationality: Kuhn’s views probably do imply that science is not objective, in three distinct ways: a) there is no strict logical relation between theory and data b) science is not progressing towards an objective truth (but see Topic 8 Realism) c) science depends in important ways on the personal decisions of individual scientists. This alone does not imply that science is not rational, as Kuhn is keen to argue. In later work, Kuhn says the personal subjective decisions of individual scientists are guided by at least five values: accuracy, consistency, broad scope, simplicity and fruitfulness. He thinks that the decisions they make are genuinely scientific and recognisably rational in spite of being personal and subjective, and the fact that scientist A might be drawn to a new paradigm because of its simplicity, while scientist B is attracted by its fruitfulness. He wishes to allow for some real scientists to act irrationally. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. Question: Is this plausible? Think about relation to Quine and logical versus pragmatic senses of rational choice of theory. If it is plausible, is science still subjective and not cumulative, even though it’s rational? Topic 8 Realism Scientific realists accept, while anti-realists deny, at least some of the following: 1 There is such a thing as scientific progress 2 There is such a thing as a unique scientific methodology 3 Scientific theories can be evaluated in terms of their truth or approximate truth, and the success of a scientific theory is evidence that it is true. 4 If a scientific theory is true, then the unobservable entities posited by the theory are explained and really exist. Constructive Empiricism (Van Fraassen’s version of anti-realism): 'Science aims to give us theories which are empirically adequate; and acceptance of a theory involves as belief only that it is empirically adequate.' Van Fraassen in Curd and Cover p1069 There are two main focuses of debate: whether the aim of a theory is truth, and whether unobservable entities really exist. Unobservable or Theoretical Entities: We posit the existence of in principle unobservables to explain observable phenomena. These are things that can’t be directly observed by human beings at all, even using instruments, such as neutrinos, protons, electrons, fields of force, quarks, photons and so on. We have to observe these indirectly, by their effect on something which is observable. In the cloud chamber we see the effect on nearby particles of a close-passing electron, and we see iron filings moving in a magnetic field, not the field itself. Realists about unobservable entities accept that such entities really exist, while anti-realists insist they are merely useful theoretical posits, and not real entities at all. Arguments: 1 ‘No-miracles’ or ‘Ultimate’ argument for realism: The success of science needs explaining, and the approximate truth of scientific theories is the best explanation of that. Criticisms: (from anti-realists) a) Circular. The argument relies on inference to the best explanation generating a view that is likely to be true, while this is exactly what the argument is intended to establish. b) There is never a clear ‘best’ explanation. There are two reasons i) Quine’s underdetermination of theory by data thesis ii) pragmatism about scientific explanation. Scientific explanation is merely contextual and for a particular purpose. Change the purpose or context, and the ‘best’ explanation changes. So there is no unique best explanation of a particular fact/event/etc. 2 Pessimistic meta-induction for anti-realism: The history of science shows that every past theory, however successful, has proven false, and many highly embedded theoretical entities have proven not to exist. So we should expect that current theories will prove to Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. be false and current theoretical entities will prove not to exist. We should accept that the aim of scientific theories is something less than truth – empirical adequacy, and the theoretical entities only have to be useful posits, not real, to be valuable to science. Simple inductive argument run on theories, and on theoretical entities. Criticisms: Too pessimistic. History of science really shows theories progressing towards truth, and theoretical entities like electrons around for a long time through substantial theory change, and that is what we should expect in future. Theoretical entities are not all the same. They have quite different statuses, and so we shouldn’t make a simple inductive inference based on all of them, to the future of all of them. And see the natural ontological attitude below. The natural ontological attitude: Arthur Fine: we can only ask local realist questions about this theory or that entity – not a global question about them all. And science does this itself. View may be questionable for theories, but Fine has an important point about entities. Take electrons – highly embedded and scientists treat them as real. Neutrinos not, and could vanish with one theory. Still treated tentatively. Up to scientists whether and when to begin treating them as real. Fields of force around a long time and always explicitly treated as theoretical tools, not real things. Would be odd to think philosophers could make a global decision that all such things are real when scientists don’t. Phyllis McKay Illari: Material for Philosophy of Science course for revision session for anxious students. 

Lecture 6 Explanation 1 Last week Last week we looked at a major aim of science – natural law. Saw there were problems with the simple regularity theory, and that introducing the idea of necessity solves some problems, but introduces new ones. Second major aim of science today – explanation. One interest is how it links with the aim of describing natural laws. QUESTIONS? 2 Explanations Start with informal notion of explanation. Sorts of things explained Why did the patient recover? Why did the ball bounce so high? Why did the chemical reaction use up that much oxygen? Why did the window break? Why did the particle decay? Why was the electron deflected left, not right? Why did the earthquake happen today? Why do Kepler’s laws work? Explain both general and particular things. Informal notion of explanation both partial and contextual. Mummy, why did the ball bounce so high? Because I pumped it up properly. Partial: Doesn’t give full story. The mother doesn’t explain why pumping up works, or talk about what the ball is made of, and the elasticity of rubber Phyllis McKay Illari: Material provided for Philosophy of Science Unit and its relation to motion – saying why the ball bounces in the first place and so on. She just gives what the asker looking for. She clears up what the child doesn’t understand. Contextual: What asker looking for depends on the context. It is usually obvious to the person asked. Earlier: why bounce now rather than this morning Other ball: why does this ball bounce better than that other old worn ball Brick: why do balls bounce rather than bricks, which don’t bounce Right answer depends on context. We call this the contrast class. Informal explanation involves an implicit contrast class. This is what the context gives you. Contrasts above are: this morning, other old ball, bricks. Some think this is all there is to explanation, even scientific explanation. [Think about reasons for holding SRT of laws, even though problems. For explanation, the position is similar. Roughly, the idea is there isn’t any further underlying structure to world. So all explanation does is answer these informal questions – clear up immediate confusion.] For now, we’re going to put aside these thoughts and look for something more solid. But there is an alternative account of explanation to fall back on if the search for something more substantial doesn’t work out. Scientific explanation: Types: (Bird p62) Why did the ball fall? 1 Causal explanations Because I knocked it off the table. 2 Explanations in terms of law Because it’s a law of nature that unsupported objects fall Why have I got an opposable thumb? 3 Darwinian explanations Because delicate manipulation is really advantageous so it’s selected for. 4 Functional explanations Phyllis McKay Illari: Material provided for Philosophy of Science Unit Because it’s really good for delicate manipulation Functional explanations aren’t really used in science, but we do use them when talking about artefacts – ie objects designed for a particular task. The answer to the question ‘Why is the hammer made of steel?’ could be ‘Because steel is hard so good for knocking nails in’. Not mutually exclusive. More than one explanation above could apply to the same case. Citing the law of gravity might show that the ball’s movement is law-governed, but it doesn’t cause the ball to fall. To give the cause, you have to cite something different. Is there anything that can’t be explained? Do accept current explanations which don’t go beyond a particular point. For example, there is no answer to the question ‘Why to large masses attract each other?’ If we don’t accept that some things do not need explained, we risk infinite regress: ‘Why a?’ ‘B’, ‘Why B?’, ‘C’ .... QUESTIONS? 3 Hempel on Explanation Now you’ve got some idea what the issue is, let’s look at an account of scientific explanation. Hempel’s approach is sometimes called the covering-law model. For Hempel, to explain a fact is to show how it can be subsumed under laws together with other particular facts – initial conditions. Basic idea – the explanation shows that it did it as a matter of law. It’s a law of nature that it act like that. Laws are what explain in science. Carl Hempel OHP Deductive-nomological. D-N model. Natural law: Unsupported objects fall at 9.8ms-2 Conditions: The ball is an unsupported object Outcome: The ball falls at 9.8ms-2 Phyllis McKay Illari: Material provided for Philosophy of Science Unit It’s deductive because it takes the form of a deductive argument. And nomological just refers to the fact that it contains laws and shows that the phenomenon is law-governed. Fairly intuitive idea. Natural laws are what explain, and they entail the thing to be explained (along with initial conditions we believe hold). So they explain why the outcome happened as it did. This gives the full story that science knows about that outcome. You can see that this involves a close relation between explanation and prediction. The only difference is in whether you are constructing the argument after the fact or not. But if all it is to explain something is to show that it was to be expected, there would be a very close link with prediction. You can use the same model of explanation for many different circumstances – you might need more laws, or more complex ones, and lots more initial conditions. You can do that. Anyone see immediate problem? Are probabilistic explanations possible? There, the existence of the explanation couldn’t entail the event happening. Actually, Hempel adapts his model for that too. Hempel still uses law in probabilistic explanation, but the law is a probabilistic one. The structure is still an argument, but the truth of explaining facts just make the occurrence of the event likely. OHP Hempel’s Probabilistic-statistical model It is very likely that radium nuclei decay within an hour of being made This is a radium nucleus just made So It is very likely that it will decay in the next hour Outcome not entailed by premises – just made likely. Phyllis McKay Illari: Material provided for Philosophy of Science Unit These are MODELS: real scientists use shorthands, just like other normal people. But the model gives the full story all the shorthand explanations are based on. To explain something to show that it was to be expected, to show that it falls under a natural law. QUESTIONS comments. Problems for covering-law model One major problem – Laws plus initial conditions can entail an outcome without explaining it. Pre-emption Consider (Bird) OHP Any human being who eats a pound of arsenic will die within 24 hours Jones ate a pound of arsenic SO Jones will die within 24 hours But Jone’s death from arsenic poisoning is pre-empted. Jones is hit by a bus shortly after consuming the arsenic. Arsenic consumption allows you to predict death, but it doesn’t explain this death. Asymmetric Explanation is asymmetric. If A explains B, it is very uncommon for B to explain A too. All and only unsupported objects fall This is an unsupported object SO This fell Phyllis McKay Illari: Material provided for Philosophy of Science Unit But this is a deductive argument. This means that the law plus the conclusion taken together entail the initial condition: All and only unsupported objects fall This fell SO This is an unsupported object So in this way D-N explanations are symmetric. The explanandum can wind up explaining one of the initial conditions, and v-v. But clearly although the first example above does explain, the second, in exactly the same form, doesn’t. So maybe to explain something is to do more than show it was to be expected. There are also problems for the probabilistic model. On that model, you can only explain probabilistic outcomes that are likely. IF something is less probable than ½, you can’t explain it. Suppose a probabilistic law tells you there’s a 1/3 chance of outcome A, 1/3 chance of outcome B, and 1/3 chance of outcome C? You can’t explain any outcome on Hempel’s model, but you still know you will get either A, B, or C. And if you are doing repeated trials you know roughly how many of each outcome you will get. So you can give some information. There may even be cases where this is all the information possible to give. Nevertheless, Hempel’s model is committed to claiming you cannot explain in such cases. If you think you can explain in such cases, maybe a scientific explanation of something doesn’t even always show it was expected. There is a general problem in accounting for scientific explanation. The issue is link to the question of whether there is a unique scientific methodology or not. If there isn’t, maybe different explanations do different things in different circumstances. Phyllis McKay Illari: Material provided for Philosophy of Science Unit What do you think? QUESTIONS? 4 Explanation, Laws and Causes Some people still want to hold on to the idea that explanation has something to do with laws. And maybe laws have something to do with explanation. There is a clear link to laws for Hempel because his model uses laws explicitly. It is different depending what you take a law to be. Remember last week – the minimal account of laws is given by simple regularity theory, and the problem is it doesn’t exclude accidental regularities. There is a stronger account saying laws give relations of natural necessity between properties – works better for how we treat laws, but problems for how you account for necessity and properties. Compare All creatures that have a heart have a liver This creature has a heart SO This creature has a liver Law of Definite Proportions: ‘Any pure chemical compound is made up of its constituent elements in definite and invariant proportions by mass.’ Eg 1 mass hydrogen to 7.94 oxygen to get water. This reaction to get water used up 1 mass of hydrogen SO This reaction also used up 7.94 mass of oxygen Difference? The first one involves an accidental regularity. The second doesn’t. It involves a real law. There may be a relation of natural necessity between being a reaction to get water, and using up exactly those proportions of hydrogen and oxygen. The link is about what it is to be water. Phyllis McKay Illari: Material provided for Philosophy of Science Unit As an explanation, the first does achieve something. It tells us stuff happens like that, regularly. But the second tells us more. It tells us what must happen, and the detail of the law, or the theory the law is part of, tells us why. Hempel’s laws are very minimal. You can tell from the examples he uses – there are lots of them. He is working with something more like the SRT than a strong claim like a necessary relation between properties. If your view is that laws involve something like necessity, there are going to be far fewer of them than there are simple regularities. Lots of regular things don’t have a law of nature about them. So if you think you need a law of nature of this sort for explanation, you are going to think there are lots of things science can’t explain. So science is complex. There is a relation between what you think about laws and explanation. But these are not all there is to the aims of science. Science also involves theory. Theory tells you what molecules are, gives models for understanding them and interpreting their behaviour. Remember the cloud chamber – theory about sub-atomic particles tells you why this line in the photograph is a positron doing its stuff. Theory may include models to tell you how to interpret laws. And might include causal claims. Don’t worry too much yet. Just start thinking about. We are going to explore this in some detail when we look at Kuhn. What we are aiming at is something ambitious. We are trying not to do a simplistic analysis of different parts of what science does, but accept in full the real compexity of what science produces for us, and see what it implies. This project is genuinely fascinating. QUESTIONS Phyllis McKay Illari: Material provided for Philosophy of Science Unit 5 Summary and Next Week Explanation. Hempel’s D_N and probabilistic models. Problems. Suggest not work, but is relation law and explanation. Just more complicated and including other things not just law or explanation. Kuhn What’s a theory? Staggering views on it. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Handout Lecture 3 How Bacon and Popper would expect Kepler to proceed Bacon Kepler used systematic data Kepler had no presuppositions Kepler produced laws from data by a relatively simple generalisation. Kepler made predictions from his laws and checked to see if true. After this process, Kepler had good reason to suppose his laws true. Popper Kepler was selective about data he used Kepler chock full of presuppositions Kepler bold and creative in producing laws This stage offers no justification for laws Kepler rigorously tested laws and rejected any that were falsified Kepler tentatively accepted a law if it withstood rigorous testing Kepler had no good reason to suppose his laws true. They were merely unfalsified. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Kepler’s Hypotheses and Laws Handout Lecture 3 Hypothesis 1 Orbit of Mars a circle round a centre C displaced from sun. Motion uniform with respect to point U (planet moves faster when nearer the sun). Draft calculations based on four observed positions of Mars in opposition. But when tested against further observations of Tycho Brahe’s it didn’t match. Rejected. Hypothesis 2 Based on Ptolemaic epicycle and deferent. Produces egg-shaped orbit. Doesn’t fit Tycho’s data. Rejected. Hypothesis 3 Mathematically equivalent to an ellipse. Through mathematical error, it didn’t work. Rejected. This figure is NOT based on a circle. By moving to this hypothesis, Kepler has ditched the circular motion assumption without directly falsifying it. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Hypothesis 4 Ellipse. Worked. Then Kepler realised it is equivalent to Hypothesis 3. Law 1: orbit an ellipse with the sun at one focus Law 2: radius vector from the sun to a planet sweeps over equal areas in equal times – i.e. the planet moves faster when nearer the sun. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Why type 3 entities might not be real Next week, we’re going to get on to the actual arguments that realists and anti-realists come out with, but they will all make much more sense if we look at some examples this week of why those arguments got started about modern science. There are some very specific problems with holding that these type 3 entities are real. They come in two varieties. The first is how we interpret the theories we have of them, and the second is how we match up theories in separate areas of science. I’m going to look at two examples of problems with interpreting the theories that we have. The two problematic entities I’m going to use are photons and electrons. The problems with them generalise to a lot of similar entities. Photons are to do with light. Theories about light use the idea of ‘photons’. Photons cannot be directly observed, but they appear in various wellsupported scientific theories, and they have also been around in scientific theories for a while. It could look like we’ve no reason for doubting that they really exist. The problem is, we’ve got two possible ways of interpreting what light, and photons are. They are incompatible, and we don’t know which one to pick. We’ve got some observational evidence to show that light is a wave. It shows interference effects in a double or single-slit experiment. In a double-slit experiment [known as Young’s double slit experiment after Thomas Young], you observe lighter and darker parts of the wall you’re shining the light through the double slit and onto. [DRAW] This has been explained on the grounds of light being a wave. When the waves interfere with each other constructively, that produces the very light patches, and when it interferes destructively, you get the dark patches. You can get the same effect with sound waves and two loudspeakers if you are careful. Phyllis McKay Illari: Material provided for Philosophy of Science Unit The problem is, we’ve also got observational evidence to suggest that light comes in particles – the particles we call photons. The earliest serious reason we had to believe that light was composed of particles was the photoelectric effect. This is the name of the phenomenon that shining light on some metals will make them emit electrons. But it was noticed that if the frequency of the incident light is below a certain minimum depending on the type of metal used, it doesn’t matter how intense the light shining on the metal is, no electrons are emitted. Einstein explained this by saying that light comes in units called photons which have a certain amount of energy. It takes one photon to knock one electron out of the metal surface. It takes a certain amount of energy to do this, depending on the metal. Since the energy a photon has is a function of its frequency [hf] photons of light below a particular frequency do not have enough energy to dislodge one electron from a particular metal plate. This was followed with more evidence. One piece is called the ‘shot effect’. Photons can be collected by a very sensitive detector which amplifies the electric signal they generate. If you feed the amplified signal to a speaker and listen to it, it sounds like it’s a sprinkle of lead shot falling on a surface. The noise is not continuous. It sounds like you can hear individual photons arriving. Another piece also shows a discrete effect which seems to show light comes in bundles. If very dim light is shone on a photographic plate. You don’t get darkening uniformly over plate. [‘Darkening’ because it’s negative – wrong way round.] You get a set of separate little spots. If you increase the light intensity, you don’t increase how dark the spots are, you increase the number of spots. It looks like each spot corresponds with a discrete particle we might call a photon. So the problem is, what’s a photon? Is it a wave? Is it a particle? If it’s real, we tend to think it must be one or the other. The two are incompatible, so it couldn’t be both. But if a photon isn’t real, if it’s just a tool for theorising, it can be both. There is no problem. Although it is worth noting that photons are light. If you can’t interpret what photons really are, this is the same as being unable Phyllis McKay Illari: Material provided for Philosophy of Science Unit to interpret what light really is. Do we really want to deny the existence of light? The second entity I want to look at is the electron. They’re even more widely used in theory than photons, and have been around even longer. But we have difficulties in saying what exactly they are. It is quantum mechanics which particularly caused problems for interpreting what an electron is. The sort of problem can be illustrated by Heisenberg’s uncertainty relations. These apply to anything that can be described as a wave. So we can apply it to electrons. According to these equations, the position of an electron can be measured in principle to an arbitrarily high precision but then its momentum has a very low precision. It is just the same in reverse. If its momentum is measured to a high degree of precision, then its position is uncertain. In fact, the two are related by quite a simple relation. Uncertainty in position times uncertainty in momentum equals approx h. Again, the problem is, what is an electron? Many people have thought that if it is real, it must have a determinate position and momentum. It hasn’t, so it cannot be real. Again, if an electron is just a tool for scientific reasoning, and not a real entity, there is no problem. There is no reason to suppose a theoretical tool couldn’t have such uncertainty relations. Questions? Do you think there is a real problem for either photons or electrons? I was doing problems we have with interpreting theories we have currently. The second sort of problem we have with interpretation is how to match up theories which describe different areas, and seem to be incompatible in some situations. Quantum mechanics describes a micro-level and classical mechanics describes a macro-level. On the micro-level there is lots of weird stuff which doesn’t appear on the macro-level. For example, the macrolevel doesn’t show the sort of uncertainty we’ve already talked about which is common on the micro-level. Phyllis McKay Illari: Material provided for Philosophy of Science Unit But what happens when the two levels interrelate? Well, you get various paradoxes it is very unclear how to interpret. Probably the most famous example is Schrodinger’s cat. The set-up is that radioactive decay, which is uncertain, causes the death of the cat, which is not uncertain. It is indeterminate whether radioactive decay has happened, but it is not indeterminate whether the cat has died. But where does the certainty come from? How do these two systems really relate? So this is a problem for understanding and interpreting theory, not entities. It is particularly acute because we are trying to apply two apparently incompatible theories to the same situation, and we cannot match the two up satisfactorily to get a complete account of what happens. In this case, can both theories be true or nearly true, or either theory, or neither? It is important not to get this point mixed-up with other stuff we’ve looked at. Some stuff it is linked with, some not. Kuhn and observation versus theory, and Quine – certainly. We’ll look at this next week. But the debate over realism is not a version of scepticism about using induction in science. Both realists and anti-realists agree that scientific theories are not certain. They’re not proved. That is not where the issue lies. The issue lies in whether scientific theories are aiming at truth, or whether unobservable entities are real. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Laws of Nature Lecture 5 1 Course so far and introduction Useful to take stock at this point: Course so far covered: Is there a distinctive scientific method? Bacon and Popper’s theories of method Various problems for theories of method particularly, the view that there is no simple relation between a chunk of theory and a chunk of data and worries about what the rationality of science consists of Conclusion: Giving an account of the method of science is a lot more difficult than it might be thought. Put this aside because it’s as far as we can get now. Change our interest from method to look at the aims of science. We’ll look at two major topics – law today, explanation next week. After getting these straight, we’ll come back to method. Understanding of the two will interlink, giving us richer resources for thinking about scientific method. Law Talked about it a lot, in passing. We’ve talked about hypotheses, theories, and even laws. We’ve complained about Popper’s particular notion of a hypothesis, and worried about Bacon’s claim that we get generalisations from data like squeezing wine from grapes. What Bacon and Popper have in mind for these laws will affect the account of scientific method they give. Now we’ll think about it explicitly. We’ve already mentioned several laws, and can add some. OHP and handout Physics: Kepler Newton Chemistry: Avogadro’s Law, law of definite proportions Economics Parkinson’s First Law Phyllis McKay Illari: Material provided for Philosophy of Science Unit Increasingly controversial as you move down the list. But are all called laws by practitioners. Laws not the only product of science. It is uncontroversial that they are one important product. It is controversial – what they are. General idea is that a law of nature identifies some fundamental property of the world. How to make this claim precise is a problem. The best way to show you this it to go through accounts of laws and look at their problems. Overview OHP Go through 2 Regularity theory of laws It is a law that all Fs are Gs if and only if all Fs are Gs OHP Ie It’s a law that all frogs are green if and only if all frogs are green. It’s a law that f=ma if and only if force always equals mass times acceleration etc. This called the regularity theory. Laws are just regularities in the world. Two things come together, like frogs and green, income increases and expenditure increases, force and ma, and as long as they always come together, that makes a law. Laws are nothing over and above that collection of regularities. Why hold this? It is the most minimal account of a law. Can be a response to various worries we’ve discussed about what science can get from observational data. We only see particulars – remember the problem of induction. The regularity theory makes the weakest possible claims about what a law is. Specifically, a law doesn’t involve anything metaphysical and hidden. It is just a collection of particulars that we do know exist. A law doesn’t identify something behind the instances that we might not know exists. So there is a point to this view. This is why some still hold it in spite of its problems. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 3 Problems for the regularity theory 1 Accidental regularities ConsiderL All people are under 10ft tall All creatures that have a heart have a liver Bode’s law proportions between planets. If inc asteroids as missing planet between Mars and Jupiter. Fitted until Neptune. Should be 388, actually 300. The regularity theory implies all these are laws. (Bode’s Law was before the planet Neptune was found.) What wrong with em? They are just accidental regularities – coincidences. Laws are not accidental. They describe a relation such that in some sense it couldn’t have been otherwise. You also have problems with universal regularities that have no instantiations at all. All zoogons are cheeky. All zoogons are not cheeky But some people think that laws are not vacuous. And two laws of nature cannot be contradictory. 2 Laws that aren’t regularities Regularities seem to demand repeated events. The simple regularity theory says a law is just a collection of its instances. To get a genuine regularity, you might think you need more than one. But some people think that laws could have a single instance. For example, we could have laws about unique events that will only happen once, like the Big Bang. Phyllis McKay Illari: Material provided for Philosophy of Science Unit A particular reading of the simple regularity theory rules them out. 3 Laws covering stuff that didn’t happen/couldn’t happen The simple regularity theory says a law is just a collection of instances. So there could be no law about events that never happen, or objects that don’t exist. (This is true whether it never happens as a matter of accidental fact, or as a matter of law.) But we do seem to have laws that describe events that haven’t or couldn’t happen. Merely haven’t happened: the laws of motion describe what would have happened if a large body of mass x and velocity y had entered our solar system on 1/1/2002. We could calculate, for example, if it would have missed the earth, or changed the orbit of any of the planets. Couldn’t happen: the theory of gravitation implies no mass will ever be under no gravitational attraction whatsoever. All masses in universe attract every other mass. Nevertheless, the theory describes what would happen to a mass that wasn’t. It would behave as a mass under no net force, or according to other forces acting on it. The simple regularity theory doesn’t allow this. If it couldn’t happen, or merely hasn’t happened, no law can describe it. So we are going to have gaps in laws that try to describe some areas where there are no actual instances. This is especially a problem for functional laws. These are laws which involve numbers – they range over indefinitely many values of the variables the laws contain. For example, the functional law F=ma covers values of force and mass we haven’t observed and which may not even exist. Similarly, Hooke’s law F=kx says that the force exerted by a spring is proportional to the extent the spring is stretched. It tells you what will happen if a spring is stretched to 100 times its normal length – although this doesn’t happen in practice. Phyllis McKay Illari: Material provided for Philosophy of Science Unit The ideal gas law PV=nRT says that the pressure times volume of N moles of gas is proportional to the absolute temperature of the gas. This tells you what the pressure of a gas will be at 1 million degrees. The simple regularity theory doesn’t allow for laws like these. So laws can cover events that have never happened, or could never happen. They are not just about what actually happens, but what would or might happen. 4 What do these problems suggest about laws? 1 Not accidental Couldn’t have been otherwise Laws not vacuous. Not contradictory. 2 laws can cover unique events 3 laws can cover stuff that never happens – either by accident, or because we think it never could happen. Put technically – laws support counterfactual claims. If a big body had entered our solar system, then .... If a body was under no gravitational attraction at all, then .... Some philosophers still want to hold to the simple regularity theory. The accept the above implications because they want to hold on to a minimal account. They try to add conditions to make the theory more sophisticated to get round the problems. They are not entirely successful. But many think these problems show the simple regularity theory is wrong. Laws can’t be just collection of instances. They must describe more than a simple regularity. What? Laws OHP What think? BRAINSTORM Phyllis McKay Illari: Material provided for Philosophy of Science Unit Some people think laws can be more solid – they can make metaphysical claims. We should put aside sceptical doubts about what science can get from the particular instances that are its data, and accept that whatever a law of nature is, it is more than a mere regularity. Laws give you what underlie regularities. This approach offers the possibility that laws will explain their instances (see next lecture). If laws are just a collection of instances, they couldn’t, because something can’t explain itself. For example, the law f=ma can’t explain why things are so if all the law is is the collection of things that are so. So what is the something more that can do all this? 5 Necessity The most commonly suggested idea is necessity. A natural law isn’t just the claim that all Fs are G, but the claim that it is in some sense necessary that all Fs are G. If something’s an F, it must be a G. This necessity, whatever it is, exists in the world. There’s a necessary relation between F and G, and this explains why Gs come along when Fs do. The necessary relation explains why the regularity exists. Examples: Laws OHP Run through meaning for Laws Kepler: a planet in our solar system must move in an ellipse with the sun at one focus, and move faster when nearer sun. On this reading, the law would mean: There is a necessary relation between being a planet under those conditions and having that motion. Newton: body experiencing no net force must remain at rest/continue with constant velocity. Phyllis McKay Illari: Material provided for Philosophy of Science Unit This then means: there is a necessary relation between the force exerted and the motion of the body. Avogadro: this litre of gas, and this other litre of gas, at the same temperature and pressure, must contain the same number of molecules. Necessary relation between how many molecules of a gas and how much space it takes up at a certain temperature and pressure. Law of definite proportions: any reaction turning hydrogen and oxygen into water must use up exactly the ratios 7.94 hydrogen to 1 oxygen. Necessary relation between being H2O molecule and how much hydrogen and oxygen you contain. If the income of a family increases, their expenditure must increase. Is this last necessary? Is it the same as the others? This approach shows why these laws are increasingly controversial as counting as laws. So this approach does deal with the problems of the simple regularity theory. It accounts for laws not being accidental. There is a necessary relation, so the situation couldn’t have been otherwise. This why accidental regularities aren’t laws. It explains why ‘Zoogons are cheeky’ is not a natural law. There is no reason to believe there is any neccessary connection. It also explains why laws can’t be contradictory. There can’t be a necessary connection between F and G AND between F and ¬G. But if you did think laws describe a necessary connection, that explains why we could have a law about events that don’t happen. There is no reason to think a necessary connection goes away just because there are no instances. Events that accidentally don’t happen: If a large mass had entered our solar system on 1 jan 2000, then .... If there are necessary connections between being a mass, and conforming to the laws of motion, that explains why we Phyllis McKay Illari: Material provided for Philosophy of Science Unit can still make that claim. We know what would have happened because we know and understand those connections. Events that don’t happen by law work similarly, such as for masses under no gravitational attraction, or springs stretched to 100 times its normal length. We can still isolate what would happen even though we know other laws prevent it’s happening. Again, this is because we think we understand the connection, and think the connection doesn’t go away just because other laws intervene. No gaps in functional laws. If the law describes a necessary connection, there is no need for gaps. We think we understand the connection even in the absence of instances. Laws support counterfactuals. And if laws are necessary connections, that accounts for them supporting counterfactuals, since we think we understand that connection, whether the event really happened or not. Just as we think there is no reason we can’t have empty laws, we also think laws can cover unique events. This account can explain this. There is no reason why there couldn’t be necessary relations instantiated by the Big Bang and no other event whatsoever. Explanation. Laws explain the regularity of their instances. They explain this by explaining the link, saying why one comes along with the other. They explain why the world is as it is. (We will look at this more next topic, on explanation.) Laws OHP Look through laws again, with these claims in mind. Can you see why they are increasingly controversial? What everyone think? DISCUSSION Problems with necessity What is necessity? Phyllis McKay Illari: Material provided for Philosophy of Science Unit Laws presumably are not logically necessary. We take it that there could, logically, have been different natural laws governing the world. This might be described as saying that there are logically possible worlds with different basic natural laws. So what does it mean to say that a law describes a necessary relation? Discussion There is also a genuine problem with how science can establish the existence of necessary relation. We never directly observe them. Why are we so confident that it really exists? And how do we establish that it exists? These problems are why some stick with the simple regularity theory, for all its problems. This is also related to what I have said about explanation. Many people think science gives explanations, so if we need more than regularities to get genuine explanations, then science must discover more about the world than simple regularities. Even if this doesn’t demand establishing the existence of natural necessity, it does involve accepting we can get more than simple regularities. We will talk more about this when we study explanation. Summary and next week We have looked at accounts of the method of science, and have now moved on to study its aims. Most agree it is a major aim of science to discover and accurately describe natural laws. We have looked at the problem of giving an account of what natural laws are. We have seen that the simple regularity theory has problems, but understood its attraction as a minimal account of laws. We have thought about introducing necessity into an account of laws to deal with the problems of the simple regularity theory. We have seen that the major problem for that approach is in giving an account of necessity itself, and how we establish its existence scientifically. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Next week we study explanation. We will think about whether explanation as also a function of a natural law, or whether finding natural laws and constructing explanations are distinct in science. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Lecture 8 Kuhn: implications of the theory 1 Last week We looked at Kuhn’s distinction between normal and revolutionary science, and thought about what he means by a paradigm. We compared Kuhn’s descriptions of normal science and of revolutionary science. This week, we will study the implications of all this for science and scientific method. 2 Kuhn and Incommensurability What exactly Kuhn means when he says that competing paradigms are incommensurable is controversial. But it is to do with his claim that we can’t compare competing paradigms in any straightforward way with an independent body of data and decide which is better. This what Kuhn is trying to get at by claiming that competing paradigms are incommensurable. Competing paradigms are not completely incomparable with each other – Kuhn is trying to get at very specific, particular problems which arise in trying to compare them. Last week I was trying to get you to see the overall structure of the theory. This week, we need to look at some important detail. There are three distinct things Kuhn might mean by ‘incommensurability’. Or maybe his account includes elements of all three, or only two. We nee to separate the three strands and understand better what Kuhn might mean by ‘incommensurability’. 2a Observations conditioned by Theory We have talked about this before in various places. Bacon: we thought it was impossible, and probably not desirable, to look at the world with no presuppositions Phyllis McKay Illari: Material provided for Philosophy of Science Unit Duhem thought that all observation in physics is theory-laden. The example he uses is that of measuring the electrical resistance of a coil. You get the answer of 2.5 ohms, he says, only by interpreting the nature and movement of many pieces of apparatus, using a group of complicated physical theories. Duhem thinks your observation is theory-laden any time you use an instrument, even a magnifying glass. Galileo used the newly-invented telescope in 1609 to record for the first time observations of mountains on the moon, sunspots, about ten times more stars than could be seen with the naked eye, that the Milky Way is just a large cluster of stars, and that Jupiter has moons orbiting it. These discoveries were treated with suspicion – many people believed that the phenomena observed were somehow created by the telescope. We make scientific observations of the very big, like Brazil, Mars, the Milky Way and the very small, like cells, bacteria, molecules and sub-atomic particles. To make these observations, we have to use instruments. Cloud chamber a sophisticated instrument for making observations that we’ve discussed before. Using it, we detect the particle but don’t directly observe it. And as we have discussed before, the observation is not independent of the theory governing the sort of thing observed. The theory about what is going on in a cloud chamber is closely related to theory about the sub-atomic particles we use cloud chambers to detect. These issues that we have talked about before is just what the claim that observation is theory-laden is about. Some philosophers think theory conditions our observations even in daily life. We don’t need to worry about this too much, but think about the example I’ve given already of perceiving a horse trotting. Even to recognise a moving patch of colour as a white horse trotting requires theory – what a horse is, and what trotting involves at least. Necker cube (see OHP) Phyllis McKay Illari: Material provided for Philosophy of Science Unit Look at the diagram. You can see the a-plane as the front, or the b-plane as the front. And you can, with a bit of practice, manage to switch between seeing it one way or another. Duck-rabbit Wittgenstein’s famous example. You can see it as a duck, or as a rabbit. The theory we say conditions these observations is not a scientific theory – it is very thin, very insubstantial. It is merely that we are accustomed to seeing representations of mathematical objects, and natural objects like animals, so you interpret these line drawings quickly as a mathematical object and an animal according to what you’re accustomed to. The ability to switch between two different ways of seeing the objects just emphasises the fact that you are doing some interpretation in taking this object to be a cube, or a duck/rabbit. We don’t have to worry about this so much, because we’re thinking about Kuhn. We are trying to figure out what he means by the claim that two paradigms are incommensurable. But one thing Kuhn might mean is closely related to these worries. Kuhn thinks that the paradigm you hold conditions your observations. So people working under different paradigms cannot straightforwardly compare their observations with observational data independent of the two competing paradigms. This is not true of people holding the same paradigm, so doesn’t apply to normal science. People working under the same paradigm will have their observations conditioned in the same way. They will be working from the same set of observational data. But in revolutionary science, two competing paradigms cannot be compared in a straightforward way with observations independent of each paradigm. A set of data, conditioned according to paradigm A, goes with that paradigm, and a different set of data, conditioned according to paradigm B, goes with that paradigm. And no observation is independent of a paradigm. So there just isn’t a paradigmindependent set of data to compare two competing paradigms to in a straightforward way. Phyllis McKay Illari: Material provided for Philosophy of Science Unit In SSR, Kuhn makes pretty strong claims along these lines – although he later moderates them a lot. He says that people holding different paradigms make different observations. Kuhn gives us two striking examples. 1 Phlogiston theory of burning. This is an old theory in chemistry. One of its claims is that burning substances give off phlogiston. (Since we no longer use the theory, the word is no longer used either.) Now we think burning substances take on oxygen. But scientists working under the phlogiston paradigm knew oxygen existed. The gas we now call oxygen had been isolated before we decided burning was really all to do with oxygen. The difference Kuhn is interested in is in how phlogiston-paradigm scientists saw oxygen versus how we – oxygen-paradigm people – see it. Priestley, who first isolated oxygen, thought oxygen was dephlogisticated air, and thought it confirmed phlogiston theory. Lavoisier thought it was oxygen, and confirmed the modern paradigm. As Kuhn claims, both scientists had got hold of oxygen, but interpreted what it was differently. The fact that they both took it to confirm their own theory does support Kuhn’s claim. In some sense, they made different observations. According to Kuhn, we think Lavoisier is right because we have the same paradigm. 2 The second example Kuhn gives is pendulum motion. Modern paradigms for motion of a pendulum is that it is repeated motion – the pendulum almost manages to repeat the same motion over and over. The Aristotelian paradigm for motion is different. They thought a heavy body is moved by its own nature from a higher position to its natural state of rest at a lower position. To them, a pendulum is a heavy body trying to fall but constrained by the chain. It is trying to get to its state of rest and only manages that with some difficulty. Again, Kuhn says these are different observations, and points out that scientists under each paradigm both took the observation of a pendulum to confirm their own paradigm. Phyllis McKay Illari: Material provided for Philosophy of Science Unit These claims of Kuhn’s are very different from the prior picture of science we’ve looked at, where you compare a single theory with a body of data, whether you want verification or falsification. For Kuhn, this can happen in normal science, but not paradigm shift. For these reasons, Kuhn thinks there can’t be crucial experiments to decide between competing paradigms. Kuhn goes so far as to say Galileo was living in a different world from Aristotle. I think this claim is best taken metaphorically. QUESTIONS What do you think? 2b Meaning change After SSR, in various essays collected as The Essential Tension, Kuhn withdrew some of his more dramatic claims, and tried to explain what he really meant by incommensurability. This is the second sort of thing Kuhn might mean. It is also strongly present in various places in SSR, although this interpretation became more prominent later, because Kuhn himself drew attention to it. It is still another reason Kuhn might have for thinking you can’t compare competing paradigms adequately with each other or with an independent body of data. When you take on a paradigm, Kuhn thinks you take on a whole new world view, and start looking at the world differently. One of the important aspects of this Kuhn points out is the language, or more specifically, the concepts, you acquire as part of the paradigm. There can be big differences between the conceptual structures of different paradigms, for two sorts of reasons. First, paradigms often use clusters of inter-defined concepts. For example, what phlogiston is is defined using words like ‘principle’ and ‘element’ which are also no longer used in the same way in modern science. To understand properly what phlogiston is, you need to acquire the whole Phyllis McKay Illari: Material provided for Philosophy of Science Unit cluster together. You can’t completely understand what each of the terms means individually. Secondly, different theories often involve conceptual disparity among terms. Take as an example the French word ‘doux’. In the French language, doux is a unitary concept. It can be applied to honey (sweet), underseasoned soup (bland), a memory (tender), or a slope or wind (gentle). But there is no single concept in English which covers all these. The concept can only be acquired by acquiring at least a large section of the language, if not all of it. Certainly, the meaning of key terms in science can change. Einstein and Newton both use the word ‘mass’, but it means something different. For Newton mass is absolute, an intrinsic property of matter. It can’t be created or destroyed. For Einstein, mass is a complex relation between the speed of light, a chunk of matter and a reference frame from which the velocity of the chunk is measured. It can be converted to energy. This is an important conceptual disparity. It means it is not possible to lay out all the claims of these two paradigms and compare them point by point. You can only compare the whole with the whole. Notice that Kuhn’s claims about language and concepts are mostly to do with comparing different paradigms with each other. But if you can’t do that, it certainly makes it more difficult to compare them to a single set of observations. On this point, part of Kuhn’s idea is that to get the concepts of a paradigm properly, you really have to be working in it. You have to take on the cluster of concepts – the whole lot – and start thinking according to it. The way Kuhn puts it is that you stop translating the language and start using it. This means you can’t take up a stance independent of the conceptual structure of one theory or another and compare the claims the paradigms make with an independent body of data. DISCUSSION What do you think? Certainly true of some terms that they are defined by role in theory – like mass and energy. These do look like concepts that get their entire meaning from how the theory that uses them defines them. But suppose we are talking about competing astronomical paradigms. Is it really true that Phyllis McKay Illari: Material provided for Philosophy of Science Unit Ptolemy and Copernicus meant something different by ‘the planet Mars’ even though one thought it circled the earth, one the sun? 2c Shifting standards This is the third possible meaning of incommensurability and third possible reason for thinking scientists can’t adequately compare competing paradigms, with each other or with an independent body of data. A paradigm includes not just instruments, but more advanced techniques and models for interpreting observations. It might also include values that tell you what counts as an important observation and what can be discounted. Consider what Chinese astrologers had observed centuries before western astronomers, because we were hampered by the belief in the immutability of the heavens. Western astronomers must have discounted countless observations by not taking them seriously, and so not recording them and pursuing them. At a deeper level, a paradigm might condition what counts as successful science at all. The move from Aristotelian science to modern science involves huge changes in method. I have already given you an example – the enormous success of Newtonian mechanics rendered his mode of explanation of gravitational attraction acceptable. Darwin. The theory was adopted along with an alteration in attitudes about what was acceptable as a scientific theory. Argument about whether it was an acceptable scientific theory was what caused a lot of the controversy. If scientific method is part of the paradigm – including in important ways not just the observations themselves, but what you ought to do with them, then you can’t compare competing paradigms with independent body of data at all. It is the paradigm that tells you what to do when you set about comparing. QUESTIONS What do you think? Phyllis McKay Illari: Material provided for Philosophy of Science Unit An issue is how far this claim plausibly goes. An alternative possible view is that maybe one revolution established modern (successful) scientific method, and inquiry according to the old method, while a precursor to real science, wasn’t really scientific at all. Question: Does Kuhn rely on all 3, just one, etc? Is it plausible that paradigm shifts all involve 1, 2, 3, two of them or all of them? Talk about in seminar just get thinking about now. 3 Rationality and objectivity Now we have an idea what the problem is: normal science is ok, but only because of the existence of a ruling paradigm. But science in paradigm shift does not look objective and rational. So does this mean science is subjective and irrational? Kuhn in later writings is very keen to explain why he doesn’t think his views imply science is irrational and subjective. Some philosophers are still uneasy. Note that objectivity and rationality are DIFFERENT. There is no reason to believe – at least not without a good deal of argument – that something couldn’t be subjective but rational, even in science. Take them separately. Objectivity How is science not objective on Kuhn’s account? There are (at least) three things you could mean by the claim that Kuhn implies science is not objective. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 1 There is no logical relation between theory and data. But remember, we have already criticised Bacon and Popper on precisely this point. It is perfectly possible that Kuhn is just right about this! If there is no such relation, we need to get on with it. We might be wrong about the sort of objectivity to look for in science. 2 Science is not progressing towards an objective truth. Yes. Kuhn implies science is not cumulative. We will talk about this when we look at realism next week. 3 Science depends in important ways on the decisions of individual scientists/groups of individual scientists. Choosing one paradigm over another involves decisions that are personal. Kuhn says they are ‘inarticulate and aesthetic’. They involve what is merely an intuitive grasp – this new idea is fruitful. This goes along with the rest of the picture Kuhn gives us of science in revolution – like political revolution, resorting to mass persuasion, training new members of the community up in the new paradigm etc. Kuhn also says that individual scientists just transfer, for personal reasons. They ‘convert’. In this way, Kuhn certainly does imply that science is not objective. The question to think about now is whether that implies that science is not rational. Rationality Not clear objectivity and rationality are the same thing. I suggest that there is no reason why science shouldn’t be subjective in the ways I’ve suggested Kuhn thinks it is, and still be rational –possibly even a rational way of finding out about the world. Kuhn thinks science is rational. It is just not right to insist that science must be rational in the way that those before him demanded. Science is not rational in the way that logical relations are rational, but this certainly not Phyllis McKay Illari: Material provided for Philosophy of Science Unit mean that science is irrational. The idea is that decisions made when the scientists choose on aesthetic, personal, subjective grounds can still be described as rational. They are still distinctively scientific grounds. Interpret Kuhn as saying that scientific rationality just doesn’t have to be rigidly prescribed, like the logical positivists thought. Most important reason why Kuhn is thought to imply that science is irrational is his claim that the decisions in science are the personal decisions of individuals. And obviously he thinks that there are certain sorts of things those decisions can’t be – because of what he thinks about incommensurability. Those decisions can’t be based on simple comparison of paradigms with each other, or with data. They involve an element like conversion – a scientist takes on the theory, values, language and standards ie he or she takes on the paradigm. To investigate whether such decisions could still be rational, and scientific, consider yourself as a working scientist in an area of science in revolution. You have to decide what to do. What is the exact nature of the situation you are faced with? Your decision CAN’T work like this, because of incommensurability: 1 based on comparison of two paradigms with body of data where observation of data not conditioned by one paradigm or other. 2 based on comparing paradigms point by point. Has to be comparison as a whole, and involve taking on the language of the paradigm, which affects how you see world according to 1. 3 based on shared standards. You have to take on the standards of the paradigm along with it. Can a decision that doesn’t work like this still be rational? You’ve been working away for years. You’ve noticed anomalies, and since you’re a good normal scientist, you’re involved in the puzzle-solving tradition. Suppose you’re working on explaining the problem with the perihelion advance of Mercury under the Newtonian paradigm. You have made very little progress over time. You are fed up. Phyllis McKay Illari: Material provided for Philosophy of Science Unit At this stage, you are starting to think this paradigm is not going to work. You can’t see how it could work, and when you chat about it neither can most of your colleagues. You hear, over coffee or in a conference or mentioned in a journal article about some bright spark called Einstein. Apparently, he thinks that space and time are not absolute. He’s got a very odd theory. But it explains the perihelion advance of Mercury. Rubbish, you think. How can space and time not be absolute? You carry on working. But the idea niggles at you. One day you decide to investigate, and you get hold of the paper and read it. You can’t understand everything just by reading it, so you start some work. You start investigating. You figure out some new experimental procedures and start asking colleagues about the new idea. In a very baby way, you have started exploring the new paradigm – you have started the work of normal science. You don’t immediately come to a grinding halt over insuperable difficulties, so you continue. At some point, you get very excited by this new idea and start doing lots of work. Before long, you’ve taken on the new paradigm and started working in it. You are no longer worried about relativistic space-time – you are now excited about how it might work and what the theory using it might show. It seems to me that the reasons here are personal and subjective and everything Kuhn said they were, but still rational and still scientific. You convert to the new paradigm because think this will do it! Your excitement is scientific. Kuhn also thinks that social and historical factors are crucial. Some people don’t like that aspect of his work. But think about the things it involves. You are influenced by colleagues, journals, conferences etc. But you are still making decisions for scientific reasons. It is true that in Kuhn’s work it is crucial that scientists form a community. It is not just you working away alone. There is a very large community of scientists, some working on your area. An element some dislike in Kuhn’s work is that he allows for individual nutters still clinging to outdated paradigms. In the context of the example we are using, that would be the Phyllis McKay Illari: Material provided for Philosophy of Science Unit Newtonian paradigm. But paradigm-shift involves all the scientists in relevant field. My suggestion here is that perhaps you should be affected by your colleagues. This is another useful source of pressure, and not something that further renders science irrational. Now, for a paradigm to take hold and paradigm shift to be complete, you have to persuade every scientist in the field to start working in it. They all have to convert, so they all have to believe for some personal reason or other that this new paradigm is more fruitful, this is the new best approach. A new paradigm persuades the relevant part of the scientific community of its worth – over and above a once-successful old paradigm. People think Kuhn allows the possibility that the entire community could be persuaded for bad reasons, and I think this is true. He allows the possibility. But I wonder – how long would it last? Could a fad really persuade the whole scientific community for any length of time for the wrong reasons? So what I am suggesting is that there are going to be good personal subjective reasons for making scientific decisions, and bad ones. And how likely is it that the entire community would be persuaded, for long, by bad ones? So long as this is entirely unlikely, then the whole process of science has a claim to rationality – and to being distinctively science. Question – does the plausibility of all this involve some implicit standards independent of a paradigm, even if those standards don’t constitute a complete scientific method? QUESTIONS Do you think Kuhn’s account implies that science isn’t rational? What about objective? Do you think they’re objectionable? Progression If there is any method independent of the paradigm – we have a paradigmindependent standard and might identify some sort of progression. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Although a political revolution looks like a radical break from one point of view, it is a familiar point that the revolution actually arises out of the old system, and a lot is preserved between the two systems. To some extent we saw this when we looked at the shift from Ptolemy to Copernicus. Copernicus is still using the deferent-epicycle system developed under the Ptolemaic paradigm, and calculating the predicted angles of the planets using a similar mathematics, and similar instruments. Relevant observational evidence might even emerge only through the operation of a paradigm, but survive the rejection of the paradigm and even be part of the reason for that rejection. Were it not for Newton the anomalous movement of Mercury might not have been noticed, but in the end the success of Einstein in explaining it while Newton couldn’t was part of the reason for the rejection of Newton’s theory. So some think that there are some observations that do stand between theories. Take electrons as an example. It is true that what electrons really are is a matter for debate. We don’t directly observe them. Do they exist? Maybe, maybe not. Much of what we now believe about electrons might come to seem false, but the truths we now interpret as being about electrons remain. Something activates cathode ray tubes and Geiger counters; something makes bodies fall to earth, even though the explanations we might produce for these things can be wildly different at different times. We will discuss these issues next week when we study realism. 4 Next week We’re going to study scientific realism – whether scientific theories aim at the truth, and whether the entities postulated by those theories are real. One of the reasons for denying realism is of course buying into the Kuhnian picture. But think about Quine and underdetermination as well. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

10 Demarcation 1 Last Week We’ve now covered most of the central topics. We’ve studied the theories of method of Bacon, Popper and Kuhn, and the implications of the views of Duhem and Quine. We’ve also looked at the major debates about the aims of science: law, explanation and realism. I hope you feel you have a more sophisticated philosophical understanding of science. We also did demarcation in the first week. We talked about astrology and whether it was a science, in a very informal way. Hopefully now you realise demarcation raises some more interesting issues. Part of the problem is whether science has got a distinctive method at all. You can see from the issues we’ve studied, especially Kuhn, that that’s a problem. We can now talk about it much more carefully. 2 Revise relevant stuff What do we want to distinguish from science and why? OHP Here again, we are looking at a pseudo-science – creation-science, this time rather than astrology. This is generally done because we want to deny the pseudo-science the standing of science, which is thought to be epistemically superior. So now we can see the demarcation question as ‘Does science have a distinctive method?’ and ‘Does it matter if it doesn’t?’ 2a Popper Popper gives the classic account of demarcation. His views on this are closely related to the stuff we’ve already done on his method. He thinks demarcation has two parts: OHP 1 A claim is scientific if it is falsifiable by empirical evidence. Something is a scientific theory if it makes predictions which are falsifiable by empirical evidence. (i.e. if it makes scientific claims.) Phyllis McKay Illari: Material provided for Philosophy of Science Unit So for example, a statement like ‘God exists’ is not scientific. It is not falsifiable by comparison with observations. But a statement like ‘light passing close to the sun will bend so that the stars appear displaced by 1.75 seconds of arc,’ is. It’s a prediction too, so the theory which generated it is also scientific. 2 Popper also thinks that the attitude and methodology of the practitioners of a study are important to whether it is a science. Popper says they must hold the theory tentatively. They must be prepared to compare it to evidence and reject it if the evidence is not favourable. They must be prepared to find that their theory has been falsified. Summary: A claim or statement is scientific if falsifiable by comparison to observational data. A theory is scientific if those who hold it hold it tentatively, try to compare it to empirical evidence, and are prepared to reject it if it does not compare favourably with the evidence. So they must make predictions from it that are falsifiable by comparison with empirical evidence. (I.e., if it makes predictions which are themselves scientific claims.) And then they must test them. So there is a strong link between what makes a theory scientific, and what makes a statement scientific. 2b Problems with Popper For some studies this comes out intuitively right: But note that Popper needs both parts to rule some pseudo-sciences not scientific. Astrology: OHP ‘Mercury is in a link to both Saturn and Pluto so people who were being defensive are now putty in your hands.’ Creation-Science: ‘The world was created in six days.’ Both falsifiable. The astrological claim could be tested by methods used in the social sciences, and we take the CS claim to have been falsified. But arguably Popper’s claims about practitioners shows them non-scientific. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Their practitioners either do not attempt to test them at all, or are at least unprepared to give them up in the face of unfavourable results of testing. So the claims about the attitude of practitioners are really important to Popper’s account – otherwise astrology and CS would come out as scientific. Popper’s account has been hugely influential and lots of people buy it. But we know there have to be problems with this view. Duhem and Quine showed that falsification in Popper’s sense cannot be crucial to what makes a scientific theory scientific. Nothing is falsifiable alone. You need at least a group of other theories and auxiliary hypotheses to generate a prediction, or if you buy Quine, you need everything you know in there in generating any prediction. Remember, the ‘unit of empirical significance is the whole of science’. So falsifiablility CAN’T be what makes a scientific claim scientific, and if that’s true, then generating scientific testable predictions can’t be what makes a theory scientific either. There is a fairly snappy way of showing falsification is just too simplistic as a demarcation criterion. OHP God exists Bacteria exist Planets exist God does not exist Bacteria do not exist Planets do not exist The left-hand ones come out as verifiable but not falsifiable, while the righthand ones come out as falsifiable but not verifiable. Whether these claims are falsifiable or not is just an accident of the logical form in which they are expressed. And clearly you can’t have one set being scientific while the other set are not. Popper’s claims about practitioners are also in some trouble. Quine shows at least that there are potentially different ways of altering beliefs when faced with anomaly. Kuhn goes further. He argues that you are entitled to hold on to a theory in the face of anomaly – because you still have faith the paradigm will work out. We agreed at least with this element of Kuhn’s work, noting examples from history of science such as the positing of the Phyllis McKay Illari: Material provided for Philosophy of Science Unit planet Neptune, and even the planet Vulcan which did not prove to exist. But it was not irrational to posit this to explain the anomalous orbit of Mercury; it was good to keep working on this problem within the Newtonian paradigm while you still thought it would solve the problem some day. Consider the history of medicine and chemistry. Their origins are occult and magical, but their practitioners persisted and they are both highly respected sciences now. So hopefully we can agree that Popper’s account of demarcation cannot be the full story. The worry is whether there is a story at all. Consider what we have studied. Kuhn may be claiming that there is no scientific method independent of particular paradigms. And you might buy anti-realism. So overall you might come to view science as just a human construct in search of empirical adequacy. Problem – if science no distinctive method, may be unable to demarcate at all. Question is: can we find a middle path between simple snappy criterion like Popper’s and being unable to demarcate science from non-science, or are we stuck? Do we think science really has no distinctive method? Remind you – 3 things Kuhn might mean by incommensurablity. 1 observations conditioned by theory, 2 conceptual disparity 3 shifting standards – method in paradigm. We’re particularly worried here about the 3rd. QUESTIONS? 3 Is Creation-science a science? 3a What is creation-science? Creation-science is just the theory that the earth was created as the biblical book of Genesis claims. So the world was created in much its current Phyllis McKay Illari: Material provided for Philosophy of Science Unit condition in six days not very long ago – i.e. among other things complete with a fossil record etc suggesting the earth is much older. Man has separate ancestry from the apes, all species were created by God and have been limited in their variation since then. This is opposed to the evolutionary view which involves claims like: the earth is much much older than that, the fossil record shows us about inhabitants of the earth at various times including times long before creationscientists place the creation of the earth, and man evolved from lower animals. A very long and gradual process of evolution accounts for the variation in other sorts of species too, and all creatures are still evolving. This issue is live. 1982 court case in the Arkansas supreme court to decide whether it was constitutional for the Arkansas Act 590, passed by the Arkansas congress, to force biology teachers in Arkansas to teach creationscience as well as evolutionary biology in biology classes. Judge Overton decided that the key issue was whether creation-science was science or religion. (It is unconstitutional in America to promote religion in a secular institution.) It was decided that creation-science is not science, it is religion, and the Act was dismissed as unconstitutional. There was a big furore surrounding the case. We can talk about the details of this case in the seminar, because they are interesting. The question now, given worries about unique method and the possibility of a simple demarcation criterion, is are we stuck? What could we say in spite of problems? Note: We are still thinking about a psuedo-science case. It is not like eg history or maths, which might have a quite different story. Everyone straight on all that? 3b What Kuhn says about demarcation Surprisingly enough, Kuhn does make some claims that involve demarcation. He says astrology is not a science, calling it a pseudo-science. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Says that astrology has no central theory – no paradigm – and no puzzlesolving tradition like normal science. So here Kuhn is in fact using his characterisation of science to distinguish science from non-science. Science is ruled by a paradigm and is concerned with puzzle-solving activity governed by that paradigm. Science in revolution is different, of course, but he has told us a fair bit about that too. We could probably extend the story. What Kuhn is saying here is that if it isn’t ruled by a paradigm and involved in puzzle-solving, it isn’t a science, so astrology isn’t a science. Note Kuhn is discriminating science from non-science not by using a scientific method. He is using characteristics that he thinks all sciences display. Sciences proceed in a series of long periods of normal science and then paradigm shift, but this is not a method that they use. You can see how you might tell the story for creation-science. If it is not governed by a paradigm and puzzle-solving, it is not science. What do you think? Does it work? It looks fairly easy to decide something is not a science in a period of normal science, but it is open to pseudo-sciences to claim that they are in revolution. It seems to me, though, that there are still available responses. In the first place, you can’t remain in revolution for your entire history. If nothing changes, we are entitled to judge you by the standards of normal science. Secondly, pseudo-scientists might claim that they are merely clinging to an old paradigm that should not have been abandoned. So again, the suggestion Kuhn makes is a bit loose. Again, it allows pseudo-sciences room to manouevre. But again, I think there are possible responses. Kuhn does still identify some characteristics common to science, and it looks possible to identify pseudo-sciences as not having those characteristics. SO might say about CS: Not normal science: There is a great deal of consensus among creation-scientists and a very powerful ruling paradigm. But there is no puzzle-solving tradition, and no Phyllis McKay Illari: Material provided for Philosophy of Science Unit attempt to make predictions and compare them with evidence. The major duty of any good creation-scientist is to attack the work of evolutionary biologists and try to show that they are wrong. There is no attempt to produce their own positive work, push out the boundaries of their paradigm and try to apply it to new areas. So don’t do what scientists in normal science do. Not an old paradigm that should not have been abandoned: This is also arguable. Remember that I said although you might think Kuhn has shown science does not progress towards objective truth, but you don’t have to say that science is irrational, or even not rational. So you can still look at creation-scientists clinging to this ‘old paradigm’ and criticise the rationality of their decisions. So we can say to the creation-scientist, we have got better theories about same stuff, that accord with other current theories: Physical theories – origin of the universe, evolutionary theories – origin of life and how there came to be so much diversity in living creatures. These theories do show a puzzle-solving tradition and a strong concern to compare their predictions with evidence. They are also fruitful, coming up with novel predictions and ideas, as opposed to creation-science, which seems to have the object of remaining always as it first was. Not fruitful. Creation-science is inconsistent with current theory. We take current theory to have refuted the claims creation-science makes, like the one about the time of creation of the earth, and how long it took. So the idea is not that CS is unscientific per se, but that it’s over and done with. It’s been replaced. In a much more complex way than Popper ever meant, it’s been falsified. And it is no longer rational (in scientific terms) to hold on to it. Remember that Kuhn says that scientists’ personal decisions are guided by the values of accuracy, consistency, broad scope, simplicity and fruitfulness. Be careful that this doesn’t imply that there is some paradigm-independent method, though. Although bear in mind if you reject Kuhn’s extreme views, you might think this. Phyllis McKay Illari: Material provided for Philosophy of Science Unit SO the idea is that you cannot say creation-science is not and never has been a science. But you can look at CS now and see that the paradigm, if it ever was a paradigm, is over. So you can see, for personal and aesthetic and perhaps ultimately subjective reasons, that it has no place in current science. It seems that Kuhn leaves space for this. Remember that Kuhn says scientists’ subjective personal decisions are guided by at least five values: accuracy, consistency, broad scope, simplicity, fruitfulness. The broader idea is that Kuhn forces you to look at pseudo-sciences and build a case by case argument that they’re each not a science. But this doesn’t mean you can’t make the distinction between science and nonscience. Think about our discussion of astrology, and how you might build a similar case there. The case would not be the same, there would be different emphases, but there could still be a case. Astrology OHP. What we said about astrology from lecture 1. Discuss. QUESTIONS What do you think? Do you think this can work? Do you think it implies there is a distinctive scientific method that we could identify and simplify all this? Maths and Metaphysics These are not pseudo-sciences. The story might be different. Some do worry about the demarcation between these and science, although not many. We haven’t discussed it. Maths makes claims like ‘the angles in a triangle add up to 180 degrees.’ Metaphysics makes claims like ‘if God exists and is omnipotent and wholly good, there can be no evil in the world.’ Here, there might be a difference of kind, not just degree. And this difference might tell you something about scientific method, even if it is Phyllis McKay Illari: Material provided for Philosophy of Science Unit very thin. For example, maths and metaphysics are not really concerned with empirical stuff at all. They’re all about relations between concepts. I just stuck this in to remind you there are other issues here. QUESTIONS Next week Seminar Then done and done. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Lecture 9 Realism 1 Last week We’ve now looked at Kuhn, understood the structure of how he sees science. Now we are worried about the rationality and objectivity of science. I showed you that there are at least three ways Kuhn's account of science might imply that science is not objective, and tried to persuade you this didn't mean science wasn't rational. This week, we’re looking at realism. This is a debate strongly affected by Kuhn's work, and Quine's. Overview OHP 2 Debate Realism consists of interrelated theses, including: OHP 1 There is such a thing as scientific progress 2 There is such a thing as a unique scientific methodology 3 Scientific theories can be evaluated in terms of their truth or approximate truth, and the success of a scientific theory is evidence that it is true. 4 If a scientific theory is true, then the observable phenomena covered by that theory are explained, and the unobservable entities posited by the theory are also explained and really exist. We’ve discussed 1 and 2 already and we will talk about more next week. 3 should be fairly straightforward. Realists think scientific theories aim at truth. Note: Realists do NOT believe current theories are true. They merely believe that they aim to be true. The anti-realist denies this. The anti-realist thinks the goodness of a theory is nothing to do with its truth or falsity. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 4 could probably do with some elaboration. The debate about realism is not only concerned with the status of scientific theories, but also with the status of postulated entities, particularly the problematic unobservable entities. Scientific theories are not just about the observable. We go about postulating other stuff to account for what we do observe. For example, scientists postulated the existence of a new planet Neptune to explain the observed perturbation in the orbit of the planet Uranus. Now a planet is not an unobservable entity, so this is not a big problem. We then looked for the planet Neptune and found it where we expected it to be. So when we talk about the problem of unobservable entities, stuff we merely haven't observed is not the problem. The problem arises because we posit the existence of in principle unobservables. These are things that can’t be directly observed by human beings at all, even using instruments. The sorts of things I have in mind are neutrinos, protons, electrons, fields of force, quarks, photons and so on. All these things we have to observe indirectly, by their effect on something which is observable. So in the cloud chamber we see the effect on nearby particles of a close-passing electron. We see iron filings moving in a magnetic field, not the field itself. These sorts of unobservables are common in science now. Big issue in the realism debate. Generally: anti-realists think scientists should stick closely to observed phenomena when making and evaluating theories. They are very suspicious about stuff we can't observe directly. Realists are more confident – they think we can move from observed phenomena to claims about stuff behind it, underlying the observed phenomena. Hopefully you recognise these motivations as very similar to the motivations we talked about in discussing minimalist views of laws and explanation. QUESTIONS? 3 Constructive Empiricism The debate on realism is a vast field, including lots of different views. To study it in enough detail but not get lost, we are going to focus on one Phyllis McKay Illari: Material provided for Philosophy of Science Unit version. The major proponent this is a famous name in philosophy of science: Bas van Fraassen. He holds that scientific theories are the sorts of things that might be true or false, but their truth or falsity is not important to them being good theories. What's important is their empirical adequacy, and because of what Quine shows, we know there will be lots of theories that are empirically adequate for the same set of data. OHP van Fraassen says himself Curd and Cover p1069 'Science aims to give us theories which are empirically adequate; and acceptance of a theory involves as belief only that it is empirically adequate.' A theory is empirically adequate if it fits the facts. Ie if it fits the observable phenomena. For the anti-realist, the rest does not matter. If there is a deeper truth or explanation etc, that does not matter. So the anti-realist thinks all theories are doing is accounting for the observable phenomena. You can tell whether a theory is good by how well it does that. You do not need to worry about whether it's also true. It is not the main aim of a theory to be true. QUESTIONS Just so you keep straight: We have talked about theories as possibly not being true or even approximately true before, when discussing Quine and Kuhn, and even Popper. Don’t get misled. There are lots of things a false theory can do. They can help us organise our ideas, and give us a new, more successful approach to old problems. They can allow us to approach data in better ways, be more precise etc. We have talked about all these things when we talked about why a paradigm was good for normal science. For a theory to do these things, it does not need to be true. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 4 The Arguments There are quite a few on both sides. These are the major ones. 4a The 'no-miracles' argument for realism and inference to the best explanation Argument The no-miracles argument is simple. It just runs: the success of science needs explaining, and the approximate truth of scientific theories is the best explanation for that success. In other words, if scientific theories weren't approximately true – there would be no explanation of their success. The realist claims the anti-realist has to say the success of science is an apparent miracle, or a huge coincidence. The realist takes scientists to be getting genuine explanations from science. That is, they get an explanation that is not just contextual, it is more solid. In spite of the fact that Quine tells us there are indefinitely many explanations that go with the same set of data, the realist thinks the scientist can, and should, still pick the best. This is a method of science called inference to the best explanation. The general idea the realist has is, the best explanation for a phenomenon is likely to be approximately true. Anti-realist response 1 Circular. You're saying realism is the best explanation of the success of science so it is probably true. But you are trying to establish that when scientists get the best explanation of eg planetary motion this likely to be approximately true. So your argument depends on prior acceptance of its conclusion – it is circular, in a question-begging way. Realist trying to establish: Planetary motion needs explaining OHP Phyllis McKay Illari: Material provided for Philosophy of Science Unit Theory T is the best explanation of planetary motion So Theory T is likely to be approximately true Trying to argue: Success of science needs explaining. Approximate truth of theories and existence of entities best explanation. So we can infer that this best explanation is true – theories are approx true and entities exist. Ie circular – same argument form. Anti-realists deny Inference to the Best Explanation gets the truth. Inference to the Best Explanation gets, at best, empirical adequacy. This is not enough to tell you that the empirically adequate explanation you have is approximately true. And realists can't use a circular argument to defend themselves. 2 No best explanation OHP If there is just no such thing as a best explanation at all, the realists’ argument can't even get going. That's in fact what many anti-realists think. There are 2 main reasons. i) Quine shows there are going to be indefinitely many total theories consistent with any set of data. You can't pick out the best one in any uncontroversial way. ii) Many anti-realists believe explanation is nothing more than pragmatic and contextual clearing up of temporary confusion. Explanation doesn’t uncover the fundamental structure of the universe. Instead, it is useful for one purpose or another. And this means that an explanation with respect to one purpose might not be an explanation for another purpose. There is no one best explanation of any fact or happening – it is always relative to a purpose. Phyllis McKay Illari: Material provided for Philosophy of Science Unit These anti-realist responses seriously weaken the realist argument. At the least, they show that the realist no-miracles argument depends on controversial theses about explanation. The realist cannot merely assume these to argue in favour of his thesis. Also the argument is possibly viciously circular. If this all goes throught, then there is no reason to believe Inference to the Best Eexplanation gets truth rather than just empirical adequacy. The most promising realist argument falls. What about the anti-realists? QUESTIONS 4b The pessimistic meta-induction Argument This argument has been very powerful and influential. It is an argument based on the history of what's happened to old theories and theoretical (unobservable) in actual science. The argument roughly runs: Every scientific theory we have had has proven retrospectively to be false. Even highly empirically adequate very useful ones with wide-ranging explanations. Newton, classical theory of electromagnetism, Dalton’s atomic theory, classical physical optics, special theory of relativity and Bohr’s view of the atom. They were all terribly successful in their time, but they are false. So the anti-realist thinks the rational thing to believe is that all scientific theories, however successful, will all prove to be false eventually. But they are useful anyway. So the right thing to think is that their falsity is not important to them being good scientific theories. Scientific theories do not, after all, aim at truth. A similar argument can be run for unobservable entities. In the history of science, we can identify theoretical (unobservable) entities which were posited as parts of now-defunct theories and have now completely vanished. We thought we could observe effects of such entities, but we were just Phyllis McKay Illari: Material provided for Philosophy of Science Unit wrong. Eg phlogiston, electromagnetic ether. They were highly embedded in the science of their day, but we no longer believe they exist. So the antirealist thinks the rational thing to believe is that none of the currently posited unobservable entities are real. They are useful posits, not real things. Just to make it clear, these are both simple inductive arguments. OHP Theories Newton’s theory was successful but false The classical theory of electromagnetism was successful but false The special theory of relativity was successful but false Bohr’s view of the atom was successful but false SO All current theories will one day prove false, however successful they are Entities Phlogiston was central to science in its day, but doesn’t exist The electromagnetic ether was central to science in its day, but doesn’t exist SO All current unobservable entities will one day prove not to exist These arguments have had a pretty powerful effect on thinking. Realist responses Pessimism. Realists accept past failure but insist that the history of science shows progression towards truth. This is why theories are so much more predictively accurate than they used to be. So the pessimistic metainduction is too pessimistic. What we should really expect based on the history of science is either success or to get increasingly close to success in the future. Realists also point out that some unobservables have survived substantial theory change. We are unlikely now ever to claim that such things as oxygen, carbon, mercury and so on atoms do not exist, even if we alter our Phyllis McKay Illari: Material provided for Philosophy of Science Unit precise understanding of what these atoms do exactly, and how they are composed. So although the theory governing them might prove false, alterations will only be slight. There are some entities we will not now decide don’t exist. QUESTIONS What do you think? Think they're equally convincing? I don't. The pessimistic meta-induction about theories is much more convincing than the one about entities. This is mainly because not all theoretical entities are alike in the relevant respects. So now it is worth emphasizing that it actually is possible to be anti-realist about theories, but not about entities. 4c The natural ontological attitude Arthur Fine thinks the whole debate about realism is wrongheaded. He points out that both realists and anti-realists see science as a set of practices in need of an interpretation. But Fine says science doesn’t need interpretation. The aims and goals of science are part of science itself. They are set locally and can only be questioned locally. So you can't ask a global question about all theories, all entities – do they aim at truth? are they real? – only this theory, that entity. And Fine thinks science does that for itself. Fine argues that we should adopt the natural ontological attitude towards science ie we should adopt the attitude towards science that science has to itself. This attitude varies locally. Science makes decisions about when to hold that theories are true, and when to ditch them for other theories, on a case by case basis. In the same way scientists make these decisions for theoretical entities. They get postulated, and they either come to be accepted as really existing, or they don’t. But again this decision is made on a case by case basis. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Fine thinks there is no sense in asking any more questions than science already does. There is no global question above and beyond all these case by case questions which are the daily work of science. QUESTIONS? It is possible that scientists make these case by case decisions about theories. Perhaps they do in some cases. But it is much more convincing that scientists make case by case judgements about whether unobservable postulated entities really exist. The status of unobservable entities varies hugely – some are very tentative, lying just on the outskirts of one theory. If that theory proved false, the entity would vanish. But some unobservable entities are very embedded, used in lots of different theories and explain lots of different phenomena. They couldn’t disappear without a major restructuring of science. For example, electrons are embedded in a lot of different theories and even different sorts of theories. Theories about physical space and matter use them, and so do explanations of chemical bonding. They are detectable with a variety of different observational techniques, not just one. It’s just not the case that electrons are theoretical entities dependent only on one theory, which could vanish with the vanishing of one theory. So anti-realists are right to a certain extent. But science makes these sorts of decisions. Some entities much more tentatively postulated than electrons. Take neutrinos as an example of a not very embedded sub-atomic particle that has a status quite different from that of an electron. We have this idea that there’s a particle with certain properties, but we’re not sure about a lot of them. We don’t know, for example, whether it’s got any rest mass. But we want to investigate because think their existence might explain the missing mass of the universe. But we don’t yet take their existence to have been established, and it is not up to philosophers of science whether neutrinos will be adopted and become embedded in theory or not. That decision is up to scientists. Phyllis McKay Illari: Material provided for Philosophy of Science Unit It is also worth thinking about how scientists use fields of force. They’re ubiquitous and have been around a long time, but they are usually explicitly adopted as a theoretical tool. Scientists themselves take the anti-realist stance to them, taking them to be mere explanatory tools and not a really existent thing. Again Fine seems to have a point here. It would be absurd for a philosopher of science to insist that these things are real when scientists do not. So there are some entities we ought to adopt an anti-realist stance on. The anti-realist is right about that. But Fine’s point is that there is no answer to the global question – are theoretical (unobservable) entities real. There are only specific questions about particular entities. QUESTIONS? 5 Next week Now looked extensively at accounts of method – Bacon and Popper arguing over induction and deduction, and Kuhn's dramatically different paradigmbased approach. We have also looked at aims – law, explanation and truth. For the last topic, we are going back to where we started – demarcation. We can now recognise this as the question whether science has single method, and whether that matters. It will serve to revise and pull together a lot of interesting strands in what we've done. Particularly bring in Kuhn, Quine and realist concerns. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Lecture 3 Popper 1 Last Week We did two things: Bacon’s method: 1 look at data systematically and without presuppositions 2 tabulate and generalise from the data 3 make predictions from your generalisations and see if they are true Problem of induction. We investigated what it is, and saw why some think it is a general problem for science and the philosophy of science. This week, we will study Popper, who thinks that induction has no place in scientific method. 2 Popper’s method 2a Motivation Popper has two important motivations. 1 He is very impressed with the problem of induction. He thinks there is no solution. 2 He is very against the verification method as abused by early psychoanalysts and Marxists. They saw confirmation everywhere and never seriously tested their theories – they certainly never attempted to falsify them. Popper thinks scientific method is not about proving theories true, but about testing to see if they are false. While Popper’s emphasis makes his theory very different from Bacon’s, we will see that they have some important similarities. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 2b The method of conjectures and refutations Popper splits ‘context of discovery’ from the ‘context of justification’. 1 Context of discovery is what happens when scientists are getting to the hypothesis. Popper thinks this is interesting to psychology etc, but is NOT part of the methodology of science. 2 Context of justification is what justifies accepting the hypothesis. Popper’s most radical claim about this is that the context of discovery is irrelevant to any justification. Instead, what justifies accepting a hypothesis is what you do with it after you’ve got it. Very different from Bacon. Bacon concentrates on Stages 1 and 2, which are all about getting to the hypothesis. Popper is trying to exclude that stage from scientific method, and the justification of the hypothesis. Why? Well, because it’s inductive. Popper’s method is new and ingenious. See how he does. Popper claims that science proceeds by a series of bold conjectures, followed by attempts to refute that conjecture. Hence his theory is sometimes known as the method of conjectures and refutations. Stage 1 Conjecture The scientist makes a bold conjecture of a hypothesis. How you get to this conjecture does not matter. It is interesting only to psychology. Popper does make some comments about this stage. He emphasises that the conjecture should be bold, and wants it to be creative. He thinks human subjectivity has a place here – in fact, it is unavoidable. Different from Bacon’s emphasis on being systematic and objective in Stages 1 and 2. Popper thinks we can’t observe (never mind generalise) without a theoretical background. Observation requires knowing What? When? How often? Phyllis McKay Illari: Material provided for Philosophy of Science Unit What can I safely ignore? The answers to these questions are related to the chosen object, task, interest, point of view or problem. Even further, Popper thinks, the descriptive language we must use to describe our observations includes words for properties, and presupposes similarities that are and are not important. This again presupposes object, task etc. Example: Don’t just go to wood, sit and watch/wander round and look. Go to study patterns of different plants and conditions affecting it: Measure its coverage and light levels etc near it and presence of predators that eat it – not other stuff. Ignore movements and eating habits of birds you have reason to believe have nothing to do with the plant. Physicists don’t sit in a lab, waiting for something to move, then pounce and measure every detail. They select something, make it move in a particular way and measure particular elements of that movement. They are unlikely to report that it was blue with green stripes apart from a slightly rusty bit where it got damp in the cupboard when the lab flooded last year. Just to reiterate: Popper accepts human subjectivity here, but this might be because this part is all the context of discovery. It is relevant to psychology, but not relevant to why you are justified in accepting the hypothesis. Stage 2 Refutation This is the crucial stuff. What you do with the hypothesis after you (or someone else) got it is crucial to scientific method. What you do with it is test it. For Popper, this means you set out to try to prove it false. That is, you deduce predictions from it, and see if those predictions are right. Popper wants you to be rigorous in this testing. This involves being inventive, making innovative predictions, and being thorough in checking whether they are correct. Phyllis McKay Illari: Material provided for Philosophy of Science Unit If the hypothesis withstands rigorous testing, it is corroborated. You are justified in accepting it tentatively. It is always subject to future falsification. Bacon does have this stage – Stage 3 – he just doesn’t emphasise it as Popper does, clearly considering Stages 1 and 2 more important. Example Do ‘all swans white’ Not matter how you get to the hypothesis. What is important is that you test it rigorously. If you find a non-white swan, your hypothesis is falsified and you must reject it. If you do not find any non-white swans, you are justified in accepting the hypothesis tentatively. Implications for eg Newton/Einstein: Doesn’t matter where either theory came from. They are bold conjectures, and were probably creative, but how they were got to is just the context of discovery. We are justified in accepting them – tentatively – because they have withstood rigorous testing without being falsified. (In their day – obviously, Newton at least is falsified now.) Summary: Popper thinks the logic of science is deductive NOT inductive. As regards his original motivation, we can see that he really has tried to eliminate induction from scientific method. He also has forced Marxists/ pschoanalysts to face facts, or at least accept that their preferred habits fall short of the method he requires. For Popper, science is not about verifying theories, but about testing them. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 2c Implications There are various problems. Today going to concentrate on one very major one. This is what Popper implies about the status of any scientific hypothesis, law, theory. On Popper’s account, all it has is that we are justified in not rejecting it. It has failed to be falsified, and so it is what Popper calls ‘corroborated’. This does NOT mean that the theory is likely to be true. It is just not false. This implication is a direct result of eliminating induction. Without it, no general theory can get any positive support. Popper’s method implies: no scientific conclusion ever gets any positive support. He cannot wriggle out of this since it is a central point of his method to eliminate induction. He cannot legitimately reintroduce inductive reasoning in any concealed way. Scientific results are constantly put to practical use. This requires that they have positive support. Every time I use a table, bridge, or plane, I positively expect it to function as before. Popper can’t allow that. Yet it is just not true that I merely fail to think the table, bridge or plane won’t work. If that was all science gave me, I might chance using a table, but I wouldn’t go near a bridge, never mind a plane. Note: Popper cannot even allow me to have a high degree of belief that the nice solid stone bridge that has been around for a couple of centuries and seems ok with cars will not collapse when I walk across it. He can allow me to have no positive expectations about the bridge whatsoever. QUESTIONS 3 Comparison of Bacon and Popper with regard to Kepler 3a What Bacon and Popper would expect Bacon: Kepler had no presuppositions, and looked at systematic data. He produced laws from data by fairly simple generalisation allowing for Phyllis McKay Illari: Material provided for Philosophy of Science Unit checking for negative instances. He made predictions from generalisations and checked to see if they were true. Kepler had good reason to suppose his laws true, or at least depend on them to a high degree. Popper: Kepler was selective about which data he used, then bold and creative in producing laws, and chock full of presuppositions the whole time. This stage of Kepler’s work offered no positive support to his laws. Kepler rigorously tested laws and rejected them if falsified. He accepted them tentatively only if they withstood rigorous testing. Kepler had no good reason to suppose his laws true. They were merely unfalsified. 3b How Kepler worked Had two laws. Actually another, but not till later. He worked on these two simultaneously: 1 The orbit of a planet is an ellipse with the sun at one focus. [i.e. it’s an oval with the sun at one pointy end.] 2 The radius vector from the sun to a planet sweeps over equal areas in equal times. [i.e. the planet moves faster at some points on the orbit than other – faster nearer the sun, and slower further away.] Tycho Brahe made his observations 1576-97. These were continuous – he measured daily. This was NOT common. Was the practice to use only measurements of noticeable points on the orbits. This was partly for convenience, partly because a circle is a very simple figure. If the orbit is a circle, you don’t need many points on the orbit to figure it out. If the orbit is more complex – even just an ellipse, you need more points. Brahe also used Phyllis McKay Illari: Material provided for Philosophy of Science Unit fantastic instruments. Had absolutely the best data available. In fact, telescope was invented shortly after, so Brahe probably had the best ever naked-eye measurements. Much more accurate than anything else available to any contemporary astronomer. Their accuracy, and the fact that they were continuous, crucial to Kepler’s success. Were very systematic, and involved no presuppositions of Kepler’s, and rather fewer of Brahe’s than any other data involved presuppositions of the collector. In 1600 after all the observing Kepler became Brahe’s assistant. Brahe allocated him Mars to study. This was a happy accident since the orbit of Mars is the most ‘eccentric’ – this just means it is the orbit most obviously not a circle. [It is the most ‘squashed’ if you think of a circle that has been sat on.] Kepler was a convinced Copernican so thought planets orbited the Sun, not the earth. There was a debate raging, so all contemporary astronomers were going to have an opinion on it. Brahe was geocentric. Kepler wanted a dynamic explanation, so stuck with heliocentric. Although he never achieved it, his desire for a dynamic explanation was very new. Up until then, all desire to study the heavens centred on how the planets moved, not why. This concern of Kepler is incredibly innovative. It is doubtful whether this aspect of his genius was fully recognised by any astronomer coming after him – right up until Newton. It was nevertheless a huge presupposition, crucial to his success. See Orbit of Mars. This shows the path Mars traced between 1580 and 1598, supposing it moved around a stationary earth. Get complicated figure, which never repeats. Every year, the figure will just get more complicated. Hopefully makes clear Kepler would have been unlikely to get to the ellipse working from the assumption that Mars orbited the Earth. Kepler started with one further important assumption; that planetary motion is circular or composed of a small number of circular motions. We will see what happens to this assumption in detail. It gets ditched, but how it happens is very interesting. Phyllis McKay Illari: Material provided for Philosophy of Science Unit So how did Kepler actually proceed? Well, he formulated a series of hypotheses and tried to get them to fit Tycho’s data. (See handout) Hypothesis 1: Orbit of Mars circle round a centre C displaced from sun. Motion uniform with respect to point U. (As this works out, the planet moves faster when nearer the sun.) Have hundreds of pages of draft calculations based on four observed positions of Mars. But when Kepler tested against further observations of Brahe’s it didn’t match. Rejected. Hypothesis 2: Epicycle and deferent. Old Ptolemaic system that can generate lots of different orbits, but applied to heliocentric system instead of geocentric. Best one Kepler got was egg-shaped. Sun at pointy end. Doesn’t fit Tycho’s data. Rejected. Here, the accuracy of Tycho’s data was crucial. Kepler was only getting errors of roughly 6, 7 minutes of arc. No other data available was accurate to more than 10 minutes of arc, so nobody would have castigated him for such small errors. Kepler knew Tycho’s data was better, though, and stuck by it. Hypothesis 3: Mathematically equivalent to an ellipse. Through mathematical error, it didn’t work. Rejected. An ellipse is not a circle, or composed of a small number of circular motions. It just cannot be modelled in that way. [Please just accept as a matter of mathematical fact!] By moving to this hypothesis, Kepler has ditched the circular motion assumption. Hypothesis 4: Ellipse. Worked. Then realised equivalent to H3. Kepler now has his first two laws. Taken him six years. He runs into publication problems, but eventually gets them published after a few years in 1609. Calculates and publishes a new set of astronomical tables, the Rudolphine tables, in 1627. These are immediately taken up for navigation, Phyllis McKay Illari: Material provided for Philosophy of Science Unit and bring him a certain amount of fame. He has managed to succeed Brahe as Imperial Mathematicus on Brahe’s death. [He also takes the opportunity of Brahe’s death to nick the data for all the planetary orbits (since Brahe has only given him Mars) to the fury of Brahe’s heirs. Who to be fair were busy nicking the wonderful equipment that they considered more valuable, denying Kepler access to it and therefore stopping the observations. Kepler knew he was sitting on gold dust and brazenly refused to hand back the data.] Kepler does allright. 3c Study of Bacon and Popper’s theories Presuppositions yes and no. Data: Brahe’s data being continuous reduced the number of presuppositions it contained, and was a huge help. Had no presuppositions of Kepler’s. BUT heliocentric and motion in circles still around. Kepler being willing to ditch them was important, but he didn’t become presuppositionless – he switched them for different presuppositions based on an alternative theory. What Kepler did with data: He worked very closely with the data – fits both Bacon and Popper. Kepler clearly puts together hypotheses and tests them rigorously, rejecting if no good. Fits Popper. But Bacon has this stage too – just less emphasis on it. How Kepler got laws For Bacon this is crucial – Popper thinks doesn’t matter to justification of laws. Clearly wasn’t simple enumeration, even allowing for checking for negative instances. The data was already in tables, but Kepler’s laws were not obvious, and clearly not inevitable. On the contrary, they took six hard years to discover. Especially the second law moves beyond what the data simply show. To get that the planets move faster at certain points on the orbit, you need the assumption that the world works along lines of Phyllis McKay Illari: Material provided for Philosophy of Science Unit mathematical simplicity. Arguably, the data do describe in some nonobvious way the shape of the orbit. But the assertion that the speed of the planets change need a move like – this is the best way to account for or explain the data. This is more than Bacon allows for. Fits Popper’s comments on bold creativity better. But the laws are NOT unrelated to the data. Popper committed to any relation to data being irrelevant, but actually the relation seems important. How Kepler truly arrived at his laws will always be speculative, but you can see that the series of hypotheses moves closer to the final one. There is an obvious link to the data that Kepler is becoming increasing familiar with. There does look to be an inductive stage, just not so simple as Bacon’s. Popper’s claim that it doesn’t matter how the conjecture is related to the data just looks suspect. Status of laws Popper could still be right, so long as the relation of the data to the laws is irrelevant to the justification of the laws. Striking difference between Bacon and Popper on the status of laws. Bacon thinks Kepler had reason to believe them true, Popper that they were merely not falsified. Kepler thinks he has reason to believe they are true. He extends his two laws from Mars to all the planets. It is the practical application of the laws in his navigation tables that really gets him fame. Question: Is Kepler right? If Kepler is right – the laws have inductive support. Popper can’t allow. Bacon can. The circles assumption: Kepler ditched this assumption by hypothesis 3. What happened? Kepler doesn’t test it. He has tried and rejected two hypotheses based on it. He gives up and tries something new. Phyllis McKay Illari: Material provided for Philosophy of Science Unit The ellipse works. He’s not going to go back to the circles assumption. He is probably going to think it false. He has not tested the circles assumption directly. He has just used it to generate specific hypotheses based on it, and tested them. Not clear he can test directly. It’s just an abstract hypothesis, about how to configure possible orbits. Could get indefinitely many specific orbits based on it. But it is an empirical claim – a false one. As it happens, the shape of the planetary orbits can’t be modeled using circles. Kepler doesn’t reject it because he has proven it false. He rejects it because he reckons it’s not doing the right thing, and then because he’s found something that works better. Popper? Wants scientific claims to be directly falsifiable. This is not, but it does look scientific. In a roundabout way Kepler has tested it. In a roundabout way, he has shown that it is false by replacing it. As a matter of historical fact, it was a very important very old assumption in astronomy that Kepler got rid of. Can Popper accept this or not? Summary Popper better on creativity, presuppositions. And looser connection with data. Bacon better on using induction to get the law. And inductive support for law. QUESTIONS 4 Summary and Next topic Popper claims science a series of conjectures and refutations. Divides context of discovery and context of justification, claiming it is only how you Phyllis McKay Illari: Material provided for Philosophy of Science Unit test hypotheses that is relevant to their justification. No induction in method. Looked at how Kepler actually went about formulating and testing hypotheses, and using systematic observational data, the importance of theoretical presuppositions to his work, the status of the resulting theory, and how some theoretical presuppositions are not directly testable. Next week – Duhem and Quine and the problems they create for falsification. For the first time we will look at one of the really serious attacks on objectivity or rationality of science. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Duhem and Quine Lecture 4 1 Last Week Popper 2 motivations: 1 remove induction from scientific method. Unpersuasive. 2 Attack the verificationist method claimed by, for example, early psychoanalysis and Marxism. Better. So Popper’s falsificationist picture is good for something. It gives prediction and testing an important place in scientific method. Today – problems for that picture. 2 Duhem The picture Duhem is attacking is: should test theories. So make predictions by deducing things you can observe from theories. Compare them with what you do observe to see if they are right. Examples: Law: All swans are white Condition: This is a swan Prediction: It is white The prediction is simply deduced from the law plus initial condition. Similarly for: Law: rest/uniform motion unless force Conditions: no net force on this body Prediction: it will stay at rest Law: If a piece of blue litmus paper is put in an acid solution, it will turn red. Condition: This is a piece of blue litmus paper I’m putting in an acid solution. Prediction: It will turn red. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Testing just involves seeing if the predicted observation occurs. While the experimental set-up you need to produce the right conditions might be complicated, the logic of what’s going on looks simple. Popper needs this picture, and so does Bacon. Both think that if an observation made conflicts with the expected predictions of the theory, the theory is falsified and should be rejected. Today we will look at a thoroughgoing attack on this picture of testing in science. First problem. Anyone see it already? The predictions are derived as the conclusions of deductive arguments. If the prediction does not hold true, the conclusion is false. This tells us that one of the premises of the argument must be false. It does not tell us which premise must be false. Actually, if the prediction does not occur, this only tells us either that the law false, OR that the initial condition does not hold. ‘This’ is not a swan, is not under no net force, or is not litmus paper/acid solution. This what Duhem pointed out. You always need more than just the theory alone to derive predictions. So you never test just the theory alone. You are always testing the theory plus the initial condition. You have all done this if you have done any science at school. If your experiment doesn’t work out as expected, what do you do? Suppose you’re measuring gravitational acceleration. You get 9.5ms-2. You’re a little surprised. Do you think Newton was wrong? No. You assume you mucked up. There are many such conditions. It is a pervasive problem. ‘Initial conditions’ are the most obvious. Testing any theory requires the set-up of experiment to be right, and the measuring equipment to be working. See above. But you need more than that. Many experiments involve ‘auxiliary hypotheses’. These are more theoretical, important background assumptions. Sometimes assumptions about measuring equipment can be an Phyllis McKay Illari: Material provided for Philosophy of Science Unit auxiliary hypothesis, because the functioning of the equipment is based on theory. For example, you study the behaviour of cells using a microscope and other optical equipment, but the microscope and optical equipment is based on a theory of light. If that theory is wrong, you might get anomalous observations. You always need the assumption – nothing intervened that we didn’t think of, or didn’t even know existed. Orbit of Uranus was originally calculated assuming there were only 6 planets. Its anomaly was explained by positing the existence of Neptune. The assumption of only 6 planets was dropped. Auxiliary hypotheses/initial conditions can get very theoretical in some areas. Cloud chamber photograph. Just lines on paper to you and me. To a particle physicist it’s a positron doing its stuff. But the theory of the cloud chamber is complicated. The line you see is the track where the particle has been – NOT the particle itself. It is marked out by water vapour in the chamber condensing onto gas particles ionised by the close passage of the charged particle. So even to ‘observe’ sub-atomic particles, the measuring equipment involves so much theory. Even worse, unlike the optical equipment for studying cells, the theory involved in the observing equipment is closely related to the objects being studied. It is just not possible to do particle physics without all this. SO: the result of any experiment only tells you that something is wrong. It might not be the theory you were testing. Implications: when should we give up a theory? When do we decide it has been falsified? Well, this all shows that whatever it is, falsification is not the simple process Popper and Bacon describe. It is much more complicated. This is what Duhem realised. First: a single anomalous result tells you nothing. But even further: A regular anomalous result acknowledged by many scientists who have detected it with different equipment might still not falsify a theory. Can always adjust something else instead of giving up the theory. Eg Neptune and Vulcan both posited to explain anomalies in orbits of Uranus and Mercury respectively. Only Neptune really existed. 1919 Phyllis McKay Illari: Material provided for Philosophy of Science Unit solar eclipse experiment vital (historically) to Einstein. Famously, got telegram from Eddington while teaching, put calmly aside and carried on. Einstein wouldn’t have given up his theory from just one anomalous result. The point is NOT just that these experimental results mean scientists can misbehave and refuse to give up bad theories. The point is it is often genuinely hard to decide when to give up a theory. And human subjectivity and creativity in making such decisions can get very important. Popper has some stuff to say. Talked about some and can talk more. Duhem – further stuff on crucial experiments leave until seminar. QUESTIONS? 3 Quine Develops and expands Duhem’s work. Only mentions Duhem once in his groundbreaking 1951 article. But you will see the link. [Quine has concerns with the philosophy of language. Philosophy of science in 1951 was closely allied with the philosophy of language in a way it isn’t now. Ignore that aspect and concentrate on stuff relevant to the philosophy of science now. This is why Quine talks of ‘statements’. Just take theories and auxiliary hypotheses, as written down, as statements. It is not an important difference to the main points.] Difference between Quine and Duhem: Duhem limited scope of his thesis to physics. Says it doesn’t apply to physiology or chemistry. Thinks maths and logic have a quite different status to science. He explicitly says that we take the axioms of Euclid as established true by common sense knowledge, and attacks non-Euclidean geometry. Quine thinks something quite different. Quine thinks the point applies to any statement whatever. He famously says: ‘The unit of empirical significance is the whole of science.’ Phyllis McKay Illari: Material provided for Philosophy of Science Unit OK, Duhem points out that deriving predictions involves theory plus initial conditions and auxiliary hypotheses and that this may include a lot of theory not directly relevant to the theory you are testing. Basic point of Quine is that any prediction, so any experiment, involves everything. In deriving any prediction, so in doing any test, you involve all of human knowledge, including maths and logic. Any time you test a theory, you test everything. The results of empirical testing affect maths and logic. Picture of human knowledge web. Experience impinges on web only at its edge. OHP/handout In the web is all our scientific theories, all of maths and logic, all our observations. Outside is data – described by simple observation statements, which themselves fall just inside the web. Only way to deduce predictions is from entire web. The web entails observation statements – predictions. Had: OHP A body will remain at rest or continue in uniform motion unless acted on by a force. There is no net force on this body. Add: The entire surroundings of the room aren’t going to move suddenly. My perceptions are working normally, not dreaming or hallucinating etc. I am arguing the right way in constructing this argument and deriving that prediction. And so on.... SO It will remain at rest. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 1 theory plus initial conditions, 2 Duhem adds auxiliary hypotheses and makes point about initial conditions 3 Quine sticks in everything else. Thinks only if everything you know is in there that you can derive prediction. See where the problem is going to be with falsificationism. You do experiment and observation conflicts with your prediction. Duhem says, you don’t know if you’ve falsified theory, initial condition or an auxiliary hypothesis. Quine – potentially, anything you know is falsified. Might not be closely related to the experiment at all. You might think your logic is wrong – you made wrong deduction so got wrong prediction. Every time you compare a theory with any data, you’re comparing all the theory, not just a part of it. Hence ‘the unit of empirical significance is the whole of science.’ Web OHP again Quine does NOT think that everything in the web is in the same position regarding experience. Quine does think you are more likely to alter some statements than others in response to an anomalous observation. Differ in status according to being closer or not to the centre of the web. Closeness is metaphorical. It is meant to model how much any statement is dependent on i.e. interlinked with the rest of the web. Eg statements about what you saw yesterday but didn’t pay much attention to, or current perceptual experience like seeing a moving figure in the distance are close to the edge. If you have an anomalous experience, these are likely to be given up. You just decide you were mistaken, and don’t have to change much else in the web. Assumptions like that your voltmeter is working, or claims like you just watched Red Rum winning the Grand National involve more theory. But they might be given up. Phyllis McKay Illari: Material provided for Philosophy of Science Unit More theory involved in saying that you are seeing a positron moving across a cloud chamber. Maths and logic are at the very heart of the web. If you change your mind about these, you are restructuring how you reason and manipulate relations between observations, theory and observations, and theory and theory. This would involve re-writing a lot of the web. So if you get an anomalous observation, what happens? You need to make the web consistent again. Most likely to alter stuff close to outside. You were wrong about what you saw, or your voltmeter is faulty. This is the least trouble. But you could challenge the theory that V=IR if you want. This would also make the web consistent again. Suppose you believe all the chairs in the room are blue, and always have been. You come in one morning and see one is red. You are most likely to think someone brought a new chair in, or you never noticed it before. But you could form the belief that chairs can change colour, or that you were hallucinating before if you want. Either move makes all your beliefs consistent. This is still link to the prediction as the concusion of a deductive argument. The prediction is still the conclusion, and if it is false, one of premises must be false. But now the premises include all of scientific knowledge. The argument doesn’t tell you what to alter. Must have other reasons for what you choose to alter. Simplicity, consistency, resulting theory is easy to understand, gives clear good explanations, or provides a good and easily manipulated conceptual framework. Quine agrees that we are very unlikely to alter logic/maths in response to anomalies. But difference between that and deciding you were mistaken about all the chairs in the room being blue is NOT OF KIND. It is a matter of degree and anything is in principle assailable. Actual examples. Logical law of the excluded middle: P or ¬P. Just says any statement is either T or F. There is no statement that is neither T nor F. Law around Phyllis McKay Illari: Material provided for Philosophy of Science Unit since Aristotle (about 300BC) and taken to be a priori true. But it has been suggested that we ditch it in response to quantum non-locality, so that we can say it is neither T nor F that a particular particle has this precise momentum and this precise position at a particular time. People do work on 3-place logic, and intuitionistic logic still refuses to take the law of the excluded middle as an axiom. Geometry: Euclid’s fifth postulate. For any straight line, there is one and only one line parallel to it drawn through a particular point. Now nonEuclidean geometry is used to describe curved space-time. Reimannian geometry takes there to be more than one line parallel to a given line drawable through one point. So we have altered, or at least considered altering statements at the very centre of the web. We can’t reject Quine’s claims out of hand. If Quine is right, the decision about what part of the theory – the web – to alter is not, and cannot be, a matter of logic. It is a human decision and must be made for other reasons. DISCUSSION Plausible? Outrageous? What sorts of reasons? 4 Implications Quine’s thesis doesn’t just affect how we see falsification. It affects verification too. It has strong implications for how we see the relation between scientific theory and data at all. You can pick anything to alter, or anything to hold unalterable. If you really wanted to, you could challenge deductive logic to make web consistent with simple perceptual anomaly. Or you could maintain your belief in the fact Phyllis McKay Illari: Material provided for Philosophy of Science Unit that all chairs in the room are blue in spite of all sorts of evidence, just by making enough changes elsewhere in the web. True? This hard stuff. Got to think about. I’ll give you some pointers. Quine’s claims that in principle you might alter anything, or hold anything fixed, are true. But the claim that any time you do science, you involve the whole web shakier. Yes, potentially the whole web might get involved if something unusual happened, but this not the usual situation. So in practice you can delimit a section you are testing. You do have to accept that your reasons for picking out these sections are not a matter of deductive logic. Does this mean they are not objective or rational? Rationality. Best not to say that science is not rational. Best to say that it turns out that the rationality of science is not a matter of deductive logic. CHAT One final important implication of Quine’s work. In principle, more than one total web implies exactly the data we have. Draw. Eg our web plus ‘zoogons exist’. Zoogons have no possible observational implications, so adding that claim to our web gets a different web with exactly the same observational consequences. Or you could get more creative and re-write the whole web with a different logic. In principle, you could generate one which has exactly the same observational consequences as ours. Deductive logic does not tell you which web is better. Phyllis McKay Illari: Material provided for Philosophy of Science Unit This is known as the underdetermination of theory by data. Given a set of data, there are indefinitely many contrary theories, each of which logically entails the evidence. So how do you choose? There must be a rational way, or science is not rational. Can’t do it deductively. Can you do it inductively, working from data back up to the theory? Will this pick out one unique web, or will the same problem recur? Summary Duhem – not test theory alone Quine test everything along with any theory and everything up for grabs when get any anomalous observation. Implications – underdetermination. And rationality in science can’t just be deductive logic. Any way out. Remember Kepler assumption about circles without falsifying because replaced with something better. Perihelion advance of Mercury explained by Einstein. One major objection to astrology being a science. Suggest anything? Neither Popper nor Bacon talk about theory competition. Will see if helps. Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Lecture 2 Overview 1 Last week 2 Bacon a Theory b Examples 3 Problem of Induction a What induction is b Problem of Induction c Responses to the problem 4 Summary Bacon’s theory Stage 1 Remove presuppositions. No speculation before data. Be as a ‘child before nature’. Systematically collect data about world – observe, experiment etc. Stage 2 Put data in tables and study. (Still systematic) Generalise from that data. Stage 3 From generalisation, make prediction about something not done in Stage 1. Go see if it’s true. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Bacon’s heat example Stage 1: Gather and describe examples of heat List: fire, sunlight, bodies rubbed violently, boiling liquid, compost and horse dung, the effects on us of fortified spirits of wine and aromatic herbs. AND gather and describe cases importantly similar but not heat List: moonlight, starlight, non-boiling liquids, will-o’-the-wisp, unheated air etc. Include different degrees of heat. Stage 2: Work out essential nature of heat Compare lists: find thing present in all positive cases, absent in all the negative cases, and varies in appropriate degrees in the different positive cases. Bacon’s result Heat is a ‘motion’ which acts on the ‘smaller particles’ of bodies – so you will produce heat if you ‘excite a dilating or expanding motion’ in any natural body. How Bacon sees his method: Bacon says the scientist gathers observations like ‘countless grapes … ripe and fully seasoned’. Then he gets laws or theories like getting the juice for the wine squeezed from the grapes. Bacon writes of this method that it: ‘shall analyse experience and take it to pieces, and by a due process of exclusion and rejection lead to an inevitable conclusion’. [My emphasis] (p249) Deduction: Set of statements such that if the premises are true, the conclusion must also be true. All men are mortal. Socrates is a man. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Therefore Socrates is mortal. All material things close to the surface of the earth accelerate towards it at approximately 9.8ms-2. This is a material thing close to the surface of the earth. Therefore, if I let it go, it will accelerate towards the earth at approximately 9.8ms-2. Induction This man is mortal. This man is mortal. This man is mortal. ….. The next man will be mortal. OR All men are mortal. When I dropped the pencil the first time, it fell to the ground. When I dropped the pencil the second time, it fell to the ground. When I dropped the pencil the third time, it fell to the ground. … The next time I drop the pencil, it will fall to the ground. OR Every time I drop the pencil, it will fall to the ground. Making the distinction Induction Deduction 1 Moves from the particular to the general Moves from the general to the particular Eg from a claim about ‘this man’, Socrates, to a claim about all men Eg from a claim about ‘all men’ to a claim about this man, Socrates. 2 Tells you something that is not contained in the premises – sometimes Tells you nothing that is not contained in the premises Phyllis McKay Illari: Material provided for Philosophy of Science Unit called ‘ampliative’ 3 The conclusion is NOT guaranteed by the premises. That is, the conclusion can be false, while the premises are all true. The conclusion is guaranteed by the premises. That is, if the conclusion is true, the premises must be true. When Europeans first traveled to The only way the conclusion can be Australia, there they were astonished to false is if one of the premises is false. discover for the first time black swans. EITHER Bess is not an elephant, OR Many observed white swans is good some elephants don’t come from evidence for the claim that all swans Africa. are white, but it is still defeasible evidence. And if the premises were true, the conclusion would be true. The sun rose 1st September The sun rose 2nd September The sun rose 3rd September .... and so on SO The sun will rise tomorrow We believe this, but what justifies that belief? Inductive argument for induction Induction has worked about cooking times for pasta Induction has worked about the sun rising every morning Phyllis McKay Illari: Material provided for Philosophy of Science Unit Induction has worked about bread always being nourishing. ...and so on SO induction will work the next time I use it OR induction will always work Russell’s Principle: (a) the greater the number of cases in which a thing of the sort A has been found associated with a thing of the sort B, the more probable it is (if no cases of failure of association are known) that A is always associated with B. (b) Under the same circumstances, a sufficient number of cases of the association of A with B will make it nearly certain that A is always associated with B, and will make this general law approach certainty without limit. Problems for Bacon’s theory Many scientific successes not mechanical – highly imaginative. Laws beyond what data could give by simple generalisation (eg Kepler’s second) Science has big jumps – not simple progression. Eg Copernicus to Kepler, Kepler to Newton, Newton to Einstein. Bacon’s method might only get general descriptions. Science wants explanation. Eg Kepler/Newton. Kepler wants to explain why planets move; Newton manages it. Presuppositions can be crucial to science. Kepler’s Copernican presuppositions, because he wants a dynamic explanation. Bacon’s Idols 1 Tribe Foundation in human nature: Phyllis McKay Illari: Material provided for Philosophy of Science Unit Postulate more regularity than really see Generalise hastily Overemphasise value of confirming instances 2 Cave From an individual’s upbringing and education 3 Market-place Meanings of words distorted by common use, which impedes concept-formation in science 4 Theatre Received dogmas and methods of various philosophies Counterinductive argument for counterinduction Counterinduction hasn’t worked for the sun rising Counterinduction hasn’t worked for bread nourishing Counterinduction hasn’t worked for cooking times for pasta ... SO counterinduction will work the next time Phyllis McKay Illari: Material provided for Philosophy of Science Unit 

Topic 1 Lecture notes Introduction to course We’re going to be studying the philosophy of science. This is important and exciting, since science might be called the new religion. It gives explanations previously unavailable, or the realm only of myth or religion. Now the discoveries of science are often treated with the same reverence previously reserved for religious objects. The study of philosophy of science worthwhile because science throws up problems that science can’t answer. They are a priori questions, calling for the examination of concepts. They won’t be answered by more investigation of the world. Philosophy of science involves critical evaluation of these problems. We will do 10 topics. Structure clear from course handout. Involves: What does and doesn’t count as science and why, which we will start now and return to in topic 10. We will also study the method of science, whether observational evidence is used support or falsify laws, and look at theories that say scientific method is more complicated than that, that social and historical aspects are crucial to scientific method. We will examine problems with these. We will also be interested in the aims of science – law and explanation, and what this amounts to, and whether science aims at the truth or is merely instrumental. All these concepts are used or assumed by scientists, often without careful thought. We are going to think about them. Being critical means we think about them carefully and try to understand them. We will also look at examples from the history of science, which will keep us from making wildly implausible claims about science. The point of a course in philosophy could never be to tell you what to think. I will tell you what major figures thought, and what is wrong with it, I will identify important issues, and guide you on how to approach these issues in seminars. Fundamentally, the rest is up to you. What is science? Looking in rest of lecture at science and non-science. So: Phyllis McKay Illari: Material provided for Philosophy of Science Unit Think about examples of sciences: Natural sciences Social sciences physics chemistry biology and their border areas sociology political science anthropology economics historiography and related disciplines Psychology a difficult one. Sometimes put in one category, sometimes in the other. Generally the boundary is quite blurred. Science is more than just physics, and more than just physics, chemistry and biology. Philosophers of science can go badly wrong if they forget that. Think about what are not sciences, and why we might want to distinguish science from them. Not sciences: maths and logic psuedosciences eg history or literary studies Reasons for distinguishing: Understanding something about method of science, or their methods. MAIN REASON: Deny non-sciences the authority of science. Think about what scientists do. Explore do things not done before. Eg probes into parts of space, or ocean. More detail than before. Eg human genome project, microscopic study of cells in body, or use of complex machines subatomic particles. Double-check own or others’ work. Huge variety of methods: Experiments. Intervene and see what’s happened. eg lengths of copper rods heated different temperatures. Systematically observe Eg position of the planets, or the coverage of a particular plant naturally occurring in a wood over the course of a year. Interviews or surveys Eg asking householders noise pollution/daily diet Phyllis McKay Illari: Material provided for Philosophy of Science Unit Clinical trials. Particular sort rigidly controlled expt. Need to get license to produce drug UK. Examination of documents, inscriptions, coins, archaeological relics and so on Not usually thought of but anthropology, and historiography use them. Describe results of their/other people’s exploration Variety of forms tables, mathematical equations, physical models papers, books etc. Explain again either own or others’ exploration and description Newton explains movements of planets by positing gravitational force attracting every massive body in the universe to every other massive body. Doctors explain why some patients die while others recover by citing differences in the general health, fitness, immune systems of patients before infection. Predict Not all, but many do. Work from descriptions, explanations. Predict what will happen when new things done. Eg ‘perturbation’ orbit Uranus not what Newton predicted. Suggested new planet particular mass, particular place. Neptune’s mass and position described before it was ever actually observed. Einstein v famous. Predicted that light from stars would bend as it passed sun, so stars appear further apart when viewed from earth. Famous solar eclipse experiments of 1919 showed this is true. Crucial to Einstein gaining ground. Scientists also spend a lot of time learning what other scientists have already done. The foundation of all their descriptions, explanations and predictions is experience. Theories are only supported by empirical evidence. Is astrology a science? What is astrology? View that stars and planets affect your current emotions and actions. Study of them can help predict your future actions and emotions. The positions of the stars and planets at your birth has an ongoing influence on your character and prospects of success. Example Cancer. ‘Sometimes whatever you say it seems to be the wrong thing, as the past few weeks have proved. But with Mercury in a brilliant link to both Saturn and Pluto next week, you’ve now got a good connection. Perhaps someone is more open to hearing what you’ve got to say, or this time is right. Either way you find people who were being defensive are now putty in your hands.’ Phyllis McKay Illari: Material provided for Philosophy of Science Unit Claim is: ‘Mercury is in a link to both Saturn and Pluto so people who were being defensive are now putty in your hands.’ Astrologers think can get better predictions with more detailed information about a particular person and their place, date and time of birth. Is astrology a science? Why not? False But many scientific theories are false. Objects inappropriate to scientific study. But psychology, and parts of evolutionary biology, studies the emotions and actions of people. Predictions inaccurate. Yes, v weak. But other studies like sociology etc. also inaccurate, especially in their infancy. Astrologers could say astrology complex in same way, unlike physics etc. Origins of medicine and chemistry are very imprecise, and the predictions of different doctors, even specialists, still vary. Astrology has magical origins. Medicine and chemistry came out of same world view. Magic not Wizard magic – idea that similarities now regarded as superficial had important links – such as the idea that Mars being in an important position at your birth means you will be a courageous bold, go-getting person since Mars is red and angry-looking. Medicine has similar roots, like prescribing something yellow like turmeric for jaundice, or powdered rhinoceros horn for impotence. Chemistry comes out of alchemy – turning lead into gold. No central body of theory all astrologers agree on But look at sociological studies, and periods in the history of science like medicine. Astrologers won’t give up their theories. True, but so do some scientists, like Newtonian tradition. Anomaly in orbit of Uranus led to discovery of Neptune. But the anomaly in orbit of Mercury wasn’t explained until Einstein. Newton’s theory was not given up until then. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Astrology inconsistent with good theories. Current physical theory says the planets never mind the stars are too far away and the wrong sort of thing to have a causal effect on us. But especially in infancy, many genuinely scientific theories look inconsistent with contemporary theories. Eg Einstein, and quantum theory. Contemporary scientists were very resistant. Competing theories are better. Have better theories of human behaviour ie psychology, cog sci, evolutionary theory Why better? More accurate predictions, more consensus and collaboration, more concern with the empirical accuracy of their theories, etc. The theories are progressive, generating novel predictions, new ideas, etc. This will imply: Astrology was once a science, when all these things were true of it. Up until after Kepler astronomy/astrology were combined. Astronomy was very prestigious and highly funded, and not just because it was important to navigation and a solid calendar – the reasons most often given today. Much of its funding was because of the prestige of astrology. More generally, what counts as a science will change over time. Note: NOT according to what people believe is a science at the time, but according to other factors above, like the existence of competing theories. How does science work? Is there a distinctive method for science? Summary Scientists explore, describe, explain, predict. This involves a huge variety of actual activities. Foundation of all this is experience and evidence. It is difficult to give a neat answer to why astrology is not a science. Some quite obvious answers would eliminate genuine sciences too. Phyllis McKay Illari: Material provided for Philosophy of Science Unit Suggestion: the answer must be in here somewhere, it is just not a simple one. It must have something to do with science’s foundation in experience and evidence, in some more complex way. Phyllis McKay Illari: Material provided for Philosophy of Science Unit