Experiencing the a priori

Experiencing the a priori Denis Seron University of Liège (BE) European Journal of Philosophy, 2021. Penultimate version. lease cite the published version. ABSTRACT: Brentano clearly asserts, in his Vienna lectures of 1887-8, that his descriptive psychology is an a priori or ‘exact’ science. Since he rejects Kant’s idea of a synthetic a priori, this means that the descriptive psychologist’s laws are analytic. My aim in this paper is to clarify and discuss this view. I examine Brentano’s epistemology in the Psychology from an Empirical Standpoint and then its later developments. I conclude with a difficulty inherent in Brentano’s psychological approach to a priori knowledge. _______________________________ According to a common objection, empiricism is self-refuting since it both requires and excludes appeal to a priori intuitions. An influential version of this objection is George Bealer’s ‘starting points argument’ (Bealer, 1992, pp. 104–108). Similar arguments are found, among others, in (Bealer, 1998; Chudnoff, 2013, p. 14 ff.). It runs as follows: [1] The fundamental principle of empiricism is that only experiences or observations can be invoked as primary evidence. [2] This principle obviously requires one to acknowledge certain ‘basic epistemic classifications’ — ‘starting points’ in Bealer’s terminology. For example, the empiricist needs to know what does and does not count as an experience or observation. [3] Starting points are not found in experience or observation. They are given (let us say) in ‘a priori intuitions’ — meaning by this ‘purely intellectual’ (as opposed to sensory) conscious episodes in which something seems necessarily true to us (Bealer, 1992, p. 102). It follows from [2] and [3] that the empiricist must invoke intuitions regarding starting points. For example, she needs to use intuitions in order to determine what does and does not count as an experience or observation. Since intuitions are not experiences or observations, the empiricist’s use of such intuitions ‘contradicts the principle of empiricism’. One possible reply to this objection is to say that the empiricist actually uses intuitions not as primary evidence, but just as a guide for scientific and philosophical inquiry. But even in this case, Bealer argues, empiricism is faced with insoluble difficulties. Consider the following dilemma: [4] The empiricist’s starting point intuitions are either reliable or not. [5] If they are not reliable, then any comprehensive theory that results from them must be highly (that is, incorrigibly) unreliable. [6] Now, suppose that intuitions about starting points are reliable. If they are, then our concrete-case intuitions that intuitions are primary evidence are reliable. Hence intuitions are primary evidence and empiricism is false. But if empiricism is false, any comprehensive theory based upon it is highly unreliable. Bealer’s conclusion is this: ‘Therefore, on both prongs of the dilemma, empiricism leads one to formulate a comprehensive theory that is highly unreliable. But, given that we can now see this, we certainly would not be justified in accepting this comprehensive theory. However, empiricism implies that we would. So empiricism is false’ (Bealer, 1992, p. 107). The German-Austrian philosopher Franz Brentano defended both a strong form of empiricism and the view that the philosopher’s primary task was to uncover a priori laws governing mental life. A particularly interesting tenet of his thought is that he challenges proposition [3] and claims that starting points intuitions are experiences. If starting points intuitions are experiences, the view that they are primary evidence does not contradict empiricism’s principle that only experiences are primary evidence. Hence, the appeal to starting point intuitions does not make empiricism incoherent; the intuition that they are primary evidence is both reliable and legitimately useable as primary evidence. Obviously, the main difficulty facing this tenet is to work out how experience — in obvious contradiction to the traditional view — could give us epistemic access to a priori truths. Roughly, Brentano’s basic idea is that thinking is a form of inner experience. Thinking the concept C means experiencing a mental phenomenon, namely a thinking, along with a ‘content’ intentionally included in it, namely C. For example, we certainly do not discover that all red things are colored by observing red things. In this sense, ‘all that is red is colored’ is certainly not an empirical, but an a priori truth (Brentano, 1970, p. 49). Such knowledge ‘does not rest upon perception, namely upon knowledge that there is something that falls under the concepts in question’ (Brentano, 1970, p. 151). Nevertheless, Brentano continues further on, ‘it is not psychologically independent of the thinking of these concepts’. It derives from experience insofar as it requires us to ‘have these concepts and innerly perceive that we have them’ (Brentano, 1970, p. 151). Since concepts are experienced, they can be observed and decomposed in parts, that is, analyzed through analytic judgments such as ‘all that is red is colored’. Conceptual analysis means to analyze concepts qua inner experiences and hence pertains to empirical psychology. In this sense, all sciences, including mathematics, ‘are empirical and, as Kant would say, a posteriori’ (Brentano, 1970, p. 49). Starting points or ‘basic epistemic classifications’ are exactly the sort of things Brentano’s descriptive psychology is about. Descriptive psychology’s primary task consists in classifying mental states into general types — ‘judgment’, ‘will’, ‘observation’, etc. — that are subject to a priori knowledge. For this reason, I focus here on Brentano’s account of psychological a priori knowledge, although much of what I say applies to mathematical knowledge as well. Brentano clearly asserts, in his Vienna lectures of 1887-8, that descriptive psychology is an a priori or ‘exact’ science. Since he rejects Kant’s idea of a synthetic a priori, this means that the descriptive psychologist’s laws are analytic. My aim in this paper is to clarify and discuss this view. I examine Brentano’s epistemology in the Psychology from an Empirical Standpoint and then its later developments. I conclude with a difficulty inherent in Brentano’s psychological approach to a priori knowledge. Historical overview Surprisingly, Brentano’s 1874 masterpiece, Psychology from an Empirical Standpoint, is very elusive on the topic of a priori knowledge. At first glance, it may even seem that Brentano here undertakes to challenge the view of a mental life being a priori knowable at all. In direct opposition to Herbart, Fechner, and Wundt, he declares in the Psychology that psychology is not an exact science. The reason he gives for this claim is that psychology is an empirical or ‘purely phenomenal’ science, namely a science committed to inner experience as its only source of knowledge (Brentano, 1924, p. 94/65). Page references are first to the German text and second to the English translation (if there is one), separated by a slash. On Brentano’s view, this entails that psychology, like the natural sciences, progresses mainly through the method of induction. He conceives of the inductive method as follows (Brentano, 1924, p. 102 ff./70 ff.): First, the psychologist obtains ‘very general laws’ (Gesetze von einer sehr umfassenden Allgemeinheit) by using what Brentano terms ‘mental induction’ (psychische Induktion); from these laws she then deductively infers more special laws, which she finally seeks to establish through direct induction (direkte Induktion). Interestingly enough, Brentano also proposes for psychology a method alternative to induction, namely the ‘historical’ or ‘inverse deductive method’ elaborated by John Stuart Mill in his System of Logic (Brentano, 1924, pp. 104–5/72). The historical method consists in ‘empirical generalization’ (empirische Generalisation), that is, in constructing general hypotheses that best explain individual observations. It thus seems that psychology, as defined in 1874, is not an exact science. It is true, however, that Brentano considers an objection to this view. Wundt, for example, claimed that psychology, although mostly inductive, was exact in some special cases, namely in the case of sensory intensity (Brentano, 1924, p. 94 ff./65 ff.). If so, the epistemological status of psychology is quite similar to that of linguistics. Everybody at that time agreed that linguistics, like psychology, was an empirical science. Nonetheless, neo-Grammarians maintained that in some special cases, namely in the case of phonetic changes, the linguist is entitled to enunciate a priori — ‘exceptionless’ — laws. On this controversy, see (Paul, 1880; Brugmann, 1904, p. 27 ff.) and the epistemological assessment in (Leroux, 2005). But Brentano explicitly rejects this line of thought in the Psychology as well as in later writings (Seron, 2012; Antonelli, 2015, p. 49 ff.). He maintains that the measurement of mental intensity by means of Fechner’s logarithmic law, by contrast with the measurement of physical magnitudes, is bound to be imprecise. He agrees with Fechner that there is a functional correlation between mental and physical intensity. But he also holds that, in the present state of psychological science, the measurement of mental intensity must be imprecise due to the extreme complexity of the physiological processes and of the role played by mental factors such as attention and memory. As Denis Fisette (2014; 2018) has showed, this account in terms of complexity echoes Auguste Comte’s classification of sciences. Exact sciences are at the bottom of Comte’s hierarchy of sciences: they are the simplest and most basic sciences. By contrast, sociology (or psychology in Brentano’s classification) is the highest and most complex science. Brentano, as I said, claims in his later work that psychology is an exact science. The clearest statements on this issue are found in the 1887-8 Vienna lecture on descriptive psychology (Brentano, 1982) and in the posthumous manuscript published in 1925 by Alfred Kastil with the title Versuch über die Erkenntnis (Brentano, 1970). However, it would be a mistake to conclude that Brentano changed his mind after 1874. This position, as I see it, is fully consistent with the assumption Brentano made in the Psychology, that psychology is not exact. Brentano’s actual claim in the Vienna lecture is that not all psychology, but only ‘psychognosy’ or ‘pure’ or ‘descriptive psychology’ (as opposed to genetic or physiological psychology) is an exact science (Brentano, 1982, p. 1/3). By contrast, genetic psychology ‘will presumably have to renounce forever any claim to exactness’ (Brentano, 1982, p. 1/3). To a certain extent, this is what Brentano already said in 1874. The laws the Psychology claims are inexact are ‘the highest laws of mental succession’ (Brentano, 1924, p. 102/70; cf. p. 94/65), that is, the laws of genetic psychology. It follows from this that, even though the exactness view is not explicitly held in the Psychology, it is at least compatible with, and at most implicitly presupposed by, Brentano’s earlier epistemology of psychology. In what sense is psychognosy an exact science? This question is explicitly raised and answered in the Vienna lecture. The propositions of exact sciences, Brentano explains, are ‘sharp’ (scharf) and ‘precise’ (genau, präzis) (Brentano, 1982, pp. 3–4/5–6), while those of inexact sciences are ‘undetermined and vague formulae’ (Brentano, 1982, p. 3/5). For example, the meteorologist’s propositions are said to be true ‘often’ or ‘mostly’ or ‘on average’. By contrast, it is true not only in most cases, but in all possible cases that the angle sum of a triangle is two right angles, or that a body preserves its velocity and direction so long as no force in its motion’s direction acts on it. In Brentano’s view, there is in this respect no great epistemological difference between the psychognost’s laws and Thales’s theorem. For example, the psychological law in virtue of which purple is between red and blue is not true ‘on average’, but in all possible cases. It is an exact law just as is Thales’s theorem. Conceptual intuition The psychognost’s laws are about a priori relations of inseparability and incompatibility. As Brentano puts it, they are about the ‘necessity or impossibility of a unification of certain elements’ (Brentano, 1982, p. 28/31). Since in his view necessity derives from impossibility, this means that a priori knowledge is knowledge of the impossibility for a concept to be included in, or separated from, another concept. Mental life presents elements; some associations and exclusions of elements are possible and some are not. Most importantly, the kind of impossibility Brentano has in view is purely conceptual. A priori incompatibility ‘becomes clear from the concepts themselves’ (leuchtet unmittelbar ein, erhellt aus den Begriffen selbst). In other words: the psychognost ‘intuitively grasps (intuitiv erfassen) general laws’ in the concepts she uses (Brentano, 1982, pp. 28/31 and 73/75). Brentano gives some examples of this in the course on descriptive psychology: the psychognost states, on the basis of conceptual intuition, that evidence is impossible anywhere outside the domain of judgments, or that a point can exist only within a spatial continuum, etc. (Brentano, 1982, p. 73/75). This should apply as well to mathematical and logical axioms. Of course, this view is fraught with major difficulties, one of which being the much debated problem of how to reintroduce modalities within the context of a strong empiricism. Another difficulty lies in Brentano’s causal construal of inference. In his lectures on practical philosophy, he views the relations of inference as necessary causal relations (Brentano, 1952, p. 283/175). Mental causality is subject to a priori laws, for example laws of inference. This view is highly problematic given Brentano’s claim that genetic psychology is not an exact science. It is commonly assumed that, for Brentano, mental causality is dealt with not in descriptive, but in genetic psychology. One possible way out from this difficulty is to say that genetic psychology deals only with psychophysical causation, while mental causation, including inference, is studied in descriptive psychology. This interpretation is not fully satisfactory, although it finds clear support in Brentano’s use of the term ‘physiological psychology’ as a synonym of ‘genetic psychology’ (Brentano, 1982, p. 1/3). Brentano’s argument on this score combines a psychological and a mereological approach to conceptual thought. It can be reconstructed as follows: A concept is something that occurs in the mind, namely the content of a thought. On Brentano’s view, all that occurs in the mind must be phenomenally conscious. Thinking a concept C means ‘innerly perceiving’ oneself thinking C. This entails that concepts are somehow objects of experiences. Now, consider an a priori judgment, for example a definition of the form ‘C = uvw’. You certainly can think C without (explicitly) thinking its definitional features uvw. However, Brentano claims that thinking C is necessarily equivalent to thinking uvw. To think C without explicitly thinking uvw means no more than to think uvw confusedly. The reason for this is that, on Brentano’s view, a complex (i.e., definable) concept is but a collection of simple concepts: if you do not think uvw, it cannot be the case that you think C. Since thinking a concept, as we have seen, is a variety of experience, it follows from this that your inner experience of your thinking C is identical with your inner experience of your thinking uvw. For example, when you think the concept of a triangle, you intuitively ‘know’ that it is a priori impossible for a four-sided figure to be a triangle. You know it just as you know (through inner perception) that you think the concept of a triangle. If you were to think of something with four sides, you would immediately know that you think another concept than that of a triangle. Like any other datum of experience, the confused thought of C can be made clear by distinguishing its parts and comparing it with other thoughts, that is, through psychological analysis. In the case at stake, the result of the analysis is expressed by the definition ‘C = uvw’. This definition is taken to be necessary in the sense that, if you know that you think C, then you must know that you think uvw. As Brentano explains: ‘Immediate a priori knowledge (…) does not rest upon perception or any knowledge of the fact that there is something that falls under the corresponding concepts. For example, an a priori proposition about the triangle is completely independent of there being triangles. However, such propositions are not psychologically independent of the thinking of these concepts. They become clear (leuchten ein) from the consideration of the concepts. This means that we have these concepts and must innerly perceive that we have them’ (Brentano, 1970, p. 151). For example, it is self-evident to me that a geometrical point can exist only within a spatial continuum, or that a point is a boundary of a continuum. Of course I do not discover this by experience: the impossibility for a point to exist outside any spatial continuum is purely conceptual. Nonetheless, the concept of a point is empirically given. When experiencing a concept you ‘intuitively grasp’ the a priori relations of inseparability and incompatibility that are constitutive of this concept. Clearly, the intuition Brentano appeals to is not an ‘intuition of essences’ that somehow enables us to ‘see’ necessary properties and relations in things. Brentano strongly emphasizes the fact that his view of a priori knowledge is fully consistent with his critique of ‘a priori evidence’ (apriorische Evidenz) (Brentano, 1982, p. 74/76; cf. Brentano, 1970, pp. 26 and 40). All that is required here is thinking, that is, experiencing concepts. See also, on arithmetic, (Brentano, 1970, p. 67): ‘As we construed arithmetic as a purely analytic science, we were careful thereby not to view it as an a priori science in the sense of resting upon a priori intuitions or concepts. We affirm instead that in arithmetic, too, all the conceptual elements are taken from perception’. This issue is exactly what is at stake in Anton Marty’s criticism of Husserlian ideation in his 1908 Untersuchungen (Marty, 1908, p. 53 ff.; Seron, 2017). To Husserl’s ideation Marty opposes what he terms ‘analytic intuition’ (analytische Einsicht). Marty agrees with Husserl that logical laws express not empirical, but a priori truths, and that these a priori truths are somehow given in a priori intuitions. For example, we intuitively know that — this is how the psychologist Marty understands the principle of excluded middle — it is impossible for a judgment to be neither affirmative nor negative. Such analytic intuitions are comparable to the ones appealed to in the theory of color. We know a priori that it is impossible for a purple color not to be between red and blue. Logic, like the theory of color, is an a priori science that is grounded in analytic intuitions. The main difference between Marty and Husserl on this point is about the nature of logical intuition. For Marty, the fact that logic is an a priori science does not entail that the logical a priori is not psychological. Against Husserl, he affirms that being an ideal science does not involve having ideal objects, namely ideal meanings or ‘propositions in themselves’ considered independently of their mental realization. The objects of a priori intuition, Marty argues, are not a priori objects, but objects of experience. The objects of logical truths are mental acts given in inner experience. The view that logical truths ‘become clear from concepts’ given in inner experience does not entail that they are (as J.S. Mill claimed) inferred from experience. Since concepts are presentations and not judgments, they cannot be used as premises for an inference. See (Brentano, 2010, 13.362–5). That is why Marty, unlike Husserl in the first edition of the Logical Investigations, holds that the logician and the descriptive psychologist are one and the same person. A difficulty The most important feature of Brentano’s account of a priori knowledge is that it is entirely psychological. As such, however, this account faces serious difficulties. I now want to point out one of its weaknesses, which I think is decisive. Another target of Brentano’s criticism, besides concept realism, is Kant. First, Brentano claims that all a priori judgments are analytic in the sense of analyzing a concept into its constituents. This implies a strong criticism of Kant’s theory of synthetic a priori. See (Albertazzi, 1996, pp. 447–448). For a defense of synthetic a priori knowledge against Brentano’s critique, see (Körner, 1987). Most interestingly, Brentano’s notion of analyticity is not restricted to sciences like logic and mathematics and should be understood in the widest possible sense. In order to criticize Kant’s view that analytic judgments do not expand knowledge, Brentano presents as a counter-example Helmholtz’s theory of harmonics (Brentano, 1970, p. 10). The decomposition of a harmonic sound into simple vibratory phenomena whose frequencies follow a mathematical law is a form of conceptual analysis. By doing so, Helmholtz just ‘clarified (verdeutlicht) the concept of a sound with respect to a range of intrinsic features without which it would not be the same concept’ (Brentano, 1970, p. 10). Put otherwise: Helmholtz’s law is an analytic judgment, that is, a judgment obtained by analyzing the empirical concept of a sound. Secondly, Brentano defends a form of concept empiricism that plainly contradicts Kant’s view that some concepts are not empirical. What Kant calls ‘pure concepts of understanding’ actually are empirical concepts, namely concepts that derive from inner experience (Brentano, 1970, p. 28). For example, ‘the origin [of the concept of necessity] resides in the domain of the perceptual presentation that we have, for example, in the case of an evident syllogism. We know analytically, by an apodictic judgment, that it is impossible for the conclusion to be false while the premises are true’ (Brentano, 1970, p. 27). The most basic difference is that Brentano, in complete opposition to Kant, believes that analytic judgments — including ‘A is A’ — expand and advance knowledge. See the unpublished lecture notes kept at the Husserl Archives in Louvain under the signature Q8, p. 207, quoted in (Rollinger, 1999, p. 40): ‘The judgment “A is A” brings, to be sure, no new concept; the presentation is not enriched. It is however another matter whether or not knowledge is enriched. A alone contains no judgment. If, however, we say “A is A”, knowledge is nonetheless enriched when one progresses from the concept to the judgment’. I owe this quote to an anonymous reviewer. Concepts are psychological data, given in inner experience. However, concepts thus conceived generally have implicit or indistinct (undeutlich) parts, since inner perception is mostly confused. Thus, the psychognost’s task is to analyze these parts so as to make them explicit. In doing so, she makes us discover properties and relations that were before implicit, that is, unknown. The key point is that conceptual analysis clarifies things that must be already present, although indistinctly, in the analysandum. The whole paradox of a priori knowledge is that it is something you both acquire and already had: ‘I do not think we could say that when something is presented confusedly or clearly, the object of the confused presentation does not include everything that is contained in the object of the clear presentation. For all relations that are noticeable in the latter are already present in the confused presentation, but were not noticeable in the same manner as in the clear presentation’. (Brentano, 1928, p. 76/56) This view is a variant of what Berlin Gestalt theorists called the ‘constancy hypothesis’ (see below). A substantially similar view on this issue is found in Meinong’s essay on psychological analysis of 1894 (Meinong, 1894, p. 348 ff.). Meinong held that the only difference between the analysandum and the analysans is that the constituents set out in the latter are known. Since the property of being known adds no new content, the analysandum and the analysans can be identical even though the analysans expands knowledge. But this leads to a difficulty, sometimes called the ‘paradox of analysis’ (Langford, 1942; Pap, 1949, pp. 448–455; Humberstone, 1997; Keefe, 2002). The problem is that the ‘intrinsic features’ that are made explicit through analysis do not seem to be phenomenologically manifest in unanalyzed inner perception. As Brentano says, analysis expands knowledge inasmuch as inner perception is confused. This important feature of Brentano’s account — uvw are discovered rather than produced by analysis — is nicely expressed in (Boccaccini, 2015, pp. 27–8) as reflecting a certain realism with respect to psychological classifications. But on the other hand, if analysis uncovers characters that are not phenomenologically manifest in inner perception, then the analysandum is not identical with the analysans. If a priori judgments are not mere tautologies, if they expand knowledge, then there must be something more in the analysans than in the analysandum — which seems to contradict Brentano’s view that they are analytic judgments. Kant’s theory of synthetic a priori is an attempt to escape this difficulty by saying that the concept of the sum of 1 and 2 is indeed different from the concept of 3, and hence that the judgment that 1 + 2 = 3 is not analytic. Brentano, however, affirms both that ‘1 + 2 = 3’ is an analytic judgment (Brentano, 1970, pp. 25, 51, 67; Brentano, 1974, pp. 154–5) and that it is informative. But if so, how can it be that ‘1 + 2 = 3’ expands knowledge, while ‘3 = 3’ obviously doesn’t? Why do we learn that all bachelors are un-married, and not that all bachelors are bachelors? Suppose that Brentano is right and that Helmholtz’s law of harmonics is analytic. Consequently, it is a conceptual truth, constitutive of the concept of sound, that all (musical) sounds have the properties described in Helmholtz’s law — or, as I will say, are ‘Helmholtzian’. On Brentano’s account, this means that someone who denies that sounds are Helmholtzian either is a fool, or uses the word ‘sound’ to express another concept of sound. The properties attributed in an analytic judgment are ‘intrinsic features without which [the concept] would not be the same concept’ (Brentano, 1970, p. 10; see above). Thus, how does Helmholtz’s law expand knowledge? What do we learn through it? Not that the phenomena called ‘sounds’ are Helmholtzian (indeed, you can call ‘sound’ a phenomenon that is not Helmholtzian) — but that the word ‘sound’ is commonly used to denote Helmholtzian phenomena. However, this account does not do justice to Brentano’s actual rationale, which is that science needs a priori classifications that are ‘natural’ (Brentano, 1925, p. 3/177; Dewalque, 2018). Brentano firmly rejects the idea that the psychognost’s laws express ‘a priori constructions’ (apriorische Konstruktionen) (Brentano, 1925, p. 28/194), that is, stipulations or linguistic conventions. Likewise, the phenomena that fall under Helmholtz’s law must be the same phenomena that were called ‘sounds’ before Helmholtz. Put otherwise: everybody understood, although confusedly, what a Helmholtzian sound was before Helmholtz discovered his law of harmonics. Analytic judgments clarify contents that are confusedly given already in unanalyzed experience. The question is what changes and what remains constant as we pass from an analysandum C to its analysans uvw — for example from ‘3’ to ‘1 + 2’, from ‘purple’ to ‘color between red and blue’, from the unanalyzed experience of a sound to the Helmholtz law, and the like. There are many possible answers to this question. A Fregean would reply that the object is identical but that the logical meaning must be different. Kant’s account seems to be that the phenomenal data are identical and that the conceptual representation is different. A Carnap-style answer to the question is to say that the relations between phenomenal data remain identical and that just the words used vary, the word ‘C’ being no more than a convenient abbreviation for a more complex linguistic form ‘uvw’. Brentano’s answer to this question may be regarded as a variant of what the representatives of the Berlin school of Gestalt psychology (Koffka, Köhler, Wertheimer) termed the ‘constancy hypothesis’. Roughly, the constancy hypothesis (Rubin, 1921; Koffka, 1922) states that, when the attentional focus shifts from one to another area of a sensory field, or from the whole field to one of these areas, the whole field itself remains constant. Applied to the issue at hand, this means that, when you analyze a concept C in its constituent parts uvw — i.e., attentionally focus on u or v or w instead of the whole C —, the constituents must be already present to consciousness before C is analyzed. The Gestalt theorists reject this view, arguing that the two experiences’ phenomenology is entirely different and hence that the constancy hypothesis is not empirically supported. This involves a rejection of the method of analysis, although only to some extent (Metzger, 1928; Kanizsa, 1979, pp. 59–61). It is to say that the analysans is somehow already contained, that is, phenomenally conscious, in the analysandum, although only ‘confusedly’. For example, psychological analysis just clarifies your confused intuition that purple is between red and blue. You already knew it before — otherwise you wouldn’t have been able to judge that the eggplant is purple —, but your knowing was confused and needed analysis to become clear and distinct. This solution may look attractive in the case of purple, but it is implausible in other cases. It is most implausible that the properties described in Helmholtz’s law of harmonics — which Brentano takes to be an analytic truth — were already phenomenologically manifest, although confusedly, before Helmholtz discovered them, or that someone who never heard of Helmholtz’s law confusedly experiences them. It rather seems that these properties were not experienced in any way, and that Brentano’s claim that they were amounts to illegitimately introducing into experience something that is actually not contained in it. One way to escape this objection is to say this: The concept of sound before Helmholtz was different from the concept of sound after Helmholtz; so the word ‘sound’ is ambiguous since it expresses both concepts, just as the word ‘vegetable’ is sometimes used to denote tomatoes and sometimes not, or the word ‘fish’ is sometimes used to denote whales and sometimes not. Thus, the analysandum of Helmholtz’s law is not the pre-Helmholtzian concept of sound, but, say, the concept of sound Helmholtz himself abstracted from his own observations. This does not solve the problem, however. As the Berlin Gestalt theorists rightly objected to ‘analytic’ psychology — including the Brentanians’ psychology —, the fact that introspection reveals such-and-such features of a given mental state does not allow you to infer that these features were already present in the state as it was experienced before you introspected it. It seems much more natural to say that before you discover them through introspection, these features are just unconscious, that is, not only unnoticed as Brentano claims, but not phenomenally manifest in any sense. In consequence, it seems more natural either to propose another solution to the ‘paradox of analysis’ than in terms of clarifying confused perception, or to abandon the view that some analytic judgments expand knowledge. Conclusion I have considered two basic claims underlying Brentano’s conception of a priori knowledge: [a] all a priori knowledge is analytic; [b] since concepts are thought and thought is a form of experience, concepts, like sensory data, are no more than data of experience. The overall idea is that most concepts are composed of simple concepts and that a priori knowledge is knowledge of these latter: analytic judgments expand knowledge inasmuch as they ‘clarify’ concepts by analyzing them into their component parts. This obviously presupposes that the unanalyzed and the analyzed concept are one and the same concept. According to [b], however, concepts are no more than data of experience. The statement that a concept C is composed of uvw makes sense only if your thinking (and hence experiencing) C involves your thinking (and experiencing) uvw even if you are not explicitly aware of uvw. Brentano infers from this that C’s constituents uvw must be already given in the unanalyzed experience, although in an implicit or confused manner. My suggestion has been that this latter idea makes highly problematic Brentano’s account of a priori knowledge as well as any other account in both mereological and psychological terms. I am grateful to Guillaume Fréchette, Charles Siewert, Arnaud Dewalque, Hamid Taieb, and Johannes Brandl for helpful comments on a previous version of this paper. Thanks also to an anonymous reviewer for important suggestions to improve the manuscript. References Albertazzi, L. (1996). From Kant to Brentano. In L. Albertazzi , M. Libardi, R. Poli (Eds.). The School of Franz Brentano (pp.  423–464). Dordrecht, The Netherlands: Kluwer. Antonelli, M. (2015). Ein unveröffentliches Kapitel der Philosophie- und Psychologiegeschichte. In F. Brentano & G.T. Fechner, Briefwechsel über Psychophysik 1874-1878 (pp. 3–73). Berlin, Germany: De Gruyter. Bealer, G. (1992). The incoherence of empiricism. Proceedings of the Aristotelian Society, Supplementary Volumes, Vol. 66, pp. 99-138. Bealer, G. (1998). Intuition and the Autonomy of Philosophy. In M.R. Depaul & W. Ramsey (Eds.), Rethinking Intuition: The Psychology of Intuition and its Role in Philosophical Inquiry. Lanham, USA: Rowman & Littlefield. Boccaccini, F. (2015). The bounds of object: The Brentano-Meinong dispute, a priori knowledge, and the power of perception. In B. Leclercq, S. Richard, D. Seron (Eds.), Objects and Pseudo‐Objects: Ontological Deserts and Jungles from Brentano to Carnap (pp. 17–37). Berlin, Germany: De Gruyter. Brentano, F. (1924). Psychologie vom empirischen Standpunkt. Hamburg, Germany: Meiner. Engl. trans. by A.C. Rancurello, D.B. Terrell & L. McAlister (1995): Psychology from an Empirical Standpoint. New York, USA: Routledge. Brentano, F. (1925). Psychologie vom empirischen Standpunkt, Vol. 2: Von der Klassifikation der psychischen Phänomene. Leipzig: Meiner. Engl. trans. by A.C. Rancurello, D.B. Terrell & L. McAlister (1995): Psychology from an Empirical Standpoint. New York, USA: Routledge. Brentano, F. (1928). Vom sinnlichen und noetischen Bewusstsein [Psychologie / Band III]. I. Teil: Wahrnehmung / Empfindung / Begriff. Leipzig, Germany: Meiner. Engl. trans. by M. Schättle & L.L. McAlister (1981): Sensory and Noetic Consciousness. Psychology from an Empirical Standpoint III. London, UK: Routledge & Kegan Paul. Brentano, F. (1952). Grundlegung und Aufbau der Ethik. Bern, Switzerland: Francke. Engl. trans. by E.H. Schneewind (2009) The Foundation and Construction of Ethics. New York, USA: Routledge. Brentano, F. (1970). Versuch über die Erkenntnis. Hamburg, Germany: Meiner. Brentano, F. (1974). Wahrheit und Evidenz. Hamburg, Germany: Meiner. Brentano, F. (1982). Deskriptive Psychologie. Hamburg, Germany: Meiner. Engl. trans. by B. Müller (1995). Descriptive Psychology. New York, USA: Routledge. Brentano, F. (2010). Logik EL80. Edited by R. Rollinger. Graz, Austria: Franz Brentano Archiv Graz. URL: http://gams.uni-graz.at/archive/objects/context:bag/methods/sdef:Context/get?mode=logik-en Brugmann, K. (1904). Kurze vergleichende Grammatik der indogermanischen Sprachen. Strassburg, Germany: K.J. Trübner. Chudnoff, E. (2013). Intuition. Oxford, UK: Oxford University Press. Dewalque, A. (2018). Natural classes in Brentano’s psychology. Brentano Studien, 16, pp. 111–142. Fisette, F. (2014). Franz Brentano et le positivisme d’Auguste Comte. Les Cahiers philosophiques de Strasbourg, 35, pp. 85–128. Fisette, F. (2018). Franz Brentano and Auguste Comte’s positive philosophy. Brentano Studien, 16, pp. 73–110. Humberstone, I.L. (1997). Two types of circularity. Philosophy and Phenomenological Research, 57/2, pp. 249–280. Kanizsa, G. (1979). Organization in Vision: Essays on Gestalt Perception. New York, USA: Praeger. Keefe, R. (2002). When does circularity matter? Proceedings of the Aristotelian Society, 102/3, pp. 253–270. Koffka, K. (1922). Perception: An Introduction to the Gestalt-Theorie. Psychological Bulletin, 19, pp. 531–585. Körner, S. (1987). On Brentano’s objections to Kant’s theory of knowledge. Topoi, 6, pp. 11–17. Langford, C.H. (1942). The notion of analysis in Moore’s philosophy. In P.A. Schilpp (Ed.), The Philosophy of G. E. Moore (pp. 321–342). Chicago Evanston, USA: Northwestern University. Leroux, J. (2005). An epistemological assessment of the Neogrammarian Movement. In D.A. Kibbee (Ed.), History of Linguistics 2005: Selected papers from the Tenth International Conference on the History of the Language Sciences (ICHoLS X), 1–5 September 2005, Urbana-Champaign, Illinois (pp. 262–273). Amsterdam, The Netherlands, and Philadelphia, USA: John Benjamins. Marty, A. (1908). Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie, Vol. 1. Halle, Germany: Niemeyer. Meinong, A. (1894). Beiträge zur Theorie der psychischen Analyse. Zeitschrift für Psychologie und Physiologie der Sinnesorgane, 6, pp. 340–385 and 417–455. Metzger, W. (1928). Certain implications in the Concept of ‘Gestalt’, The American Journal of Psychology, 40/1, pp. 162–166. Pap, A. (1949). Elements of Analytic Philosophy. New York, USA: Macmillan. Paul, H. (1880). Prinzipien der Sprachgeschichte. Halle a. S., Germany: Niemeyer. Rollinger, R. (1999). Husserl’s Position in the School of Brentano. Dordrecht, The Netherlands: Springer. Rubin, E. (1921). Visuell wahrgenommene Figuren. Studien in psychologischer Analyse, 1. Teil. Copenhagen, Denmark: Gyldendal, 1921. Seron, D. (2012). The Fechner-Brentano controversy on the measurement of sensation. In I. Tanasescu (Ed.), Franz Brentano’s Metaphysics and Psychology (pp. 342–365). Bucharest, Rumania: Zeta Books. Seron, D. (2017). Marty, Husserl, and the (psycho)logical a priori. In G. Fréchette & H. Taieb (Eds.), Mind and Language — On the Philosophy of Anton Marty (pp. 309-323). Berlin: De Gruyter. 15