Embodied Minds in Action: chapter 3

Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 101 3 Essentially Embodied Agency I: Actions, Causes, and Reasons Let us not forget this: When I ‘‘raise my arm,’’ my arm goes up. And the problem arises: What is left over if I subtract the fact that my arm goes up from the fact that I raise my arm? Ludwig Wittgenstein¹ FN:1 The act of raising the arm is a complex event, constituted out of a causally linked pair, the trying and the arm rising, which are . . . ‘‘made for one another.’’ Brian O’Shaughnessy² FN:2 3.0 Introduction For healthy ordinary people in ordinary situations, arm-raising seems like the simplest thing in the world. But philosophers are not ordinary people. And philosophers of action, in particular, have correctly observed that there is an obvious and categorical difference between (i) my deliberately raising my arm to wave to a friend, and (ii) my arm’s uncontrollably rising in a Dr Strangelove-like spasm, even if the overt body movements are indiscriminable. Indeed, as we mentioned in the Introduction, the problem of action is how to give an adequate account of the categorical difference between the things we intentionally do, or intentional actions, and the things that just happen to us, or mere bodily events. ¹ Wittgenstein, Philosophical Investigations, 161e , §621. ² O’Shaughnessy, ‘‘Trying (as the Mental ‘Pineal Gland’),’’ 70. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 102 102   : , ,   It turns out that it is not at all easy to say what the categorical difference between an intentional arm-raising and a mere arm-rising really consists in. Now the we and us in action are conscious, intentional, motile, egocentrically-centered and spatially oriented, thermodynamically irreversible, suitably neurobiologically complex living organisms—minded animals. So our goal in this chapter and the next is to say as precisely as possible what, for minded animals, the categorical difference between intentional actions and mere bodily events really is. More precisely, this chapter explores the neurophenomenological, conceptual, and metaphysical connections between intentional actions, causes, and reasons. In Section 2.1, we spell out and criticize classical causal theories of action in a general way. But as we point out, non-causal theories of action are also unacceptable, since they implausibly substitute teleological reasons-explanations for the basic causal facts that actually bring about intentional actions. Our response to this dilemma is to develop a non-classical but still causal theory of action—what we call the Essentially Embodied Agency Theory. In Sections 3.2 to 3.5 we motivate this theory of action by focusing specifically on Davidson’s classical causal theory and then developing four fundamental worries about it. Here is a quick Coming Attractions preview of the Essentially Embodied Agency Theory of action. As we see it, every classical causal theory inserts a vitiating metaphysical or temporal gap between antecedent mental causes and consequent body movements. Now for our purposes, a ‘‘body movement’’ in a creature minded like us is an integrated series of dynamic endogenous events involving both ‘‘covert’’ neurobiological processes as well as ‘‘overt’’ behavioral processes normally arising from these processes. Or more precisely put, body movements are of two importantly different but closely related kinds: (1) covert body movements are internal neurobiological processes that occur between the vital organs and the muscle tissue/skin interface, and that normally begin prior to overt body movements, and (2) overt body movements are external behavioral processes, normally arising from and accompanying neurobiological processes, that begin at the muscle tissue/skin interface and engage with the outer world. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 103 103 On our view, what distinguishes intentional actions (e.g., arm-raisings) from mere bodily happenings (e.g., arm-risings) are the causally efficacious operations of a certain mental activity of an essentially embodied intentional agent throughout the entire time in which some of that agent’s body movements covertly arise neurobiologically and then display themselves overtly and behaviorally. All of these body movements are the agent’s intentional body movements. So if we assume that the Essential Embodiment Thesis is true, then since the causally efficacious mental activity of the agent is necessarily and completely neurobiologically embodied, and since that mental activity is also synchronous with the complete two-part process that encompasses both the relevant covert neurobiological process as well as the relevant overt behavioral process, it follows necessarily that there are no vitiating metaphysical or temporal gaps whatsoever between the mental activity of the agent and her intentional body movements. Or more briefly and imagistically put, the conscious intentionalitylo of the agent and her intentional body movements fit together as seamlessly as W.B. Yeat’s dancer and her dance: FN:3 O body swayed to music, O brightening glance, How can we know the dancer from the dance? ³ 3.1 Classical Causal Theories of Action, and Beyond FN:4 Classical causal theories of action say that what distinguishes intentional actions from mere bodily events is essentially a difference in the causal origin of those bodily events. The bodily events are the same in both cases, and intentional actions are supposed to be the ones brought about by antecedent causes in some categorically different manner. There are three classical causal theories of action. On the agent-causal view, the causal antecedence is metaphysical but not temporal. A pure mental substance, or ‘‘agent-cause,’’ which exists outside the series of natural events, and thereby is naturally undetermined by those events, is supposed to bring about the relevant bodily event in an incompatibilistically free way.⁴ ³ Yeats, ‘‘Among School Children,’’ verse viii, 245. ⁴ See, e.g., Chisholm, ‘‘Human Freedom and the Self ’’; Clarke, ‘‘Agent Causation and Event Causation in the Production of Free Action’’; and O’Connor, Persons and Causes. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 104 104   : , ,   FN:5 FN:6 FN:7 On the volitional-causal view by contrast, the causal antecedence is temporal. A mental event of conscious willing at one time is supposed to cause a later bodily event.⁵ By another contrast, on the Davidsonian causalist view, which has dominated the philosophy of action since the 1970s, the causal antecedence is both metaphysical and temporal. Reasons are supposed to be causes, and an action is a physical event ea that is caused by a ‘‘primary reason.’’ A primary reason, in turn, is a psychological pair consisting of a belief and a desire that together ‘‘rationalize’’ or teleologically explain ea .⁶ The mental properties of this psychological pair are strongly supervenient in an ‘‘anomalous’’ way—i.e., in accordance with Davidson’s Principle of the Anomalism of the Mental, which says that there are no strict deterministic psycho-physical laws—on fundamental physical facts about some earlier physical event ec , that in turn naturally causes ea under strict deterministic physical laws. And the psychological pair, considered as a single mental event of selfconscious deliberative intention—call it ‘‘Me’’—is numerically identical with ec .⁷ Following Arthur Danto, we accept the classical distinction between (i) basic acts, and FN:8 (ii) non-basic acts.⁸ Basic acts occur whenever an intentional agent performs a particular sequence of intentional body movements and no other acts are performed, and non-basic acts are acts that involve some basic acts but are not identical to those basic acts. Thus, e.g., someone waves to a friend (non-basic act) by raising her arm (basic act). Our analysis of action will focus primarily on basic acts. It seems clear to us that all classical causal theories of action—whether agent-causal, volitional-causal, or Davidsonian—ultimately alienate the conscious intentionalitylo of the agent from the intentional body movements ⁵ See, e.g., O’Shaughnessy, ‘‘Trying (as the Mental ‘Pineal Gland’).’’ In Ch. 3 we will argue that a volitional, trying-based account of action coheres perfectly with a non-classical causal theory of action. ⁶ See Davidson, ‘‘Actions, Reasons, and Causes,’’ 27. ⁷ See Davidson, ‘‘Mental Events’’; and Davidson, ‘‘Thinking Causes.’’ ⁸ See Danto, Analytic Philosophy of Action, 31. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 105 105 that are supposed to be the immediate effects of conscious intentionalitylo in basic acts. This is because all such accounts imply that whenever a basic act occurs, there is some sort of vitiating gap—whether metaphysical, temporal, or both—between conscious intentionalitylo and its immediate bodily effects. In agent-causal theories, the alienation is the result of a vitiating substancedualist gap into which a mysterious Cartesian causal interaction must be inserted between transcendent mental substances and fundamentally physical events. In volitional-causal theories, the alienation is the result of a vitiating temporal gap into which deviant causal chains can always be inserted between earlier mental events and later physical events involving body movements, thereby making those later body movements unintentional. For example, someone tries to raise her paralyzed arm and fails, but her simultaneous frustrated desire to move her arm accidentally triggers a nearby brain scanner, which accidentally triggers someone else’s Blackberry, which accidentally connects with the digital control system of a tractor-beam ray gun on Mars, which accidentally zaps her arm perfectly into place above her head. This is what you might call causal deviance with an altitude. And in Davidson’s theory, in addition to the same vitiating temporal gap that is always open to deviant causal chains, the alienation is also the result of a vitiating property-dualist-without-substance-dualist gap into which upwards determination relations must be inserted between the causally efficacious properties of the physical event ec and the strongly supervenient mental properties of the conscious intentional mental event Me that is numerically identical with ec . Because the mental properties of Me are strictly upwardly determined by the causally efficacious properties of ec , and because Me is numerically identical with ec , the mental properties of Me do not have any causal efficacy apart from the properties of its underlying physical base, and must be causally inert or epiphenomenal. This is very unfortunate for Me. (Boom, boom, two in a row.) But whatever the origins of the classical causalist gap, the result is always the same. The causal autonomy of the agent’s conscious intentionalitylo undermines the causal efficacy of her conscious intentionalitylo , and thus intentional action is not adequately explained. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 106 106   : , ,   FN:9 FN:10 FN:11 FN:12 Some action theorists, finding serious problems with the Davidsonian account, have advanced non-causal, teleological theories of action.⁹ Carl Ginet, for example, suggests that for a teleological reasons-explanation to be true, the action must be accompanied by an intention with the right sort of content: The subject intended of that action that by it she would A.¹⁰ His view suggests that the mere presence in the agent of an intention about her A-ing is sufficient for that intention’s being explanatory of her action (i.e., her A-ing). However, it seems clear that our body movements might merely accidentally coincide with our desires or intentions, without those movements’ being explained by these desires or intentions. If, to use the example described above, an intentional agent’s arm-movements were actually and accidentally caused by a tractor-beam ray gun on Mars, it would be highly implausible to say that the agent’s frustrated desire to move her arm explained those movements.¹¹ Indeed, unless desires and intentions play a direct causal role in the production of body movements, it seems that the conscious, intentional animal does not act as a genuine agent, but is instead under the control of outside forces. Also, without appealing to causal facts, it is very difficult to make sense of what makes it true that an agent acted in pursuit of one goal rather than another, or for some reason rather than another. The natural answer to the question of what makes it true that an agent acted in pursuit of one goal rather than another or for some reason rather than another is that the mental event or process that explains the particular action in question is the one that ‘‘figures suitably in the etiology of the action or of [the subject’s] completing that action.’’¹² Therefore a causal theory makes much better prima facie sense of what it means to perform body movements and act for the sake of some goal, than any non-causalist view does. But the $64, 000.00 question is: Can one be a causalist about action without also being a classical causalist? Our answer is: Yes, but only if the classical causalist gaps have been closed up tight from the start. So our response to the dilemma that both classical causal theories and non-causal theories are manifestly inadequate is to present a non-classical but still causal theory of action that is designed to rule out the various vitiating gaps associated with classical causal theories ⁹ For some non-causal theories of action, see Anscombe, Intention; Sehon, ‘‘An Argument Against the Causal Theory of Action Explanation,’’; and Sehon, ‘‘Connectionism and the Causal Theory of Action Explanation.’’ ¹⁰ Ginet, On Action. ¹¹ Mele, Motivation and Agency, 46. ¹² Ibid., 40. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 107 107 and also to account for the full range of action-types ordinary agents perform. This is the Essentially Embodied Agency Theory of action. The two basic features of the Essentially Embodied Agency Theory are (i) that it explicates intentional body movement in terms of synchronous trying and its active guidance of the agent’s motile living body, and (ii) that it explicates trying and its active guidance in terms of desire-based emotions or emotionsd . The present chapter and Chapter 4 concentrate on the first element. For us, trying and its active guidance, like all mental states, events, or processes of a consciousnesslo , are necessarily and completely neurobiologically embodied. This enables us to solve the problem of action by holding that intentional actions are not mere bodily events of any kind, but instead are essentially embodied events of a certain kind, namely those inherently involving a synchronous trying and its active guidance of the agent’s own motile, living animal body. For when we combine the Essential Embodiment Thesis with the thesis that intentional actions are brought about by synchronous trying and its active guidance, it directly follows that the conscious intentionalitylo of the agent and her intentional body movements in basic acts are not only temporally in sync, but also are metaphysically connected as closely as possible, short of strict identity, by virtue of their intrinsically reciprocal, or intrinsically two-way, non-logical or strong metaphysical a priori necessitation. These points about non-identity and intrinsic reciprocity require particular emphasis. The relation of strict identity, whether construed as either (i) type-identity (identity of properties or universals), (ii) token-identity (identity of individuals or particulars, a.k.a. ‘‘numerical identity’’), or (iii) essential identity (identity of kinds), can always be construed as a reductive relation, whenever it is represented by means of two terms that differently name or describe the same thing. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 108 108   : , ,   For given the right theoretical backdrop (say, a scientific essentialist theory of physical microstructure), then one term t1 (say, ‘water’) can then be used to name or describe something that is ‘‘nothing but’’ or ‘‘nothing over and above’’ what is named or described by the other term (say, ‘H2 O’), by virtue of the strict identity relation between them (in this case, essential microphysical identity). But even if a non-identity between the referents of the terms can be somehow demonstrated, then a certain kind of fairly robust reduction is still possible if one or both of the properties, individuals, or kinds is logically strongly supervenient on the other. This is because the supervening individual or properties can then be held to be ‘‘fully determined’’ or ‘‘fully fixed’’ by the properties of its corresponding physical supervenience base, whether by logical or analytic necessity alone (a priori physicalism or reductionism), or by logical or analytic necessity together with causal laws (a posteriori physicalism or reductionism). Supervenience, in turn, can be either (i) one-way (asymmetric), as, e.g., in the ‘‘upwards’’ determination of temperature properties on the mean molecular motion of the particles comprising the material bodies, gases, or liquids that have temperature, or else (2) two-way (bilateral), as, e.g., in the ‘‘back-and-forth’’ mutual determination of force and the product of mass and acceleration according to the classical Newtonian equation F = ma. Furthermore it seems that even if there are some other ways in which reduction is possible (e.g., finegrained logically necessary equivalence of properties, or perhaps logically necessary co-extension), all of these will also entail either strict identity or at least logical strong supervenience. So the disjunction consisting of either strict identity, or finegrained logically necessary equivalence, or logically necessary coextension, or one-way logical strong supervenience, or bilateral logical strong supervenience would seem to be necessary and sufficient for physicalist reduction. If this is correct, then since the relation we are positing between the conscious intentionality of the agent is at once a relation of non-logically necessary co-extension (and thus is neither finegrained logically necessary equivalence nor logically necessary co-extension), nonidentity (and thus is neither type-identity, token-identity, nor essential Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 109 109 identity), and reciprocal intrinsicness (and thus is neither non-asymmetric nor only extrinsically symmetric, as in one-way or two-way logical strong supervenience), it is therefore nonreductive, no matter how modally airtight that relation may otherwise be. More precisely then, on our view fundamental mental properties (involving conscious intentionalitylo ) on the one hand, and certain corresponding fundamental physical properties on the other, are (i) non-logically or strongly metaphysically a priori necessarily coextensive in all and only living organisms of a suitable level of neurobiological complexity, (ii) non-identical, and (iii) reciprocally intrinsic properties of those very organisms. Or in other words, those organisms are essentially mental-and-physical. As we will see in Chapters 6, 7, and 8, when looked at from a metaphysical standpoint, this non-logically or strongly metaphysically a priori necessary co-extension together with reciprocal intrinsicness, but without identity, means that the corresponding fundamental mental properties and fundamental physical properties in animals of a suitable level of neurobiological complexity are at once mutually irreducible and yet also fused. And that thesis, in turn, allows us to answer Wittgenstein’s deep question, ‘‘What is left over if I subtract the fact that my arm goes up from the fact that I raise my arm?,’’ correctly and directly. The correct, direct answer is: That would be like trying to subtract the dance from the dancer—but you cannot know the dancer from the dance! More precisely, we think that it is non-logically or strongly metaphysically a priori impossible to subtract the fact that my arm goes up from the fact that I raise my arm. In my intentional action of raising my arm, the two facts become non-logically or strongly metaphysically a priori non-detachable. So, given the essential embodiment of minded animals, there simply cannot be and thus never is a metaphysical or temporal gap between our conscious intentionalitylo and our intentional body movements. When looked at from a philosophy-of-action standpoint however, the very same non-detachable but non-reductive relation of mental-physical property fusion is also interpreted by us as a synchronous causal relation Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 110 110   : , ,   of effective desiring, willing, or trying-and-its-active-guidance. As we said in Chapter 1, the essential embodiment of consciousnesslo entails the Essentially Embodied Cogito: I desire, therefore I am. In Chapters 3–4, we will argue that effective desiring is the foundation of willing, and also that willing is the same as trying and its active guidance of body movements. If so, then the Essentially Embodied Cogito is also in effect a causal-intentional Cogito: I effectively desire, therefore I am simultaneously intentionally moving my body. Under favorable endogenous and exogenous conditions, just by effectively desiring to move my living body, I thereby also simultaneously will my own intentional body movements, which is the same as the fact that my trying and its active guidance simultaneously self-determine my intentional body movements. In this way, both the causal autonomy and the causal efficacy of the conscious intentionalitylo of the agent are jointly secured. FN:13 Much contemporary philosophy of action begins with the more or less explicit assumption that it is always possible to construct a broadly Humean, belief-desire based, instrumental reasons-explanation for action.¹³ For example, Jane’s desire for a doughnut and her belief that a doughnut is in the cupboard explain why she walks over to the cupboard. This standard account of action entails that intentional actions are done because the agent has a certain belief-desire pair that explains the action by giving an instrumental teleological rationalization of it. A version of this broadly Humean story has been very influentially defended by Davidson, whose central claim, as we have said, is that a primary reason for an action is its cause. As we have also said, a primary reason consists of a belief-desire pair that insrumentally teleologically rationalizes a certain physical event ea , the action. So whenever someone does something for a reason, he has some sort of pro-attitude toward actions of a certain kind and believes that his action is of that kind. To know a primary reason for action is to know the intention with which the action was done, and also to know how that action is coherent with certain traits, both long and short termed, of a rational agent. Davidson points out that a person can have a reason for action, perform the act in question, and yet not act on the basis of this reason. If a reason is to explain an action, ¹³ See, e.g., Hume, Treatise of Human Nature, book II, part III; and Blackburn, Ruling Passions: A Theory of Practical Reasoning. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 111 111 the agent must perform the action because he had that reason. Davidson regards such instrumental teleological rationalizations of behavior as a type of ordinary causal explanation, so that the ‘because’ in ‘‘Mary went to the fridge because she wanted a beer’’ is a causal ‘because’. We do not wish to dispute in any way Davidson’s claim that actions are suited for causal explanation. But we do also hold that an adequate causal explanation of an action cannot be secured just by placing an action in the context of an instrumental teleological rationalizing reason that strongly supervenes in an anomalous way on a causally efficacious physical event, precisely because this does not show how the conscious intentionalitylo of the agent can have both causal autonomy and causal efficacy. Instead, in order to secure the fusion of causal autonomy and causal efficacy in agency, we must tell a metaphysically plausible story about the causality of the action from the inside —where ‘from the inside’ means both from the first-person standpoint of the agent as a conscious, intentional (and in some cases, also a rational human) animal and equally also from the endogenous standpoint of the agent as a motile, situated, forward-flowing, complex living organism dynamically embedded in and dynamically engaged with the natural world. Indeed, this metaphysically plausible ‘‘complete insider’s story’’ about the causality of the action is precisely what we need if we are ultimately to combine our theory of action in Chapters 3, 4, and 5 with our metaphysics of agency in Chapters 6, 7, and 8. In order to motivate our Essentially Embodied Agency Theory of action, we will work through four serious worries about Davidson’s theory of action. Here are capsulized versions of the worries. First, Davidson’s theory of mental events, together with his actiontheory, jointly entail that the instrumental teleological rationalizations to which he appeals do not truly refer to mental causes at all. By his own admission, the reasons that explain action are causally efficacious only by virtue of their token-identity with physical events. Second, it appears that the possession of a reason is not in and of itself sufficient for action: some further mental effort or exertion is required on the part of the agent if an intentional body movement is to take place. Third, we think that Davidson’s theory cannot adequately account for the full range of actions carried out by ordinary intentional agents, and that it errs by narrowly concentrating on actions associated with instrumental rationality. One obvious version of this worry is the objection that many Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 112 112   : , ,   non-human animals and young human children, none of whom can be plausibly taken to be self-conscious and deliberative agents capable of forming or recognizing instrumental reasons for their actions, are nevertheless minded animals who can act intentionally. But even when we focus exclusively on agents who are capable of self-conscious deliberative action via instrumental reasons—e.g., rational human animals, or real persons—it seems clear that there are several types of action that fall below Davidson’s radar and which do not actually require self-conscious deliberative actions via instrumental reasons. Here we will consider (i) pre-reflective or spontaneous actions, (ii) akrasia, or so-called ‘‘weakness of the will,’’ which we will appropriately re-name impulsiveness of the will, and (iii) so-called ‘‘desire-independent,’’ or non-instrumental, reasons for action. Fourth and finally, we will show how the well-known worry about deviant causal chains poses a fundamental problem not just for Davidson’s theory, but for all classical causal theories of action. 3.2 Against Davidson 1: Reasons are Epiphenomenal FN:14 FN:15 The first serious problem with Davidson’s theory of action emerges in conjunction with his well-known solution to the mind–body problem, Anomalous Monism.¹⁴ In an accordance with the Principle of the Anomalism of the Mental, Davidson denies the existence of strict deterministic psychophysical laws, but also claims that psychological events are token-identical with certain physical events. He also adopts the Principle of the Nomological Character of Causality, according to which there exists a closed and deterministic system of strict laws into which all causally related events, when appropriately described, fit.¹⁵ It follows that if psychological events are to cause physical events, then there must be strict deterministic physical laws that govern these causal relations. However, in that case mental events ¹⁴ See Davidson, ‘‘Thinking Causes.’’ ¹⁵ Mele, ‘‘Introduction to The Philosophy of Action,’’ 5. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 113 113 cannot cause physical behavior by virtue of their mental properties—which are, after all, only extrinsic or accidental, external properties of the physical events with which those mental events are token-identical, even if those physical events can be shown to have correct, instrumentally and teleologically illuminating intentional descriptions—but instead only by virtue of the fundamental physical properties of the physical events with which mental events are identical. In this way, because reasons as psychological event-tokens have to be identical to physical events in order to cause our actions, it cannot be the case that they are causally efficacious as mental. Insofar as all the real causal work goes on at the underlying physical level, the instrumental teleological rationalizations of which Davidson speaks turn out to be merely ways of informatively and usefully re-describing action. Or otherwise put, reasons can have causal relevance because they provide illuminating descriptions of the bodily physical events—descriptions that are perhaps accessible in no other way than through certain teleological and instrumentally rational concepts whose content, due to semantic holism, is irreducible to mechanistic physical concepts—but they have no causal efficacy. A type does not have causal efficacy just because one or more of its tokens has causal efficacy. A type has causal efficacy if and only if at least one of its tokens has causal efficacy and the type is an inherent or intrinsic property of that token. But on Davidson’s account the mental properties of physical events are at best extrinsic or accidental, external properties of those events, and all of their intrinsic properties are fundamentally physical. Thus Davidson has not shown us that reasons can actually do anything. For X to be able to do something, presumably, requires that (or at the very least, has as a sufficient condition that) either (i) X is a simple singular event¹⁶ that is a nomologically sufficient condition of a physical event, FN:16 or (ii) X is a simple singular event that belongs inherently or intrinsically (i.e., as a necessary proper part) to a complex singular event that is a nomologically sufficient condition of a physical event. ¹⁶ For definitions of the notions of simple event, complex event, singular event, and compound event, see Section 6. 1 below. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 114 114   : , ,   FN:17 FN:18 And Davidson has not shown us that reasons as reasons can satisfy either of these conditions. Moreover, according to Jaegwon Kim’s Explanatory Exclusion Principle or EEP, ‘‘two or more complete and independent [causal] explanations of the same event or phenomenon cannot exist.’’¹⁷ Complete explanations are self-contained and require no other concepts or principles in order to apply to the relevant event or phenomenon. Or in other words, complete explanations are self-sufficient. Independent explanations are complete and also rule out other logically distinct concepts or principles from applying to the relevant event or phenomenon at the same time and in the same respects. Or in other words, independent explanations are both selfsufficient and unique. So given Kim’s EEP, the obtaining of a complete and independent physical causal explanation excludes any complete and independent mentalistic causal explanation. Furthermore the actual existence of the physical causal event confers epiphenomenality, or causal inertness, on any corresponding mental event whose properties are strongly supervenient on the fundamental physical properies of the underlying physical event. Therefore if, as Davidson’s theory entails, all the real efficacious causal work is being done by fundamental physical properties and events, then it follows directly from Kim’s principle that the instrumental teleological rationalizations appealed to by Davidson are ‘‘causes’’ only because the mental events that constitute reasons are identical with physical events. In the Davidsonian world, the psychological and rational facts, as psychological and rational, have no causal efficacy whatsoever, even if they do have explanatory causal relevance.¹⁸ For Davidson, ultimately, we need to appeal to reasons and mental events rather than merely to physical events, only because reasons-talk has epistemic and pragmatic force. For example, if we say that Mary got off the couch because she wanted a beer and believed that a beer was in the fridge, we are informatively and usefully re-describing her body movements in terms of her ¹⁷ See Kim, ‘‘The Myth of Nonreductive Materialism,’’ 268. ¹⁸ The very same problem applies to the sophisticated Davidsonian account of mental causation offered by MacDonald and Macdonald in ‘‘The Metaphysics of Mental Causation.’’ At most their account shows that the causal relevance of reasons is not ruled out by Kim’s exclusion worries. But this does not show that reasons are themselves causally efficacious. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 115 115 primary reason. But these body movements already have a nomologically sufficient physical cause, so they already have a complete and independent causal explanation. Therefore our illuminating (i.e., informative, useful) re-description of it cannot have any substantive implications for actioncausation. Davidson’s theory of action, in effect, falsely substitutes the epistemology and pragmatics of causal explanation for the metaphysics of action-causation. In this connection, then, it is very important to distinguish carefully between (i) explaining why some action happened, i.e., describing the agent’s reasons and other motivations, and (ii) explaining how some action happened, i.e., describing the causal process that actually brought about the action. FN:19 FN:20 When we ask why someone acted as he did, we want to have an illuminating interpretation, ‘‘a new description of what he did which fits it into a familiar picture’’¹⁹ so that we can make sense of his behavior. Thus, if we want to explain why Mary got up from the couch and walked towards the fridge, we are likely to provide a narrative and cite her desire for a cold beer. But on the other hand, if we ask how her desire for a beer got her up from the couch and walking towards the fridge, the story is going to be quite different. The teller of the how-story must describe an efficacious causal link between Mary’s beliefs, desires, intentions, and her intentional body movements. For us, the natural and obvious place to look for this efficacious causal link is in the dynamic neurobiological processes and overt body movements of essentially embodied conscious, intentional mindslo . But merely to place Mary’s action in a ‘‘wider social, economic, linguistic, or evaluative context,’’²⁰ by supplying a reason for her action, seems to offer little or nothing whatsoever towards a proper characterization of the agent-centered (first-personal, endogenous) causally efficacious process that actually produced her action. ¹⁹ Davidson, ‘‘Actions, Reasons, and Causes,’’ 33. ²⁰ Ibid. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 116 116   : , ,   3.3 Against Davidson 2: Reasons are Insufficient for Actions FN:21 Much more must be said about the mental causes of intentional action than is offered by Davidson’s theory. As we have just seen, the Davidsonian theory invokes desires, beliefs, and self-conscious deliberative intentions—in short, reasons—as causes of action. But in fact it is evident that these factors are not in and of themselves sufficient for action. Mary could want a cold beer, believe that a cold beer is in the fridge, and intend to get one. But obviously it does not necessarily follow that she will actually get up from her comfortable seat on the couch to do just that. She could, consistently with the possession of a complete Davidsonian reason, and without any mental breakdown or pathology of volition whatsoever, simply continue to be a couch potato. Those of us who are fond of cold beer and couches alike know this to be all too obviously true. Sometimes the couch just wins out. So as Searle has pointed out, intentional causation is radically unlike billiard-ball causation in several crucial respects. Even if desires and intentions are present, they are not yet sufficient to compel the agent to act. Necessarily there is what Searle calls (somewhat misleadingly, as we will argue in a moment) a ‘‘Gap’’ between intentions and action, such that the intentional agent is able either to choose or not choose the object of her intention.²¹ Or as T. S. Eliot more darkly and poetically puts it: FN:22 Between the idea And the reality Between the motion And the act Falls the Shadow.²² But one person’s Gap and another person’s Shadow can also be a third person’s Time to Dance. What we mean is that if beliefs, desires, and selfconscious deliberative intentions are not themselves sufficient for action, then this is precisely the point at which it seems natural and plausible to say that the volitional phenomenon of trying enters in. No matter what her desires, beliefs, and intentions, Mary will never intentionally get up off the couch to go to the fridge for a cold beer if she does not even try to get up. ²¹ Searle, Rationality in Action, 231. ²² Eliot, ‘‘The Hollow Men,’’ verse V, line 31. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:23 FN:24 11:28am Page 117 117 And whenever she does in fact intentionally get up off the couch to go for a beer, it is always fundamentally because of her trying to do so, together with whatever desire-based reasons she might also have for acting. In any case, we think it would be a big mistake to think of Searle’s Gap or Eliot’s Shadow as a nomological fissure in which the deterministic or statistical laws of nature somehow fail to hold, and into which we must insert an agent-cause or some other radically indeterministic source of libertarian free will. Instead, we think that the so-called Gap or Shadow between intentions and acts is far more adequately understood in terms of dynamic systems theory (DST),²³ as merely a far-from-equilibrium, or unstable, phase in the natural processes of the essentially embodied life of a minded animal. Now by the notion of a ‘‘far-from-equilibrium, or unstable, phase’’ we mean an ongoing situation in natural processes such that very small changes in the initial conditions of events can lead to unpredictably large effects. Examples would include the Big Bang, black holes, the straw that broke the camel’s back, the shout that triggered the avalanche, the flat tire that caused traffic gridlock all over Manhattan, the large effects of small changes in the environment on the weather, the large effects of small environmental and external stimuli on the internal states of living organisms, the large effects of small environmental and external stimuli on the neurobiological and overt intentional movements of minded animals, and so- on. Far-from-equilibrium or unstable phases are thus dynamic periods in which the nomological causal architecture of nature—the complete set of general and more specific natural causal laws, all the way down to actual events—is still inherently in process of formation as regards its finegrained or hyper-finegrained structure. Here we need to pause briefly to define our terms. In this connection the notion of ‘‘roughgrainedness’’ means that a given concept, property, proposition, or law is identical with any other concept, property, proposition, or law that correctly applies to all the same actual and possible objects or states of affairs. When roughgrainedness holds specifically for causal laws of nature, this means that their application is invariant with respect to reversals in the direction of time, or in other words that they presuppose symmetry.²⁴ So roughgrained causal laws are symmetry-based causal laws. ²³ Here we are going to deploy, in an anticipatory and intuitive way, some concepts of DST. See Section 7.3 for more details. ²⁴ See, e.g., Van Fraassen, Laws and Symmetry. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 118 118   : , ,   FN:25 FN:26 FN:27 The notion of ‘‘finegrainedness,’’ by contrast, means that even two concepts, properties, propositions, or laws that correctly apply to all the same actual and possible objects or states of affairs can still significantly differ in their internal structures. When finegrainedness holds specifically for causal laws of nature, this means that their application is variable with respect to reversals in the direction of time, or in other words that they presuppose thermodynamic asymmetry.²⁵ So finegrained causal laws are asymmetry-based causal laws. And the notion of ‘‘hyper-finegrainedness’’ means that even two concepts, properties, propositions, or laws that correctly apply to all the same actual and possible objects or states of affairs, and also share the same internal structure, can still significantly differ in how they are presented to or evaluated by a living organism—e.g., a plant, or a non-human animal, or (most interestingly), a conscious, desiring, willing, rational living human organism. When hyper-finegrainedness holds specifically for causal laws of nature, this means that their application is highly variable with respect to reversals in the direction of time, non-equilibrium, non-linear, dissipatively structured, and naturally purposive, or in other words that they presuppose thermodynamic self-organizing complexity.²⁶ So hyper-grained laws are not only asymmetry-based causal laws, they are also highly context-sensitive causal laws. Finally, both finegrained and hyper-finegrained causal laws of nature are ceteris paribus laws, which is to say that they are modally robust causal rules of nature that specify necessary connection only against a determinate backdrop of special actual conditions or constraints whether world wide or contextual.²⁷ One very simple actual example of a natural process constrained and governed by a hyper-finegrained, assymmetry-based causal law in a far-fromequilibrium or unstable phase of nature would be the highly idiosyncratic and fairly silly-looking 2.5-fingered search-and-smash typing movements by which I just quickly hammered out the last sentence of the just-previous paragraph on my Dell Latitude D810 laptop’s keyboard. I violated no prevailing roughgrained or symmetry-based general laws of physics, whether deterministic or statistical, in order to do this—but at the same time ²⁵ See, e.g., Prigogine, Being and Becoming: Time and Complexity in the Physical Sciences. ²⁶ See, e.g., Nicolis, and Prigogine, Self-Organization in Nonequilibrium Systems. ²⁷ See, e.g.., Rupert, ‘‘Ceteris Paribus Laws, Component Forces, and the Nature of Special Science Properties.’’ Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 119 119 my body movements were neither necessitated by the roughgrained or symmetry-based general laws of physics together with the settled facts about the past nor were they merely the more or less random result of some complex natural statistical sequence of actually past chance events. Those body movements did not just happen to me. Silly-looking as they were, they were uniquely mine. They were intentional. I performed them. For better or worse, they were really and truly up to me. It is crucial to recognize that far-from-equilibrium or unstable phases are perfectly consistent with whatever roughgrained or symmetry-based general deterministic or statistical laws of nature there already are. No roughgrained general deterministic or statistical law is ever violated by an unstable phase and what happens in it. But, in particular, consistency with roughgrained general deterministic laws is not the same as entailment by roughgrained general deterministic laws. Because of their inherently ‘‘in-process’’ character, such phases are not deterministic in the logical or causal sense of classical LaPlacean determinism. That is, the finegrained or hyper-finegrained asymmetry-based natural causal architecture of what occurs during unstable phases is not logically or causally entailed by facts about the past together with roughgrained general deterministic laws of nature. Something with new natural causal powers dynamically emerges and makes a novel determinate contribution. At the same time, however, precisely because uniquely new determinate finegrained or hyper-finegrained natural causal architectures are actually being formed during such phases, such phases and what happens during them are also not strictly indeterministic, or the mere logical or causal result of accumulated antecedent random facts, or chance, together with roughgrained general statistical laws. Hence an event or process can be nondeterministic, in the sense of being not completely deterministic, without also being indeterministic, in the sense of being completely indeterministic. This is because even though such events or processes could not have been predicted in advance, at least in principle, they could be predicted from the inside, as the event or process is actually unfolding, which is to say that, at least in principle, they could be predicted from the standpoint of the uniquely new determinate finegrained or hyper-finegrained causal architecture that is occurrently in process of self-production, and which ultimately stabilizes the natural disequilibrium or instability. Another way of putting this is to say that the event or process is naturally purposive or naturally teleological. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 120 120   : , ,   FN:28 Adopting this insider’s in-process standpoint—the standpoint, in effect, of what-it-is-like-to-become-something, the internal standpoint of a naturally purposive dynamic system—is of course not practically feasible in the case of the Big Bang, black holes, traffic jams, the weather, and living organisms like plants. No matter how much a cosmological physicist tries to imagine her way into ‘‘what-it-is-like-to-become-the-Big-Bang,’’ no matter how much a meteorologist tries to imagine her way into ‘‘what-itis-like-to-become-a-thunderstorm,’’ and no matter how much a botanist tries to imagine her way into ‘‘what-it-is-like-to-become-an-oak-tree,’’ she is always restricted to an external, or outsider’s, explanatory standpoint But in the case of intentional body movements, the essentially embodied agent can predict her own body movements from the insider’s in-process standpoint, precisely because she herself occupies the standpoint of her own dynamic consciousnesslo and intentionalitylo , and precisely because she herself is synchronously willing those body movements. Intentional causation is simply what-it-is-like-to-become-an-intentional-body-movement. In the intentional act of dancing, the dancer consciously becomes her dance. Wittgenstein puts this crucial point perfectly: ‘‘when people talk about the possibility of foreknowledge of the future they always forget the fact of the prediction of one’s own voluntary movements.’’²⁸ In short, then, on our view there are three categorically different kinds of events or processes: (1) completely deterministic events or processes, i.e., events or processes that are logically or causally entailed by roughgrained general deterministic laws plus antecedent facts, (2) completely indeterministic events or processes, i.e., events or processes that are logically or causally entailed by roughgrained general statistical laws plus antecedent facts, and (3) natural causal singularities, which are events or processes that are neither logically or causally entailed by roughgrained general deterministic laws plus antecedent facts nor logically or causally entailed by roughgrained general statistical laws plus antecedent facts, and also exert ²⁸ Wittgenstein, Philosophical Investigations, 162e , § 629. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 121 121 their finegrained and hyper-finegrained causal powers consistently with whatever roughgrained general deterministic or statistical laws there are. We hold that some events and processes in the natural world are completely deterministic, e.g., the acceleration of falling bodies due to gravity, and systematic changes in the perihelion position of Mercury. To that extent, we are semi-determinists. And we also hold that some events and processes are completely indeterministic, e.g., quantum mechanical phenomena in the microphysical world, and coin-flipping sequences in the macrophysical world. To that extent, we are semi-indeterminists. But we also hold that some events and processes in the natural world are natural causal singularities, e.g., the Big Bang, black holes, traffic jams, the weather, and the biological processes and endogenously produced overt movements of living organisms. Amongst the natural causal singularities are what we call self-determining events and processes—i.e., intentional body movements. So to that extent, and most importantly, we are also self-determinists. Sometimes Wittgenstein’s brilliant remark is read as support for compatibilism—the thesis that free will and determinism are consistent (weak compatibilism), and that both exist (strong compatibilism, a.k.a. ‘‘soft determinism’’). But as we have just seen, it is arguable that Wittgenstein was driving at something else altogether. An event or processes’s self-determining intentional predictability from the inside, as the event is actually unfolding does not entail complete determinism, even though it also rules out complete indeterminism. So because they are based on dynamic instability, self-determining events or processes are not completely deterministic; but because they are also predictable from the inside, self-determining events or processes are equally not completely indeterministic. For us then, ‘‘causal self-determination’’ means that an event or process has neither a closed future (as in universal natural determinism) nor an open future (as in universal natural indeterminism) precisely because, consistently with all the roughgrained general deterministic or statistical laws there are, it spontaneously, consciously, and intentionally naturally creates its own future by what it actually does in the ongoing, forward-flowing present. The larger metaphysical story about the mind–body relation and mental causation that is making up the backdrop to our remarks here will be spelled out in detail in Chapters 6 to 8. But just to present the essentials Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 122 122   : , ,   of our metaphysical story here as a preliminary soundbite, we will argue in those chapters that the correct description of such semi-deterministic, semi-indeterministic, and self-determining natural events is that they are (i) natural causal singularities, which is to say that they are uniquely new nomologically sufficient causes of physical events, that are also consistent with whatever prevailing roughgrained general deterministic or statistical laws there are, (ii) dynamically emergent, which is to say that they generate causally efficacious truly global or inherently dominating intrinsic structural properties of dynamic systems that are neither reducible to the intrinsic non-relational properties of their parts nor strongly supervenient on the intrinsic non-relational properties of their parts together with all their extrinsic relational properties, and (iii) intentional body movements of essentially embodied intentional agents. In this way, far-from-equilibrium or unstable natural dynamic phases in the life of an essentially embodied intentional agent are ‘‘Times to Dance,’’ and thereby spatiotemporal sites for the manifestation of natural creativity, of which intentional agency is only one special kind—although, obviously, the natural creativity of intentional agency is of great importance for minded animals, whether rational or non-rational, and whether human or non-human. In any case, our main point here is that we should think of a Davidsonian self-conscious deliberative intention as a normatively empowering but not causally overpowering state in a minded animal that cannot cause an action all by itself, and that instead requires some additional mentalistic factor to solidify the initial conditions of events and help constitute a new regime of natural stability. This new regime of natural stability, in turn, is nothing more and nothing less than a natural causal singularity in the essentially embodied agent’s neurobiological processes and overt body movements that, along with this additional mentalistic factor, jointly constitute the intentional action. The crucial additional mentalistic factor for constituting intentional action, we shall argue at some length in Sections 4.1 and 4.2, is trying and its active or initiating guidance of the agent’s own living animal body. But for the moment the crucial point is that if we are to tell an adequate Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:29 FN:30 11:28am Page 123 123 causal story about action, we cannot treat an action merely as a body movement that stands in an illuminating instrumentally and teleologically rationalizing relation to certain belief-desire pairs. We have to say just how the action itself happens, from the inside, as that event is actually unfolding in a naturally purposive way, by appealing to some mentalistic causal factor beyond Davidsonian reasons. As Searle correctly points out, choosing and deciding are themselves partially grounded in the presupposition that Davidsonian reasons are not in and of themselves causally sufficient for our actions.²⁹ If they were sufficient, then obviously we would always act once the appropriate beliefs and desires were in place, and nothing further would be required. But even given the appropriate beliefs and desires, it is a plain fact that we do not always act intentionally, and this in turn is not in every case just because we are prevented from acting by unpropitious circumstances, or compelled to act by coercion or overwhelming outer forces. Sometimes we just shy away from choice: here we either neglect to try or refuse to try. So there is in Searle’s or Eliot’s terminology as we have said, a Gap or Shadow—or as we think it should be more accurately described, a far-from-equilibrium or unstable phase, a Time to Dance, or a spatiotemporal site for the manifestation of natural creativity—in the dynamics of intentional action. But there is not only one kind of far-from-equilibrium or unstable phase in the dynamics of intentional action. Indeed, it appears that there are at least three different kinds:³⁰ (1) There is an instability between desiring to do something and deciding to do it. To want ice cream and believe that it is in the fridge is not yet to decide to go get some. I can just put off deciding. (2) There is another instability between deciding to do something and actually trying to do it. For example, I may decide to get out of bed many minutes before I actually make the effort to do so, or decide to get out of bed and then not make any effort whatsoever to follow up on this decision. (3) And there is also another instability between beginning to try to do something and carrying out that task to its completion.To see this, note that trying to run a marathon does not occur in one fell swoop. ²⁹ Searle, Rationality in Action, 71. ³⁰ Ibid. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 124 124   : , ,   To accomplish this task, even professional athletes typically need to try for over two hours. And, as all long distance runners intimately know, it is possible to give up at any point before the end of the race. FN:31 Some further factor is needed to stabilize each of these kinds of action-based instability. But it is not going to be reasons that do it. To be sure, Davidson admits that in order to be causally effective, reasons must be appropriately connected to resulting actions, and there must be no deviant causal chains at work. However, Davidson does not say enough about the nature of this appropriate connection, nor indeed does he ever acknowledge that there is an inherently action-based instability that needs to be stabilized—much less, three different kinds of instability in need of stabilization. If it is trying that is additionally required in order to cause intentional actions, then we will need a detailed account of the nature and causal role of trying. As a methodological bridge to such an account, Searle’s distinction between ‘‘prior intentions’’ and ‘‘intentions-in-action’’ is instructive. While prior intentions are the plans that one often has before undertaking some action, the intention-in-action is the intention one has while actually performing the action.³¹ In many cases, we first form a prior intention, and then perform the whole action, which consists of the intention-in-action together with the overt intentional body movement. But in cases of actions that are not premeditated, there is no prior intention, but only an intentionin-action. Therefore some actions are done intentionally even though the actor has formed no prior intention and no prior plan or self-reflection is involved. Such a view seems to reflect a synchronous causation model: Searle seems to be saying that intentions-in-action are how an agent is in touch with her body during and throughout basic action. But whether or not Searle himself would want to frame it this way, this is the account we will adopt. Our idea is that the mental causes of action—the conscious, intentional mental states, events, and processes of trying and its active guidance—are synchronous with the neurobiological processes and overt body movements that essentially embody those mental activities, and, indissolubly together with those mental activities, jointly constitute intentional actions by virtue of the diachronic instantiation of truly global or inherently dominating, intrinsic structural, irreducibly mental properties ³¹ Searle, Rationality in Action, 44. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:32 11:28am Page 125 125 of the agent’s living animal body. In the intentional act of dancing, then, the conscious mind of the dancer simultaneously structures her dance. It is true that there is a purely superficial, although real enough, time-lag between the time at which the relevant neurobiological process starts in the brain—as indicated, e.g., by neural imaging devices—and the time at which the relevant overt body movement starts at the muscle tissue/skin interface. But this time-lag is entirely due to the fact that it takes time for a neurobiological process to be propagated through all the vital organs and vital systems and become fully displayed as an overt body movement. And thus this time-lag is not a temporal difference between an antecedent mental cause that expires before (or just as) its effect begins, and a later bodily event that is its effect, as in classical volitional-causal theories of action. On the contrary, on our view the mental cause is right there at the very beginning of the intentional action in the form of a conscious state, event, or process of trying that is essentially embodied in neurobiological processes; it is right there throughout the development of the neurobiological process in the form of trying’s active or initiating guidance as it controls the overt intentional body movements that arise from that neurobiological process and accompany it; and it is still right there at the end of the relevant sequence of neurobiological and overt movements, as contained in trying’s active guidance of the agent’s own body through her successful completion of the entire action. Therefore the mental cause is synchronous with all of the intentional movement’s constituent phases. To return again to the example from Yeats, in the intentional act of dancing the dancer’s conscious mind simultaneously structures the whole dynamic natural event of her dance. As we have said, we will recapitulate and argue for this doctrine at some length in Sections 4.1 and 4.2, and then give a metaphysical analysis of it in Chapters 6, 7, and 8. In this connection, Mele argues that intentional body movements are causally initiated by proximal intentions, that is, ‘‘intentions to A straightaway,’’³² and that these proximal intentions also play a persisting role in causally sustaining the relevant bodily motions. Sensorimotor feedback indicates whether one’s body is moving according to plan or whether things are veering off course and require correction of bodily motions. A plan embedded in the agent’s persisting proximal intention is what provides the ³² Mele, Motivation and Agency, 54. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 126 126   : , ,   instructions for such corrections, so that intention can play a central role in the causal guidance of ongoing bodily motions. We think that Mele’s thesis that a persisting proximal Davidsonian intention can sometimes causally sustain action by providing an empowering plan is quite plausible, and that it improves Davidson’s theory. Nevertheless, we also need to be careful not to construe intentional body movement in an overly intellectualist way, and also not to think that persisting proximal Davidsonian intentions are universally necessary for intentional body movement. 3.4 Against Davidson 3: Actions without Reasons FN:33 FN:34 Davidson’s theory holds that intentional action is always and essentially a self-conscious or self-reflective and deliberative affair, carried out by means of causally responsible instrumental reasons. But one obvious objection to this view is that many non-human animals and all neurobiologically normal human infants—none of whom can be plausibly taken to be self-conscious or self-reflective, deliberative agents capable of forming or recognizing instrumental reasons—are nevertheless also minded animals who can act intentionally. Of course, we who are members in good standing of the Universal Community of Rational Animals or Real Persons can cognitively project instrumental rationality onto their body movements, and admit them as more or less permanent associate members. But as Dennett has pointed out, the same cognitive projection of instrumental rationality could in principle be extended to all sorts of machines and other non-animals.³³ So even if this cognitive projection is both informative and useful to us, it does not at all follow that minded non-human animals and human infants actually operate by means of instrumental rationality, unless we have independent reason to believe that they possess the psychological capacities that would support this sort of activity. Notoriously, Davidson holds that only ‘‘talking animals’’—i.e., linguistically competent animals—are capable of thought and intentional action.³⁴ Moreover, he argues as if the conditions for our ascribing instrumental rationality to talking animals are the same as the conditions for their actually having instrumental rationality. But short of an independent and sound argument ³³ Dennett, The Intentional Stance. ³⁴ See Davidson, ‘‘Thought and Talk.’’ Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:35 11:28am Page 127 127 for anti-realism—the thesis that the truth-conditions of our judgments depend essentially on the conditions under which those judgments can be asserted by us—Davidson is not entitled to this identification. And the available evidence very strongly suggests that while many non-human animals and all normal human infants are indeed conscious, intentional agents, nevertheless they are not capable of practical reasoning, since this also requires logical reasoning capacities and conceptual capacities, which are intrinsically bound up with linguistic competence,³⁵ which of course they do not possess. So contrary to Davidson’s theory of action, it seems far more plausible to hold that many non-human animals and all normal human infants operate as conscious, intentional agents fundamentally on the basis of desire-based emotions, together with whatever else it takes to cause actions. Moreover, even when we focus exclusively on agents who are capable of self-conscious or self-reflective, deliberative action via instrumental reasons—rational animals or real persons, whether human or nonhuman—there are at least three important types of action that Davidson fails to acknowledge, none of which actually requires self-conscious or self-reflective, deliberatively formed or recognized instrumental reasons for action. Let us now look at these three types in turn. (3.4.1) Pre-Reflective or Spontaneous Actions. FN:36 FN:37 Having said what we just said, we clearly need to explore further the role that intentions typically are believed to play in action. Davidson claims that a person does not perform an action unless his movement occurs as a result of his relevant beliefs and desires, and that an individual is the agent of an act if and only if what he does can be described under an aspect that makes it intentional.³⁶ But what does it mean to say that all action is intentional under some description? We cannot suppose that whenever an agent acts intentionally, he necessarily goes through a process of self-conscious or selfreflective, deliberative instrumental practical reasoning.³⁷ Davidson notes that an individual may intend to do X without having decided to do X, deliberated about X, reasoned about X, or formed an intention to do X. What Davidson calls ‘‘pure intending’’ is an unconditional judgment that ³⁵ See Hanna, Rationality and Logic, chs. 4–6. ³⁷ Davidson, ‘‘Intending,’’ 85. ³⁶ Davidson, ‘‘Agency,’’ 46. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 128 128   : , ,   FN:38 FN:39 FN:40 FN:41 an action of a certain sort is desirable, and he claims that such judgment can occur without practical reasoning, action, or consequences.³⁸ He also holds, however, that if someone acts with an intention, he must have attitudes and beliefs from which, had he been aware of them and had sufficient time, he could have reasoned that his action was desirable. Note that, according to Davidson, while one does not act intentionally when one makes a mistake, one nonetheless acts. This is because in cases where I make a mistake and fail to achieve my goal, I nevertheless intentionally do something. It is simply that I have done something with the intention of achieving a result that is not forthcoming. Here the question of whether an act was intentional or not quickly becomes very muddled. We believe that some clearer distinctions concerning intention and intentional doings ought to be made. First, Harry Frankfurt introduces an illuminating, although terminologically somewhat clumsy, distinction between ‘‘intentional action’’ and ‘‘intentional body movement.’’³⁹ According to Frankfurt, intentional action is self-conscious or self-reflective, deliberative action resulting from instrumental reasoning based on desires; but the occurrence of intentional body movement need not be intended by the agent by way of either self-conscious or self-reflective forethought or reflective assent.⁴⁰ This distinction captures Davidson’s recognition that even when I fail to achieve what I intend in a self-conscious or self-reflective, deliberative way, I may nevertheless intentionally do something or other. Unlike Frankfurt, we use the term ‘intentional action’ more broadly so that it includes all cases of intentional body movement, as well as all cases of self-conscious or self-reflective, deliberative action. Moreover, as we shall argue below, not all self-conscious or self-reflective, deliberative actions motivated by desires are based on instrumental reasons. But at the same time we fully endorse the basic upshot of Frankfurt’s distinction between the two types of intentional action: it is possible for an agent to do something intentionally, without doing it as a result of self-conscious or self-reflective, deliberative intentions. Along these same lines, Michael Bratman points out that there is a distinction to be made between ‘‘intentionally A-ing’’ and ‘‘intending to A.’’⁴¹ To illustrate this, he asks us to imagine a marathon runner who ³⁸ Davidson, ‘‘Intending,’’ 101. ³⁹ Frankfurt, ‘‘The Problem of Action,’’ 79. ⁴¹ Bratman, ‘‘Two Faces of Intention,’’ 179. ⁴⁰ Ibid., 74 Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:42 FN:43 FN:44 FN:45 11:28am Page 129 129 gradually wears down her shoes over the course of a race. He suggests that while she runs the marathon, she can intentionally wear down her sneakers even if she did not intend to wear them down.⁴² This is because what one intends to do is a matter of one’s future-directed conduct and what one plans to do, while intentionally moving one’s body need not be a matter of plans and explicit purposes. More generally, it seems quite obvious that minded animal intentional agents do many things intentionally without in any way self-consciously or self-reflectively and deliberatively intending to do them. When one walks to work, e.g., although each of one’s steps along the way are intentional movements, it seems clear that each step does not require its own distinct, self-conscious or self-reflective, deliberative prior intention, plan, or goal.⁴³ If our efforts always required this degree of self-consciousness or selfreflection, then the fluidity of behavior that we so often see would surely be mostly absent. On our view, then, the paradigmatic and universal sort of intentional action is intentional body movement, and not self-conscious or self-reflective, deliberative intentional action. Whenever we engage in intentional body movements, we intentionally act by moving our bodies—but we need not intend to move them in the sense that we self-consciously or self-reflectively plan to move them. Second, it seems that we need not have any Davidsonian reason whatsoever in order to move our bodies. To see this, consider the class of ‘‘intrinsically motivated actions’’ that are performed only for their own sake and not for the sake of some further goal or purpose.⁴⁴ For example, a person who absent-mindedly hums or drums her fingers on the table normally does so intentionally, even though she has no further goal or ‘‘end-directed intention’’: the reason for action is simply an intrinsic desire, fundamentally based on our primitive bodily awareness, to hum or drum. Otherwise put, the agent hums or drums just because she suddenly just feels like doing it at that very moment. While Mele admits that some intentional movements are not done for reasons that involve a belief component of the sort required by Davidson,⁴⁵ he believes that this is merely a technical problem for Davidson’s account that can be remedied by granting that intrinsic desires to A are themselves reasons for A-ing. But is intrinsically motivated action really intentional in the Davidsonian sense? That is, even supposing ⁴² Ibid., 199. ⁴³ Mele, Motivation and Agency, 205. ⁴⁴ Ibid., 71. ⁴⁵ Ibid., 73. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 130 130   : , ,   counterfactually that the agent had been self-consciously and reflectively aware of what he was doing, would he necessarily have judged that his action was desirable? Our answer is: No. It seems clear to us that intrinsically motivated actions are done for no instrumental reason at all, and that this is sufficiently shown by the dual fact that (i) the agent acts pre-reflectively or spontaneously without a selfconscious or self-reflective, deliberative intention, or because the agent suddenly just feels like doing it at that very moment, and (ii) the action would not necessarily have been judged desirable either at the moment of action or upon reflection by the agent performing it. FN:46 For obvious reasons, let us call all such actions pre-reflective or spontaneous actions. Here are some examples: humming the faintly annoying Speedy Muffler jingle that you heard on radio and television commercials circa 1976; drumming your fingers on the desk while talking on the telephone; idly wiggling your toes while reading a book; suddenly frowning when the sun goes behind the clouds; biting your fingernails while working on your tax returns; throwing your cell phone across the room in a fit of anger; jumping up and down as an expression of excitement; throwing your hands in the air while freestyle hip-hop dancing; and (making an abrupt Nietzschean shift from the dionysian to the tragic) rolling in the clothes of one’s dead wife as an expression of grief. Rosalind Hursthouse calls these ‘‘arational actions,’’ and very plausibly claims that they are all intentional actions explained by their intrinsically resulting from occurrent emotions—or as we would put it, by occurrent desire-based emotions.⁴⁶ Hursthouse’s term ‘arational’ is arguably a misnomer, however, in that such actions are done intentionally by agents who are fully sapient and sane. To be sure, there is a sense in which such actions, considered as act-tokens, could be judged to be less than optimally rational from the normative standpoint of instrumental rationality. In that context, the act-token would not necessarily be judged to be in the agent’s best self-interest. But unless the normative standard of optimal act-token ⁴⁶ Hursthouse, ‘‘Arational Actions,’’ 58. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:47 FN:48 11:28am Page 131 131 instrumental rationality must be privileged over every other sort of rationality (and it seems plausible to us that it need not be⁴⁷), then because in all other respects these acts are rational, they are authentically rational. The everyday phenomenon of losing one’s temper is a perfect example. (We are not of course denying that it is possible to have ‘‘an anger management problem’’—i.e., to have a pathological tendency to experience uncontrollable fits of temper in inappropriate circumstances. We are here talking about normal anger.) To run a minor variation on a famous observation by Hume,⁴⁸ ’tis not contrary to reason to prefer the non-negligible loss of utility consequent upon smashing my expensive cell phone to bits, to my not flinging it across the room. In such cases, we will be strongly inclined to say that while the action was intentional, and while the agent was fully sapient and sane, nevertheless she did not do it for any further goal or purpose. There is no need to ascribe a suitable self-conscious or self-reflective belief or desire, precisely because the very fact that the agent was in the grip of some strong desire-based emotion adequately explains the action. She is a rational animal who acts that way just because she suddenly just feels like doing it at that very moment. End of story. Here someone might object that the agent desires to have and to express her desire-based emotion, and also believes that her action will express it. However, it seems very unlikely that in every or even most such cases the agent has a distinct self-conscious or self-reflective desire that is separate from the desire just to, e.g., fling a cell phone across the room. Rather than acting in order to express an emotion, it seems more reasonable to suppose that the agent’s action just is the expression of a certain essentially embodied, pre-reflective desire-based emotion: in this case, being completely pissed off. No doubt it is true that a desire-based emotion itself can count always as a reason for some action or another, and that someone in a desire-based emotional state will tend to have the further self-conscious or self-reflective beliefs and desires typically associated with that state. But it seems obvious that in many cases such beliefs and desires do not count as any sort of instrumental means-ends rationalization for an agent’s actual desire-based emotional behavior. So in desire-based, emotionally-driven, pre-reflective or spontaneous action, there need not be some associated ⁴⁷ See, e.g., Hanna, Rationality and Logic; and Searle, Rationality in Action. ⁴⁸ See Hume, Treatise of Human Nature, book II, part III, section iii, 416. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 132 132   : , ,   FN:49 FN:50 FN:51 occurrent or even dispositional belief that acting in such a way will serve some goal, or even bring one pleasure.⁴⁹ Consider again being completely pissed off and then throwing your cell phone across the room. Surely you do not do so with the belief that you will advance some goal. In fact, given the high probability that you will destroy your expensive and highly stylish little communicator, you clearly have much more reason not to throw it. And in fact if you reflectively backed off and considered your own desire-based emotion dispassionately, you probably even would desire not to throw your cell phone across the room. Similarly, consider the individual who tears out pages of his mathematics textbook in frustration. In carrying out actions such as these, agents do not have any instrumental purpose or aim at some goal. It is not the case that they desire A and believe that performing B is the best way to achieve A. Moreover, it is not the case that their purposes and aims are bizarre, or that their reasons are bad, but rather simply that any instrumental reasons in Davidson’s sense for doing that act simply either do not exist, or at least are psychologically suspended. After all, what instrumental purpose could an agent possibly have in tearing out the pages from his mathematics textbook? After doing so, and in a different mood, the agent in all likelihood would judge that his action was not desirable and that it was unlikely to help him do any better at maths. In other words, there is nothing ‘‘to be said for’’ the action ‘‘from the agent’s point of view’’⁵⁰ —except, and this will make all the difference, that it is uniquely the agent’s own desire-based emotion that is being expressed. This is particularly clear in the case of pre-reflective or spontaneous acts of self-expression. Consider again freestyle hip-hop dancing, and someone’s suddenly throwing her hands in the air. She acts that way precisely because she wants to give bodily expression to the desire to throw her hands in the air. In effect, she is trying on this desire-based emotion for size, so that the act is then a bodily picture of that very desire-based emotion. This nicely supports Wittgenstein’s famous remark in the Philosophical Investigations, often misinterpreted as behaviorism: ‘‘The human body is the best picture of the human soul.’’⁵¹ In his philosophical terminology, carried over from the Tractatus, a ‘‘picture’’—as opposed to an ‘‘image’’—captures the internal ⁴⁹ Hursthouse, ‘‘Arational Actions,’’ 63. ⁵⁰ Davidson, ‘‘Actions, Reasons, and Causes,’’ 32. ⁵¹ Wittgenstein, Philosophical Investigations, 178e . Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:52 FN:53 11:28am Page 133 133 structure of a real fact.⁵² So if the body is the best picture of the soul, then the intentional agent’s body in motion is internally structured by the agent’s effective desire to move her body in just that way. The freestyle hip-hop dancer gives immediate bodily expression to her desire to dance, and thereby shows that pre-reflective or spontaneous intentional action is possible. So while Davidson believes that an agent who wants to do X also believes that he ought to do X and that it is desirable to do X, the examples we have described in the last few paragraphs clearly show that his depiction of desire as instrumentally evaluative in nature puts too much emphasis on the role of self-conscious deliberation and instrumental reasoning in producing intentional movement. On Hursthouse’s view, which we share except for the terminological quibble we mentioned earlier, the cases described above form a single class of actions that are neither intentional actions in Davidson’s preferred full-blown sense nor unintentional body movements—they are neither planned nor accidental. This, again, is the class of pre-reflective or spontaneous actions, which inherently involve intentional body movements in the sense that the agent controls her own body. The body movements are neither random nor the result of inner or outer compulsion. The agent is fully sane and her rational capacities are all online. As David Velleman aptly puts it, the action is seemingly ‘‘effortless’’⁵³ —although as we will argue in Chapter 4, this effortlessness is perfectly consistent with its also being the result of a certain kind of trying. It is true that, as act-tokens, such pre-reflective or spontaneous actions would not necessarily be judged to be perfectly rational by the normative standards of instrumental rationality. But they are still authentically rational intentional acts. It is just that this sort of rational intentional act is the intrinsic result of a desire-based emotion. Our conclusion is that for rational minded animals, pre-reflective or spontaneous actions are both rationally and intentionally done, and therefore are intentional actions. The Davidsonian theory of intentional action is too narrowly committed to self-conscious or self-reflective, deliberative instrumental rationality. The intentional movements of an essentially embodied rational agent can be authentically intentional without her having any further purpose or primary reason in Davidson’s sense. ⁵² See Wittgenstein, Tractatus, 39–43, props. 2.1 to 2.225; Wittgenstein, Philosophical Investigations, 101e , §301. ⁵³ Velleman, ‘‘The Way of the Wanton.’’ Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 134 134   : , ,   (3.4.2) Akrasia and Impulsiveness of the Will FN:54 FN:55 FN:56 FN:57 Cases of akrasia, or so-called ‘‘weakness of the will,’’ likewise show that the causes of action need not involve reasons in the way that Davidson’s theory implies. Davidson holds that if an agent wants to do X more than he wants to do Y and is free to do either, then he will intentionally do X if he does either X or Y intentionally.⁵⁴ He also holds that if an agent judges that it would be better to do X than do Y , then he wants to do X more than he wants to do Y . However, he notes that these two theses are inconsistent with the existence of akratic actions. In order to address this difficulty, Davidson suggests that while the akratic person regards one course of action as better (for a reason), he nevertheless does something else (for a reason).⁵⁵ That is, while the akratic person judges that all things considered, it would be better to do X than Y , he can nevertheless unconditionally judge that doing Y is better than doing X. And while it is only unconditional judgments that can actually bring about intentional action, irrational unconditional judgments may sometimes lead to akratic action. Davidson asserts that ‘‘there is no paradox in supposing a person sometimes holds that all that he believes and values supports a certain course of action, when at the same time those same beliefs and values cause him to reject that course of action.’’⁵⁶ But by reflecting on our own actions and by looking around at everyday life, it seems to us that akratic actions are happening all the time. So we find it strange to suppose that agents’ all-considered-judgments conflict with their unconditional judgments on a regular basis, for this would suggest that a serious form of cognitive and practical dissonance, or irrationality, is an everyday occurrence. On the contrary, everyday life seems to be awash with cases of non-irrational akratic action in which fully sapient, sane agents unconditionally judge that it would be better to do A than B, are free to do A, and yet spontaneously intentionally do B.⁵⁷ For example, Theresa can fully believe that it would be better not to smoke, be entirely free from compulsion, be perfectly rational and sane, yet just haul off and smokes anyway. Furthermore, the fact that her smoking is an intentional action does not entail that she intends to act akratically, for Theresa may intend ⁵⁴ Davidson, ‘‘How is Weakness of the Will Possible?,’’ 23. ⁵⁵ Ibid., 34. ⁵⁷ Mele, Irrationality: An Essay on Akrasia, Self-Deception, and Self-Control, 43. ⁵⁶ Ibid., 41. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:58 11:28am Page 135 135 and plan not to smoke, but then just smokes anyway. Similarly, while an unwilling drug addict, also perfectly rational and sane, but struggling against his addiction, may neither want to desire heroin nor self-consciously and deliberatively intend to take heroin, and only do so as a result of his addiction, nonetheless his movements as he sticks the syringe into his arm are intentional movements, spontaneously (although in this case of course unfortunately) chosen by him. These acts are purely hedonic, but neither egoistic nor self-interested. He knows that his heroin addiction is a truly terrible thing for him personally. So we are saying that actions can be perfectly sane and rational, intentional, akratic, and neither egoistic, nor in what the conventional wisdom takes to be the agent’s best interest, nor even in what the agent himself takes to be his own best self-interest. This again highlights the fact that self-conscious or self-reflective, deliberative plans and purposes that result from the desires rational agents deem best, are not in any way necessary for intentional movement. We can act pre-reflectively and spontaneously. But it also discloses the deeper and perhaps surprising fact that rational agency and so-called weakness of the will are intrinsically connected. We also can act impulsively, against either what is, or appears to be, own best interest. And thank god for that, since it is obvious that acting in the service of either what is or appears to be our own best interest is not always the best way to act. In this way, to take akrasia seriously is to open up our conception of rational agency. More precisely, cases of akrasia reveal that an agent’s current Davidsonian or self-conscious or self-reflective, deliberative intentions can be easily defeated or deflected by occurrent opposing wants and desires, and that making an unconditional judgment is not a causally sufficient condition for performing a particular action. Because one’s self-conscious deliberative evaluation of a particular desire need not match the motivational force of that desire, it is obvious that instrumental reasons and self-conscious or self-reflective, deliberative intentions do not tell the full story.⁵⁸ For example, while Janet may think it best to satisfy her desire to write a philosophy paper this afternoon, her desire to go out drinking with her friends may simply turn out to have much more motivational force, in which case she will carry out the impulsive act of going out drinking with ⁵⁸ See also ibid., 84. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 136 136   : , ,   FN:59 FN:60 her friends. Similarly, people sometimes impulsively fall passionately and truly in love with one another entirely against their own better judgment. Such cases also show that the desires we do judge, or would judge, to be worse often exhibit much greater motivational force than the desires that are in line with our self-conscious deliberative instrumental judgments, prior intentions, decisions, and plans. So this is in direct conflict with Davidson’s claim that necessarily if one judges it would be better to do X than Y , then one wants to do X more than one wants to do Y .⁵⁹ In this way, whereas Davidson holds that desire and self-conscious or self-reflective, deliberative instrumental evaluation are intrinsically linked, we believe on the contrary that the everyday, widespread fact of akrasia makes it very likely that a narrowly instrumental rationalist view of action is simply incorrect. But this is not an attack on the very idea of rational agency. Given the widespread fact of akrasia, it is not the case that we are really less rational than Davidson takes us to be, nor is it the case that we are somehow fundamentally irrational, nor even that we are fundamentally arational. It is rather that our rationality, like our consciousness, is essentially embodied and thereby grounded on desire-based emotion, and so intrinsically open to akrasia. In this way, our rationality is intrinsically dynamic and impulsive: that is, we are rational animals whose nature is such that, when push comes to shove, we can just haul off and ignore, override, or reject our selfish or best or all-things-considered self-conscious or self-reflective, deliberative instrumental reasons. This is perhaps most obvious and plausible to us when adopting such instrumental reasons would lead to morally impermissible acts, as in the standard Kantian counterexamples against ethical egoism, where the impulse to ignore, override, or reject our selfish or self-interested instrumental reasons is a necessary condition of moral autonomy.⁶⁰ But the same point holds in certain consequence-independent cases in which both we ourselves and everybody else would in fact be much better off if we adopted an instrumental reason and acted that way. Consider, e.g., (to use a minor variant on Bernard Williams’s famous example of ‘‘George the chemist’’) an unemployed chemist, call him George∗ , who impulsively refuses to take a job doing chemical and biological warfare research, even though ⁵⁹ Davidson, ‘‘How is Weakness of the Will Possible?’’ 23. ⁶⁰ See Kant, Groundwork of the Metaphysics of Morals, sections I–II. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:61 FN:62 11:28am Page 137 137 his family badly needs the money, and even though he will actually then be in a position to make the research go more slowly than other keener candidates.⁶¹ In such cases, adopting the instrumental or consequentialist reason as our own would violate our fundamental projects and our personal integrity, and so we feel deeply in our guts and hearts that we must ignore or reject it, and so act on an impulse contrary to or without instrumental reasons, and independently of the consequences. In other words, we non-instrumentally plump for that action. This sort of non-instrumental, non-consequentialist impulsiveness or ‘‘plumping’’ is not irrationality or arationality, precisely because it preserves our psychological authenticity (the internal coherence of our desires, beliefs, volitions, and actions), which in turn is a necessary condition of our rational autonomy. Akrasia is thus a partial expression of our deepest freedom of the will. If this is correct, then it is not accurately described as any sort of volitional failure. So akrasia should not be thought of as ‘‘weakness of the will,’’ which seems to reflect a much more modern, sanctimonious, and sternly moralistic notion than the ancient Greeks actually had in mind, but instead as impulsiveness of the will. To be sure, free will itself is not the same as impulsiveness of the will. The complete fact of free will is much more than that, arguably including negative freedom (freedom from preventative checks or overriding compulsion), positive freedom (freedom to choose or do what I want), psychological freedom (the subjective experience of being negatively and positive free), deep freedom (transcendental freedom or original ‘‘up-to-me-ness’’), moral responsibility, moral authenticity or integrity and rational autonomy (self-legislation).⁶² But at the same time, free will nevertheless includes impulsiveness of the will as a necessary condition. (2.4.3) Desire-Overriding Reasons. In a closely-related way, what seems to us to be the obvious fact of noninstrumental reasons for action also poses serious problems for Davidson’s account of action. Following a broadly Kantian line, Searle, e.g., defends the possibility of non-instrumental rationality and maintains that there are motivations that lead agents to do things that they would honestly say they ⁶¹ See Smart and Williams, Utilitarianism: For and Against, 97–8. See also Sartre, ‘‘Existentialism is a Humanism’’; and Williams, Moral Luck. ⁶² See Kane, A Contemporary Introduction to Free Will. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 138 138   : , ,   FN:63 FN:64 do not want to do.⁶³ One can eat carrot sticks, brush one’s teeth, or behave morally even when one seemingly (or at least initially) has no desire to do so. Searle argues there are things we find ourselves obligated to do whether we want to at that moment or not, and that in virtue of certain speech acts, we create ‘‘desire-independent’’ reasons. When a speaker makes an assertion, he creates a non-instrumental reason for accepting the logical consequences of what he has just said.⁶⁴ Likewise, when one makes a promise, one commits oneself to carrying out what one has promised, so that the making of various commitments is built right into the structure of speech acts. Searle believes that insofar as such commitments create non-instrumental reasons for action, obligations create a species of action in which occurrent desires do not play a central causal role. Correspondingly, it may seem that some actions are fully explained by evaluative beliefs about what one should do. Louise may invite Marty to the party simply because she believes it is the right or appropriate thing to do, and not because she wants to invite him or would enjoy seeing him there. These sorts of non-instrumental reasons appear to pose a challenge to the standard Humean–Davidsonian account, which sees all actions as explicable by pairs of desires and means-ends beliefs. While we fully agree that the Humean–Davidsonian account requires significant revision, we also believe that Searle’s characterization of noninstrumental reasons as ‘‘desire-independent’’ is significantly misleading. In this context, it seems that the notion of ‘‘the desire-independence of a reason’’ can mean either (1) a reason that is exclusive of any and all desires (strong desireindependence), (2) a reason that is underdetermined by any and all desires even if it is always associated with desires (moderate desire-independence), or FN:65 (3) a reason that is underdetermined by certain first-order desires, even if it is always associated with desires (weak desire-independence).⁶⁵ ⁶³ See Searle, Rationality in Action, ch. 6; Kant, Groundwork of the Metaphysics of Morals; and Kant, Critique of Practical Reason. ⁶⁴ Searle, Rationality in Action, 173. ⁶⁵ The notion of underdetermination in this connection is most easily explicated as non-supervenience, such that X underdetermines Y if and only if Y does not supervene on X. See Section 1.1. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 139 139 The strong desire independence, or desire-exclusion, of a reason means that the reason bears no relation whatsoever to any and all desires. For example, a divine or angelic being might have a reason for acting that is strongly desire-independent. The moderate desire-independence, or universal desire-underdetermination, of a reason means that the reason bears no intrinsic or necessary relation to any desires whatsoever, although it could also still bear extrinsic or contingent relations to some or even all of the agent’s desires. For example, an indeterministic agent cause or a radical existential voluntarist, who is somehow capable of making choices or decisions while standing apart from all his own desires, might have a reason for acting that is moderately desire-independent. But the weak desire-independence, or specific-desire-underdetermination, of a reason as we will understand it, means only that the reason is not necessarily related to certain first-order desires. This leaves open the possibility that reasons are still always justified and motivated by facts about our desires, even though we are not always justified or motivated to act by a special class of first-order desires. For example, we could sometimes be justified and motivated to act by a special class of higher-order desires (say, the desire to be moved at time t by a non-selfish, non-egoistic, non-self-interested, non-hedonistic, and consequence-independent concern for others) that are directly in opposition to certain first-order desires (say, any selfish, egoistic or self-interested, hedonistic, or consequence-driven first-order desire at t), and so adopt a reason that is justified and motivated by facts about higher-order desires, but not justified or motivated by facts about all first-order desires. So while desire-independence in the first sense entails desire-independence in the second and third senses, the converse is not the case. More generally, it seems that Searle is implicitly equivocating between these three logically distinct kinds of desire-independence, and also that most of what he says is in fact consistent with weak desire-independence, and thus consistent with a hierarchical desire-based approach to practical reasons. But in order to sort things out properly, we will first have to say something about our own general approach to the nature of reasons. It seems plausible to hold that reasons are (or are provided for agents by) facts that normatively support (and thus justify or motivate) beliefs and intentional aims or actions, and do not merely cause or mechanically trigger those beliefs, aims, or actions. It remains possible, however, for reasons to be Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 140 140   : , ,   at least proper parts of mental causes of those beliefs, aims, or actions—e.g., in self-conscious deliberative intentional actions. Reasons for beliefs are epistemic reasons, and reasons for intentional aims or actions are practical reasons. Now there may be deep connections between epistemic reasons and practical reasons. For example, Searle has argued that epistemic reasons are a sub-class of practical reasons—and if so, then epistemic reasons are nothing but practical reasons for undertaking the intentional act of believing. For the rest of this sub-section, however, we will concentrate exclusively on practical reasons. Internalism about reasons says that reasons both justify and motivate our intentional aims or actions. So all practical reasons are internal reasons. Internalism normally entails a desire-based theory about the nature of justifying reasons. According to that theory, justifying reasons are (or are provided for agents by) facts about the desires of persons. By contrast, Externalism about reasons says that while all reasons justify our intentional aims or actions, nevertheless at least some and perhaps all reasons fail to motivate our aims or actions. So some or all practical reasons are external reasons. Externalism normally entails an objective-value-based theory of the nature of justifying reasons, according to which justifying reasons are (or are provided for agents by) facts about the ends or objective values recognized by persons, and not by facts about their desires. The primary philosophical virtue of Internalism about reasons is that it offers a very plausible account of action. But its primary philosophical vice is that it cannot account for desire-overriding non-instrumental or consequence-independent justifying reasons, since it is normally assumed that all desire-overriding non-instrumental, consequence-independent justifying reasons must also be desire-independent. Correspondingly, the primary virtue of Externalism about reasons is that it can account for desireoverriding non-instrumental, consequence-independent reasons. But its primary philosophical vice is that it cannot plausibly account for how justifying reasons can cause actions if they are not based on desires. The view of reasons we want to defend is a non-standard form of Internalism about reasons that also fully account for desire-overriding non-instrumental, consequence-independent justifying reasons. It does this by holding that all reasons are (or provided for agents by) facts about the desires of persons, including not only all instrumental, consequence-dependent reasons, but also some desire-overriding non-instrumental, Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 141 141 consequence-independent reasons. And this, in turn, is because at least some reasons express some essentially non-selfish non-egoistic or non-self-interested, non-hedonistic, and consequence-independent desires that can themselves override our selfish, egoistic or self-interested, hedonistic, and consequence-dependent first-order desires. So we call our view about practical reasons Desire-Overriding Internalism. More explicitly, according to Desire-Overriding Internalism about reasons, Thesis A: while all reasons are both justifying and motivating, and all reasons are (or are provided for agents by) facts about the desires of persons, nevertheless it is also the case that Thesis B: while many or even most reasons are instrumental and consequence-dependent, at least some reasons are desire-overriding non-instrumental, consequence-independent reasons, precisely because they are (or are provided for by) facts about some essentially non-selfish, non-egoistic or non-self-interested, non-hedonistic, and consequenceindependent desires of persons. FN:66 But how can Desire-Overriding Internalism be true? How can some of our desires ground desire-overriding non-instrumental, consequenceindependent reasons? The answer is that if one adopts, as we do, a version of the hierarchical desire model of the will and personhood first developed and defended by Frankfurt,⁶⁶ then Desire-Overriding Internalism about reasons is perfectly coherent and possible. But before we unpack this idea, we will need to pause very briefly to make some distinctions and acquire some terminology. A desire is a felt need for something, or a preference for something, or a wish for something. Desires and wants can be taken to be essentially equivalent. To desire X is to want X; to desire to X is to want to X; and conversely. Now according to Frankfurt (and we fully agree), some conscious animals have not only first-order desires but also effective first-order desires. Effective first-order desires are desires that move (or will move, or would move) the conscious animal all the way to action. An effective first-order desire, that is, is the same as a conscious animal’s will. For example, I have an effective first-order ⁶⁶ Frankfurt, ‘‘Freedom of the Will and the Concept of a Person.’’ Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 142 142   : , ,   desire for a cold beer that gets me off the couch and all the way to the refrigerator, the bottle opener, and finally to guzzling that beer, and so that is what I have willed, or is my will, on that occasion. First-order desires may or may not also be accompanied by second-order desires. In having a second-order desire, I want (not) to want X, or to want (not) to want to X. For example, I may get a beer from the refrigerator and drink it because I effectively desire one, but also at the same time I might want not to want that beer, because I know that drinking it (and its friends) will make me podgy and sleepy, when I should on the contrary be skinny and alert. Now suppose, counterfactually, that my wanting not to want that beer actually stops me from getting up off the couch. In this way, at least some of my second-order desires can be directed to the determination of precisely which first-order desire is the effective first-order desire, or my will, and such desires are second-order volitions. This in turn leads to an account of personhood framed primarily in terms of desires and the will, and only secondarily in terms of rationality: Real persons are all and only the essentially embodied conscious, intentional creatures—i.e., minded animals—capable of having second-order volitions. The capacity for having second-order volitions also automatically confers a capacity for instrumental rationality on any creature that has that capacity. But it does not thereby confer a capacity for self-conscious or self-reflective, autonomous, non-instrumental rationality. According to our account, then, necessarily all real persons are essentially embodied rational animals, but not all real persons are self-conscious or self-reflective, autonomous, non-instrumentally rational animals. For example, normal small children between the ages of two and five (a.k.a. ‘‘toddlers’’) are obviously not self-conscious or self-reflective, autonomous, non-instrumentally rational animals, although they are real human persons. Here we differ substantively from Frankfurt. He holds that some rational animals are not persons because they are ‘‘wantons’’ and (temporarily or permanently) incapable of having second-order volitions. This seems wrong. On our view, essentially embodied rationality is inherently related to intentional agency, and entails real personhood. Frankfurt also holds that small children and all non-human animals are wantons. This too seems wrong. Toddlers, e.g., are obviously not self-conscious or self-reflective, autonomous, non-instrumentally rational animals. But they are also certainly meta-conscious (i.e., they are conscious of their own consciousness), Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 143 143 instrumentally rational (i.e., they very often know exactly what they want, and intentionally pursue their goals with great intensity and stubbornness), and highly willful (i.e., their wills are highly impulsive). Hence toddlers are capable of second-order volitions, and are real persons. And the same seems true of at least some non-human animals, e.g., Great apes. In any case, the crucial point is that if we adopt the hierarchical desire model of the structure of the will, then we can hold that it is effective firstorder desires that always move us to action, so that Thesis A above—which says that all reasons are both justifying and motivating, and grounded on desires—will be obviously true with respect to our effective first-order desires. On the hierarchical desire model developed by Frankfurt and also adopted by our Desire-Overriding Internalism, some special second orderdesires, namely second-order volitions, are directed to the determination of precisely which first-order desires are to be effective. But as DesireOverriding Internalists, we also adopt the following doctrine, which we call the Desire-Overriding Desires Thesis: The hierarchical volitional constitution of every self-conscious or selfreflective, autonomous rational animal is such that some second-order volitions are not only able, under successful volitional conditions (which Frankfurt calls ‘‘freedom of the will’’), to determine just which first-order desire is the one that moves us on that occasion, but also can override an occurrent first-order desire that would otherwise have motivated the agent to action on that occasion, either by (1) impulsively generating a new first-order desire in order to substitute it for the first-order desire that would otherwise have effectively moved the agent to action on that occasion, or else (2) impulsively super-charging another relatively motivationally weak occurrent first-order desire in order to select it to be the effective first-order desire instead of the would-be effective first-order desire. This impulsive desire-overriding activity would occur, e.g., in a case in which a self-conscious, autonomously rational agent has a motivationally forceful occurrent first-order desire to embezzle some money without Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 144 144   : , ,   fear of detection and thereby gain some significant personal benefit (say, paying for his law school education), and also significantly benefit others (he also intends to give a large donation to ). But then he recognizes a non-instrumental, consequence-independent reason for not embezzling the money, despite the risk-free and significantly beneficial personal and social consequences of embezzlement, which then evokes a successfully overriding second-order volition to be moved by a first-order desire to be an honest person. And this in turn happens either (i) by impulsively generating a new first-order desire to be an honest person or (ii) by impulsively super-charging an already-existing but less motivationally forceful desire to be an honest person. But more precisely, just how can the self-conscious or self-reflective, autonomously rational animal actually either impulsively generate new first-order desires or impulsively super-charge existing ones, in this desireoverriding way? It seems clear enough that instrumental, consequence-dependent first-order desires can be either impulsively generated or impulsively super-charged merely by saliently presenting or re-presenting their intentional contents to a rational animal. This principle is of course an axiom of consumer advertising. In this way, e.g., while driving along the highway you may not begin to impulsively want a chocolate milkshake until you see a billboard with pictures of them. Or while sitting on your comfortable couch in front of the television set you may not begin to impulsively want a brand-new BMW until you watch a television commercial about them. Alternatively, seeing the billboard or watching the TV commercial may simply impulsively reinforce and strengthen a pre-existing desire to drink a chocolate shake or buy a new BMW. Now we need only make this process reflexive or self-applying, and hierarchical. Since every second-order desire (say, my wanting to want a chocolate shake) includes the intentional content of a first-order desire (in this case, my wanting a chocolate shake) within its own content, second-order desires can either impulsively generate new first-order desires or impulsively super-charge pre-existing first-order desires in essentially Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 145 145 the same way as the billboard or TV commercial cases, that is, just by making the content of those first-order desires salient to oneself. So, under the right conditions, I can either make myself impulsively want a chocolate shake or impulsively super-charge a pre-existing desire to want a chocolate shake, by simply wanting to want a chocolate shake, and thereby getting an impulsive first-order desire to want a chocolate shake to be the effective one. This self-prompting of impulsive effective first-order desires by means of second-order volitions—surely—happens all the time. But how will this work in the case of the first-order desire to embezzle, which is then overridden by a non-instrumental, consequence-independent second-order volition? Here one might ask oneself, Do I really want to want to embezzle that money? Won’t I be just a dishonest [insert here your most personally meaningful profane expletive for characterizing a bad person] if I do this? And isn’t my intention to give some of the money to  just a pathetic attempt to hide my disgusting dishonesty from myself?, and this might either impulsively generate a new first-order desire to be an honest person or impulsively super-charge a pre-existing desire to be a honest person, and thereby get the first-order desire to be an honest person to be the effective one. Of course, the same volitional effect can also be brought about through the practical imagination. You can, e.g., imagine yourself both resisting the awful temptation to embezzle and then later giving a large donation to  as a kind of penance, and then love that idealized image of yourself so much that it either impulsively generates a new first-order desire to be an honest person or impulsively super-charges a pre-existing one, and thereby get a first-order desire to be an honest person to be the effective one. So let us assume that there are at least some conscientious, self-conscious, autonomously rational people who are sometimes tempted to do risk-free bad things, but still manage to fight off those temptations and impulsively motivate themselves to be morally good instead of morally bad. In the natural order of things, the volitional successes of such conscientious people rarely reach the local newspapers, radio news, or television news, since by hypothesis these volitionally successful—and in Frankfurt’s terminology, ‘‘free’’—agents never in fact do the bad things, and also rarely tell anyone else about their temptations and inner struggles, since that Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 146 146   : , ,   FN:67 would be intensely embarrassing. As they say, no good deed ever goes unpunished. In any case, if our assumption about the existence of some conscientious people is correct, then the self-prompting impulsive production of desire-overriding non-instrumental, consequence-independent effective first-order desires by means of second-order volitions therefore happens sometimes too. None of this will make any sense unless we can explain how it is that someone has a desire-overriding non-instrumental, consequenceindependent second-order volition to want to want to be honest despite the intense temptation to embezzle. On our view, this desire-overriding non-instrumental, consequence-independent second-order volition can be explained by postulating the existence of a universal and innate desire-based emotional disposition in self-conscious, autonomously rational animals to be moved at least sometimes by non-selfish non-egoistic, non-self-interested, nonhedonistic, and consequence-independent effective first-order desires. For lack of a better name, let us call this universal and innate emotional disposition to be moved at least sometimes by non-selfish, non-egoistic, non-self-interested, non-hedonistic, and consequence-independent effective first-order desires, the desire for self-transcendence. The desire for selftranscendence, as we understand it, is a fundamental felt need for some form of self-abnegation, self-denial, self-discipline, and self-effacement in our lives, independently of the consequences. It demands that we at least sometimes delay, reject, or sublimate the satisfaction of occurrent first-order desires, and it also demands that we sometimes widen our outlook from the narrowly selfish, to the standpoints of significant others, and even to a synoptic standpoint encompassing all persons, and that in so doing we do not pay any sort of attention to the calculation of consequences. In this latter respect it is quite close to what Hume calls sympathy,⁶⁷ and so this important Humean moral emotion could be usefully regarded as a sub-species of the desire for self-transcendence. In any case, the ultimate aim of the desire for self-transcendence is to integrate and unify the self in a better and more complete way by sometimes rigorously disciplining one’s current self, regardless of the consequences, and so its upshot is that to the extent that we can satisfy this desire, we can impulsively overcome ⁶⁷ See Hume, Treatise of Human Nature, book II, part I, section XI. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:68 FN:69 11:28am Page 147 147 the incoherence, inauthenticity, narrowness, egoism, and calculating enslavement to consequences that constitutively characterizes our current selves. Self-transcendence, as we are understanding it, clearly seems to play an essential role in morality, religion, mysticism, cults, martyrdom, sainthood, artistic genius, scientific genius, philosophical genius, certain kinds of athleticism (e.g., long distance running, and many other high-performance sports), and also in seemingly bizarre personal projects such as that of Simeon Stylites, who lived for many years on top of a pillar.⁶⁸ It also includes the highly perverse kind of non-selfish non-egoistic, non-selfinterested, non-hedonistic, and consequence-independent desire described famously by Hume, and that we alluded to earlier in this section, namely that it would not be contrary to reason for me to prefer the destruction of the world, including of course the total destruction of myself along with everyone else, to the scratching of my finger.⁶⁹ Moreover and more radically, it seems impossible to conceive of any self-conscious or selfreflective, autonomous rational animal that is completely incapable of feeling the desire for self-transcendence. This becomes much clearer when we note, as the perverse Humean desire shows, that the desire for selftranscendence does not necessarily imply self-transcending goals that are morally good, morally permissible, or even particularly nice. Indeed it is a striking, surprisingly widespread, and occasionally tragic fact that self-transcending values can also be highly immoral or just highly perverse. Why else would it be true that many otherwise ordinary and decent-seeming people will often go well out of their way, usually in a completely self-destructive and apparently unself-satisfying manner, and more generally for no instrumentally good reason at all, just to be horrid brutes to other people? Anyone who has worked in an academic department at a university for a few years, or in a law office, or in a business corporation, or even just at a fast food place, knows this to be all too true. The recently popular British and then American TV program The Office brilliantly displays this striking fact about rational human nature in a highly humorous way. ⁶⁸ It is of course possible to have deep philosophical worries about the moral defensibility and value of the desire for self-transcendence. See, e.g., Nietzsche, Beyond Good and Evil; and Wolf, ‘‘Moral Saints.’’ ⁶⁹ See note 65 above. There is another reading of Hume’s remark, to the effect that he is saying that it would not be contrary to reason to want to avoid scratching my finger even though the rest of the world would be destroyed. But then Hume is just talking about mere titanic selfishness, which presumably no one would ever think of sharply contrasting to instrumental rationality. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 148 148   : , ,   FN:70 FN:71 FN:72 More dramatically and sublimely, however, the novels of Fyodor Dostoevsky,⁷⁰ which are filled with characters who are what Lillian Hellman very aptly called ‘‘sinner-saints,’’⁷¹ also brilliantly reveal that even the most selfish, egoistic, selfish, hedonistic, calculating, and wicked people care very deeply about self-transcending values, and are likely to reveal this in times of greatest personal crisis and stress—such as being condemned to death by a firing squad and pardoned at the last moment, a truly terrifying event that Dostoevsky himself actually experienced, and which changed his life.⁷² If we are correct, then it is also a consequence of our view that if a minded animal lacks any capacity to have a desire for self-transcendence, or experiences a significant disruption or distortion of this capacity, then he is to that extent non-rational. Sociopaths, e.g., seem to be clear examples of human beings who are significantly damaged in that way. In any case, the main point we are driving at is this. If we posit a universal and innate desire-based emotional disposition to, and thus a fundamental felt need for, self-transcendence in all self-conscious or self-reflective, autonomous rational animals, then not only is Thesis A above—which just re-states standard internalism about reasons—obviously true. It is also the case that Thesis B above—which says that while many or even most reasons are instrumental and consequence-dependent, at least some reasons are desire-overriding non-instrumental, desire-independent reasons, precisely because they are (or are provided for by) facts about some essentially non-selfish non-egoistic, non-self-interested, non-hedonistic, and consequence-independent desires of persons—is obviously true. This is because Thesis B applies directly to all the would-be selfish, egoistic or self-interested, hedonistic, and consequence-dependent effective first-order desires that are overridden by successful second-order volitions expressing the desire for self-transcendence. If all of this is correct, or even just roughly correct, then the way is open to allowing for reasons that are independent of certain first-order desires that an agent has prior to his recognizing non-instrumental and consequenceindependent reasons, while at the same time acknowledging that all reasons are (or are provided for agents by) facts about the desires of real persons. ⁷⁰ See, e.g., The Brothers Karamazov, Crime and Punishment, The Devils, and The Idiot. ⁷¹ Hellman, ‘‘Introduction,’’ to Hammett, The Big Knockover, viii. She is referring specifically to Hammett. ⁷² See Dostoevsky, The House of the Dead. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:73 FN:74 11:28am Page 149 149 According to Searle, to recognize a reason for acting is already to recognize a reason for wanting to accept it. So it seems to us that Searle’s socalled ‘‘desire-independent’’ reasons in fact generate second-order volitions that inherently express the desire for self-transcendence in some way or another, and thereby motivate the self-conscious, autonomous rational agent to act by getting a desire-overriding first-order desire (whether new or super-charged) to be the effective one.⁷³ For example, according to Kantian ethics, an individual’s second-order volition to desire to keep a promise is derived from the fact that she recognizes she has made a promise, together with a universal, innate desire-based emotional disposition that is automatically triggered by that recognition. This universal, innate desire-based emotional disposition, which Kant calls respect (Achtung), generates the desire be moved to action by non-selfish, non-egoistic or non-self-interested, non-hedonistic, consequence-independent, and morally correct first-order desires.⁷⁴ Thus Kantian respect, like Humean sympathy, and like Humean finger-scratching world-destroying perversity, is another important sub-species of the innate desire for self-transcendence. Of course, an individual might recognize her moral obligation to keep a promise and still not act on it. The second-order volition generated by her recognition of her obligation together with the higher-order moral emotion of respect may fail to determine the desire that is effective in producing action, and thus be volitionally unsuccessful (or in Frankfurt’s terminology, ‘‘unfree’’). For Kant however, for a person to recognize her moral obligation is thereby also to recognize a desire-based and yet also desire-overriding justifying and motivating reason, co-grounded in the moral emotion of respect, for action. What we are claiming, then, is that for self-conscious or self-reflective, autonomous rational animals (a class which would include all ordinary sane, sapient adult human beings), at least some non-instrumental, consequenceindependent reasons, as recognized under the appropriate conditions, trigger the universal, innate dispositional desire for self-transcendence, and thereby give rise to corresponding second-order volitions to be moved by the appropriate first-order desires. So we are saying that by their very nature, self-conscious, autonomous rational animals have the innate dispositional ⁷³ Searle, Rationality in Action, 176. ⁷⁴ See, e.g., Kant, Groundwork of the Metaphysics of Morals and Kant, Critique of Practical Reason. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 150 150   : , ,   FN:75 FN:76 FN:77 desire for self-transcendence built right into all their thought and action. This deep-seated pull towards self-transcendence is not only very different from other latent or standing desires that an intentional agent may have, but it also seems to be an essential part of what makes us us.⁷⁵ No doubt this is a controversial claim. After all, consider the case of Eve, who suffers from clinical depression and at first glance appears to be utterly unmotivated to do what she believes she is morally required to do. Suppose, e.g., that Eve recognizes that she is morally required to help her uncle, and yet utterly lacks the effective motivation to do so. Mele asserts that such cases are examples of the ‘‘problem of listlessness,’’ and that in such instances, agents completely lack motivation to fulfill their obligations.⁷⁶ Nevertheless, we find it difficult to imagine that Eve, even though she is clinically depressed, has no motivation whatsoever to help her uncle. To be sure, her motivation to stay at home is ultimately motivationally significantly stronger, because by hypothesis it is the effective one. Still, because also by hypothesis Eve recognizes her moral obligation to help her uncle and because she is a self-conscious or self-reflective, autonomous rational animal, this recognition necessarily yields a second-order desire to be moved by a first-order desire to help her uncle. Thus she wants to be moved by a first-order desire to help him. But because of her clinical depression, some disruption or distortion in the essentially embodying neurobiological basis of her desire-based emotions contingently prevents or undermines the generation or super-charging of a first-order desire to help him. Thus her desire-overriding non-instrumental, consequenceindependent second-order volition is wholly unsuccessful.⁷⁷ Yet that does not imply its non-existence as a psychological capacity. Indeed, the abject failure of her desire for self-transcendence in this context precisely implies its existence as a psychological capacity and vividly points up the pathological fact of her clinical depression. Similarly, evaluative beliefs can play a direct causal role in impulsively producing new first-order desires or in impulsively super-charging the motivational force of other occurrent but relatively motivationally weak ⁷⁵ See Korsgaard, The Sources of Normativity. ⁷⁶ Mele, Motivation and Agency, 111. ⁷⁷ In at least some of the cases in which agents are completely unmotivated to do that which they recognize they are morally obligated to do—although this is not Eve’s problem—this can be understood as a severe deficiency in rationality, and as a kind of insanity. So we will be strongly inclined to regard such individuals as sociopaths or psychopaths. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:78 11:28am Page 151 151 first-order desires. Along these lines, Michael Smith proposes that all intentional action is mediated by the overall tendency of our psychology towards personal coherence, and by the presence of a generic desire for this sort of coherence.⁷⁸ Among ordinary rational agents, there is some sort of necessary connection between believing an act to be desirable and having at least some motivation to perform the act. In cases of maximal consistency, the agent’s evaluative judgments will be in line with his or her emotional and motivational states. We think that Smith’s notion of generic desire for personal coherence is plausible, and also that it is closely related to what we have called ‘‘authenticity,’’ but would also want to locate this desire more fundamentally in the innate dispositional desire for self-transcendence. So if this is correct, then the Smithean desire for coherence would count as another important sub-species of it, along with Kantian respect, Humean sympathy, and Humean finger-scratching world-destroying perversity. In this connection, and now to borrow and slightly modify an example of Searle’s, suppose that an individual recognizes the validity of a logical proof. Because the proof demonstrates that one chapter of her dissertation is utterly confused and mistaken, she does not want to accept it. However, once she recognizes that the argument is valid and sound, she has a reason for accepting it and some sort of desire to accept it. So she will come to have this desire impulsively and in spite of herself, although of course she may repress or suppress it. But what makes this possible? It seems that if upon recognizing the validity and soundness of the proof she had no desire whatsoever to accept it, then she would be deeply irrational in some way. Note that it is not that she has a desire to be logically consistent in her beliefs and desires in the same way that she has a desire to be intelligent or witty. Rather, it is part of her very psychological nature and her existence as a rational being that she feels the deep need for some sort of consistency between her beliefs and desires quite apart from her egoistic interests, self-interest, hedonic interests, or her attention to consequences. Similarly, once a person recognizes that she has made a promise to write a letter of recommendation for one of her students by a certain date, she must as a matter of consistency come to have some sort of desire, however motivationally weak, to keep that promise. If she does not, her beliefs and desires will be utterly and completely out of joint, and then presumably ⁷⁸ Smith, The Moral Problem, 32. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 152 152   : , ,   FN:79 FN:80 she already is, or will become, unable to make sense of herself. As we all know first-hand, however, self-conscious or self-reflective, autonomous rational animals are not perfectly rational or perfectly moral beings. In fact, we very frequently screw up in little, medium-sized, or colossal ways. This means that although an individual recognizes the proof as valid, she may also impulsively fail to accept it. Similarly, although a person’s recognition that she has made a promise must issue in a second-order volition to be moved by a first-order desire to keep her promise, this first-order desire need not be effective in action. The agent in question may very well end up impulsively going to the movies, then to the pub, and finally home to fall exhaustedly into bed, instead of writing that letter of recommendation. One last point should be made in this connection. The Desire-Overriding Internalism about reasons that we are advocating, for all its radicalness, is still in one absolutely crucial respect importantly weaker than the more standard internalistic and instrumentalist view of reasons defended by Davidson and many others. Davidson says that if an agent unconditionally judges that it would be better to do X than to do Y , then he wants to do X more than he wants to do Y .⁷⁹ So on his view, if someone sincerely believes he ought to do something, then his belief must show itself in his behavior, his inclination to act, and his desire. Such a view reflects Davidson’s belief that the natural expression of a desire is instrumentally evaluative in form. Someone who wants to do X believes that he ought to do X or that it is desirable to do X. For example, an individual who honestly believes that it is desirable to stop smoking has some pro-attitude toward his stopping smoking; feels some inclination to stop smoking; and will do so provided that nothing stands in the way, he knows how, and he has no contrary values or desires.⁸⁰ Thus for Davidson, pro-attitudes express instrumental value judgments that are at least implicit. But while we agree that an unconditional belief that one ought to do something does ordinarily generate some sort of first-order desire to do that thing, we do not follow Davidson in holding that this desire must produce a corresponding action. The everyday widespread existence of cases of akrasia or (as we think of it) impulsiveness of the will clearly show that evaluative judgments need not match up with the motivational force of desire-based emotions. And actions done for non-instrumental, ⁷⁹ Davidson, ‘‘How is Weakness of the Will Possible?,’’ 23. ⁸⁰ Davidson, ‘‘Intending,’’ 86. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 153 153 consequence-independent reasons based on the innate dispositional desire for self-transcendence equally clearly show that instrumental, consequencedependent reasons significantly underdetermine intentional agency. Since actions done for non-instrumental, consequence-independent reasons, in addition to being impulsive, are often also pre-reflective or spontaneous actions, our Desire-Overriding Internalism about reasons and the fact of pre-reflective or spontaneous actions naturally go hand-in-hand. 3.5 Against Davidson 4: Deviant Causal Chains Again The last and certainly most extensively discussed problem for the classical causal theories of action is the possibility of deviant or wayward causal chains. As we noted at the beginning of the chapter, and as Frankfurt has pointed out, classical causal theories of action say that FN:81 FN:82 a bodily movement is an action if and only if it results from antecedents of a certain kind. Different versions of the causal approach provide differing accounts of the sorts of events or states which must figure causally in the production of actions. The tenet they characteristically share is that it is both necessary and sufficient, in order to determine that an event is an action, to consider how it was brought about.⁸¹ But if belief-together-with-desire, or a Davidsonian reason, produce action in the wrong way, or accidentally, then the body movements that result are a mere causal effect of them rather than an intentional act caused in response to them.⁸² For example, there is the case of the unfortunate rock climber who, because it is so painful to hold the rope that supports his partner, naturally desires to rid himself of the weight of his partner and knows that loosening his grip on the rope would do the trick. But, tragically, this belief so unnerves him that he loosens his hold, and drops his partner. And there is another case of someone who intends to spill his drink at a party in order to signal to some accomplices to begin a robbery. Although he has not yet committed any crime as he stands there sipping his drink, the thought of doing so makes him so nervous that his arm trembles and he spills his drink, thereby initiating the robbery. In these cases, it is highly ⁸¹ Frankfurt, ‘‘The Problem of Action,’’ 70. ⁸² Audi, Action, Intention, and Reason, 17. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 154 154   : , ,   FN:83 implausible to hold that the agent’s dropping his climbing partner or spilling his drink is an intentional act, precisely because he does not actually control the arm-movement that dropped or spilled. Hence the dropping and the spilling were unintentional body movements, not intentional movements.⁸³ One might think, however, that these cases are quite different from ‘‘normal’’ cases of unintentional body movement—e.g., someone who trips over a curb in the dark or loses his balance when a train suddenly jerks forward. More specifically, it seems that both the unfortunate rock climber and trembling robber bear some degree of causal responsibility for their unintentional movements, even if they are not strictly speaking morally responsible. This is because these body movements do result from their desires, but not in the proper way, that is, as a result of trying and its active guidance. It is true that one would no doubt hold the trembling robber morally responsible for setting up a state of affairs in which something like this could happen, and for having wicked intentions, even if his unintentional body movement is something for which he is not strictly speaking morally responsible. But when one trips over a curb in the dark or loses his balance when the train jerks forward, on the other hand, this is not in any way a result of desire, so it seems that there is not even any causal responsibility in such cases, much less moral responsibility. This calls for a brief remark on the notion of unintentionality and the ascription of it in ordinary language. Often we will correctly say that someone does or did something unintentionally—say, tripping over a curb in the dark—but this should not be taken to mean that there is such a thing as unintentional basic acts, or ‘‘unintentional intentional action.’’ Tripping over the curb in the dark is an unintentional body movement, but not strictly speaking an act of any sort. Moreover, sometimes even in cases other than those of unintentional body movements, we will also correctly say that someone does or did something unintentionally. But here it seems that we are really talking about unintended effects or side-effects of our basic actions and intentional body movements, whether foreseen or unforeseen, and not about our basic acts and intentional body movements themselves. For example, we might say that I acted unintentionally in shooting someone by accident or by mistake, even though the trigger is intentionally pulled by me in both cases. Or again we might say that I acted ⁸³ Frankfurt, ‘‘The Problem of Action,’’ 70. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   FN:84 FN:85 11:28am Page 155 155 unintentionally by killing some innocent civilian bystanders with a missile when the target was a military one, even though, again, the missile-firing button is pressed intentionally by me. And so on. In such cases, we certainly do ascribe causal responsibility to the agents, and sometimes ascribe moral responsibility too. Nevertheless, the agent might also use the fact that the effects were unintended (say, in the case of shooting someone by accident) as part of an excuse in order to avoid accepting moral responsibility, blame, or punishment.⁸⁴ In any case, examples involving deviant causal chains lead to a general worry about classical causal theories of action, as Mele observes, because whatever psychological causes are deemed both necessary and sufficient for a resultant action’s being intentional, cases can be described where, owing to a deviant causal connection between the favoured psychological antecedents and a pertinent resultant action, that action [i.e., that body movement] is not intentional.⁸⁵ It is very important to note that the standard examples of deviant causal chains hold whether, as in the agent-causal theory, we place the agent outside of time, or as in the volitional-causal theory and the Davidsonian theory, we place the agent inside the series of mental or physical events. For even if we assume that the very idea of timeless causal agency for minded animals actually makes sense, such a timeless agent could then also always produce the relevant action through mere nervousness. Supra-temporality, presumably, is in and of itself no protection against the subjective experience of anxiety, as in the unfortunate rock climber and trembling robber cases. This shows that every classical causal theory of action is incomplete, for it cannot make proper sense of the causal connection that must obtain between a mental antecedent and a body movement in order for action to count as intentional. As a minimal condition of philosophical adequacy, then, any causal theory of action must be fully equipped to explain precisely what goes wrong in cases of causal deviance. We believe that our own non-classical causal theory, grounded in trying and its active guidance, smoothly explains these types of cases. The unfortunate climber never actually tries to loosen his hold, and the ⁸⁴ See, e.g., Austin, ‘‘A Plea for Excuses.’’ Moral responsibility, blame, and punishment are clearly not the same. Person X can hold person Y morally responsible for doing something bad, while at the same time also forgiving Y, thereby either not blaming Y or at least ceasing to blame Y. And X can hold Y morally responsible and blame Y, while also reasonably refusing to punish Y. ⁸⁵ Mele, ‘‘Introduction to The Philosophy of Action,’’ 6. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 156 156   : , ,   trembling robber never actually tries to signal to his accomplices. So any body movements resulting from their beliefs and desires—or indeed from any other causal antecedent—by a deviant causal chain will automatically be both uncaused by the agent and also unintentional, and thus not be a counterexample to our theory. Moreover, even if my frustrated trying to raise my paralyzed arm accidentally neurally triggers a strange causal mechanism that later brings about the rising of my arm in a very bizarre way—say, by triggering a signal to a ray-gun on Mars that sends a tractorbeam back to earth and levitates my arm—nevertheless we could not correctly say that my unintentional body movement is actually caused by me, since of course by hypothesis I was trying to raise my arm, and was not trying to bring about the rising of my arm. The rising of my arm merely happens to me. On our view, an arm-raising necessarily requires a synchronous trying and its active guidance, which in turn requires essentially embodied engagement. But during the time when my paralyzed arm rises by means of the tractor beam, I am not actually trying to raise it, nor am I actively guiding its movement. In fact, part of the causal chain leading to body movement is utterly detached from my body, and so it is obvious that this is something that merely happens to me. Therefore deviant causal chains can be easily accommodated by our non-classical causal theory of action. In this connection, we think that it is very important to distinguish carefully between (1) deviant causal chains and (2) non-standard causal mechanisms. A non-standard causal mechanism is a causal process that produces a certain effect, but is not normally deployed for the production of that effect, or occurs relatively infrequently in the normal course of nature. Deviant causal chains accidentally bring about an effect by means of some or another non-standard causal mechanism. Yet a non-standard causal mechanism can also be used to bring about an intentional body movement. For example, if I discover that my left arm is paralyzed and want to raise it, I can move it into position using my right hand and right arm. The movement of my right hand and right arm is a basic intentional act, and the movement of Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008   : , ,   11:28am Page 157 157 my left arm is a non-basic intentional act brought about by the intentional body movement of my right hand and right arm. I thereby raise my left arm by means of a non-standard causal mechanism. This is not, however, a deviant causal chain, since the effect is brought about non-accidentally even if in a non-standard way. In this chapter we have developed a full-scale critique of Davidson’s highly influential theory of action. We also sketched the outlines of a decisively post-Davidsonian and non-classical causal theory of action, the Essentially Embodied Agency Theory. More precisely, we argued that reasons in the Davidsonian sense are never mental causes of actions, even though every intentional action has a mental cause. Instead, the mental causes of basic actions are synchronous, essentially embodied, pre-reflectively conscious effective first-order desires, or tryings that actively guide intentional body movements. Furthermore, even though every intentional action is normatively supported by internal reasons, provided for agents by facts about their desires, many of these reasons are neither self-consciously nor self-reflectively recognized, and many of these reasons are not instrumental reasons. Finally, even when a reason for action is self-consciously or self-reflectively recognized, it is frequently a non-instrumental reason. Of course, even beyond thoroughly criticizing Davidson’s theory and handling the problem of deviant causal chains, our non-classical causal theory of action must also respond directly to the three other basic worries about the Davidsonian theory. Most importantly, our theory must provide independent positive grounds for claiming that trying and its active guidance explain intentional action. To begin the development of this positive account, in the next chapter we will look closely and critically at Frankfurt’s guidance-based theory of action and O’Shaughnessy’s theory of trying. Robert Hanna and Michelle Maiese run03.tex V1 - August 8, 2008 11:28am Page 158