Libertarian defense of free will

I conclude that because a viable model of free-agent action requires that an agent exercise control over her individualized and rational decision-making, such a model 7 Wolf argues thro

Your unpublished thesis, submitted for a degree at Williams College and administered by the Williams College Libraries, will be made available for research use You may, through this form, provide instructions regarding copyright, access, dissanination and reproduction of your thesis

- The faculty advisor to the student writing the thesis wishes to claim joint authorship

in this work

In each section, please check the ONE statement that reflects your wishes

1 PUBLICATION AND QUOTATION: LITERARY PROP~ERTY NGHTS

A student author automatically owns the copyright to hishe]: work, whether or not a copyright symbol and date are placed on the piece The duration of U.S copyrighl on a manuscript and Williams theses are considered manuscripts is the life of the author plus 70 years

- I/we do not choose to retain literary property lights to the thesis, and I wish to assign them immediately to Williams College

Selecting this option will assign copyright to the College This in no way precludes a student author from later publishing hisher work; the student would, however, need to contact the

Archives for a p e r ~ s s i o n fornl The kchives worlld be free in this case to also grant permission

ro another reseucher to publish smdI sections from the thesis Rarely would there be any season for the Archives to grant permission to another party to publish the thesis in its entirely; if such a situation arose, the k c h i v e s would be in touch wihln the author to let them know that such a request had beel; made

- Uwe wish to retain literary property rights to the thesis for a period of three years, at which time the literary property rights shall be assigned to Williams College

Selecting this option gives the author a few years to make eexclusive use of the thesis in tip-coming projects: articles, later research, etc

I/we wish to retain literary property rights to the thesis for a period of 3 4 years, or until my death, whichever is the later, at which time the literary property rights shall be assigned to Williams College

Selecting this option ailows the author great flexibiiity in extending or shortening the time of hisher automatic cop jiright pesiod Some students a e interested in using their thesis in g-aduate school work In this case, it would make sense for them to enter a number such as '20 years' in the blank, and line out the words 'or antii my death, whichever is the later.' En any event, it is easier for the Arciiives to administer copyright on a manuscript if the period ends with the

individual's death our staff won't have to search for estate executors in this case but this is entirely up to each student

IT ACCESS

The Williams College Libraries are investigating the posting, of theses online, as well as their retention in hardcopy

J Williams College is granted permission to maint n and provide access to my thesis

in hadcopy and via the Web both on and Selecting this option allows researchers around the worlb to access the digitd version of ~ f f campus f yoilr 1

- Williams College is granted permission to maintain and provide access to my thesis

in hardcopy and via the Web for on-canipus use only

Selecting this option allows access to the digital version of your work from the on-camps

network only

- The thesis is to be maintained and made available in hardcopy form only

Seiecbng this option alIows access to your work only from the hardcopy you submit Such access pel-tarns to the entirety of your work, i~~c?uding any media that it comprises oi- ~ncludes

Il[I COPYING AND DISSEMINATION

Because theses are listed on FRANCIS, the Libraries receive numerous requests every year for copies of works IUwhen a hardcopy thesis is duplicated for a researcher, a copy ofthe release fonn always

accompanies the copy Any digital version of your thesis will include the release fonn

J Copies of the thesis may be provided to any researcher

-

Selecting this opi~on al~cws any researcher to request a copy from the M71liiams College Lbranes,

or to make one froin an eiectromc verc;ion

- Copying of the thesis is restricted for - years, at which time copies may be provided to any researcher

This option alIows the ai~thor ro set a time limit on copying restrictions During rhis period, an

electronic version of the thesis will be protected against duplication

- Copying of the thesis or portions thereof, except as needed to maintain an adequate number of research copies available in the Williams College Libraries, is expressly

prohibited The electronic version of the thesis will be protected against duplication

Selecting this ophioli allows no reproductions to be made for researchers The electronic verslon

of the thes-is will be protected against duplication This option does not dis-allow researchers from reading!viewing the work in either hxdcopy or digital form

Signed (student author)

Signed (faculty advisor)

Date accepted < 2 ; ; t & L

A LIBERTARIAN DEFENSE OF FREE WILL

by

D CAROLINE ANDERSON

Professor Melissa Barry, Advisor

A thesis submitted in partial fulfillment

of the requirements for the

Degree of Bachelor o:f Arts with Honors

in Philosophy

WILLIAMS COLLEGE

Williamstown, R/lassachusetts

May 22,2006

I would sincerely like to thank Professor Barry for her help on t h s project It was her free will tutorial that first led me to pursue this topic and her constant support that kept me going through the seemingly endless lobjections Her insights and philosophical eloquence are unmatched I would also like to thank Professor Cruz for serving as an excellent second reader for this thesis and for his willingness to support me despite my radically different views Professors Dudley and Gerrard were also kind enough to provide initial feedback and objections to help me strengthen my argument The philosophy department in general has been incredibly supportive during my time at Williams; my philosophy courses have consisitently been the most exciting, challenging, and rewarding classes I've taken while here Thanks so much also to Rosie Smith for taking the time out during exams to read through my final draft and suggest valuable revisions

On a more personal note I'd like to thank my family and friends for seeing me through this sometimes stresshl process and fior their constant support Thanks so much

to Janaki O'Brien for putting up with all the philosophical jargon and being willing to spend a disproportionate amount of time chatting about my free will obsession Thanks

to Lucky for helping me keep my perspective, to my dad for getting me excited about bizarre theoretical notions in the first place, to] my mom for giving me hugs and making

me pots of coffee and to Ariel for being such a great role model Finally, I would like to thank my boyfriend Philipp Huy for being more understanding and caring than I ever could have hoped

Introduction

Although it may seem that all theorists engaged in the free will debate are addressing the same question, namely, "do we exercise free will," they in fact pursue a number of significantly different approaches In order to reveal the conflicts among the views, it must be made explicit what kind of evidence is used to support them, along with the goals and motivations of the theorists who hold them Historically, the two main approaches have been the ethical and the mletaphysical The driving question of the former is, "what conception of freedom must be in place to justify our praise and blame attitudes," while the driving question of the latter is "do we have the capacity to freely choose our own actions, regardless of the innpact the answer will have on our moral practices."1 The ethical approach may additio~nally be characterized as pragmatic, asking what would be lost were we to abandon our current praise and blame practices, but it would still use the answer to this question to address the underlying ethical question mentioned above The metaphysical appraach, on the other hand, uses conceptual argumentation, such as concerns of logical coherence and intelligibility, and in some cases scientific data, to discover what is really the case about how and why we make the decisions we make and perform the acts we pe~rform.3

In this paper I use a metaphysical rather than an ethical or pragmatic approach because the truth of what underlies our decision-making does not merely determine

whether we ought to be praised or blamed for our actions, but also issues ranging from the rationality of o w actions, our abilities to detect and act on what is really the case in the world, and, perhaps most important, o w capacity to defend the bases on which we make decisions according to objective criteria If it can be proven that there is no possibility that we choose our actions in any significant way, the notions of genuine normative and rational agency will have to be abandoned, along with many responsibility-dependent ethical practices as well

Even if I do focus on metaphysicid concerns for the reasons just described, the primary question that I seek to address in this paper is actually not the traditional metaphysical one Rather, I pursue a diagnostic metaphysical approach, meaning that I examine the current framing of the free-will debate and attempt to show why this framing ought to be abandoned Traditionally, the question of free will is framed causally: First,

it must be true that our actions are either c:ausally determined or they are not From this,

it initially seems that the truth of causal r~ecessitation would most threaten our free will because it would limit o w range of choices to one inevitable outcome But, many have argued, its alternative would be far worse If we were not causally determined, then how would we control o w interactions with the world? A lack of causality seems to entail randomness and we would certainly not be in control of o w actions if those actions were random Therefore, the argument goes, if determinism is false, we most definitely do not have free will If determinism is true, then at least we might The two possibilities that result from this line of argumentation are randomness and necessitation

Given this framing of the debate, tlhere are three main positions taken by theorists

of free will The first division is between compatibilists, who believe that free will and

determinism can be reconciled, and incompatibilists, who think that if the truth of determinism can be established, free will must be an illusion The incompatibilists are then divided into two further groups: hard determinists, who argue that there is no way to defend -Free will given the causal order that empirically holds in the world, and libertarians, who argue that, given the falsity of determinism, we do have responsibility- conferring hee

In this paper, I present two arguments First, I argue that though the compatibilist position does protect agent action from the threat of randomness, it does not protect it from the equally problematic threat of ~rbit~rariness, despite its acceptance of causal necessity Second, by offering my own model of non-necessitated free agent action, I defend the intelligibility of libertarianism, showing that the falsity of determinism does not entail either randomness or arbitrariness in agent action This emphasis on arbitrariness over randomness represents the first revision that I suggest to the traditional form of the debate Randomness implies a complete lack of connection between two events, concepts, truths, and so on I consider randomness to be an inappropriate term to characterize the state of affairs under libertarianism because an utter lack of connection between two things need not be assumed just because there is no causal connection between them As long as there is any sort of non-purely-coincidental relation between two entities, whether conceptual, contextual, clr of some other sort, the term 'random' no longer applies Arbitrariness, in contrast, as I use it here, is a more useful term because it entails a lack of signzficance in the connection between two entities, even given a context

of non-randomness

4

To require that genuine free will is responsibility-confizrring is not meant as an ethical statement It is meant instead to establish the hee will in question as being robust enough to render actions the agent's own

While compatibilists and libertarians agree that a lack of significance in agent action would threaten its status as free, they have disagreed on what features are necessary to preserve significance for the agent The form of this disagreement hinges on the interpretation of two notions, namely those of choice and control I will take each notion in turn The call for choice in a model of flee-agent action is typically referred to

in two ways The first claims that if an agent indeed had a choice pre-action we must be able to say of her post-action that she could have done otherwise Because Harry Frankfurt's compelling Dr Black thought experiment reveals a weakness in this conception of choice, I avoid it in this paper.5 The second, more promising conception is the principle of alternative possibilities, PAP This principle can be regarded in three ways First, some compatibilists require that the agent lack alternative possibilities pre- action in order to be considered free To avoid the awkward construction 'non-PAP', I will simply refer to this position as deterministic This deterministic view is defended on the grounds that any form of PAP would disrupt the possibility of acting rationally, a view that I will criticize in Section I11 of Chapter 1 Second, other compatibilists argue for what I will call weak PAP This requires only hypothetical alternatives It claims that, had an agent wanted something different from what she in fact wanted, she would have faced a different alternative than the one that she actually faced.6 Third, libertarians argue that free will requires a range of genuine alternative possibilities right up to the moment of action onset in order for the agent to be considered free This position I will

call strong PAP The alternative possibilities l.he agent faces under it are termed genuine because they are not hypothetical alternatives, as is the case under weak PAP, but rather alternatives that the agent indeed faces They are not, in other words, just possibilities that she would have faced had she wanted or chosen differently

The notion of agent control accepted lsy various camps in the debate is directly related to the notion of choice that they accelpt Determinists, for example, who reject both weak and strong PAP, argue that the agerit need not have control over her actions in order to be free Instead, t h s view argues that the objective relations that exist in the world ought to wholly structure the agent's behavior An ideally free agent on this view

is one who is caused by normative reasons to behave in rational ways Because this view claims that freedom would be preserved neither by strong PAP nor by the chance for acts

to arise from the individual psychology of the agent, I will refer to it as mechanistic The second view of the role of agent control and the one taken by supporters of weak PAP is that a free action must merely be the agent's own in some significant way This view denies, in other words, that the freest action is the action shaped entirely by the objective relations that hold in the world Instead, it claims that an agent's action is free if it arises from her own psychology Agent control is emphasized on this view in its stipulation that the agent must be able to shape the nature of her own psychology through a process

of self-modification Because this view relies on a neo-Hurnean notion of self, which considers the agent's deeply held contingent desires to be constitutive of who she is, I refer to it by its traditional name, the deep-self view I, as a libertarian defender of strong PAP and the third view of agent control, support the intuitions that motivate both the mechanistic and the deep-self views' of agent control That is, I argue both that the neo-

Humean is right that for an action to be freely chosen by the agent, it must reflect her as

an individual and that the determinist is right that for an agent to be able to freely act, she must be able to act rationally However, I defend a different conception of how these two criteria can be fulfilled by a model of fYee-agent action that combines both individuality and rationality

In this paper, I defend strong PAP and individualized rational agent control as necessary features of free-agent action in the following way First, in Chapter 1, I argue that due to his acceptance of a neo-Hurnean notion of self and its associated conception

of reason, Frankfurt's model of agent free:dom fails to fulfill his own criteria That is, it fails to fulfill the requirement that an agent's action stem from who she is in a significant sense and the requirement that she be able to shape the kind of person she becomes I argue that the process of self-modification suggested by Frankfurt requires an ability to respond to and evaluate not merely desire-based reasons, but desire-independent normative reasons as well I then, in the second part of Chapter 1, introduce Wolfs view

as the determinist view that is most promising in enabling the agent to act according to such normative reason^.^ I argue against Wolfs view that it does not render the agent free because it does not actually provide her with the ability to act rationally Because Wolf argues that rational action is necessary for free action, I object to her view on its own terms I conclude that because a viable model of free-agent action requires that an agent exercise control over her individualized and rational decision-making, such a model

7

Wolf argues throughout much of Freedom within Reason (1990) and "Asymmetrical Reason" (1980) that the most moral agents are those who are caused to act the right way by objectively weighted reasons When pressed to explain how it is that some agents are caused to behave rightly and others wrongly, Wolf abandons her compatibilism and posits psycho1ogic:al indeterminism For the purposes of h s paper I evaluate Wolfs central reason view, which defend:; a determinist model, despite the fact that she does not support this view consistently throughout her argumentation

must be incompatible with both neo-Humeani and mechanistic notions of self In the second and third chapters, therefore, I defend an alternate conception of self as well as propose a new libertarian model for agent deliberation and action in the context of strong PAP

Before proceeding, I would like to offer another word on what motivates my methodology I pursue the diagnostic metaphiysical approach I have just laid out using conceptual arguments rather than neuroscientific data to defend my position T h s does not reflect a denial of the crucial role that eimpirical research plays in addressing this question I am, on the contrary, convinced that theorists in the fields of psychology and neuroscience, metaphysics and philosophy of science, must work in concert to adequately address the free will question This paper, however, aims to overturn both the traditional assumption that determinism can rescue algent action from arbitrariness and the assumption that the falsity of determinism entails randomness These assumptions have been defended on logical and theoretical rather than empirical grounds and can, as I hope this paper will show, be undermined using the very sorts of arguments that appeared to establish them in the first place This brings my argument into direct dialogue with the compatibilist position in a way that scientific data likely would not It is for this reason that I do not address the implications of available empirical data in the context of this paper

This paper emphatically denies that the free-will debate ought to center around event-causal matters Though the majority of both compatibilist and libertarian approaches treat the question causally, I assert that turning to causality will not provide

us with the conceptual answers we seek There is no wide consensus to date on a

defensible model for event causality Moreover, the identification of brain states with mental states has not been philosophically ratified, nor has the categorization of mental states as events Questions of what sort of causal interaction could be taking place between non-physical elements like reasons and physical events like actions have not been answered What we may intuit about basic billiard-ball causality and its power to render the interaction among reasons, motivations, and actions intelligible has not been sufficiently scrutinized to justify limiting the possibilities for models of agent action to being either random or necessitated Throughout the paper, I focus on the conceptions each view under consideration implicitly accepts of agents, reasons for action and rationality An exploration of these notions reveals the true source of tension in compatibilist models and suggests a new libertarian model that is more convincing than those currently available

Chapter 1: Two Cornpatibilist Approaches

Introduction

In this chapter I present two characteristic compatibilist views, first Harry Frankfurt's deep-self view and second Susan FVolfs reason view.* I explicate each view, revealing the notion of self that it accepts and how it is used to deal with the features of free agency introduced in the introduction, n m e l y PAP and agent-control Frankfurt's view relies upon a neo-Humean conception of self, meaning one that is entirely reducible

to the agent's contingent psychological states; This notion of self assumes that agent actions cannot get off the ground unless motivated by desires Wolfs view, on the other hand, conceives of the self as normatively capable, able to recognize reasons in the world independent of the agent's psychological elements I will object to Frankfurt's view of self on the ground that it causes self-shaping, a necessary criterion for his agent's ability

to choose freely, to run into an infinite regress According to Frankfurt's own criteria, therefore, his notion of self renders the agent's action not free In section 111, I use the collapse of reasons to causes under universal event-causality to reject Wolfs argument for the reliance of rationality on determinism

I draw two conclusions fiom Frankfurt and Wolf First, I support Frankfurt's emphasis on the importance of an action's origination in the agent herself Second, I agree with Wolf that the three processes necessary for free action are recognition of desire-independent reasons, acting out of such reasons, and self-conceiving according to

See Frankfurt, Harry "Freedom of the Will and the Colncept of a Person" (1971); Wolf, Susan

"Asymmetrical Freedom7' (1980), Freedom within R e a s ~ g (1990), "Sanity and the Metaphysics of

Responsibility" (1987)

such reasons Unlike Wolf, however, I argue in chapter 3 that these three processes can preserve freedom for the agent only if they operate under strong PAP

Before turning to Wolfs and Fr,ankfurtYs specific compatibilist views, I first examine what their common commitmer~t to universal event-causality entails Every compatibilist position, regardless of how iit manifests itself, is founded on the assumption that the world operates universally accordling to the relations of event-causation For the human agent, this means not only physical determinism, but psychological determinism

as In order to determine how the truth of this assumption would affect accurate models of agent action, we must first understand how the thesis of universal event- causation seeks to make sense of occurrences in the world in general A loose sketch of the event-causal model, which is provideld in Section I of this chapter, reveals why its supporters assume that its only viable alternative is randomness This section makes explicit the view towards agent action that the model of event causation takes, reducing reasons to causes and making the self passive The goal of this chapter is to demonstrate that reason's justificatory and normative potential, to which I will argue the agent must have access in order to act freely, is greatly reduced by the event causal model

Section I The Event-Causal Story

According to the thesis of universal event-causation,1° the world consists of a network of elements that are bound togetlher according to the regular, patterned laws of

9

In other words, it is not merely the events that occur in the external world, like the acts performed by the agent, that operate deterministically, but the operation of the agent's internal mental life as well, including every feature of her phenomenological experience

10

I do not reference particular event-causation supporters because the sketch that I draw here does not reflect the particularities of any one theorist's viewpoint In the respects crucial to the claims that follow in the next sections, disagreements over the details will not affect the argument that I present

nature As active stimuli run through the web, they trigger shifts in it, causing it to constantly reconfigure From the minutest to broadest level, this reconfiguring occurs in necessitated ways

When it comes to the causal interaction of a human and her environment, every external element that has come into causal contact with her has had an effect on her Although each effect can be described on a nimber of different levels, from the micro- molecular to the macro-psychological, there is only one referent for these descriptions The term 'human,' on this view, refers tlo the present, past and future internal configurations of a particular distinguishable region of the larger web, whose infinite reconfigurations are causally linked through time Various large-scale effectors for the internal configurations of humans are heredity, upbringing, spatio-temporal positioning, and the individual's unique bundle of past experiences and external stimuli Every element in the surrounding environment is intieracting with that internal configuration at all times, some of these effects more noticeable than others

In contrast to our ordinary notion of the human self as operating actively, in ways that objects like billiard balls cannot, on purely event-causal terms there is no great division between the self as a set of structural configurations and the external environment We may describe ourselves as being fundamentally distinct from other things and beings, but the web of causal relations makes no such distinction In more technical terms, the status accorded to hwna~is on the universal event-causal model is ontologically no different from the status of any other set of causally linked configurations The only relation that one thing, being, or component can have to another

is causal A different and distinct set of structural properties may apply to entities like

humans, but these properties do not accord humans any active properties fundamentally different from the passive structural properties of other types of entities Instead, our behavior merely reflects our given structures in patterned ways."

According to the event-causal model, then, behavior does not reflect active capacities, but rather structural tendencies On the level of action, when the agent is exposed to a set of external stimuli that have sufficient force to significantly change the internal configuration of the agent, this network of stimuli sends a causal disruption throughout the web, stimulating the links in the human's internal web and triggering a necessitated response from it On the micro-level, these disruptions consist in neurophysical interactions, but the compatibilist usually characterizes this sort of causally necessitated response as an action and describes it in the following way The contingent psychological configuration of the self, which is a network of causal connections including beliefs, desires, tastes, preferences, commitments and ideals, interacts causally with the stimuli that come in A response is triggered that is characteristic of the structural components of that particular, uinique causal web This web, recall, constitutes that agent's self Just as it is characterj~stic of a ball, given its structural and causal makeup, to roll when it is pushed, it is characteristic of selves who have been exposed to traumatic situations to respond violently when placed in tense environments

Compatibilists, because their entire view on human action is predicated on the universality of this causal model as an accurate description of how the world works, have

no option but to conclude that any interaclion between two elements that does not follow this model must be random The indeterminacy and randomness of the movement of

-11

My alternate model of agent action will emphasize the activity and non-contingent unity of the agent, two properties that are not available, I claim, on a purely event-causal model

quantum particles supports this conclusion The idea is that the physical laws of nature always operate causally, so if the relations among objects, events, facts and persons are not causal, then they are also not held together by the laws of nature and, therefore, there

is no non-coincidental way that they could relate It may be the case that there is more than one law of nature, so the way that event-causality manifests itself will appear different depending on the sorts of structures involved and the particular forces acting on them, but if agent action is not random, then the basic causal model must be held constant

Because any relation between two elements must be causal on this model, a particular reason for action, which is typically considered to be the relation between a motivation and an action, must also operate causally One need merely provide the structural dispositions of the agent and the c:xternal stimuli affecting her at time tl in order to fully explain her action taken at time: t2 The reason for the action, then, is the agent's given causal tendencies and external iBctors beyond her control If one were an omniscient observer and could, at time tl perceive every single link on the causal web and know the laws of nature dictating the ways in which the external stimuli trigger the structural dispositions of that agent, one would be able to predict with infinite precision and accuracy the state of affairs for the causal web at tz This is because the universal event causal model denies the existence of gelmine alternative possibilities Under event causation the reason for action x must be restated as "the reason for the state of affairs at t2," making it equivalent to "the state of affairs at tl given the laws of nature in effect for this particular system."

Section II Harry Frank$urt7s Deep-Self View

I now turn to Frankfurt's deep self view, which defends a notion of free will that

it takes to be compatible with the universality of event causality, i.e with the falsity of strong PAP.12 Recognizing the tension that arises between causal necessity and the intuitive need for agent authorship13 in a plausible model of freedom, Frankfurt accepts a neo-Hurnean notion of the self According to this view, the self consists in a hierarchy of psychological elements, including, most crucially, desires, by which every agent act is motivated In this section I argue that the notion of self that Frankfurt accepts, in large part due to its grounding in weak PAP and psychological contingency, does not succeed

in rescuing Frankfurt's agency model fronn the accusation of randomness that drove him

to reject strong PAP in the first place In so arguing I show how the reduction of reasons that follows from the universal event-causality model, which I present in the previous section, renders Frankfurt's own appeal to the agent's deep-self unintelligible

The motivation for Frankfurt's view is the seeming incompatibility between determinism and the intuition that for an agent to have freely chosen her action it must

have her as its causal source rather than some element alien to her If, for example, an agent is the victim of a neurosis, like kleptomania, she is acting on desires that have been implanted in her, of which she does not approve and on whch she does not want to act The reason, Frankfurt points out, that we consider this agent not to have acted freely is not because she did not face genuine alternatives, but rather because the desires on which she did act were not her own Even if, hypothetically, it had been the case that the agent

could have refrained from stealing, she did not act freely if the cause of her actually

l2 See Frankfurt (1971)

13

Agent authorship refers to an action's having its source in the agent

having stolen was the kleptomania rather than her deep self, i.e who she is as a person Strong PAP does not, Frankhrt concludes, contribute to her freedom

By the same token, if the hypnotic or neurotic is the epitome of one who is not psychologically free, then the agent whose actions are wholly the product of her own self-affirmed hierarchy of desires must be the epitome of one who is psychologically free.14 Even if such an action, one caused by her deep self, was inevitable, it still would

qualify as her action in the most significanl possible way because it was the direct

product of who she is as a person Frankfurt thus rejects the need for choice in agent freedom and replaces it with the principle of the authorship of the action by the agent's deep self In order to make this picture of thle wholly free agent clear, though, we must establish what Frankfurt considers the agent, as the source of action, to be, along with what he means by a hierarchy of desires We must also understand by what process and

on what basis he believes agents can undergo this sort of affirmation of desires

Frankfurt takes a reductive view of the agent, considering her to consist solely of

a causal network of psychological elements, desires being the ones necessary to motivate action On the first-order level, an agent experiences immediate, passing desires that are not particularly important to her sense of self, to who she is as a person On the second- order level, the agent has deeper desires, such as the desire to want to have a certain first- order desire or the desire not to act on one she has An agent's total network of desires and psychological elements constitutes her self, while her deeper desires, because they

14

Frankfurt emphasizes psychological as opposed to mere physical freedom because physical barriers constrain one's freedom of action, not one's freedom of will The notion that a person's freedom is "a matter of doing what one wants to do does capture at least part of what is implicit in the idea of an agent who acts freely It misses entirely, however, the peculiar content of what is implicit in the idea of an agent

whose will is free," (1971), p 331 In actuality, Frankfivt claims, freedom of action and will are so sharply distinguishable that it is possible for a person deprived of the freedom to do what they want to do, for a paralyzed person, for instance, to enjoy freedom of the will

represent the kind of agent she considers herself to be and wants to be, constitute the agent's deep self An agent acts with free will, according to Frankfurt and other hierarchical-desire theorists, so long as the agent's acts result from desires approved of by her deep self.15 When an agent is "not only free to do what he wants to do," but "also free to want what he wants to want[,] it seems that he has, in that case, all the freedom it

is possible to desire or to con~eive."'~ It ,does not matter to the deep-self theorist if the agent is causally determined to perform a certain action In fact, it is all the better if she

is determined by her deep self to so perfomn.'7

Frankfurt claims that the agent comes to condone certain of her first-order desires and to reject others of them through the: "capacity for reflective self-evaluation," by which she decides whether her first-order desires are worth having or not.18 The free agent is not one who blindly moves through her environment doing whatever happens to occur to her, but rather one who criticizes her own tendencies and chooses to be the sort

of person she becomes This ability on the agent's part to evaluate and judge the worthiness or appropriateness of the motivations for her actions is, I argue in chapter 3, absolutely essential to her freedom As I argue in two ways below, however, Frankhrt7s conception of a self that can act only out of contingent psychological elements like desires renders this sort of evaluation impossible I argue first that the source of an agent's action can be considered her "owrt' only in an arbitrary way because, due to the nature of event-causality, its source can in fact be traced back to the necessitating effects

of infinite antecedent events I then argue that arbitrariness arises on the level of deep self-modification as well, which is a process ncxessary to resolve the first infinite regress

I argue that self-modification fails because contingent desires do not provide the kind of basis on which judgments of the appropriateness of a particular desire can non-arbitrarily

be made Both charges of arbitrariness, I must emphasize, relate to Frankfwt's rejection

of strong PAP

Because Frankfurt is intent on freeing the agent's will from surface-level psychological constraints such that her acts have their source in the volitions of her deep self, he must address the formation of the agent's deep self, the source of its content If

he fails to do so, he will not have shown why there is a difference in significance between the agent's surface-level and deep desires Frankfurt's view hinges on the claim that if an action is caused by the agent's deep desires, vvhich comprise the only self that she could possibly be, then the source of the action is in the agent But if the thefts of a kleptomaniac do not count the agent herself as their source because they are caused by her kleptomania, which is a force alien to her, Frankfurt must explain what the source of the agent's deep desires are such that they are significantly hers in the way that her neurotic desires are not If the nature of the agent's deep self is implanted in her at birth, for instance, and she is causally determined by whatever sort of volition-hierarchy her heredity has bestowed on her, then just as the source of her self is in some alien element, namely her heredity, so too is her action Just as her kleptomania-desires are implanted

in her, perhaps genetically, so too are her non-neurotic desires Unless there is some way for the agent to shape her own nature as a desirer, then, it seems that the distinction

between the source of a neurotic agent's action and the non-neurotic agent's deep self amount to the same thing, namely arbitraryN genetic factors

Some have argued that the only way to resolve the infinite regress that arises when the ultimate source of the agent's nature is considered is the ability to self-create, that is, the agent's ability to choose her ovvn nature Self-creation would require that the agent be able to create her own desires in order for her actions to stem from her in the way demanded The doctrine that "every event has a cause" entails that an agent's action, which is an event, is caused by the agent's desires This doctrine equally entails, however, that those desires are events that are caused, meaning that the ultimate source of the agent's action is not in the desire-evenlts, but rather in whatever in turn caused them The fact that the agent's creation of her own desires would also be a caused event means that the act of self-creation, the agent's catsing herself to be a certain way without in turn being caused by some other event, is impossible, as many, Galen Strawson chiefly, have pointed out.20 Therefore, Frankfurt's model must suggest a different way to stop the regress to infinite antecedent causal events that arises when the source of the agent's deep self is examined

Frankfurt attempts to resolve this conflict by reflecting that an agent has not only first-order desires, like the desire to eat a chocolate, not only second-order desires, like the desire to diet and therefore refi-ain from eating the chocolate, but also third-order desires, to lose weight, for example, along, with fourth-, fifth-, sixth-order desires, and so

l9 These factors are not random because they are controlled from the standpoint of evolutionary science or genetics, but they are arbitrary from the standpoint of agent control

20

Strawson, Galen "The Impossibility of Moral Responsibility" (1994)

on.21 The more consistently affirmed by the hierarchy a particular desire is, the more genuine to the agent it becomes "When a person identifies decisively with one of his

first-order desires, this commitment 'resounds' throughout the potentially endless array

of higher orders."22 When the agent associates strongly with these desires, particularly as the hierarchy extends and she reaches a level at which her core desires do not conflict, there is no sense in talking about her as something distinct from them The process of self-modification that Frankfix3 introduces attempts to use this resounding of desires to the deepest level of the self in order to affirm these deep desires as being the agent's own,

to affirm her deep self as being something of her own choosing or, more reasonably, of her own approving This way, Frankfurt can claim that when an action originates in such deeply resounding desires, regardless of their ultimate causal source, the charge of

arbitrariness that stemmed fi-om an infinite regress of antecedent events does not arise

Because Frankfurt reduces his agent to a collection of contingent motivational states that operate event-causally, however, he does not award her the kind of conscious control that would enable her to non-arbitraril,~ evaluate her own deep motivations in the way that his theory suggests This is for two reasons: first, because Frankfurt only enables the agent to evaluate according to further desires, that is to say contingent psychological elements, and second, because of his rejection of strong PAP I take each failure in turn

The resonance that Frankfurt claims secures the agent's deep desire as being of her own choosing supposedly involves a process of self-modification whereby she is able

to affirm or reject a particular desire as appropriate or inappropriate The agent's deep

21 I do not mean to imply that Frankfurt directly aims h!s herarchy view at resolving the problem of self- creation

22 Frankfurt (1971), p 332

desires, in other words, are not merely those that do not happen to conflict with any others Rather, in affirming a desire as being her own and constitutive of her deep self, the agent is making a claim about whether that desire is worthy of being acted upon When an agent claims that she acted out of a deep desire, she is claiming to have acted as the kind of person not just who she considers herself to be, but also as the kind of person who she fundamentally wants to be When a desire does not reflect this deep conception

of who she would like to be, an agent is alienated from the desire When she acts on an alienated desire, the agent has not acted freely When a kleptomaniac claims, for example, that she does not want to be a kleptomaniac, that she wants to obey the law, she

is making an evaluative claim that obeying the law is more worthy as a cause of her action than disobeying it is

But the basis for this evaluation, I argue, is arbitrary Because Frankfurt has required that a desire motivate every actioin of the agent's including her deliberation, one desire can only be evaluated on the basis of yet another one Desires, though, are not capable of recognizing what is normatively the case If it is true that obeying the law is better than shoplifting, desires are not the: sorts of things that will be able to recognize that truth Another way of saying this is that evaluating one desire according to another, which is the only option available to an agent who is constituted by her set of desires, merely introduces an infinite regress of desires, which, because no normative-evaluative apparatus is introduced to distinguish one as more appropriate than another, means that the condoning of one desire and the rejecting of another is arbitrary; it depends only on whatever complex of desires the agent already happens to have If Mike wants to suntan but at the same time wants healthy skin, the only means he has for evaluating the former

as less worthy than the latter is on the basis of some further deeply resonating desire, such as the desire to live a long life That Mike happens to want to live long does not establish the content of the desire to suntan as unworthy, though It merely means that the desire to live a long life has strong motivational force And as strongly motivating desires like eating chocolate and sleeping late indicate, the fact that a desire has strong motivational power does not directly correlate with the appropriateness of its content.23 For an agent to be able to assess her desires as being worthy or unworthy she instead needs the ability to evaluate a desire or a source of action based on reason, on rational or normative criteria

Without even addressing Frankfurt's rejection of strong PAP, it is clear that an agent who is able to self-evaluate only according to desires will not be capable of assessing one desire as worth having and another as not, thus rendering the neo-Humean notion of self incompatible with free will The insistence on the operation of desires as event-causal and the rejection of strong PAP makes matters worse When Frankfurt posits an agent who, according to her deep desires, approves of a particular action, he neglects to address the sense in which her feelling of approval is itself necessitated Every step in the process of self-modification is iiecessitated, including the outcome of the supposed assessments of particular desires and the resonance of particular desires throughout her sense of self Because it is contingent desires that cause the agent to identify with certain of her desires and not with others, it is arbitrary whether the agent happens to be constituted to identify with worthy desires or not

23 In fact, Frankfurt himself, when pressed to explain how it is that one desire can be established as more worthy than another without appealing to independent, objective criteria, concedes that the afflnning of one desire by another actually amounts to the agent's "identifying" with one desire and not another "Reply to Michael E Bratman" (2002)

When the neo-Humean claims, therefore, that she can assess the content of the agent's desires as correct or incorrect, superior or inferior, well or poorly founded, independently of the agent's contingent experience, she is neglecting the sense in which the content of the desires and the content of the evaluation also operate event-causally Typically, the content is considered to share identity with the agent's psychological states, i.e brain states If the brain states interact with necessitation, then the evaluative contents that share identity with them do as well This means that the basis for evaluation is event-causality and the laws of nature that regulate it, not the genuine normative status of the desires and the actions they motivate as being worthy or not

The neo-Humean agent, because she functions as a bundle of contingent dispositions with no strong PAP and no normative-evaluative apparatus, is unable to establish truths and genuinely defend her convictions or assessments in the way that a unified rational agent could If a neo-Humean agent is faced with the equation a + 6 = 10 and is asked what the value of a is, the question will strike this agent causally Her outputting the response 4 will be the result of a causal interaction between the question as external stimulus and the various beliefs and desires she has accrued from experiences in math classrooms, among many others Although the answer she is causally necessitated

to output is correct, at least according to thle rational agents who are able to evaluate these things, because her contingent motivatio~ial state necessitated her answer, she has no grounds on which to claim that her answer is convincing according to a set of independent criteria, as the rational agent might do The only criteria available to her are contingent psychological elements When asked the question, "why did you answer 4?," she will have to relate the various beliefs and desires, psychological motivations, that

were causally sufficient to produce that answer We will be able to ask of her "what is it about you, about yow past and your psychology that caused you to utter that number?" The answer "I answered 4 because it is th~e correct answer" will not be genuinely available to her because that 4 is the non-contangent correct answer was not the reason on which she acted, rather her psychological contingencies were

For the rational agent, in contrast, who is capable of recognizing accuracies and inaccuracies, of evaluating answers on the basis of their approximate likelihood, the answer will be precisely what is unavailablie to Frankfurt's neo-Humean, namely, "I

answered 4 because 4 is what a actually does equal and, hrthennore, I answered 4 because according to rational and revisable standards of judgment and consistency, there

is no reason why 4 should be false." Unlike the neo-Humean agent, such a rational agent would be capable of recognizing truths in the world and acting on those truths She would not be limited to her own contingenl psychology for the motivation to act or calculate in a certain way This is the sort of agent that Susan Wolf defends in the view that follows

Section III Susan Wolfs Reason View

My criticism of Frankhrt's deep-self view and acceptance of weak PAP establishes two key points First, some sort of self-modification process is necessary to resolve the infinite regress that is introduced with Strawson's demand for self-creation Second, on both the level of action and the level of self-modification, fieedom is incompatible with the neo-Humean self, particularly due to the fact that the neo-Hwnean can be motivated to act only by his contingent desires

Susan Wolf agrees that the neo-Hwnean self and its reliance on desire-dependent reasons render it incompatible with freedom In order to gain a non-arbitrary stopping point in the infinite regress of the evaluation of reasons for action, Wolf considers such reasons to be not desires, but rather true normative features of the external world In the math case, for example, in order to prevent the source of the agent's answer from spiraling into an infinite regress of further psychological elements, she claims that the correct answer itself, not some feature of the agent, causes the agent to produce it To provide normative reasons with the power to produce rational, free deliberation and action, which she agrees with me is not possible on the neo-Hurnean model, Wolf rejects both strong and weak PAP, arguing that such normative reasons must be deterministic They must have objective weights and directly cause agents to behave rationally.24 In this section I agree with Wolf that Frankfurt's model of self-modification calls for normative grounding, but disagree with her thesis that rationality requires the falsity of strong PAP

Wolfs assumption that determinism is necessary to secure agent freedom is motivated as follows She assumes that the only alternative to determinism is indeterministic randomness This alternative not only seems to deny the agent the ability

to cause her own actions but would also seem to alienate an agent's beliefs and values from what is true about the world Wolf does not understand why it would be desirable for an agent to have the option to disregard the normative reality around him.25 Just as

we cannot break physical laws by sheer willpower, we will naturally (and to our benefit)

24

Much of Wolfs argumentation focuses on her ethcal interest in defending praise and blame practices, depending on the individual case I do not focus on issues of praise and blame case by case, though, because I am more interested in using her view to show how rationality requires both normative reasons and the room for choice provided by strong PAP

25 Wolf (1990), p 55

be connected in necessitated normative ways with our external en~ironrnents.~~ We will

be determined by what is going on around us and by the perceptual and cognitive apparatuses with which we are born and which develop over time in response to external stimuli If determinism determines us to be rational or to act based on relevant reasons that truly exist beyond us, then the falsity of determinism would pose a far greater threat

to our accountability than the truth of it would For this reason Wolf rejects the requirement of radical autonomy that she attributes to Kant and ~ a r h - e ~ ~

There are three components of Wolfs view that stem from the motivation to ground an agent's actions in the world as it is The first is the ability to recognize reasons that there are, the second is the ability to act on them, and the third is the ability to structure oneself according to them Our system of responsibility, Wolf observes, holds agents morally accountable for their actions based on reasons that there are about what is right and what is wrong.28 "When we imagine an agent who performs right actions, it seems, we imagine an agent who is rightly determined: whose actions, that is, are determined by the right sorts of interests, and whose interests are determined by the right sorts of reason^."^' For an agent to be justifiably held accountable in this way, she must

be able to recognize and act on these external reasons This is the basis of Wolfs reason view.30 For an agent to be the appropriate recipient of blame attitudes, she must be able

26 "No one would suggest that freedom from empirical evidence is a valuable lund of freedom to have when forming scientific beliefs Having one's beliefs shaped by the world, then, is not all bad, and, more to the point, it is not always a qualification on or a hindrance tlo one's freedom and responsibility," Wolf (1990),

to perceive the world accurately The clondition of responsibility for Wolf, then, is a particular ability or capacity, namely the "complete possession of normal adult faculties

of reason and observation," which she labels ''sanityV." The primary division between potentially culpable agents is, in Wolfs view, between those who are sane and those who are not, as we have just established If the agent is able to identify a reason that is morally right and is able to act on it, then she is sane If, on the other hand, the agent's environment and background have been so severely limited or distorted that she is unable

to recognize reasons there are for the right kind of action or is unable to act on the identification of those reasons, she is insane.32

In addition to the abilities to recognize and act on reasons, the third necessary feature for freedom is the ability to self-conect An agent must be able to amend the individual reasons she has for acting according to the objective reasons that she identifies

in the norms around her.33 Wolf argues that "the ability to evaluate ourselves sensibly and accurately, and the ability to transfonm ourselves insofar as our evaluation tells us to

do so"34 allows her view to avoid the question of the ultimate source of an agent's deep self When a compatibilist claims that an agent's deep self determines his actions, the question of the source of the agent's deep self arises, as we saw with Frankfurt Using external, objective values to correct his ovvn externally-shaped deep self, the agent under Wolfs view is able to, in a sense, author his own behavior without having to self-

considerations to the agent, such that no one course of action is more praiseworthy or moral than any other See Wolf (1990), p 130

re ate.'^ Wolfs process of self-correction is, therefore, the counterpart to Frankfurt's proposed self-modification

Having set up the basic capacities necessary for freedom, namely the ability to recognize objective reasons, the ability to act on them, and the ability to structure oneself normatively using self-correction, Wolf makes two assumptions about what is required for such abilities to be attributable to the agenlt These are that rationality requires causal necessitation and that normative reasons must be fully determinate if they are to be objective I now consider each assumption in turn, demonstrating that they mistakenly deny the crucial interdependency between strong PAP and the notions of both rationality and normative reasons I also argue that VVolf's rejection of strong and weak PAP disqualifies her from characterizing her determined agent as having the "ability" to undergo any of the three processes she deems necessary for freedom

A natural way to accommodate the requirement that a responsible agent have the capacity to choose her actions according to reasons there are in the world is to posit the deliberative process undergone by the agent as the actual cause of the ensuing action The deliberative process will be one of identifying and weighing various reasons for one action or another; it will consist of a process we might call practical reasoning Richard Double, for example, argues that

[The most rational part] of our decision making is the deterministic part, e.g., our

consideration of the pros and cons of various options, strategies, and so forth In terms

of control, one is victim to any considerations that simply come to one

(indetenninistically or deterministically) until one can rationally reflect on and evaluate

point that, for an agent's action to be genuinely rational, it must be determined by the determined outcome of the deliberative process The idea is that only under causal determinism can the defensible reasons lead to the deliberative process that will in turn cause the actions those reasons "favor." If determinism does not hold, it is only pure coincidence if the reasons happen to map onto the action or the next deliberative stage accurately, and coincidence cannot be considered rational A rational action, both Double and wolf agree,37 must be deterministically caused by the right reason.38

Wolfs insistence that the ideal free and rational agent is determined both

physically and psychologically by the reasons there are, objectively, in the world, to perform the right action, reveals that she considers neither choice nor control to be necessary for agent freedom.39 An agent we regard as ideally rational, Wolf claims, is one who has a perfectly honed ability to recognize and perform the right act in every situation Ideal rational actions, in turn, are those caused by the True and the Good, meaning that the reasons there are for a particular action come ready-made with determinate weights Because the right reasons are external to the agent, and because the agent does not control the reasons' given worth, the agent who is perfectly rational in every situation does not choose, nor does she control, her behavior Instead, the objective reasons cause her behavior in that she is psychologically determined to recognize in them

37

Galen Strawson makes a similar assumption in hns hard determinist piece "The Impossibility of Moral Responsibility" (1994)

38

A tension arises here between the two compatibilist viewpoints discussed Frankfurt emphasizes that an

action must have the agent as its source in order to be considered her own, while Wolf argues that an action must have a reason as its source in order to be considered rational How, then, can the compatibilist render the source of an action both the agent's psychologically contingent self and an objective reason? I argue

only using a rational capacity that enables the agent to pick out the reason and act on it

39

Wolf actually directly considers the ideal moral argent because she is primarily concerned with the ideal according to whch we orient our system of praise and blame As stated in the beginning of this section, I

am more interested in the compatibility of rationality and determinism than morality and determinism

T h s does not betray Wolf's view because she would agree that the ideal moral agent is the rational ideal as well

what the ideal course of action would be." It is in no sense up to her whether a particular reason ought to be acted on over another

Denying that an agent be able to choose or control her actions in any way, however, is problematic On a model where each entity, even if it can be loosely described as having certain response-features or tendencies in common with other entities,

is fundamentally passive, merely manifesting these tendencies as a product of its given structure, the notion of normativity, of what sort of response would be appropriate for all members of entity type x, collapses Given one particular member of that set at a particular moment in time, there is not a range of reasons that could inform a set of possible decisions because the only "reason" or cause that is available to inform the

"decision" or effect that will come about at the next moment is the "state of affairs at

time t as oriented towards this individual."

The compatibilist will likely object that the psychological description of reasons

as being objective and expressing what it would be appropriate to do play a much stronger role than I give them credit for here It is not, they would say, that event causality operates on the level of the neurological, independent of and not influenced by the level of the psychological Rather, the level of the psychological plays just as cmcial

a role as the level of the physical They are two sides of the same coin One specific thesis that compatibilists offer to support this notion that the levels of the neurological and the psychological are entirely interdependent is the strict identity between brain states and mental states This view, often called psychologism, asserts that the human's mental state at time tl is the event that causes, given various external triggers, the event of her mental state at tz Features of the external situation the agent is in are event-causal

40 Wolf (1980), p 156

stimuli that interact with the physical event that is the agent's static mental state of believing that p, for example An accurate explanation of why the agent performed action x would have to reveal the event-causal interactions that took place among her mental states

I insist, in contrast, that there is a fundamental tension between these two levels and that the former, the level of psychological experience and intention, does not accomplish what its compatibilist supporters want it to accomplish when it is interpreted

as operating event-causally The level of the psychological relies on language of rationality, worthiness, and a regress-stopping deep self If the mental states associated with this level share identity with physical brain states and are therefore to be correctly interpreted as being events, then the relevant laws of nature along with the structures of the events involved drive the agent's deci~sion.~' Although we may describe the choice post-action as having involved the favoring of an aesthetic reason over a moral one, according to necessitating causality, giver1 the events, i.e mental state x at time tl, only the mental state y is possible at time t2 Because an omniscient observer could predict with infinite accuracy the state of affairs at t2 given the state of affairs at tl, the mental state 'a at t2' simply does not and cannot exist Though the agent experiences the act she faces as a choice, it is not one in the eyes of the observer In order to predict the state of affairs at t2, the observer does not need to conceive of the "choice" facing the agent as being one of reason x over reason y because in the eyes of the observer such an assessment of greater or lesser worthiness does not genuinely exist or play a role The argument, in short, is that if the omniscient observer need not conceive of time t~ as

41

See Jonathan Dancy's Practical Reality, (2000) See especially h s objection to Donald Davidson, p

162

representing a genuine choice of aesthetic reason x or moral reason y, then we cannot assume that conceiving of the choice this way plays a causal role in the outcome of the agent's "deliberation."

If the world is universally deterministic, fwtherrnore, every action, not just actions

we may describe as being rational, is causally necessitated The reason for every action

is the cause that necessitated it and therefore every action has its source in a "reason." There is therefore no way to distinguish the rationally caused from the irrationally caused, given that the language of genuine choice is fundamentally ungrounded on this view The rational, I claim contra Double, is distinguished from the irrational through the selection of one option among many for the right reasons, rather than choosing on the basis of the more appealing but wrong ones The strong version of PAP is necessary for the possibility of rationality because genuine alternative possibilities enable a rational choice to be made over an irrational one.42 If the deliberative process is instead the workings of a mechanism, which has been programmed to output rigidly patterned actions, the source of any given action is the program, not whatever reason we may externally claim makes sense of the action Equipped only with the set program and the externally-derived input information, we will be able to determine the outputs that will be produced by the mechanism without having to take any reason into account Given this,

the source of two different actions produced by two different mechanisms will be the same in evaluative status, namely their program plus the given inputs Even if, given the same situation, one mechanism outputs the action "save drowning child" and the other

42

As Searle notes, "rationality is possible only where irrationality is possible, and that requirement entails the possibility of choosing between various rational options as well as irrational options," Rationalitv in Action, (2001), p 17

outputs "hum Yankee Doodle Dandy," one will not qualify as rational and the other as irrational because the full explanation for both will be the same

Just as rationality cannot be made sense of on a model of necessitating causality, neither can intention, which is the internal commitment to a particular course of action This is because setting out to do a particular thing implies the favoring of it over other options It implies that the agent has selected this action and has a justification that makes the choice defensible A justification for the action would provide support for the agent's intention pre-action, support that would render the choice intelligibly defensible not just according to the individual's psychological structure but according to an objective rational standard Such justification cannot be reduced to a causal explanation

of action, which is simply that the internal and external causal network was such and such

at time t Thus, universal event-causation, which Wolf here requires for her ideal rational agent, radically eliminates the possibility of the agent providing genuine justification for his response One agent's response is elqual in status to another's because the single available explanation is that it was casually necessitated to be so If all beliefs are caused

by internal and external stimuli, then an individual cannot be accused of holding a belief irrationally, or without sufficient justification This is because the basis on which she holds that belief is identical in kind to the justification that any other agent could potentially offer for holding his belief, namely, "I was caused to so believe," whether mechanistically by her given cognitive program as Wolf would have it or contingently by her given complex of desires and beliefs as Frankfurt claims

The previous paragraphs argue that rationality and determinism are incompatible, thus rejecting Wolfs characterization of the processes of the recognition of objective

reasons and acting according to them as rational I next argue that the notion of ability, the ability to self-correct specifically, also collapses under determinism." This shows that Wolf cannot characterize her mechanistic agent as having the ability to self-correct,

to shape herself, according to the reasons there are, but will have to instead reduce

"ability" to the tendency to be "reliably caused to act in a certain way." Because Wolf, like Frankhrt with the process of self-modification, proposes the ability to self-correct as necessary for attributing the agent's actions to her in any significant sense, I show that under determinism whether or not an agent undergoes this process and whether she does

so success~lly will be arbitrary Just as the ol~jective reasons deterministically cause the agent to recognize and act on them, they also cause her to restructure according to them

If the right reasons are the causal source of rational agent action, failure to self-correct reflects the right reasons' failure to cause the agent's self to restructure appropriately An agent has no basis on which to choose her own mechanistic self, nor does she choose how objective reasons affect the self she has Whether or not an agent happens to have a reasons-responsive structure that is caused regularly to self-correct is not up to her, but rather a matter of forces beyond her control It is not the case that every sane agent is

corrected by reasons while others are not

On the level of free moral action, the notion of the ability to undergo practical, theoretical, and moral reasoning is also debilitated by this model of reliable causation

On a causal model, there is no privileged, active entity that is producing reactions independent of necessitating influence Rather, every shift in the causal network as a

43 The claim that PAP is necessary for the notion of ability is an argument that Wolf agrees on when it comes to the agent who fails to behave rationally, but an argument that she bizarrely rejects when it comes

to her ideal moral agent I do not consider t h s asymmetry here

whole qualifies only as a response to what has come before it, not as an action Just as it

is inappropriate to consider a ball to have acted rightly or rationally in rolling down a h11 after all, it wouldn't be a ball if it didn't exhibit ball-qualifying behaviors we cannot consider a psychological configuration of causal links with a certain structure to be fi-ee given that it actually exhibits that structure when causally provoked

It is this very position in fact that forces Wolf to commit to psychological indeterminism for the sane agent who fails to behave rationally If it is the case that deterministic deliberation causes the agent to perform the appropriate action according to the reasons' objective weights, the reasons will be responsible for the agent's behavior, which they, not the agent, cause If the compatibilist has been satisfied to reduce the notion of ability to mere reliable causation, he cannot provide an explanation for the culpability of the agent who happens to have been caused unreliably Even though compatibilists may initially fail to perceive the significance of the collapse of the notion

of ability into the notion of being reliably caused, making clear what ability cannot mean

on the event-causal model makes the preservation of responsibility-conferring ability to

be defended in chapter 3 more obviously significant, particularly in cases where the agent fails to behave rationally

This section presents Wolfs reason view, which seeks, rightly, to ground the agent's actions normatively and rationally Wolf rejects the neo-Hurnean notion of self championed by Frankfurt and critiqued in the last section, but pushes her notion of agent into a mechanistic extreme Considering reasons to be mere causes, I argue, actually strips them of their normative status I have thus argued that Wolfs three essential freedom-conferring processes of recognizing reasons, acting on them, and self-correcting