UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Agentive experience compatibilism Deery, Oisin 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_fall_deery_oisin.pdf [ 476.55kB ]
Metadata
JSON: 24-1.0074120.json
JSON-LD: 24-1.0074120-ld.json
RDF/XML (Pretty): 24-1.0074120-rdf.xml
RDF/JSON: 24-1.0074120-rdf.json
Turtle: 24-1.0074120-turtle.txt
N-Triples: 24-1.0074120-rdf-ntriples.txt
Original Record: 24-1.0074120-source.json
Full Text
24-1.0074120-fulltext.txt
Citation
24-1.0074120.ris

Full Text

Agentive Experience CompatibilismbyOis??n DeeryB.A. Philosophy and English, The National University of Ireland, Galway, 1993M.A. Philosophy, University College, Cork, 2006A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE STUDIES(Philosophy)The University Of British Columbia(Vancouver)August 2013c? Ois??n Deery, 2013AbstractLibertarians, who think that freedom is incompatible with determinism and we arefree, claim that their view is descriptively right about how we view ourselves asagents who are free to do otherwise. Much of what compatibilists, who think thatfreedom is compatible with determinism, have written in recent years about thefreedom to do otherwise has consisted in attempts to deflate these sorts of libertar-ian claims. Philosophers on each side thereby make claims about the nature of ourexperience and beliefs. These are empirical claims, which can be illuminated byempirical methods. In experiments that I ran with Matt Bedke and Shaun Nichols,our participants described their experience of being free to do otherwise as incom-patibilist across a range of conditions. Compatibilists may dismiss these resultsby insisting that people are mistaken in their introspective judgments. Or theymight insist that people only seem to have incompatibilist beliefs about freedomand determinism; in fact, these beliefs are about the compatibility of freedom andfatalism. I argue that both these compatibilist claims are false, at least in the formsthat they currently take. Instead, I argue that compatibilists should concede thatpeople?s experience and beliefs are in part libertarian, but can still be accurate ifdeterminism is true. First, I assume that experiences of being free to do otherwisehave phenomenal content that is inaccurate if determinism is true, just as libertar-ians claim. Then I argue that such experiences also have phenomenal content thatis accurate if determinism is true. So an experience with libertarian content canbe accurate under determinism. Second, I argue that implicitly libertarian beliefscan be accurate, even assuming determinism. Only when one makes an explicitiiincompatibility judgment is one?s belief false. Thus, implicitly libertarian beliefsare not incompatibilist. My view does not provide a full compatibilist theory ofthe ability to do otherwise. Still, on my view our experiences and beliefs concern-ing such freedom are consistent with determinism. My view also explains anytemptation we may feel to judge our experience and beliefs as inconsistent withdeterminism.iiiPrefaceChapter 2 is based on a co-authored paper with Matt Bedke and Shaun Nichols,with Ois??n Deery as first author. The identification of the research program andthe primary design of the experiments conducted as part of the research were un-dertaken by Ois??n Deery. The training-to-crirerion for determinism that we usedin the experiments was introduced by Shaun Nichols, and was subsequently devel-oped by all three authors. Data analysis on the experiments conducted in Study 1was performed by Matt Bedke. All subsequent data analysis on studies 2 and 3was conducted by Ois??n Deery. Preparation of the manuscript was performed byall three authors.At the time of submission of this dissertation, Chapter 2 has been publishedin D. Shoemaker (ed.), Oxford Studies in Agency and Responsibility: Vol. I. Thismaterial is reprinted with the permission of Oxford University Press. It has beenaltered in various ways for the sake of maintaining consistency with the rest of thedissertation.For the experiments conducted as part of the research reported in Chapter 2,approval was obtained from the Behavioural Research Ethics Board at the Uni-versity of British Columbia. The Certificate Number of the Ethics Certificateobtained is H10-01096.ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 How to Think about Free Will . . . . . . . . . . . . . . . . . . . 11.2 Determinism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.3 Free Will . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.4 The Experience of Freedom . . . . . . . . . . . . . . . . . . . . . 261.5 Outline of the Project . . . . . . . . . . . . . . . . . . . . . . . . 312 Phenomenal Abilities: Incompatibilism and the Experience of Agency 322.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 332.1.2 The Experience of the Ability to Do Otherwise . . . . . . 342.1.3 Previous Work on the Phenomenology of Free Will . . . . 372.2 Study 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402.3 Study 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462.4 Study 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51v2.5 General Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 562.5.1 Incompatibilism . . . . . . . . . . . . . . . . . . . . . . . 562.5.2 The Ability to Do Otherwise . . . . . . . . . . . . . . . . 572.5.3 Misinterpreting One?s Agentive Experience . . . . . . . . 592.5.4 Deliberation Compatibilism . . . . . . . . . . . . . . . . 612.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633 Is Agentive Experience Compatible with Determinism? . . . . . . . 653.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653.2 Agentive Experience . . . . . . . . . . . . . . . . . . . . . . . . 673.3 Horgan?s View . . . . . . . . . . . . . . . . . . . . . . . . . . . . 723.4 Problems with Horgan?s View . . . . . . . . . . . . . . . . . . . 783.5 Problems with Horgan?s Contextualism . . . . . . . . . . . . . . . 823.6 An Alternative Compatibilist Proposal . . . . . . . . . . . . . . . 863.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904 Against an Argument for Libertarianism . . . . . . . . . . . . . . . 924.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 924.2 Compatibilist Strategies . . . . . . . . . . . . . . . . . . . . . . . 934.3 An Analogy with Color . . . . . . . . . . . . . . . . . . . . . . . 954.4 Libertarian Content . . . . . . . . . . . . . . . . . . . . . . . . . 994.5 Compatibilist Content . . . . . . . . . . . . . . . . . . . . . . . . 1014.6 The Two-Stage View . . . . . . . . . . . . . . . . . . . . . . . . 1054.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1075 Indeterminism, Experience, and Compatibilism . . . . . . . . . . . 1095.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1095.2 Do Indeterminist Beliefs Have Their Source in Experience? . . . . 1125.3 A Forward-Looking Model . . . . . . . . . . . . . . . . . . . . . 1195.4 Etiology of Indeterminist Experience . . . . . . . . . . . . . . . . 1235.4.1 Prospection is Forward-Looking . . . . . . . . . . . . . . 123vi5.4.2 Prospection Can Be Experienced . . . . . . . . . . . . . . 1265.4.3 Prospected Choices Are Free Variables . . . . . . . . . . . 1275.4.4 Indeterminist Experience and Judgment . . . . . . . . . . 1325.5 Belief in Indeterminist Freedom is Not Justified . . . . . . . . . . 1345.5.1 Radical Compatibilism . . . . . . . . . . . . . . . . . . . 1365.5.2 Contextualism Revisited . . . . . . . . . . . . . . . . . . 1385.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1446 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1466.1 What the Dissertation Achieves . . . . . . . . . . . . . . . . . . . 1466.2 New Evidence on the Descriptive Question . . . . . . . . . . . . . 1496.3 Against a Compatibilist Error Theory . . . . . . . . . . . . . . . . 1496.4 An Alternative Compatibilist Proposal . . . . . . . . . . . . . . . 1516.5 The Source of Libertarian Content . . . . . . . . . . . . . . . . . 1546.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158A Survey Materials for Chapter 2 . . . . . . . . . . . . . . . . . . . . . 172viiAcknowledgmentsTo my wife, Katherine Bailey: thank you for your encouragement and continuedbelief in me. I do not know whether I can ever repay my debt of gratitude to you;but I will try.I thank my advisor, Paul Russell, and my dissertation committee, includingDerk Pereboom (Cornell University) and Murat Aydede for their advice and guid-ance. I thank Matt Bedke, also on my committee, and, at the University of Ari-zona, Shaun Nichols, for their critical input to Chapter 2. At the University ofArizona, I thank Terry Horgan, Michael McKenna, and Uriah Kriegel for theirsupport during my stay there, and Jenann Ismael for her inspiring ideas. Thanksto Mark Balaguer at California State University, Los Angeles, for his suggestionsregarding Chapter 4, and to Peter Railton at the University of Michigan, Ann Ar-bor, for correspondence that proved crucial to Chapter 5. For advice on how towrite philosophy, my thanks to Dominic McIver Lopes at UBC. Finally, I thankmy examination committee, especially Jonathan Jenkins Ichikawa and Eddy Nah-mias, who (in addition to those listed above) gave me valuable comments for myfinal revisions.I was supported at UBC by a Graduate Entrance Scholarship, a UBC GraduateFellowship, a Tina and Morris Wagner Foundation Fellowship, and (for one year)by a Four-Year Doctoral Fellowship. I was also supported by a Pacific CenturyGraduate Scholarship, a joint award administered by UBC and The Ministry ofAdvanced Education, Province of British Columbia. In 2010, I was supported bya Doctoral Fellowship awarded by the Social Sciences and Humanities ResearchCouncil of Canada. In 2011, I held a Visiting Graduate Fellowship at CornellUniversity.viiiChapter 1IntroductionI?ve visited thirty-one inhabited planets in the universe. Only onEarth is there any talk of free will.? Kurt Vonnegut, Slaughterhouse-Five1.1 How to Think about Free WillTake ?free will? to mean the freedom to choose among alternative possibilities,1and ?determinism? to mean that, given the past and the laws of nature, there is, atany given time, just one physically possible future.2 Indeterminism is the oppositeof this thesis: holding fixed the past and laws, there is more than one physicallypossible future.Do we have free will? To answer this question, we must clarify what we meanby ?free will,? even in the sense of being free to choose among alternatives. Isfreedom compatible with determinism? Or with indeterminism? Let us focus onthe first question.There are several reasons why one might be interested in free will. First, onemight be interested in how to distinguish actions done from weakness of will1 I discuss other senses of ?free will? later.2 I also discuss the thesis of determinism in greater detail later.1from actions performed under compulsion. Here, the question whether freedomis compatible with determinism may not even arise. Alternatively, one might beinterested in the sort of freedom-relevant control required for moral responsibil-ity, quite apart from whether such control demands that agents be able to chooseamong alternatives.3 Finally, one might take it as a basic insight that being free tochoose among alternatives for action, and thus being able to do otherwise, is fun-damental to the notion of free will. Moreover, one might think that our possessingsuch an ability is confirmed by our experience of choosing. In this dissertation, Iam concerned exclusively with the latter reason for being interested in free will,as I explain below.There are a number of standard positions that one can occupy about free will.Incompatibilists think that free will is incompatible with determinism. For in-compatibilists, the fact that determinism entails that there is only one physicallypossible future means that we lack alternatives for action. So, if free will requiresbeing able to choose among alternatives, free will is inconsistent with determin-ism. Some incompatibilists?the libertarians?maintain that we have free will,and as a consequence they think that determinism is false. The task for libertari-ans, as we shall see, is to show how indeterminism is supposed to help with freewill.4 Other incompatibilists?traditionally called hard determinists?think thatwe lack free will because determinism is true. Today, though, few philosophersare willing to commit to the truth of determinism. Instead, the current inheritorview of hard determinism is hard incompatibilism, whose advocates maintain thatfree will is incompatible with determinism, but is also probably incompatible withindeterminism. As a result, most likely we lack free will whether determinism istrue or false.53 Many philosophers now think that free will?at least when thought of as the sort of freedomor control required for moral responsibility?does not require having alternative possibilities. Idiscuss this issue later.4 Since I focus on compatibility with determinism, I will not discuss further the questionwhether free will is compatible with indeterminism, other than by way of a few brief comments inthe next paragraph.5 There is also the ?impossibilist? view, according to which free will (in the sense required forresponsibility) is impossible, since it is ruled out by both determinism and indeterminism (e.g.,G. Strawson 1994).2By contrast, compatibilists think that free will is consistent with determinism.As a result, if free will requires being able to choose among alternatives, thencompatibilists must show how choosing among alternatives is possible, assumingdeterminism. Typically, compatibilists also argue that libertarian views are inco-herent, or deviate from a naturalistic account of the world, or fail to show howindeterminism provides for more control than determinism does. Compatibilistswho press this latter point may further argue that indeterminism hinders control.After all, one might think that for agents to control their decisions, their decisionsmust flow smoothly from their beliefs and desires. Indeterminism threatens thispicture. For, when we hold fixed the entire history of an indeterministic world(including the internal states of the agent) right up until the moment of the agent?sdecision, what the agent ends up deciding is indeterminate right up until the mo-ment of her decision. This is so despite the agent?s beliefs, desires, and reasons fordeciding one way rather than another. Such ?randomness? threatens to undermineagents? control over their decisions, and thus free will?or so compatibilists claim.Traditionally, compatibilists have also thought that determinism is true, since theythought that undetermined events are uncaused, which seemed absurd. Given thatevents are caused, compatibilists thought, events are determined. Today, however,most compatibilists accept that indeterminism might be true. Despite this, they in-sist that agents have free will only if indeterminism plays no control-underminingrole in the production of decisions. Moreover, even if determinism is true, that isno threat to freedom.I focus primarily on the question whether free will is compatible with deter-minism. This question?the compatibility question?has come to dominate theliterature on free will in recent decades. Many philosophers think that the way toanswer this question is to show whether there is a possible world in which deter-minism is true, yet in which an agent is able to do otherwise. If so, compatibilismis vindicated; otherwise not. That is not primarily how I will approach matters.At least at first, I will focus on the ?natural? compatibility question, which is a de-3scriptive question about people?s pre-theoretic belief-tendencies about their ownand others? agency. The natural compatibility question asks whether people beginas compatibilists, prior to their consideration of philosophical theories. Of course,even establishing whether people are natural compatibilists in some respect (forinstance, about the freedom to do otherwise) would leave open the question howwe ought to think about our agency in that respect?about whether we should betheoretical compatibilists, say. Even so, it may inform the approach that we takewhen addressing that question.Whether people are natural compatibilists in some respect is straightforwardlyan empirical question, which I will call the descriptive question. Once we have an-swered this question, we then address the substantive question: are people in factagents of the sort that they tend to believe they are (in a given respect)? Finally,there is the prescriptive question, which asks how we should best theorize aboutfree will (in a particular respect), given how we have answered the descriptive orsubstantive questions. If the answer to the descriptive question is that people tendto believe that they have a sort of freedom that requires conditions G, H, and Kto be satisfied, yet it is reasonable on independent grounds to think that only K isactually satisfied, then our best prescriptive theory of freedom may have to departfrom our pre-theoretic beliefs.6There are a number of underlying issues here?related, for example, to philo-sophical methodology, the reference of concepts, and so on?which I want tobracket. In what follows, I mostly adopt the approach of methodological natural-ism.As I see things, methodological naturalists (henceforth, ?naturalists?) are pri-6 Others who think about the compatibility question in roughly this way are Nichols (2008),Vargas (2013), and Balaguer (e.g., 2004). On Balaguer?s way of viewing matters, the what-is-free-will question is a semantic or conceptual question about the meaning of ?free will.? UnlikeBalaguer, I avoid putting things in semantic terms, since ?free will? (or ?freedom?) most likely doesnot have a single meaning. As with any term that has been in our language for some time, thereare many (perhaps overlapping) senses of the term, and it seems unlikely that any one of these isthe meaning of that term.4marily concerned with explaining aspects of the world. To this end, they beginwith a target phenomenon. In order to establish what the phenomenon?s charac-teristic features are, naturalists rely on a combination of (i) ordinary judgmentsor belief-tendencies about supposed cases of the phenomenon, (ii) other rele-vant background judgments or beliefs, and (iii) supplementary empirical evidence.Naturalists think that we should establish what the contents are of the judgmentsmentioned in both (i) and (ii) empirically, by finding out what people?s belief-tendencies actually happen to be. Empirical methods also bear on this procedurein another way, by enabling us to determine whether agents? judgments have beenproduced by reliable mechanisms. If not, such judgments may be explained awayby an error theory. Also taken into considerations from the outset is (iii) supple-mentary evidence not directly related to establishing what our beliefs are or to as-sessing their reliability, yet which is otherwise relevant to the target phenomenon(see e.g., Paul 2010 for an overview of this method).We then build an explanatory theory of the phenomenon. In doing so, wetreat all our judgments, as well as our explanatory hypotheses, as defeasible inlight of new findings. As we proceed, each part of our theorizing is constrainedat each step by empirical evidence. In this way, we arrive at theories of the targetphenomenon, and we decide among them (if necessary) by inference to the bestexplanation.In the chapters that follow, I conclude that people?s experience of freedomis, in a sense, incompatibilist. People tend to experience possessing (and thusto believe that they possess) a sort of freedom to do otherwise that they cannotpossess if determinism is true. As a result, if determinism is true there seemsto be a gap between the sort of freedom that people experience possessing andthe sort of freedom that they do possess. Despite this, I defend a compatibilistanswer to the prescriptive question about experiences of freedom, by showinghow such experience is, in an important sense, usually veridical (and thereforecompatibilist), even assuming determinism. I also propose an explanation forhow incompatibilist experiences are generated, and in such a way that we are5not warranted in believing libertarianism simply on the basis of experience, assome libertarians have suggested that we are. I also maintain that our freedomto do otherwise is compatibilist. In this way, I defend natural incompatibilismabout the descriptive question, but normative compatibilism about the prescriptivequestion.7In the next two sections, I say more about the items whose compatibility isat issue in debates about free will: determinism and freedom. After that, I saysomething further about each of the descriptive and prescriptive questions, as wellas outlining why I focus on the experience of freedom. In the last section, I pro-vide an overview of the thematic and argumentative structure of the subsequentchapters.1.2 DeterminismThe thesis of determinism that concerns us is nomological determinism. This isthe thesis that all events are determined to occur by natural?usually physical?laws. As Kadri Vihvelin notes (2011; I rely heavily on Vihvelin?s article in whatfollows), this is a contingent thesis about the laws of nature. It says, first, thatthe laws are not probabilistic. Deterministic laws entail exceptionless regularities,such as that all Fs are Gs?i.e., Fs have an objective probability of 1 of beingGs.8 By contrast, probabilistic laws state that Fs have an objective probability ofless than 1 of being Gs. Second, determinism says that the totality of the laws isall-encompassing?the totality of the laws applies to everything in the universe,not just to a part of it. So, determinism is defined as follows: the conjunctionof a complete statement of the non-relational facts of the world at a time, t, witha complete statement of the laws of nature, entails all other non-relational truthsabout the world at times other than t, including truths about future human deci-7 I do not directly address the substantive question; I leave it aside as work for another day.8 Although determinism, as defined here, does not require a non-Humean account of laws,I will assume that the laws are non-Humean, as I explain in the next paragraph. I reassess thisassumption in Chapter 5.6sions and actions.9 The main consequence of this thesis is that at any given timethere is only one physically possible future, given the past and the laws?there isjust one way that it is physically possible for the world to unfold.10Even when it comes to all-encompassing deterministic laws, there is a furtherimportant distinction to be made between Humean and non-Humean laws.11 Thedifference lies in whether the truth of a law is established by events that fail todisconfirm it, or whether something else establishes the truth of a law, so thatevents conform to it. The first account of laws is Humean, while the second isnon-Humean. This is important, since on a Humean account if determinism istrue then a law that determines whether an agent, S, performs an action, A, at aparticular time, t, does not ?settle,?12 as John Perry puts it (2004), whether S A-s att, since S?s A-ing at t is part of the sequence of events that establishes the truth of alaw, not something that is settled by it. According to Perry, on a Humean accountof laws, ?What we do is up to us; laws are merely descriptions of what we do thatwill end up being true once human activities are complete. Laws determine, butdo not settle? (2004: 239). By contrast, on a non-Humean account, if determinism9 Vihvelin (2011) identifies three ways a world might be indeterministic or non-deterministic:(i) some of its laws may be probabilistic, (ii) the totality of its laws may not be all-encompassing,or (iii) it may have no laws. Vihvelin only calls worlds that are non-deterministic in the firstway ?probabilistic worlds,? since these worlds have probabilistic laws. Worlds that are non-deterministic in the second way she calls ?partly lawless,? while those that are non-deterministicin the third way are ?lawless.?10 I will say more in Chapter 5 about whether this is a good way to understand determinism. Itis, at any rate, the standard way.11 For broadly Humean, non-necessitarian accounts of the laws of nature, see Lewis (1973;1983; 1994) and Earman (1984). For non-Humean, necessitarian accounts, see Armstrong (1978;1983), Dretske (1977), and Tooley (1977).12 Here is how Perry explains what he means by ?settled.? He outlines three sorts of propositions:(1) those that are made true by events (e.g., in the way that the proposition ?I was born in Dublin?is made true by the event of my actually having been born in Dublin), (2) those that are not madetrue by events (e.g., Pythagoras?s theorem), and (3) those that while not yet made true by eventsnonetheless have their truth value settled. If the propositions describing the laws of nature are notmade true by events (i.e., if the laws are like Pythagoras?s theorem, and thus are non-Humean),then if descriptions of the laws of nature and the facts of the past entail another proposition, p,about the future, the truth value of p is settled, in Perry?s parlance, even though p has not yet beenmade true by events.7is true then the laws together with the facts of the past not only entail facts aboutthe future, but also settle them.Many philosophers, Perry included, find Humean accounts of laws unconvinc-ing. As Perry puts it, ?It seems to me much more plausible that . . . [a] . . . law getsat something (or some things) about the universe that explains why things con-form to the law and it has no disconfirming instances? (2004: 240). I will assumea non-Humean viewpoint, since it prevails in the literature on free will. Futureevents are settled, in Perry?s parlance, by the laws; events conform to the laws,rather than the laws? being made true by the sequence of events.13It is also important to distinguish determinism from fatalism (see e.g., Bern-stein 2002). Unlike determinism, fatalism does not count our deliberations asnecessarily exerting any causal influence on events. Fatalism is the thesis thatwhatever is fated to happen will happen no matter what one does, or?as Petervan Inwagen puts it??no matter what choices and decisions one makes? (1983:25, n. 3).14 Determinism, by contrast, implies that one?s deliberations and actionsdo exert an influence on events.Along with fatalism, there are several other theses with which determinismmight be confused. It is useful, therefore, to note what else determinism does notsay. First, determinism does not say anything about the predictability of events. Inparticular, it does not say that, in virtue of their being deterministic, events are easyto predict. Chaos or complexity theory says that events in complex deterministicsystems are difficult to predict. Yet, according to some interpretations of quantummechanics, certain behaviors of indeterministic systems are easy to predict (seee.g., Earman 2004).Second, determinism is not a thesis about causation. Here, I mean two things:(a) causation need not be deterministic, and (b) determinism need not?and ar-guably should not?be stated in terms of causation at all. Both points are im-portant, since especially in the older literature on free will (e.g., Hobart 1934;13 Again, I revisit this issue in Chapter 5.14 Even indeterministic fatalism seems to be coherent, at least when the thesis applies only tosome events.8Ayer 1954) it was often assumed that determinism just is the thesis that everyevent has a cause. By contrast, it is now widely accepted that the laws may beprobabilistic, and causation may be probabilistic too: events might have causeseven if determinism is false.15 The conceptual possibility of indeterministic cau-sation was noted decades ago by Elizabeth Anscombe (1981/1971). Accordingto Anscombe, we often claim to know whether one event caused another, evenwhen we do not know whether the second event was determined by the first event.Moreover, the concept of causation seems to be of ?giving rise to,? whereas theconcept of determinism is of ?having to give rise to.? So indeterministic causationseems conceptually coherent.16I maintain that we can (and arguably should) think of determinism in non-causal terms in any case?the prevalence in the free-will literature of talking about?causal determinism? notwithstanding.17 John Earman argues that we should dropthe term ?causal? from discussions of determinism, since to speak of ?causal de-15 Two points are worth noting here. First, Vihvelin observes that, ?[I]t might be true thatevery event has a cause even if our world is neither deterministic nor probabilistic. If there can becauses without laws (if a particular event, object, or person can be a cause, for instance, withoutinstantiating a law), then it might be true, even at a lawless or partly lawless world, that every eventhas a cause? (2011). Second, Vihvelin also points out that it is unclear ?whether determinismentails the thesis that every event has a cause. Whether it does so or not depends on what thecorrect theory of causation is; in particular, it depends on what the correct theory says about therelation between causation and law? (2011).16 Of course, the possibility that indeterminism might be true goes back even further, to thedevelopment of quantum mechanics in the 1920s. According to quantum indeterminacy, a physicalsystem does not have any determinate state that determines the values of its measurable properties.The system has only a defined set of probable states. On the Copenhagen interpretation of quantummechanics, this probability is objective. The process of measuring the properties of the systemresults (somehow or other) in just one of these objectively probable states becoming instantiated.In effect, indeterminacy is built right into the fabric of the universe. On this interpretation, therecan be no laws in the sense of exceptionless regularities stating that all Fs are Gs (i.e., with aprobability of 1). (On other interpretations of quantum mechanics, such as the ?hidden-variables?or the ?many-worlds? interpretations, the apparent indeterminacy in quantum mechanics is merelyapparent and can be accounted for deterministically.) Yet, while an indeterministic theory ofcausation may rely on, say, the truth of the Copenhagen interpretation of quantum mechanics,quantum mechanics stands in no need of a theory of causation. In this way, as I explain below,whenever we speak of determinism or indeterminism in quantum mechanics, we are speaking interms of an entailment notion of determinism, not a causal one?where entailment is understoodbroadly enough to include mathematical consequence.17 A caveat to this claim will be introduced in Chapter 5.9terminism? is just ?to explain a vague concept?determinism?in terms of a trulyobscure one?causation? (Earman 1986: 5).18 If instead we think about deter-minism in entailment terms, then determinism requires only that (i) there be awell-defined state or description of the world at a time, t, and (ii) there be excep-tionless laws of nature that are true at all times and places. If (i) and (ii) logicallyentail complete descriptions of the world at times other than t, then determinism istrue; otherwise not. Although (i) is false if the standard interpretation of quantummechanics is correct (see footnote 16), and (ii) is false if the laws of nature areprobabilistic (or if there are no laws), all that is required to articulate the thesisof determinism other than (i) and (ii) is the notion of entailment, which should beunderstood broadly enough to include mathematical consequence as it appears intheoretical physics.19Third, and finally, determinism neither entails nor is entailed by physicalism,the thesis that everything is physical or is (in some sense) necessitated by the18 See also van Inwagen (1983: 65).19 A slight wrinkle here is that, as Derk Pereboom notes (unpublished handout), Frankfurtcases appear to show that preceding conditions might entail that an agent, S, performs an action,A, without playing any part in actually bringing it about that S A-s: the preceding conditionsentail that S A-s, yet they do not cause A (and so they do not causally determine A). This issupposed to be a problem for the entailment view of determinism since we are meant not to assumedeterminism in Frankfurt cases, yet the entailment view appears to do just that, if S?s A-ing isentailed by the preceding conditions. Thus, we should prefer a causal formulation of determinism.In response, recall my definition of determinism: the conjunction of a complete statement of thenon-relational facts of the world at a time, t, with a complete statement of the laws of nature,entails all other non-relational facts about the world. However, the preceding conditions to whichPereboom alludes are, it seems to me, relational facts. So, by adopting an entailment view ofdeterminism, it is not true that one begs the question against incompatibilists in Frankfurt cases.To explain: Consider that it is a hard (non-relational) fact relative to 2p.m. that Jones eats noodlesat 1p.m. By contrast, it is a soft (relational) fact relative to 2p.m. that Jones eats noodles for twohours after having started to eat them at 1p.m. Now, the relevant fact in a Frankfurt case at t1is meant to be something like this: If Black sees a prior sign at t2 indicating that Jones is aboutto choose to B at t3, then Black will intervene to ensure that Jones A-s at t3. That seems like arelational fact. Thus, it is not the kind of fact that features in a statement of determinism. So,even if such a fact entails that Jones A-s at t3, it does not thereby illicitly assume determinism.More importantly, the facts appealed to in the thesis of determinism are stated in the languageof physics: they are physical facts. The fact that supposedly entails that Jones A-s at t3 is not aphysical fact. Thus, even if such a fact entails that Jones A-s at t3, it does not thereby illicitlyassume determinism.10physical. Peter van Inwagen (2000) thinks that the problem of free will is soabstract that it must arise in roughly the same form even in a world inhabitedonly by immaterial beings: either such beings obey deterministic laws, or they donot.20 Conversely, there are also possible worlds in which determinism is false, yetphysicalism true (our own may be such a one). However, the form of nomologicaldeterminism that is generally thought to threaten human freedom is stated in termsof physical laws.Henceforth, whenever I speak of determinism I will have in mind nomologicaldeterminism, which I take to be a thesis best stated in terms of logical entailmentand physical laws.21 At bottom, this thesis says?minimally?that there are noobjectively probabilistic physical laws governing our world, and, as a result, ob-jectively probabilistic events do not occur.22 The main consequence of this thesis(if it is true) is that there is only one possible way for the actual world to unfold,physically speaking.Why worry about determinism? After all, the standard interpretation of quan-tum mechanics is indeterministic.23 On this view, our world is indeterministic atleast in the sense that some of its physical laws are objectively probabilistic. If so,determinism is false and is no threat to freedom. One response to this proposalis to say that even if quantum indeterminism is true, human behavior may stillbe ?near-determined?:24 it may be that quantum indeterminacies make no differ-ence at the macro level of humans or neurons (or whatever else). In this way,physics might be indeterministic, while neuroscience, for instance, is neverthelessnear-deterministic.20 Clearly, however, these laws will not?however they are to be stated?be stated in terms ofphysical theories.21 Again, an amendment of this way of thinking about determinism will be introduced in Chap-ter 5.22 A minor caveat: some non-Humean, necessitarian accounts of laws are driven by a theoryof dispositions, ?according to which dispositions have their causal powers essentially. . . . Laws,then, are entailed by the essences of dispositions? (Carroll 2011; Cf. Bird 2005). On this view,causation is a more fundamental notion than law. Thus, causation may enter into an account ofnomological determinism via the back door, so to speak?that is, via a dispositional account ofnatural laws. In what follows, I leave aside this technicality.23 See footnote 16 above.24 The term ?near-determined? is introduced in this context by Honderich (1993).11There is an important distinction to be made here. First, it might be thatquantum-level indeterminacies do not have any effect at the neural level, so thatdeterminism is false when stated in terms of physical laws, yet true when statedin terms of neuroscientific laws. In that case, neural determinism is true. So thereis no need to speak of near-determinism.25 Second, it might be that quantum-level indeterminacies do, as a matter of fact, have effects at a neuronal level, butnot anything like as often?or in such a way?as to allow for indeterminism indecision-making.26 In that case, neural near-determinism is true. Even so, it ap-pears that we have nearly as much reason to worry about near-determinism as wedo to worry about determinism, since neural near-determinism presents almost asmuch of a prima facie threat to free will as does determinism, yet is much morelikely to be true.Alternatively, the world might be near-deterministic in the sense of being im-perfectly deterministic. That is, the universe (or our region of it) might be deter-ministic for long periods of time (say, millions of years), yet occasionally theremight occur spontaneous particle-creation events that violate physical determin-ism (Cf. Hoefer 2010). Thus, determinism could be false as a general thesis aboutour universe. Nevertheless, our world would be sufficiently deterministic to posea prima facie threat to free will.25 To see how quantum indeterminism might effect decision-making, see Dennett (1978: 46?47). To paraphrase: Imagine that my correctly answering No to an easy Yes/No question either bypushing a pedal with my left foot or by pushing a button with my left finger (where either actionwould register the correct answer) depends on the output of a quantum-indeterministic randomizer.Say I want to answer No. Now the randomizer outputs an instruction telling me whether I shouldpress the pedal or push the button. This output?and therefore my action of, say, pressing thebutton?is (per hypothesi) undetermined, even if intrinsically I am a deterministic system. Thequestion is whether such amplification of indeterminism from the micro- to the macro-level ispossible without such a device.26 Mere indeterminism, even if it occurs at the moment of choice, seems insufficient for freewill, even for incompatibilists.121.3 Free WillAccording to one way of thinking about human agency, free will is moral freedom,or the ?freedom-relevant condition necessary for moral responsibility? (Campbellet al., 2004: 2). We possess such freedom when we control our actions in thestrongest manner necessary for being morally responsible for them.27 In otherwords, to judge an agent, S, as morally responsible for performing an action, A, itmust have been the case that S, when she A-ed, exercised this necessary form ofcontrol over her A-ing.In thinking about the sort of control required for moral freedom, it is stan-dard to distinguish between leeway and sourcehood views. Leeway views requirethat an agent be able to do otherwise in order to be morally responsible, whilesourcehood views do not. Sourcehood theorists argue, instead, that an agent mustin some relevant sense be the source of her actions in order to be responsible forthem, where this requirement is spelled out in a way that does not require that theagent have been able to do otherwise. On another way of thinking about matters,free will is (or requires) being able to do otherwise. Call such freedom modalfreedom.28 It is a further question, on this taxonomy, whether moral freedom re-quires modal freedom. Traditionally, it was held by almost all parties to debatesabout free will that modal freedom is a necessary condition on moral freedom.The existence of sourcehood views demonstrates that this is no longer the case.Notice, too, that we can also think about free will independently of respon-sibility, solely in terms of human freedom and abilities. By analogy, considerour interest in personal identity. One reason that the topic of personal identity isimportant to us is because of its relevance for moral responsibility. We want toknow what the criteria are for an agent?s remaining the same person over timebecause we want to be able to connect the person whom we apprehend today for27 Thus, there could be conditions sufficient for the control required for moral responsibilitythat are stronger than this.28 Here, I follow the example of Holton (2010).13the crime of murder to the same person (according to the criteria) who committedthat murder yesterday. However, this is hardly the only reason we are interestedin personal identity. Another reason is that we want to know whether we are thekind of creature that we conceive ourselves to be. This question presses on us inits own right. It may turn out that there are no viable criteria for personal iden-tity. Such an answer would impact what we believe about ourselves, and mayundermine a view of ourselves that we value, independently of any further con-siderations regarding moral responsibility.29 Likewise, I suggest, modal freedomis important to us independently of moral freedom. Perhaps we tend to think thatwe possess a sort of freedom or ability that, in fact, we lack. If so, this may impactwhat we believe about ourselves, and it may undermine a view of ourselves thatwe value, perhaps deeply, and this may be so quite independently of any furthervalue-relevant considerations to do with responsibility.If we do think about free will independently of responsibility, just in terms offreedom and abilities, then we are interested solely in modal freedom, which isarguably more fundamental than moral freedom anyhow, since, as Campbell andcolleagues note (2004), the difficulties regarding modal freedom can be statedwithout reference to moral responsibility.30 In what follows, I am concerned onlywith modal freedom.A traditional analysis of free will, conceived as modal freedom, begins asfollows:(1) S has free will only if S is able to do otherwise than she does.Here, modal freedom is claimed to be (at least) a necessary condition on free will.Of course, if free will just is modal freedom (as e.g., van Inwagen 2008 maintains),29 See Parfit (1971; 1987) and Velleman (2006) for such views about personal identity.30 One difficulty regarding modal freedom is characterizing what it is to be able to do otherwise.I should also note that it is not obvious whether modal freedom can be characterized withoutreference to moral notions. Van Inwagen (2000: 17?18) characterizes what it is for an agent, S, tobe able to A partly in terms of whether S is in a position to promise that she will A. Yet, promisingis a moral notion.14then the ability to do otherwise is both a necessary and a sufficient condition onfreedom. Thus:(2) S has free will iff S is able to do otherwise than she does.In either case, the freedom at issue is the ability to decide between mutually in-compatible courses of action. If we think about free will independently of moralresponsibility?as what I have been calling modal freedom?then whatever wethink about the requirements of moral responsibility, we will still want to knowwhether we are free to do other than we do (and thus whether we have free will).As I mentioned earlier, in order to answer this question we must first establishwhether being free to do otherwise is compatible with determinism. One influ-ential compatibilist thought is that freedom should be understood in contrast toconstraint, or coercion. According to this proposal, an agent is able to do other-wise just in case, if she had chosen, or wanted, or tried to do otherwise, then shewould have done so (Cf. Moore 1912). By contrast, incompatibilists think thatbeing able to do otherwise requires being free to do something other than whatone does, all prior conditions (including one?s beliefs, desires, etc.) remaining thesame.We are now in a position to see why ?free will? is actually an unhelpful term.When we ask whether an agent is free in a given situation, our answer depends onwhat we mean by ?free.? Even in debates about free will, it may be unclear whetherwe mean moral freedom or modal freedom (or: whether the latter is a conditionon the former). I suggest that instead of talking about free will, we should insteadtalk about the various more fine-grained distinctions and notions that are of inter-est in such debates. When asking compatibility questions, I think we do better toask about the consistency of determinism and moral responsibility, or sourcehoodcontrol, or the freedom to do otherwise, and so on. Once we distinguish thesequestions, we may answer some of them differently than others. To the questionwhether the conditions of moral responsibility are compatible with determinism,15our answer might be ?Yes,? while our answer to the question whether the freedomto do otherwise is compatible with determinism might be ?No.? Thus, we couldbe compatibilists about responsibility, yet incompatibilists about the freedom todo otherwise (or vice versa).31 Whether we are natural compatibilists about anyof these notions and determinism will, once we hold fixed the definition of deter-minism, depend entirely on our presuppositions (or belief-tendencies) about theother phenomenon.32An account of the ability to do otherwise is required for any adequate accountof ourselves as agents. Even sourcehood theorists, who deny that the freedom tochoose among alternatives is required for moral responsibility, still presumablythink that sometimes we are able (in some sense) to do otherwise.33 We need anaccount of such freedom. One of the first questions that arises in developing suchan account is:The freedom-to-do-otherwise-compatibility question: Is the freedom to do oth-erwise compatible with determinism?One powerful reason for thinking that the freedom or ability to do otherwise is notcompatible with determinism is the consequence argument. This argument relieson the notion of an agent?s lacking power over a fact. For an agent, S, to lack31 See e.g., Fischer (1994, 2007, 2013) and Fischer and Ravizza (1998) for a version of theformer sort of view in the philosophical literature.32 When it comes to the intuitions people have about free will (or moral responsibility, or theability to do otherwise, or determinism, etc.), it is worth noting that, psychologically-speaking,one should be hesitant to conclude categorically that any individual person either does or does nothave incompatibilist intuitions. According to empirical results obtained by Deery and colleagues(in preparation), people?s pre-theoretic folk intuitions about free will are often naturally both in-compatibilist and compatibilist. Deery and colleagues found that when free will is understood asan agent?s being the ultimate source of her actions, for instance, participants in their studies agreedwith incompatibilist statements. Yet, participants also agreed with compatibilist statements of thisconception of free will, and to a similar extent. Thus, all else being equal, when respondents werenot placed in experimental situations that required them to resolve an explicit conflict betweenopposing intuitions, they possessed both incompatibilist and compatibilist intuitions about beingthe ultimate source of their actions.33 A deeper issue is whether incompatibilist sourcehood theorists require alternatives covertly,by requiring indeterminism.16power over a fact, A, is for S to be unable to act in such a way as to ensure thefalsity of A. Furthermore, if S lacks power over A, and lacks power over whetheranother fact, B, follows from A, then S lacks power over B. Powerlessness trans-fers from a fact to its consequences. This principle is central to the consequenceargument, as the argument?s name suggests. Recall that determinism implies thatS?s action at t is a consequence of the distant past and the laws.34 If S lacks powerover the distant past and the laws, and S lacks power over whether her action at tfollows from the past and the laws, then S lacks power over her present and futureactions.There is a vast literature on this argument, which I want to bracket. Ac-cording to the way in which I suggest that we approach the freedom-to-do-otherwise-compatibility question, any answer to this question will depend on whatpeople actually tend to believe about their abilities. In other words, we mustfirst answer the descriptive question about people?s relevant belief-tendencies.Holding fixed the definition of determinism, the answer to the freedom-to-do-otherwise-compatibility question?understood as the natural compatibilityquestion?depends solely on the answer to this descriptive question. In this way,we find out out whether people are natural compatibilists about freedom. So theanswer to the freedom-to-do-otherwise-compatibility question depends on the fol-lowing question:The freedom-to-do-otherwise-descriptive question: What sort of freedom dopeople tend to believe they have?How should we answer this question? One way is simply to ask people abouttheir belief-tendencies. Another way, which I will adopt, is to ask people abouttheir experience of being free to do otherwise. Appeals to the experience of free-dom have a long history in free-will debates, going back at least to Thomas Reid(1788), who argued that our experience of incompatibilist freedom justifies belief34 More carefully, determinism implies that a statement describing an action of S?s is a conse-quence of a conjunction of a statement of the laws and a statement of the facts of the world at atime in the distant past.17in libertarianism. Compatibilists, too, have appealed to our experience of freedomin order to deny such incompatibilist claims: they insist instead that our expe-rience carries with it only compatibilist commitments. In Chapter 2, I presentevidence showing that, as it turns out, people describe their experience of free-dom as incompatibilist. Before turning to these issues, however, it will be helpfulto say a few words about the sort of freedom that is at stake.First, we need to distinguish alternative possibilities from the freedom to dootherwise. Having the latter entails having the former, but not vice versa. Even ifindeterminism is true and we have indeterministic alternatives, we might lack theability to do otherwise. Imagine that you are in a runaway car, in which the steer-ing mechanism and brakes are not working. Here, there might exist alternatives?the car might go, even indeterministically, in various directions?but you lack theability to make it go in any direction at all, or even stop. This point also appliesif we assume determinism, at least on the assumption that there are compatibilistalternative possibilities. (I will call both compatibilist and incompatibilist alterna-tives metaphysical, in order to avoid begging the question against compatibilistsby assuming?as is done sometimes?that only incompatibilist alternatives aremetaphysical.)We must also distinguish metaphysical from epistemic possibilities. Unlikethe former, the latter are uncontroversially compatibilist. For an event to be epis-temically possible is just for an agent not to know whether that event will occur.Consider again the example of the runaway car. Given that you do not knowwhether the car will go left or right, its going in either direction is epistemicallypossible, even if one direction is disallowed by the (deterministic or indeterminis-tic) laws of nature.Regarding ?can,? I will use ?can do otherwise? as predicative of an ability(which may or may not be compatibilist) to choose among metaphysical, notmerely epistemic, alternative possibilities. This usage differs from that which18indicates moral or legal permissibility.35 Using ?can? in a permissibility sense,you might insist that ?You can?t do that!? where (for instance) I am in the processof stealing a car. What you mean is that my stealing the car is morally or legallyimpermissible.36There are several other common senses of ?can? that differ from the one inwhich I am interested. One of these is predicative of a disposition, another of apower, and yet another of a capacity or general ability. I will take each of thesein turn.When we say that a cup can break, what we mean is that it is disposed to breakwhen struck. This usage of ?can? attributes a disposition. On some views, a poweris not essentially different from a disposition: a power is simply a disposition toproduce a specific manifestation under certain circumstances. Thus, dispositionsare causal properties, which are identical to their causal bases, e.g., being fragilejust is having a certain physical structure. In this sense, a wineglass has a power,in being fragile. Some dispositions, however, clearly count as powers in a differentsense. Consider a steel hammer, which has a power in the aforementioned senseby having a high tensile strength: a disposition to deform in a certain way onlyunder high strain. Yet it also has a power in another sense: it has the power tosmash wineglasses, or ceramic cups, under certain conditions. A ceramic cupmight also have a power in this sense in relation to a wineglass. In this sense of?power,? a hammer is an Aristotelian agent. Something (e.g., a steel hammer) isan Aristotelian agent if whenever it acts on a patient (e.g., a wineglass) the changeoccurs in the patient.3735 Cf. van Inwagen (1983: 8).36 If I respond by saying, ?Oh, can?t I?? then I am predicating of myself an ability. One way tocash out the sort of permissibility talk discussed in the text along the lines of the possibility talkoutlined earlier is to say that the full meaning of ?S can (in the permissibility sense) A? depends onthe facts with which S?s A-ing is compossible, where these are moral or legal facts, or whateverapparently fact-like things make up the content of morality, law, rationality, etc. (Cf. Hobbsunpublished.)37 See e.g., Gill (1980).19There is a sense in which humans are agents that is distinct from the way inwhich hammers are agents. For instance, when we say that someone can speakGaelic, we are ascribing to her a general ability?a power relating her to a poten-tial action (speaking Gaelic is active, whereas understanding it is merely passive).Having an ability of any sort is to have what van Inwagen calls ?a power to orig-inate changes in the environment? (1983: 11). It seems that our originating suchchanges is (partly) what differentiates us as agents from hammers. We, unlikehammers, have abilities to do things, or to make things happen. On this taxonomy,it remains open whether powers or abilities just are, or are somehow analyzablein terms of, dispositions.38 This also highlights a way in which it is different toattribute an ability to an agent than it is to say that something can happen. Afterall, even if I can be hit by a meteor, we do not want to say that I have the abilityto be hit by a meteor.39General abilities are compatible with determinism: I retain the general abilityto jump up and down on one foot at time t, without?in another sense?being ableto exercise it just then, perhaps because I am asleep. Determinism is compatiblewith my having, at t, a general, but at that time unexerciseable, ability to jumpup and down on one foot.40 Only specific abilities are at issue in debates aboutfree will. In particular, we are interested in agents? specific ability to A at t, or torefrain from A-ing at t.Plausibly, general and specific abilities41 lie on a continuum, with the mostgeneral sort of ability at one end, and the specific ability to act and to refrain at theother. What happens in moving from the most general to the most specific abilityis that the facts relevant to whether we actually possess the ability attributed tous become more narrowly specified. So, by analogy with David Lewis?s (1986)38 See e.g., Vihvelin (2004), Smith (2004), and Fara (2008), for optimistic views in this regard.See Clarke (2009) for a sobering appraisal of such optimism.39 I owe this point to James Hobbs (unpublished).40 Note that unexercised specific abilities are also compatible with determinism; see below.41 See Honore? (1964) and Mele (2003) for somewhat different discussions of general and spe-cific abilities.20contextualist account of ?can,? most generally it is true that I am able to speakJapanese, even if I do not know how to speak Japanese.42 Clearly, I am intrinsi-cally constituted in such a way that I posses the properties that would allow me tospeak Japanese in the right circumstances: nothing is preventing me from learningit. In this sense, I am able to speak Japanese, whereas an orangutan is not. Thiscontrasts with a less (but still) general sense of ability. Assuming that I do notknow how to speak Japanese, when I am among tourists and someone inquires ofme whether I am able to speak Japanese, it is (as it happens) correct for me to re-ply, ?No.? After all, I have not taken the time to learn it. Even having been calledupon to speak Japanese, and despite my suitable intrinsic constitution, no matterhow motivated I am to speak Japanese I still lack the specific ability to speak aword of that language.We all possess both general and specific abilities to do things that we never do,and perhaps never will do. After all, even if I never learn how to speak Japanese,nevertheless I am still able (in the sense outlined above) to speak it. More specifi-cally, if I am on holiday in Spain and someone offers me criadillas for lunch, I amable to eat them, although it is certain that I will refuse. Often, we try and fail todo things that we possess a specific ability to do. For instance, I am able to play?Recuerdos de la Alhambra? on guitar, and I possess the specific ability to do soright now. Yet, it is a difficult piece, and so I might try and fail on this occasion(Cf. Austin 1956).What is a specific ability? We might think that my having a specific abilityto play the guitar at a time, t, is for me (a) to have a general ability to do so, and42 Lewis (1986) has a contextualist view of ?can,? according to which when we say that anagent, S, can perform an action, A, we are saying that S?s A-ing is compossible with certain facts.Here, whether it is true that S can A may depend on something left implicit?the set of facts withwhich S?s A-ing is compossible. Lewis?s view is contextualist since the meaning of ?can? doesnot, by itself, determine what facts are relevant; the additionally relevant facts are determined bycontext. So while all uses of ?can? share a semantic element?they express compatibility withcertain facts?the precise meaning of particular use of ?can? depends on something else, whichLewis calls ?context.? So, ?S can A? means that S?s A-ing is compatible with certain facts, wherethe relevant facts depend on the stringency with which ?can? is used. I discuss Lewis?s contextualistproposal in Chapter 3.21(b) to have an opportunity to exercise my general ability at t. Naturally, I retainthe general ability to play the guitar even when I am miles away from one. If aguitar were brought before me at t, then I would gain an opportunity to exercise mygeneral ability. Thus, I might possess the specific ability to play the guitar at t. Yeton this view, my specific ability to play the guitar at t and to refrain from playingthe guitar at t turns out to be obviously compatible with determinism. In orderfor the view to account for the ability at issue in questions about whether modalfreedom is compatible with determinism, we would need to say more about what itmeans to have an opportunity, and we cannot stipulate that having an opportunityis an incompatibilist notion, since doing so would beg the question against thecompatibilist.Perhaps, then, an agent, S, has a specific ability to A at t only if (i) S has ageneral ability to A at t; (ii) S has an opportunity to A at t; and (iii) holding fixedthe past and the laws (including S?s motivations regarding her opportunities asthey are at t), S can exercise her general ability to A at t. If S has such an abilityregarding two actions, A and B (where A 6= B), at t, then S has the specific abilityto do otherwise at t. Here, (i) and (ii) are neutral on the compatibility issue. Tolocate the point of contention between compatibilist and incompatibilist, we addthat even holding fixed S?s motivations regarding her opportunities as they are att, S can exercise her general ability to A at t and her general ability to refrain fromA-ing at t by B-ing instead. Compatibilists think that, assuming determinism, S isfree to do otherwise. Incompatibilists think that S is free to do otherwise only ifdeterminism is false.How are we to think about this disagreement between compatibilist and in-compatibilist? After all, one might think it obvious that if determinism is true andthere is only one physically possible future, then no one is free to do otherwisethan they do. Accounting for how agents are able to do otherwise in a determin-istic world is one of the deepest challenges that compatibilists face. There havebeen many proposals, and as many failures. Before mentioning some of these,22it may be helpful to lay a little groundwork in order to show what such a theorywould have to look like.To begin with, Alfred Mele (2003; 2006: 17?25) introduces a useful dis-tinction between ?simple abilities? (?S-abilities?) and ?intentional abilities? (?I-abilities?), each of which may be general or specific, depending on whether onehas the opportunity to perform the action in question. I will restrict discussion tospecific abilities. I might have the S-ability to roll a six when I roll a die, in asense in which I never have an S-ability to lift a bus, for instance. My having oncethrown a six is, as Mele puts it (2006: 18), conceptually sufficient for my havingthe S-ability to throw a six right now. Yet it is not conceptually necessary thatone already have done something in order to possess the S-ability to do it, sinceeven an isolated native of the Amazon rainforest who is given a die for the veryfirst time presumably also has the S-ability to throw a six. Thus, an agent S has aspecific S-ability to perform an action A at a time t iff there is a possible world inwhich S A-s at t.43By contrast, I possess the I-ability to roll a die right now as long as I have adie to hand, but not the I-ability to roll a six. After all, I am not plausibly ableintentionally to roll a six, and I do not intentionally roll a six just because on someoccasion I intended to do so and I got lucky and did. So an agent S has a specificI-ability to A at t iff there is a relevant possible world in which S intentionallyA-s at t (where this entails that S has an S-ability to A, but is not entailed by S?spossessing an S-ability). The idea is that we ?graft? onto an account of a specificS-ability an account of intentional action, so that S?s A-ing also counts as S?sexercising an I-ability.In trying to understand the disagreement between compatibilists and incom-patibilists about the specific ability to do otherwise, we might ask what worldswe should consider when assessing whether S has an I-ability to perform morethan one action at a time. As Mele puts it, ?One way to see the disagreement43 Likewise, S has a general S-ability at t to A at t iff there is a possible world in which S A-s atsome time.23between incompatibilists and compatibilists about determinism and being able todo otherwise is as a disagreement about what worlds are relevant? (2006: 21). Forincompatibilists, ?all and only worlds with the same past and natural laws as W[the actual world] are relevant; they hold the past and the laws fixed? (21). Forcompatibilists, by contrast, worlds with different pasts and natural laws than theactual world are relevant to whether S is able to perform or refrain from perform-ing a given action at time t. Thus, assuming that in the actual world W an agent Sdoes not perform some action A at time t, Mele suggests that we might define anintentional libertarian ability (L-ability) in the following way: An agent S has, att, the L-ability to A intentionally at t in W, the actual world, iff there is a possibleworld with the same past and laws as W in which S A-s intentionally at t (Cf. Mele2006: 19).Compatibilists think that there is a sense in which agents are able to do oth-erwise that is consistent with determinism. They think that worlds with differentpasts and laws than the actual world are relevant to judging whether agents havethe ability to do otherwise at a given time in the actual world. Their accountsdiffer mainly in the way that they go about making use of such worlds. For in-stance, David Lewis (1981) argues if S A-s at t, nevertheless S had the specificability to refrain from A-ing at t in the actual world, as long as there is a possibleworld in which S?s not A-ing at t is permitted by a small local exception to thelaws. This is the ?local miracle? view. On the ?backtracking? view, if S A-s at t,nevertheless S had the specific ability to refrain from A-ing at t, as long as thereis a possible world in which the laws are the actual laws, but events in that worldprior to S?s A-ing are sufficiently different from those of the actual world to allowfor S?s refraining from A-ing (Bennett 1984; Cf. Saunders 1968). For advocatesof the compatibilist ?conditional analysis? (e.g., Moore 1912), S is able to A andto refrain from A-ing at t iff were S to choose to A at t, S would A at t, and wereS to choose to refrain from A-ing at t, S would refrain from A-ing at t.44 Some44 This analysis fails, since the following conditional might be true: If S chose to refrain fromA-ing, then S would refrain from A-ing. Yet, S might be unable to refrain from A-ing becauseshe is unable to choose to refrain (Lehrer 1968: 32). There are recent revivals of this conditional24compatibilists understand the ability to do otherwise less stringently, by insistingthat S is able to A and to refrain from A-ing at t, even if, holding fixed the lawsand the state of the world at t, S is unable to exercise her general ability to refrainfrom A-ing at t.One might worry that if we characterize the specific ability to do otherwise asI did earlier, then it is difficult to see how compatibilists and incompatibilists aredisagreeing. Instead, they seem merely to be talking past one another. Recall, Isuggested that an agent S has a specific ability to A at t only if (i) S has a generalability to A at t; (ii) S has an opportunity to A at t; and (iii) holding fixed the pastand the laws (including S?s motivations regarding her opportunities as they areat t), S can exercise her general ability to A at t. If S has an ability in this senseregarding two distinct actions at t, then S has the specific ability to do otherwise.The point of contention between compatibilists and incompatibilists is meant tobe (iii). Yet, if compatibilists think that worlds with different pasts and laws thanthe actual world are relevant to whether agents are able to do otherwise at a giventime in the actual world, but incompatibilists deny this, then each party is talkingpast the other.One way to sharpen this worry is to use Angelika Kratzer?s (1977) univocal se-mantics for the modal terms ?can? and ?must.? This approach is similar to Lewis?s(1986) contextualist account of ?can,? mentioned earlier. According to Kratzer,?can? plays a similar role wherever it appears. She captures this similarity by uni-vocally treating ?can? as an existential quantifier over worlds that are restrictedby a contextual ?in-view-of? clause.45 Thus, when I say, ?You cannot move thepawn three spaces ahead,? what I mean is, ?In view of the rules of chess, youcannot move the pawn three spaces ahead.? That is true. However, once the con-textually restricting ?in-view-of? clause is altered, what I say might be false. Forinstance, if I mean, ?In view of your physical constitution, you cannot move theor dispositionalist view (e.g., Vihvelin 2004; Fara 2008). See also Lehrer 1976 for a differentcompatibilist view.45 ?Must,? meanwhile, is treated by Kratzer as a universal quantifier over a contextually restrictedset of worlds.25pawn three spaces ahead,? presumably that is false. A view of this sort spelled outin terms of one?s being able to do otherwise at a time could also be made univo-cally to treat ?is able to? as an existential quantifier over worlds restricted by thefollowing ?in-view-of? clause: ?In view of the actual past and the actual laws, Sis able to A at t or to refrain from A-ing at t by B-ing instead.? One might theninsist that unless the compatibilist and incompatibilist agree on this restriction onthe set of worlds to focus on, they are not treating ?is able to? univocally, and soare not disagreeing.46The strongest compatibilist position about the ability to do otherwise is one onwhich the compatibilist agrees to this contextual restriction on the set of worldsthat are relevant. In the chapters that follow, I tell a compatibilist story aboutthe experience of being able to do otherwise, and I will suggest a compatibilistaccount of such an ability that from the outset agrees that this is sometimes anappropriate sort of restriction to have in place when assessing claims about theability to do otherwise.47For now, however, let me return to the natural compatibility question, which isa descriptive question about people?s actual belief tendencies regarding their free-dom. In what fallows, I suggest that agentive experience is relevant to addressingthis question.1.4 The Experience of FreedomAccording to the way in which I have suggested we should address the questionwhether modal freedom is compatible with determinism, we must first ascertainwhat people?s belief-tendencies are regarding freedom. In other words, before wecan address46 I am indebted to Alex Grzankowski for suggesting this way of putting things.47 See Chapter 5 (section 5.5.1) for details about how my compatibilist view differs from moretraditional compatibilist views.26The freedom-to-do-otherwise-compatibility question: Is the freedom to do oth-erwise compatible with determinism?we must first answerThe freedom-to-do-otherwise-descriptive question: What sort of freedom dopeople tend to believe they have?The compatibility question we are asking here is whether people are free in theway that they tend to believe they are, assuming determinism.48 This is one of thecentral questions that grips people when they first consider their agency in relationto determinism. After all, when people first entertain the compatibility question,they do not think about the transfer-of-powerlessness principle that lies at the cen-ter of debate about the consequence argument, or the complex Frankfurt-cases atthe center of debates about whether responsibility requires alternative possibilities(Cf. Frankfurt 1969). Rather, they are struck by the thought that determinism robsthem of their freedom to do otherwise. (Or so I argue in Chapter 2.) People arestruck by the fact that determinism implies that there is only a single physicallypossible future: ?If that?s right,? people think, ?then I?m not free in the way that Ithought I was.?The crucial question, then, is the freedom-to-do-otherwise-descriptive ques-tion, on which the answer to the compatibility question depends. Furthermore, thefreedom-to-do-otherwise-descriptive question is an empirical question. If peopletend to believe that they possess an incompatibilist freedom to do otherwise, thenthe answer to the compatibility question will be that people are natural incompati-bilists about such freedom. It will be a further question, of course, whether peopleare actually libertarian agents. That, too, is an empirical question, the answer towhich depends on what our best scientific theories say about the nature of theuniverse and human decision-making. This is what I earlier called the substantivequestion. I leave this question aside as work for another day. I assume that it is48 Or ?near determinism?; see section 1.2 above.27unlikely that people are libertarian agents, and I further assume (at least for ar-gument?s sake) that determinism is true. Thus, I address only the descriptive andprescriptive questions, and only in relation to the question whether the freedom todo otherwise is compatibilist.How should we address the descriptive question? One way is simply to goout and empirically measure people?s belief-tendencies about the freedom to dootherwise. In other words, we might gauge people?s intuitions about being freeto do otherwise. Some empirical work has already been done on this issue byexperimental philosophers, although with inconclusive results (e.g., Nichols 2004;Turner and Nahmias 2006).I noted earlier that some philosophers think that our being free to do other-wise is confirmed by our experience of deliberation and choice. In section 1.3, Inoted that incompatibilists have long appealed to this sort of experience in orderto explain belief in libertarian free will, and also to justify such beliefs. Mostcompatibilists, of course, deny such claims. Instead, they claim that our experi-ence carries with it only compatibilist commitments. This suggests another wayin which we might gauge whether people?s commitments about freedom are com-patibilist: try to find out whether people?s experience of freedom is veridical,assuming determinism:The experience-compatibility question: Is people?s experience of being free todo otherwise veridical, assuming determinism?Just as with the freedom-to-do-otherwise compatibility question, the answer to theexperience-compatibility question depends on our answer to a related descriptivequestion:The experience-descriptive question: What sort of freedom to do otherwise dopeople experience having?What I am proposing is that experiences are as legitimate a starting point as intu-itions are for investigating people?s commitments about freedom. Leaving aside28questions about the justificatory status of intuitions, intuitions appear at least tobe reliable indicators of how things seem to be to a person (Cf. Bealer 1998).Furthermore, having the intuition that p often motivates one (to some degree) toaccept the content p.49 Thus, one may have a tendency of some strength to be-lieve that p. By doing empirical work on the freedom-to-do-otherwise-descriptivequestion, philosophers treat intuitions as reliable indicators of how things seem tobe to people, regarding their modal freedom. Is there another reliable indicator ofthis sort? Yes, there is. Another way in which things might seem to be a certainway is experientially. For instance, a stick in the water might seem to be bent:this seeming is experiential. So, there are different ways that something mightseem to be the case. As a result, there there are different ways in which one mightbe motivated to believe that a content, p, is veridical. Intuitions (as ?intellectualseeming-states?) are one way, while experiences (?experiential seeming-states?)are another.50I will assume that people?s introspective reports about their experience canbe taken at face value. In assuming this, I follow Uriah Kriegel (forthcoming)in steering a course between ?introspective dogmatism,? according to which in-trospection is infallible, and ?introspective skepticism,? according to which intro-spection is utterly unreliable. Kriegel defends ?introspective moderation,? whichis the view that introspection has above-chance reliability. Kriegel thinks thatintrospection is at least as trustworthy as a normal adult?s sense of smell, for in-stance. After all, he says, my smelling raspberries makes it more probable thatthere are in fact raspberries nearby than if I do not smell them; and when there areactually raspberries nearby, it is more likely that I will smell them than if thereare none. Likewise, my introspecting a certain phenomenology, P, makes it more49 This motivation may, in some cases, be based on one?s relative competence in an area. SeeSymons (2008) and Bedke (2008) for useful discussions of intuitions in philosophy. See alsoJenkins (forthcoming).50 Here, I am deliberately running together both perceptual and introspective ?seeming? states.There is much to be said (and that has been said) about the verb ?to seem? and other ?appearwords? (see, e.g., Chisholm 1957; Brogaard 2013). However, these issues are orthogonal to mypresent concerns.29likely that I have P than if I do not introspect it; and my having P makes it morelikely that I will indeed introspect P than if I do not have P. Of course, for in-trospection (as for smell), this claim about reliability may hold only for normalsubjects under normal conditions. It might not hold for elusive phenomenologies.For instance, there may be ?fringe? phenomenologies, such as my peripheral senseof my legs pressing against my chair. When focused upon, this phenomenologyloses its fringe quality. Thus, it seems not to be introspectible, since it disappearsas soon as I turn my introspective attention upon it. Yet, once we grant thesecaveats, it seems plausible that introspection does indeed have above-chance reli-ability.If someone introspectively reports having an experience as of being free to dootherwise that has libertarian content, the default position that I will adopt is togrant that they indeed have such phenomenology. If experiential seeming-statesdrive belief in much the same way that intuitions do, then someone?s introspectivereport about her experience of being free to do otherwise is roughly as reliable anindicator of her being disposed to believe that she possesses such freedom as therelevant intuition would be. In other words, if people report having experiencesthat are libertarian in nature, then, in the absence of countervailing considerations,we may assume that people actually do tend to believe that they possess libertarianfreedom.51One item stands in need of clarification. When we ask whether someone?sexperience as of being free to do otherwise is compatible with determinism,52 weare asking whether her experience is veridical, assuming determinism. We arenot asking whether the experience itself is compatible with determinism, even if ithas libertarian content. After all, presumably any experience whatever, including51 I discuss these issues further in the chapters that follow.52 I will not always adhere to the locution ?the experience as of being able to otherwise.? Often,for ease of exposition, I will simply say ?the experience of being able to otherwise.? Unlessotherwise indicated, the latter phrase should be read in the former way. In other words, experiencescan be in error.30a libertarian one, is compatible with determinism, given that we could be deter-mined to have that very experience.531.5 Outline of the ProjectHere is how I will proceed. In Chapter 2, ?Phenomenal Abilities: Incompatibilismand the Experience of Agency,? I present the results of a series of studies that Iconducted together with Matt Bedke and Shaun Nichols (2013), which indicatean incompatibilist answer to the experience-descriptive question. We found thatparticipants in our studies tended to report having an incompatibilist experienceof being free to do otherwise. In Chapter 3, ?Is Agentive Experience Compatiblewith Determinism?? I anticipate a compatibilist objection to these results; namely,that people?s incompatibilist reports about their experience are prone to error in aparticular way. I defend the results described in Chapter 2 against an objection ofthis sort developed by Terry Horgan (e.g., 2011). In place of Horgan?s compati-bilist view, I propose an alternative prescriptive strategy for compatibilists, whichI outline at the end of Chapter 3 and develop at length in Chapter 4, ?Against anArgument for Libertarianism.? In Chapter 5, ?Indeterminism, Experience, andCompatibilism,? I explain how people get to have indeterminist experiences offreedom, and thereby incompatibilist beliefs about their freedom to do otherwise.Yet, the account I offer falls short of justifying belief in libertarianism. On thecontrary, I explain how such experience is consistent with compatibilism aboutmodal freedom, and thus I defend compatiblism about being free to do other-wise.54 In Chapter 6, I review the dissertation, showing how its various strandshang together.53 Mele (1995: 133?37, 246?49) makes this point.54 Chapters 2?5 were prepared in the first instance for submission to refereed journals. Evenso, these chapters (together with the introductory and closing chapters) form a coherent documentthat provides a systematic account of my research project, as per FoGS requirements at UBC. Thedevelopment of my argument throughout the following chapters is presented, in form and content,as a unified whole.31Chapter 2Phenomenal Abilities:Incompatibilism and the Experienceof AgencyIncompatibilists often claim that we experience our agency as incompatible withdeterminism, while compatibilists challenge this claim. In this chapter, I reporta series of experiments that I conducted together with my colleagues Matt Bedkeand Shaun Nichols. These experiments focus on whether the experience of hav-ing an ability to do otherwise is taken to be at odds with determinism. Participantsin these studies described their experience as incompatibilist whether the deci-sion was (i) present-focused or retrospective, (ii) imagined or actual, (iii) morallysalient or morally neutral. The only case in which participants did not give incom-patibilist judgments was when the question was explicitly about whether one?signorance of the future was compatible with determinism. This lends empiricalsupport to claims made by incompatibilists about the experience of agency, whilealso showing that compatibilist accounts of ability are inadequate to the reported32phenomenology. These results also inform recent debates about the presupposi-tions of deliberation.12.1 Background2.1.1 IntroductionAgents act. They buy detergent at the store, they go to work, they celebrate holi-days, they cheat on their taxes. Sometimes we hold agents morally responsible forwhat they do, or what they fail to do, meting out credit and blame as the occasionmerits. In typical cases, when agents act they are thought to have an ability to dootherwise. This is a point on which most parties to the free-will debates agree.When it comes to characterizing the ability to do otherwise and asking whetherthis ability is compatible with determinism, however, there is no consensus.In the ensuing debates, the experience as of having an ability to do otherwiseoccupies a central role.2 Many libertarians, for instance, maintain that the abilityexperienced is incompatible with determinism (C. A. Campbell 1951; O?Connor1995). Of course, some compatibilists have challenged this idea (Mill 1865;Gru?nbaum 1952; Nahmias et al. 2004). Despite the centrality of the phenomenol-1 At the time of submission of this dissertation, Chapter 2 has just been published as: Ois??nDeery, Matt Bedke, and Shaun Nichols (2013), ?Phenomal Abilities: Incompatibilism and theExperience of Agency,? in D. Shoemaker (ed.), Oxford Studies in Agency and Responsibility:Vol. I, 126?150. It is reprinted here with the permission of Oxford University Press (UK), and hasbeen altered to maintain consitency with the rest of this dissertation (e.g., additional material hasbeen added at the beginning and end of the chapter).Thanks to Eddy Nahmias, Dana Nelkin, Paulo Sousa, David Shoemaker, Uriah Kriegel, andBertram Malle for comments on earlier drafts. Many thanks also to participants at the New OrleansWorkshop on Agency and Responsibility (November 3-5, 2011) and two anonymous reviewers atOUP for suggestions and comments.2 For some of the literature on the ability to do otherwise, see Moore (1912), Berofsky (2002),Joseph Campbell (2005), Perry (2004), Vihvelin (2004), Smith (2004), Fara (2008), Fischer(2008), Clarke (2009). The claim that moral responsibility requires alternative possibilities hasbeen disputed since Frankfurt (1969). However, it is still widely contended that free will requiresbeing able to do otherwise.33ogy of agency in all this, there has been strikingly little work on its characteristics.Of particular significance, there is almost no empirical work on whether the ex-perience of agency involves a phenomenology of being able to choose amongalternative possibilities or whether people take their agentive experiences to haveincompatibilist elements.3This paper reports a series of experiments that investigates the phenomenologyof agency. To anticipate, my colleagues and I found remarkably consistent resultsacross three sets of studies: participants regarded their experience of the ability todo otherwise as incompatible with determinism. Now that any suspense has beenspoiled, I will locate the issue in the broader literature.2.1.2 The Experience of the Ability to Do OtherwiseLet us characterize determinism as follows: a statement of the facts of the worldat an instant together with a statement of the laws of nature entail all truths aboutthe world, including those about future human actions.4 Granting that we oftenfeel that we have an ability to act other than we do, for present purposes incom-patibilists think that the experience as of an ability to do otherwise is incompatiblewith determinism, while compatibilists think the opposite.There are a number of influential introspectors on both sides of this issue. JohnSearle is a representative incompatibilist:[R]eflect very carefully on the character of the experiences you haveas you engage in normal, everyday human actions. You will sensethe possibility of alternative courses of action built into these expe-riences. . . that we could be doing something else right here and now,that is, all other conditions remaining the same. This, I submit, is3 One exception in the recent literature is a paper by Nahmias and colleagues (2004), whichI discuss in section 2.1.3. Although my colleagues and I challenge their experimental results, weare indebted to them for pioneering the investigation. I also draw on their scholarship in settingout some of the historical statements below. See also Monroe and Malle (2010).4 In the experiments below, my colleagues and I present this idea in terms of causation to makeit more intuitive and accessible.34the source of our own unshakeable conviction of our own free will.(1984: 95)Similarly, Keith Lehrer has claimed that the incompatibilist ?accurately describeswhat I find by introspecting, and I cannot believe that others do not find the same?(1960: 150). Even such a paradigmatic compatibilist as David Hume (1960/1739)agrees with this sentiment when he writes, ?There is a false . . . experience . . . ofthe liberty of indifference? (Bk. II, Part III, II).The appeal to an incompatibilist phenomenology plays a particularly impor-tant role in libertarianism. Many libertarians maintain both that we experienceour agency as incompatible with determinism, and that this experience providesreason to think that our agency defies determinism. C. A. Campbell writes:Everyone must make the introspective experiment for himself: but Imay perhaps venture to report. . . that I cannot help believing that itlies with me here and now, quite absolutely, which of two genuinelyopen possibilities I adopt. (1951: 463)Campbell goes on to argue that, unless we have good reason to doubt the impres-sion that ?it lies with me? which of two possibilities I adopt, we should accept theimpression to reflect the truth. Timothy O?Connor makes this move as well. First,O?Connor describes the character of the experience of decision-making:[T]he agency theory is appealing because it captures the way we ex-perience our own activity. It does not seem to me (at least ordinarily)that I am caused to act by the reasons which favor doing so; it seemsto be the case, rather, that I produce my decision in view of thosereasons, and could have, in an unconditional sense, decided differ-ently. . . (1995: 196)Next, O?Connor says that we should take these experiences to reflect somethingimportant about the nature of decision-making:35Such experiences could, of course, be wholly illusory, but do we notproperly assume, in the absence of strong countervailing reasons, thatthings are pretty much the way they appear to us? . . . Skepticismabout the veridicality of such experiences has numerous isomorphsthat, if accepted, appear to lead to a greatly diminished assessmentof our knowledge of the world, an assessment that most philosopherswould resist. (1995: 196?197)A number of compatibilists have challenged the basic phenomenological claim.These compatibilists deny that we experience our agency as incompatible withdeterminism. John Stuart Mill, for instance, writes,Take any alternative: say to murder or not to murder. . . If I elect toabstain: in what sense am I conscious that I could have elected tocommit the crime? Only if I had desired to commit it with a desirestronger than my horror of murder; not with one less strong. Whenwe think of ourselves hypothetically as having acted otherwise thanwe did, we always suppose a difference in the antecedents: we pic-ture ourselves as having known something that we did not know, ornot known something that we did know; which is a difference in theexternal motives; or as having desired something, or disliked some-thing, more or less than we did; which is a difference in the internalmotives. (1865: 285)On Mill?s view, the feeling of the ability to do otherwise is always contingenton our supposing that the situation prior to the decision was somehow different.Adolf Gru?nbaum repudiates any incompatibilist element with equal vigor:Let us carefully examine the content of the feeling that on a certainoccasion we could have acted other than the way we did. . . Does thefeeling we have inform us that we could have acted otherwise underexactly the same external and internal motivational conditions? No,36. . . this feeling simply discloses that we were able to act in accord withour strongest desire at that time, and that we could indeed have actedotherwise if a different motive had prevailed at the time. (1952: 672)Gru?nbaum?s last sentence here gestures at the payoff of denying the phenomeno-logical claim of incompatibilist agency: if Mill and Gru?nbaum are right, thenthe feeling of being able to do otherwise is consistent with determinism, and thiswould undercut a crucial motivation for libertarianism.This situation might seem to be a dialectical stalemate (Cf. Fischer 1994: 84).However, these philosophers are making general claims about the nature of ourexperience of agency. These are empirical claims, and they can be illuminated bytaking up empirical methods.2.1.3 Previous Work on the Phenomenology of Free WillMy colleagues and I are not the first to recommend a more systematic investigationthat is partly empirical. Nahmias and colleagues (2004) suggest that we find outhow people actually tend to describe their agentive experience (what they call thephenomenology of free will), including their experience as of being able to dootherwise:Taking a cue from recent empirical work on ?folk intuitions,? we thinkthe best way to understand the phenomenology of free will?if thereis one?is to find out what ordinary people?s experiences are like. Ifthis is not possible, philosophers? competing introspective descrip-tions will remain in yet another free-will stalemate. (164)Nahmias and colleagues undertook this task in survey studies. Their studies ap-pear to lend some support to the idea that the phenomenology of agency is com-patibilist. However, these studies have significant shortcomings, so let me brieflydescribe one of those studies, and then identify what is lacking.In one study, Nahmias and colleagues pitted compatibilism and incompatibil-ism against each other directly. The study was based on ?competing libertarian37and compatibilist accounts of our experience of the ability to choose otherwise?(174). Their survey asked participants to imagine (or recall) an experience ofmaking a difficult choice:Imagine you?ve made a tough decision between two alternatives.You?ve chosen one of them and you think to yourself, ?I could havechosen otherwise? (it may help if you can remember a particular ex-ample of such a decision you?ve recently made).Which of these statements best describes what you have in mind whenyou think, ?I could have chosen otherwise??A. ?I could have chosen to do otherwise even if everything at themoment of choice had been exactly the same.?B. ?I could have chosen to do otherwise only if something hadbeen different (for instance, different considerations had cometo mind as I deliberated or I had experienced different desires atthe time).?C. Neither of the above describes what I mean. (2004: 175?76)The majority of the participants gave the response that fits with compatibilism(i.e., B).While this study is clearly focused on an issue that divides compatibilists andincompatibilists, there are a number of limitations to the study. First, participantsare told to think of a decision and then told to think something else about thedecision: that they could have done otherwise. It is thus unclear whether theirinitial recollection actually carried with it a sense of an ability to do otherwise.So if people make compatibilist judgments about these decisions, it might be be-cause they are considering cases in which the phenomenology of the ability to dootherwise is absent.38Second, the key question is about experiences sometime in the past, ratherthan present-focused experiences where the phenomenology of agency is actuallypresent and thus presumably more accessible.Third, Nahmias and colleagues asked participants about difficult decisions,and this presents the opportunity to interpret ?could have done otherwise? in con-founding ways. Consider Martin Luther?s decision to renounce his writings or bedeclared an outlaw and heretic. Legend has it that, after praying and consultingwith advisors for a day, he said, ?Here I stand. I can do no other,? thereby reaffirm-ing his writings. Luther might have chosen B in Nahmias?s survey. But if he did,we should not conclude that there is no sense of ?could have done otherwise? thatcaptures some aspect of Luther?s phenomenology and that is incompatible withdeterminism. For Luther could have responded as he did to express his commit-ment to his cause, a commitment that would only change if the considerations be-fore him and his reasons for breaking with the Roman Catholic Church presentedthemselves differently. This commitment-expressive meaning of ?could not havedone otherwise? is consistent with other senses of ?could have done otherwise??consider whether Luther thought it was up to him whether to renounce his views?that might or might not be incompatible with determinism. Difficult decisions aresubject to confounds like this, so the above survey does not cleanly address thequestion whether there is some aspect of the phenomenology of agency that is intension with determinism.Fourth and last, it is not clear whether the participants really understand theintended meaning of ?even if everything at the moment of choice had been exactlythe same? or ?only if something had been different.?My colleagues and I wanted to run more comprehensive studies that fix theseshortcomings. The result was the following three studies, which share a com-mon structure. First, participants were asked whether they had an experience asof the ability to do otherwise when faced with a simple decision. Next, theywere given a description of determinism. Of course, we did not use the term?determinism,? since that might have conjured up unwanted associations in par-39ticipants. Rather, we used a technical term??causal completeness.? To addressconcerns about comprehension of the materials, the familiar psychological tech-nique of training to criterion was adopted, thus participants were asked a seriesof questions that tested and, if necessary, corrected, their understanding of deter-minism. Participants who passed the training were asked about the compatibilityof their experience with determinism. In study 1, this question focused on botha first-person, present-focused experience in a hypothetical deliberative contextand a past-focused judgment about such a situation. Study 2 explored the phe-nomenology of actual rather than imagined choices. And study 3 tested whetherepistemic phenomenology?the phenomenology of uncertainty?feels incompat-ible with determinism.2.2 Study 1OverviewIn the first study, Bedke, Nichols and I had participants imagine a decision aboutwhether to go left or right on a sled. In one condition, the sledding scenario wasset in the future; in the other condition, the scenario was set in the past. After read-ing the scenario, participants in condition 1 were asked whether they had a feelingof an ability to do otherwise; participants in condition 2 were asked for a retro-spective judgment about whether they could have done otherwise. Participantswho affirmed feeling (or having) an ability to do otherwise were directed to thetraining section in which causal completeness (i.e., determinism) was explainedto them. Participants who passed the training were reminded of their affirma-tion regarding the ability to do otherwise and asked about consistency with causalcompleteness.The prediction my colleagues and I made was that when asked about the phe-nomenology of imagined decision making, participants would tend to affirm a40feeling of an ability to do otherwise and also regard this feeling as incompatiblewith determinism; but when asked for a retrospective judgment about the abilityto do otherwise, we predicted that participants would be less likely to treat theability to do otherwise in an incompatibilist way.5MethodParticipants:84 participants were initially recruited online through the Mechanical Turk(MTurk) website.6 The survey itself was conducted using SurveyMonkey. 2 par-ticipants did not complete the survey. They were excluded from the analysis.Materials:Each condition had three parts.Part 1: The ability to do otherwiseParticipants were presented with a vignette and a question about the ability to dootherwise. For condition 1, this went as follows:Please read the following passage, and answer the questions that fol-low as best you can:Imagine that you are sledding down a snowy path on a mountainside.Your sled has a steering mechanism that allows you to control thedirection of the sled. Below you is a fork in the path with snow builtup in the middle, and you can tell that, if you don?t direct your sled5 See e.g., van Inwagen 1983 (8?13) for an overview of other uses of ?can.?6 MTurk is a website supported by Amazon.com (https://requester.mturk.com/mturk/welcome)that provides users the opportunity to fill out surveys for modest compensation. Recent workindicates that the data gathered through MTurk is at least as reliable as that gathered throughstandard psychology pools composed of undergraduates (see Buhrmester et al. 2011).41one way or the other, the contours of the mountain will channel youand your sled either to the left or to the right.Ability QuestionConsider how things seem to you as you approach the fork in the path.In particular, consider what it?s like to decide which way the sled willgo.Please indicate your level of agreement with the following statement:When deciding which way the sled will go, it feels like I can eithergo to the left or go to the right.Participants were asked to indicate their agreement with this statement on a 7-point scale (1=disagree completely; 7=agree completely).For condition 2, the vignette was the same except that participants were askedto imagine that the sledding episode occurred many years ago. And instead ofgetting a response regarding a phenomenology of the ability to do otherwise, weasked them to indicate agreement with a statement about a past ability to havedone otherwise: ?I could have gone right instead of left.?Part 2: Training on determinismThe aim was to focus on participants who had a phenomenology of the abilityto do otherwise, so only participants who indicated a positive level of agreementto the first questions (5 or higher) were directed to the training section. Here,participants were given a detailed explanation of causal completeness, summedup as follows: ?According to causal completeness, everything that happens isfully caused by what happened before it. This is true from the very beginning ofthe universe, so what happened in the beginning of the universe fully caused whathappened next, and so on right up until the present. Causal completeness holdsthat everything is fully caused in this way, including people?s decisions.?42Participants were then given two kinds of cases to illustrate the phenomenon.In one case, they were told that an earthquake fully caused the volcanic eruptionat Mt. St. Helens,7 and they were then told, ?According to causal completeness,if we could somehow replay the entire past right up until St. Helens erupted onMay 18, 1980, then St. Helens would once again erupt at that time. Anotherway to put this is to say that all the events leading up to the eruption made it sothat the eruption had to happen.? In another case past events, feelings and beliefsled to Obama?s decision to pick Joe Biden as his running mate, and participantswere told ?According to causal completeness, if we replayed the past right upuntil Obama?s decision?including everything that was going through Obama?smind?then Obama would once again make exactly the same decision. That is,all the events leading up to Obama?s decision (including everything that was goingthrough Obama?s mind), made it so that it had to happen that Obama would pickBiden.?8We then tested comprehension of causal completeness. First, we asked partic-ipants to indicate whether the following was true or false:According to causal completeness, St. Helens would have erupted onMay 18, 1980 even if there had been no earthquake.Participants who answered ?True? (the incorrect answer) were corrected, andgiven an explanation of the right answer. These participants were then given asimilar question to see if they had absorbed the training. If they answered incor-rectly yet again, they did not move on to answer the compatibility question, asthey were deemed to have insufficient comprehension of causal completeness.7 This is an oversimplification of the geological facts which my colleagues and I adopted toease the load on participants.8 In defining determinism?our causal completeness (CC)?as meaning ?everything that hap-pens is fully caused by what happened before it,? some might think this consistent with certainindeterminist conceptions of causation. But to say that events are fully caused is meant to avoidthis reading?being fully caused suggests that nothing extra-causal is needed to help settle events.My examples aid the preferred interpretation. I say, e.g., that under CC, ?it had to happen thatObama would pick Biden.?43Participants who passed this first kind of question either on the first or secondtry were given another true/false question to test for comprehension:According to causal completeness, if a week from now Barack Obamadecides to have soda with dinner, all the events leading up to thatdecision will make it the case that he has to decide to have a sodawith dinner.The objective here was to test for and correct overly weak interpretations of causalcompleteness. Those who answered ?False? (the incorrect answer) were correctedand given another chance at a similar question. If they answered incorrectly yetagain they failed the training and did not answer the compatibility question. Par-ticipants who passed both kinds of questions either on the first or second try weredeemed to have adequate comprehension of determinism, and these participantsmoved on to the third part of the study, the compatibility question.Part 3: ConsistencyAfter successful completion of the training, in both conditions participants weretold to recall their agreement with the statement regarding the ability to do other-wise (from Part 1 of the survey). E.g., in condition 1, they were told:Now, recall that you previously agreed with the following statement:When deciding which way the sled will go, it feels like I can eithergo to the left or go to the right.Following this, they were asked the compatibility question. In condition 1, thisread as follows:Compatibility QuestionConsidering this previous statement and your understanding of causalcompleteness, please indicate your level of agreement with the fol-lowing:44Even though it felt like I could either go to the left or go to the right,if causal completeness is true there is something mistaken about howthat decision felt to me.In condition 2, the compatibility statement was:Even though I said I could have gone right instead of left, if causalcompleteness is true there is something mistaken about what I said.Agreement was indicated on the same 7-point Likert scale as was used for theAbility Question, and an answer above 4 was taken to be an incompatibilist an-swer.ResultsOf the 34 participants who started condition 1, 33 completed it. Of these, 31 in-dicated a phenomenology of an ability to choose among possibilities and all but4 of them passed the training section.9 The remaining 27 participants gave a meanresponse of 4.93 on the compatibility question, which differed significantly fromthe midpoint of the scale, t(26) = 2.65, p = .014. That is, participants tended tointerpret their agentive experience as being incompatible with determinism.In condition 2, of the 50 participants who started the survey, 49 completed it.Of these, 47 indicated an ability to choose among possibilities and all but 2 ofthem passed the training section.10 The remaining 45 participants gave a meanresponse of 5.24 on the compatibility question, which differed significantly fromthe midpoint of the scale, t(44) = 5.05, p < .001.A t-test comparing conditions 1 and 2 showed no significant difference,p = .448. So participants tended to be just as incompatibilist about retrospective9 Nine participants required correction, and ultimately passed the training section. The re-sponses of those who required correction did not differ from those who answered correctly withouttraining (p > .2).10 Seven participants required correction, and ultimately passed the training section. Again,there were no differences between those who required correction and those who didn?t.45judgments of their ability to do otherwise as they are about their current expe-rience as of being able to do otherwise. This first study provides evidence thatpeople do indeed judge that their experience of deciding is inconsistent with de-terminism, in the sense that the experience is somehow mistaken or non-veridicalif determinism is true. It also suggests that the effect is robust across retrospectiveand present-focused cases.2.3 Study 2OverviewOne major limitation of study 1 is that it involved merely imagined choices. Thisinserts a distance between the actual phenomenology of decision-making andjudgments about that phenomenology. As a result, study 2 introduced conditionsin which agents actually make decisions. In addition, study 1 focused on deci-sions that have no moral weight. My colleagues and I thus added a condition instudy 2 in which the decision does have a moral element. So this study comprisesthree conditions to test for any effect from actual choices or from morally salientchoices. We also introduced two innovations to the study?s design.First, the vignette was more ?choicey?. It struck us that in situations suchas sledding down a hill often we don?t have a salient experience as of decidingwhich way to go. We just go one way or the other. Second, we wanted to address apotential worry about the use of the word ?mistaken? in the compatibility question.For example, condition 1 from the first study asked:Even though it felt like I could either go to the left or go to the right,if causal completeness is true there is something mistaken about howthat decision felt to me. (Emphasis added.)One worry about this wording was that participants might misinterpret it as ask-ing whether they were mistaken in thinking that their experiences felt a certain46way, rather than as asking whether there would be something mistaken about thecontent of their felt experiences.11 The solution was to replace the above wordingwith a wording of the following form:Even though it felt like I could either choose to X or choose to Y, ifcausal completeness is true then I couldn?t really have chosen differ-ently than I did.12With these modifications, condition 1 presented participants with an imaginedchoice among two very similar charities, condition 2 presented participants withan actual choice among two similar charities, and condition 3 presented partici-pants with an actual morally salient choice among two charities, one for endan-gered trees, another for children?s cancer treatments.MethodParticipants:155 participants were initially recruited online through the Mechanical Turk(MTurk) website. The survey itself was conducted using SurveyMonkey. 21 par-ticipants did not complete the survey, or indicated that they had recently taken a?very similar? survey.13 They were excluded from the analysis.11 Thanks to Lucas Thorpe for this objection.12 Two reviewers worried that, with causal completeness in mind (earlier described in terms ofevents that ?had to happen?) participants fix on one reading of the modal ?couldn?t really havechosen differently? and on that reading they give an ?incompatibilist? response, whereas the de-scription of the their phenomenology might invoke a different reading of the modal that wouldnot merit an incompatibility response. Of course, the key issue is whether participants feel theirphenomenology wouldn?t be veridical if CC were true. The question in study 3 is designed to firstrefer to the participants? reports on their phenomenology?that it felt like they could choose X orchoose?and this helps subjects to focus on that modal content and whether it would be veridicalif CC were true. Further, study 1 asked the compatibility question using different language thatavoids this worry. My colleagues and I get the same incompatibilist results there.13 Studies 2 and 3 were run after study 1, and some of the conditions in studies 2 and 3 wererun serially, so we excluded participants who indicated they had taken a very similar survey tominimize the influence of having previously taken one of our surveys.47Materials:As in study 1, this study had three parts.Part 1: The ability to do otherwiseFor condition 1 of this study, I asked participants to imagine deciding betweentwo charities for endangered trees.Imagine that you have $0.50 to donate. You have two options:Donate to a foundation that protects the endangered tree Cas-tanea Dentata.ORDonate to a foundation that protects the endangered tree UlmusDentata.14These are your only two options.Condition 2 was similar except that participants were given an actual choice. Par-ticipants were told that they had $0.50 to donate to one of the two tree charities.Participants were informed (truly) that we would actually donate this money towhichever charity they chose. Participants read:You have $0.50 to donate. We, the researchers, will actually donatethis money for you whichever way you decide.14Castanea Dentata and Ulmus Dentata are the names of the American Chestnut and the Amer-ican Elm, respectively. They are endangered species in North America. The charities used wereThe American Chestnut Foundation (http://www.acf.org/), and Trees Winnipeg: Coalition to Savethe Elms (http://www.savetheelms.mb.ca/).48Participants were then presented with the same option language as in the imaginedcondition, and each option appeared as a radio button at the bottom of the page.Finally, condition 3 presented participants with a morally salient choice be-tween a foundation that protects the tree Castanea Dentata or (2) and The Child-hood Cancer Foundation,15 on the assumption that people tend to think that savingdying children has greater moral weight than saving endangered trees.In all conditions, after being given the imagined or actual choice, participantswere asked a question about the ability to do otherwise. For instance, in condi-tion 3, participants were asked to indicate their level of agreement (on a 7-pointscale) with the following statement:When deciding which option to choose, it feels like I can eitherchoose to donate to the endangered tree Castanea Dentata or chooseto donate to the Childhood Cancer Foundation.16(In conditions 2 and 3, participants were subsequently required to make a choicebetween the charities.) As in study 1, only participants who agreed with theability-to-do-otherwise statement proceeded to the training.Part 2: Training on determinismThe training section was the same as that used in study 1, and once again onlythose who passed the training proceeded to the compatibility question.Part 3: ConsistencyThe compatibility question was adapted for the new cases. For example, in con-ditions 1 and 2, participants were asked to indicate agreement (on a 7-point scale)with this statement:15 http://www.candlelighters.ca/16 There is a potential concern here about this phrasing, which must be forestalled. SupposeI am determined to choose p. It follows that I can choose p. And you might think it furtherfollows that I can choose p or choose q, for this follows from the simple logical principle ofdisjunction introduction. In that case the ability to choose p or choose q is clearly compatible withdeterminism. However, participants report an incompatibilist phenomenology as of an ability tochoose p or q, which suggests that they are not reading ?can choose p or can choose q? in thiscompatibilist way.49Even though it felt like I could either choose to donate to CastaneaDentata or choose to donate to Ulmus Dentata, if causal completenessis true then I couldn?t really have chosen differently than I did.ResultsOf the 50 participants who started condition 1, 42 completed it and had not re-cently taken a very similar survey (3 had). Of these, 38 indicated a phenomenol-ogy of an ability to choose among possibilities and all but 3 of them passed thetraining section. The remaining 35 participants gave a mean response of 5.60 onthe compatibility question, which differed significantly from the midpoint of thescale, t(34) = 6.08, p< .001. The results of an imagined choice are consistent withthe results of condition 1, study 1, if not stronger by virtue of the more ?choicey?vignette.In condition 2, of the 48 participants who started the survey, 42 completed itand had not recently taken a very similar survey (4 had). Of these, 39 indicateda phenomenology of an ability to choose among possibilities and all but 2 ofthem passed the training section.17 The remaining 37 participants gave a meanresponse of 5.78 on the compatibility question, which differed significantly fromthe midpoint of the scale, t(36) = 6.85, p < .001. That is, participants were againincompatibilist about the phenomenology, this time of an actual choice.In condition 3, of the 57 participants who started the survey, 50 completed itand had not recently taken a very similar survey (3 had). Of these, 43 indicated aphenomenology of an ability to choose among possibilities and all but 3 of thempassed the training section.18 Most of the remaining 40 participants (90%) opted17 Six participants required correction and successfully passed the training section. Therewas no difference between the responses of those who required correction and those who didn?t(p > .2).18 Six participants required correction and successfully passed the training section. Again, Ifound no difference between the responses of those who required correction and those who didn?t(p > .2).50to donate to the Childhood Cancer Foundation, as expected on the assumption thatthe cancer charity would be regarded as more morally salient. The 40 participantsgave a mean response of 5.85 on the compatibility question, which differed sig-nificantly from the midpoint of the scale, t(39) = 7.66, p < .001. Once again,participants reported incompatibilist phenomenology?this time with a morallysalient choice.ANOVA testing showed no overall effect of condition among condi-tions 1, 2, and 3, F(2, 111) = .254, p= .776. So there appears to be no effect pro-duced by making the condition an actual choice, or by making the choice morallysalient.The results of study 2 show that people report incompatibilist phenomenologyof agency for actual choices. Indeed, whether the decision is set up as an imaginedone or an actual one does not affect the degree to which participants interpret theiragentive experience as being incompatible with determinism. The results alsoshow that whether or not the decision is morally salient doesn?t affect the degreeto which participants interpret their agentive experience as being incompatiblewith determinism. So the results of previous studies seem to extend to the moraldomain, where issues of responsibility loom large.2.4 Study 3OverviewOne possible concern with the previous studies stems from the way in which thethe key compatibility question is phrased. Notice the use of an ?even though?locution in the following:Even though it felt like I could either choose to donate to CastaneaDentata or choose to donate to Ulmus Dentata, if causal complete-ness is true then I couldn?t really have chosen differently than I did.(Emphasis added.)51Although this seems to be a natural phrasing of the question, one might thinkthat ?even though? primes the participant to agree with the statement, which inthis case is an incompatibilist response. The final study drops this potentiallytroublesome phrase and also tests whether the phenomenology of epistemic un-certainty differs from the phenomenology of being able to do otherwise in termsof compatibility with determinism. Condition 1 again presented participants withan actual choice among two options and tested whether they would continue toreport having an incompatibilist phenomenology as of being able to do otherwise.Condition 2 focused on epistemic phenomenology.MethodParticipants:106 participants were initially recruited online through the Mechanical Turk(MTurk) website. The survey itself was conducted using SurveyMonkey. 15 par-ticipants did not complete the survey, or indicated that they had recently taken a?very similar? survey. They were excluded from the analysis.Materials:The vignette and first question for condition 1 read as follows.Part 1: The ability to do otherwiseIn both conditions, participants were told that they would have a chance to win5 cents if they picked the right button. The text went as follows:At the bottom of this page, there are two buttons, labeled H and V.Each option is currently available for you to choose. In a moment,we?ll ask you to choose just one of them. For this survey, only one ofthe buttons will give you an extra $0.05 (as bonus payment on MTurk)52if you choose it. But we won?t tell you which button it is?you?ll haveto make a choice and find out.But don?t decide just yet.First, consider how things seem to you as you face your decision. Inparticular, consider what it?s like to decide which option to choose.In condition 1, participants were asked to indicate agreement (on a 1?7 scale) withthe following statement:When deciding which option to choose, it feels like I can eitherchoose H or choose V.Condition 2 was the same except that we dropped the modal ?can? and askedparticipants to ?consider what it?s like to wonder which option you?ll choose.?Participants were then asked to indicate their level of agreement with a statementabout epistemic phenomenology:When wondering which option I?ll choose, it feels like I don?t knowfor sure before I select a button which button is the bonus button.As in study 1, only participants who agreed with the ability-to-do-otherwise state-ment proceeded to the training.The two available options?H and V?appeared at the bottom of the screen,with a radio button representing each option. Participants were not told whetherthey had chosen the bonus button (H) until after they had answered the compati-bility question.Part 2: Training on determinismThe training section was the same as that used in study 1, and again participantsonly proceeded to the compatibility question if they passed the training.53Part 3: ConsistencyThe compatibility question was adjusted for the new cases. In condition 1, partic-ipants were told:Now, recall the button-choosing situation. You previously agreedwith the following statement:When deciding which option to choose, it feels like I can eitherchoose H or choose V.Considering this previous statement about how things felt to youbefore your choice and your understanding of causal completeness,please indicate your level of agreement with the following:If causal completeness is true, then I couldn?t really have chosen dif-ferently than I did.In condition 2, participants were reminded that they agreed with this statement:When wondering which option I?ll choose, it feels like I don?t knowfor sure before I select a button which button is the bonus button.They were then asked to indicate their level of agreement with the following:If causal completeness is true, then I knew for sure before I selecteda button which button was the bonus button.The aim was to test whether participants distinguish the sort of alternative possi-bilities they reported themselves as experiencing in other conditions from clearlycompatibilist alternative possibilities, which have to do simply with our ignoranceof the future.54ResultsOf the 53 participants who started condition 1, 47 completed it and had not re-cently taken a very similar survey (2 had). Of these, 44 indicated a phenomenol-ogy as of there being alternative possibilities in the situation and all but 3 ofthem passed the training section.19 The remaining 41 participants gave a meanresponse of 5.34 on the compatibility question, which differed significantly fromthe midpoint of the scale, t(40) = 4.54, p < .001. That is, participants once againdemonstrated a strong tendency to interpret their agentive experience as beingincompatible with determinism.In condition 2, of the 53 participants who started the survey, 44 completed itand had not recently taken a very similar survey (8 had). Of these, 39 indicateda phenomenology of an ability to choose among possibilities and all but 1 ofthem passed the training section.20 The remaining 38 participants gave a meanresponse of 2.66 on the compatibility question, which differed significantly fromthe midpoint of the scale, t(37) = ?5.23, p < .001. That is, participants tendedto regard their phenomenology of uncertainty about the future as compatible withdeterminism. A t-test between conditions 1 and 2 showed that results differedsignificantly between these two conditions, t(76) = 6.85, p < .001.This final study provides yet further evidence that people do indeed judge thattheir experience of deciding is inconsistent with determinism, in the sense that theexperience is non-veridical if determinism is true. At the same time, people tend tothink that the feeling of not knowing what will happen is perfectly consistent withdeterminism. This suggests an appropriate sensitivity to the fact that ignorance isnot incompatible with determinism.19 Eight participants required correction, and passed the training section. There was a significantdifference in responses between those who required correction and those who didn?t. Those whorequired correction reported that their phenomenology was incompatible with CC (M = 4.25) butto a lesser degree than those who answered these questions correctly the first time (M = 5.60),p = .057.20 Nine participants required correction, and passed the training section. There were no statisti-cally significant differences in responses between those who required extra training and those whodidn?t (p = .2).552.5 General Discussion2.5.1 IncompatibilismThese results have implications for several issues concerning free will. Perhapsmost importantly, these studies seem to vindicate the incompatibilist descriptionsof the experience of being able to do otherwise suggested by Campbell, O?Connor,and Searle. By the same token, the results run counter to the compatibilist de-scriptions of our experience suggested by Mill, Gru?nbaum, and Nahmias and col-leagues. The design of these studies left it open for participants to describe theirexperience as involving the ability to do otherwise, while allowing them to inter-pret this ability however they wished. The results indicate that the people in thepopulation that was tested tended to judge that their experience was incompatiblewith determinism.The results also address a concern that has plagued recent work on intuitionsabout free will. Nahmias and Murray (2011) contend that people give incompat-ibilist responses in previous experiments simply because people misunderstanddeterminism. This is an important concern. But rather than merely testing tosee whether people misunderstand determinism, my colleagues and I attackedthe comprehension issue directly by exploiting the familiar technique of train-ing to criterion. And we did not find any widespread confusion of determinismand bypassing. Part 1 of the training controls for confusion between determin-ism and fatalism. And the majority of participants reported that the accuracy oftheir experience as of being able to do otherwise is inconsistent with determinism,correctly understood. Across all the studies, the percentage of participants whodidn?t make it to the compatibility question due to failing the training section wassmall, at 6.15%. When we look at those participants who answered part 1 of thetraining incorrectly?that is, at those who did initially confuse determinism withfatalism, and who were directed to the follow-up training question?the percent-age was small compared with Nahmias and Murray?s results: Only 20.68% ofparticipants initially made this mistake. Of those who initially made the mistake,5685.71% answered the follow-up training question correctly. Thus, fewer than 3%of participants continued to confuse determinism and fatalism after training. Andthose who required correction did not respond in any significant way differentlyfrom those who didn?t.212.5.2 The Ability to Do OtherwiseMuch of the free-will debate, since at least Hobbes, has been about an ability todo otherwise. One influential compatibilist thought is that the notion of the abilityto do otherwise should be understood in contrast to constraint or coercion. Theidea is that an agent is able to do otherwise just in case, if she had chosen, orwanted, or tried to do otherwise, then she would have done so (Cf. Moore 1912).There are also recent versions of such a ?conditional analysis? of the ability todo otherwise. According to Kadri Vihvelin (2004), for instance, an ability to act(or not to act, which is simply to be able to act in another way) is analyzablealong something like the following lines: an agent can ? at t1 (say, raise her handat t1) just in case were she to choose to ? at t2, and her body stayed workingnormally and nothing interfered with her, she would ? at t2.22 In other words,Vihvelin holds that ?persons have abilities by having intrinsic properties that arethe causal basis of the ability? (2004: 438). So Vihvelin thinks that an abilityto act is a disposition, or a bundle of dispositions. And, as she points out, ?noone denies that dispositions are compatible with determinism? (2004: 429). Afterall, even if determinism is true, glass is still fragile?i.e., it has the disposition tobreak if struck.2321 Again, with the exception of study 2, condition 1, and study 3, condition 1. (See footnotes16 and 19.)22 Vihvelin?s exact formulation is as follows: ?S has the ability at time t to do X iff, for someintrinsic property or set of properties B that S has at t, for some time t after t, if S chose (decided,intended, or tried) at t to do X, and S were to retain B until t, S?s choosing (deciding, intending,or trying) to do X and S?s having of B would jointly be an S-complete cause of S?s doing X?(2004: 438).23 For similar accounts, see Smith (2004) and Fara (2008). Questions persist (see e.g., Clarke2009) about whether any ?dispositionalist? account is an adequate analysis of the ability to act,and thus of the ability to act otherwise.57Other compatibilists embrace an epistemic reading of ?can do otherwise.? Onthis view, to maintain that I can go left or right is simply to note that it is epis-temically open whether I will go left or right. J. J. C. Smart argues that this is anatural way to interpret the expression ?could have done otherwise? even outsidethe sphere of action. When I say, ?the plate fell, and it could have broken,? I amnot, says Smart, committing myself to any claim about determinism. Rather, whatI am saying is that, before the plate completed its fall, for all I knew, the platewould break (1961: 298). Similarly, perhaps when I say that Oswald could havedone otherwise, all I?m saying is that, before Oswald pulled the trigger, for allanyone knew, he wouldn?t pull the trigger. If I?m merely making a claim aboutepistemic possibilities, then there is no conflict with determinism.By contrast, incompatibilists think being able to do otherwise (in the relevantcontexts) means being able to do something other than what one does, all priorconditions (including one?s desires) remaining the same. This ability is presumedto be a matter of fact, not something about our epistemic access to facts.At least insofar as the relevant notion of the ability to do otherwise is reflectedin the experience as of being able to do otherwise, my results suggest that thecompatibilist accounts fail. Across three studies, participants tended to interprettheir agentive experience in terms of an ability to do otherwise, and they inter-preted that ability incompatibilistically. Concerning the traditional compatibilistanalysis, the results equally undercut old and new versions. After all, participantswere allowed to describe their experiences as involving the ability to do other-wise or not, where they were free to interpret this ability however they wished.Participants then judged that this ability?the one they had been allowed to inter-pret however they wished?was incompatible with determinism. The epistemiccompatibilist account is also undermined by these results. Participants gave com-patibilist judgments about the case of ignorance about the future (study 3, condi-tion 2), indicating that they do have an appreciation that the feeling of uncertaintyis consistent with determinism.58It seems reasonable to conclude that the notion of ?can do otherwise,? at leastwith respect to one?s decisions, is naturally interpreted in ways that contravenethe most familiar compatibilist approaches in the philosophical literature. Whenparticipants attend to their experience while they consider future events, their us-age of ?can? tends to reflect a sense of metaphysical openness that is incompatiblewith determinism.2.5.3 Misinterpreting One?s Agentive ExperienceObviously, the fact that people interpret their agentive experience as incompati-bilist doesn?t show that people actually have an incompatibilist ability to do oth-erwise. Terry Horgan argues that people might be mistaken in their interpretationof their own phenomenology. He allows that people might regard their agentiveexperience as incompatibilist:When one attends introspectively to one?s agentive phenomenology,with its. . . [representational]. . . aspects of freedom. . . and when onesimultaneously asks reflectively whether the veridicality of this phe-nomenology is compatible with causal determinism. . . , one feelssome tendency to judge that the answer to such compatibility ques-tions is No. (Forthcoming)24But Horgan notes that we must distinguish between the content of our experienceand the content of judgments. The former kind of content Horgan dubs ?presen-tational content,? and it. . . is the kind that accrues to phenomenology directly?apart fromwhether or not one has the capacity to articulate this content linguis-tically and understand what one is thus articulating, and apart fromwhether or not one has the kind of sophisticated conceptual repertoire24 For Horgan, this representational ?aspect of freedom? is what I have been calling the experi-ence as of being able to do otherwise.59that would be required to understand such an articulation. (Forthcom-ing)By contrast, ?judgmental content? is the kind of content associated with linguisticarticulations. Of course, we make judgments about our phenomenology, and sowe can have judgmental content that aims to capture our presentational content.The key point here is that it is possible for our judgments about the (presentational)content of our experience to go awry.That said, those judgments are at least prima facie evidence of the nature of thepresentational-cum-phenomenal content, so one would need some positive reasonto think that participants have systematically misinterpreted the nature of theirphenomenology. Further, even if it is granted arguendo that the presentationalcontent of agentive experience is (in the first instance) compatible with determin-ism, and that reports to the contrary count as mistaken interpretations, still, thefact that people judge the experience incompatibilist would be significant. Forone thing, when considering how best to understand the notion of the ?ability todo otherwise,? in many cases what will be of primary importance is how peoplethink about their ability to do otherwise, and that is clearly judgmental. Secondand more interestingly, judgmental content can feed back into presentational con-tent. It is well known that what one judges about a situation can affect one?s per-ception of the situation. Horgan recognizes this, and he notes that the distinctionbetween presentational and judgmental content isn?t always sharp: ?it may verywell be that the two kinds of content can interpenetrate to a substantial extent? (inpress). As a result, even if the presentational content of agentive experience is, inthe first instance, compatibilist, that doesn?t mean that the presentational contentremains compatibilist. It might be that the incompatibilist judgment shapes thepresentational content.2525 Note that Horgan?s notion of ?presentational content? is not simply ?raw feels? with nopropositional content. For everyone would concede that insofar as we have incompatibilist phe-nomenology, it must be presented at a level with greater conceptual sophistication than is providedby raw feels. Horgan is explicit about the possibility of rich conceptual resources being implicatedin presentational content: ?It is plausible. . . that humans can have presentational contents the pos-602.5.4 Deliberation CompatibilismA final issue that might be illuminated by these results is the debate over the pre-suppositions of deliberation. Some philosophers have maintained that deliberationcarries with it a presumption of genuinely open possibilities of an incompatibilistvariety. Richard Taylor writes, ?I cannot deliberate about what to do, even thoughI may not know what I am going to do, unless I believe that it is up to me whatI am going to do? (1983: 38?39). And this ?up to me? is incompatible with de-terminism. Van Inwagen makes a similar point: ?[I]f someone deliberates aboutwhether to do A or to do B, it follows that his behavior manifests a belief that it ispossible for him to do A?that he can do A, that he has it within his power to doA?and a belief that it is possible for him to do B? (1983: 155).On the other side, we find ?deliberation compatibilists,? who maintain thatdeliberation contains no such presuppositions. Tomis Kapitan begins his paper(which would become the locus classicus for deliberation compatibilism) thus:By deliberation we understand practical reasoning with an end inview of choosing some course of action. Integral to it is the agent?ssense of alternative possibilities, that is, of two or more courses ofaction he presumes are open for him to undertake or not. (1986: 230)Kapitan goes on to argue that the presumption of openness does not require meta-physical openness, but only epistemic openness.26 A number of philosophers havefollowed Kapitan in developing compatibilists accounts of the presuppositions be-hind deliberation (e.g., Nelkin 2004, Pereboom 2008).session of which require (at least causally) a fairly rich repertoire of background concepts that canfigure in judgmental states.? For instance, ?One can have presentational experiences, for instance,as-of computers, automobiles, airplanes, train stations? (Horgan forthcoming).26 According to Kapitan and other deliberation compatibilists, there are other conditions, too.In particular, Kapitan maintains that deliberation carries a presupposition of efficacy, which hecharacterizes roughly as follows: ?an agent presumes that his ? -ing is an open alternative for himonly if he presumes that he would ? if and only if he were to choose to ?? (1986: 234). See alsoPereboom (2008: 288). I leave this complication aside since it doesn?t affect my point.61Insofar as deliberation compatibilism claims that deliberation is not as a mat-ter of fact experienced as having incompatibilist presuppositions, the studies re-ported here indicate that this position is mistaken. This does not decide the disputeconcerning deliberation compatibilism, but it does show that one should distin-guish three versions of deliberation-compatibilism:(1) people?s beliefs about their current deliberations are compatible with deter-minism;(2) people?s beliefs about their current deliberations are not compatible withdeterminism, but they can be adjusted to be compatible;(3) people?s beliefs about their current deliberations are not, and cannot be ad-justed to be, compatible with determinism, but we can conceive of a rationalbeing whose beliefs about deliberation are compatible with determinism.The results of the studies that my colleagues and I ran suggest that the firstversion of deliberation compatibilism is false. People?s beliefs about their de-liberations are incompatibilist. The second version?that our actual experiencesare incompatibilist but revisable?is an interesting possibility, but it remains anopen question whether it is possible to revise this aspect of our experience. Untilwe know more about what generates the incompatibilist experience, it is hard toknow whether it can be modified. One possibility is that the incompatibilist ex-perience is generated in a way that is not cognitively penetrable (see e.g., Bayne2011). That is, it might be that even if we form the explicit high-level belief thatdeliberation is theoretically compatible with determinism, this will not eradicateour experience of our deliberation as incompatibilist. The third version of delib-eration compatibilism?that we can conceive of rational creatures who deliberateas determinists?is not under any threat from our results. But if it turns out tobe impossible for us to be such rational animals, that might undercut some of theinterest of deliberation compatibilism.622.6 ConclusionThe experience as of being able to do otherwise has long been central to debatesabout agency and free will. Libertarians appeal to this experience as evidence thatdeterminism is false; compatibilists reject the libertarian accounts of the characterof the experience. Despite the pivotal role of experience in these arguments, theexperience itself has received scant attention. The studies reported here are anattempt to advance the issue. My colleagues and I found consistently incompat-ibilist judgments about the nature of the experience as of being able to do other-wise. This lends support to the phenomenological claim of libertarians, thoughI am not inclined to take the phenomenology of indeterminism as evidence thatagency isn?t determined (as will become clear in chapters 3?5). The results ofthese studies also suggest that existing compatibilist interpretations of the notionof ?ability to do otherwise? are not adequate to people?s reported experience as ofbeing able to do otherwise. Finally, the results that I have reported here also speakto the presuppositions of deliberation. What my colleagues? and my studies indi-cate is that as a matter of fact our experience of deliberation features metaphysicalopenness (that is inconsistent with determinism). While this does not decide thedispute between deliberation compatibilists and deliberation incompatibilists, itdoes make salient the possibility that deliberation compatibilism requires an ac-count of deliberation that is explicitly revisionist with respect to people?s actualexperience of deliberation.As a result, the answer to the descriptive question about people?s experienceof modal freedom appears to be incompatibilist. We saw in Chapter 1 that thedescriptive question is an empirical question about people?s pre-theoretic belief-tendencies about their own and others? agency. We also saw that there are at leasttwo ways to address this question: (1) ask people about their intuitions or be-liefs regarding modal freedom, or (2) ask people about their experience of beingmodally free. In this chapter, I have focused on the second method, concerningexperiences of freedom. As a matter of fact, people actually tend to describe theirexperience of modal freedom (i.e., of being able to do otherwise) as inconsistent63with determinism, in the sense of being non-veridical if determinism is true. Onthe further plausible assumption that experiences of modal freedom partly drivepeople?s beliefs (or intuitions) about freedom, these results provide support forthe view that people?s pre-theoretic belief-tendencies are at least partly incompat-ibilist.In the next chapter, I defend these empirical results against an error theory forincompatibilist judgments about experience developed by Terry Horgan. I arguethat compatibilists should simply grant that the answer to the descriptive ques-tion about experiences of modal freedom is incompatibilist. Even so, I proposea prescriptive compatibilist story about the veridicality of experiences of modalfreedom. I argue that despite granting (at least arguendo) that our experience ofmodal freedom has libertarian content, our experience might still be veridical un-der the assumption of determinism. I sketch this view at the end of Chapter 3 anddevelop it more fully in Chapter 4. Finally, in Chapter 5, I develop a plausible(but deflationary) etiological story about the source of libertarian experiences andbeliefs.64Chapter 3Is Agentive Experience Compatiblewith Determinism?Many philosophers think not only that we are free to act otherwise than we do, butalso that we experience being free in this way. Terry Horgan argues that such ex-perience is compatibilist: it is veridical even if determinism is true. According toHorgan, when people judge their experience as incompatibilist, they misinterpretit. However, while Horgan?s position is attractive, it incurs significant theoreticalcosts. I sketch an alternative way to be a compatibilist about experiences of freeagency that avoids these costs while also exhibiting considerable advantages of itsown.3.1 IntroductionThe waiter offers you ice-cream. ?Chocolate or vanilla?? he asks. Each flavor isdelicious, but you know you should only choose one. You hesitate. It feels likeyou are free to choose vanilla. Nevertheless, it also feels like you can refrain fromchoosing it?say, by choosing chocolate instead. It feels like you are free to dootherwise.Is this experience accurate, or veridical, assuming determinism? Many in-65compatibilists, who think the freedom to do otherwise is incompatible with deter-minism, have thought it is not.1 They think that we experience having a freedomthat is inconsistent with determinism. According to John Searle, for instance, ourexperience amounts to the feeling that ?we could be doing something else righthere and now, that is, all other conditions remaining the same? (1984: 95). Someincompatibilists?the libertarians?even go so far as to maintain that our experi-ence in this regard is evidence that we possess an incompatibilist freedom (e.g.,O?Connor 1995).Compatibilists think that the freedom to do otherwise is consistent with deter-minism (Moore 1912; Vihvelin 2004; Fara 2008). If the freedom we experiencepossessing is compatibilist, then our experience of being free to do otherwise isaccurate, assuming determinism. For instance, compatibilists often suggest thatwe experience freedom conditionally: in the above example, as long as we arefree from constraint, coercion, and an addiction to vanilla ice-cream (say), ourexperience is that we are free to choose chocolate if we want (or try) to do so, andsimilarly regarding vanilla (Cf. Mill 1865; Gru?nbaum 1952; Nahmias et al. 2004).If that is right, then it undermines a key motivation for libertarianism?the viewthat being free to do otherwise is inconsistent with determinism, and we have suchfreedom. After all, if the nature of our experience is compatibilist, then libertari-ans cannot argue from the incompatibilist nature of experience to our possessingan incompatibilist freedom.A somewhat different compatibilist strategy is to grant that introspectionseems to reveal that experience is incompatibilist, yet insist that introspection isnot reliable in this domain. Terry Horgan adopts this strategy (forthcoming, 2012,1 Traditionally, the sort of freedom required for moral responsibility was thought to be thefreedom or ability to do otherwise. Many philosophers?both compatibilist and incompatibilist?have now abandoned this idea. They think that the sort of freedom or control that is required forresponsibility is that one be the appropriate source of one?s action, where this does not requirebeing free to do otherwise. As I outlined in Chapter 1, call the freedom condition on responsibilitymoral freedom, and the ability to do otherwise modal freedom. Even if moral freedom does notrequire modal freedom?and some compatibilists continue to think it does (e.g., Vihvelin 2004;Fara 2008)?the question whether modal freedom is compatible with determinism presses on usin its own right.662011, 2007). Horgan agrees that people often think that their experience is incom-patibilist. However, he argues that even when people judge their experience asincompatibilist, actually it is compatibilist: people misinterpret their experience.By spelling out how this happens, Horgan provides an error theory for incompat-ibilist judgments about experience.Horgan?s compatibilist strategy is attractive and has important theoretical ad-vantages. Yet it also incurs theoretical costs. After considering Horgan?s position,I sketch an alternative way to be a compatibilist about the experience of freedomthat avoids incurring these costs. On my view, even if we take people?s incompat-ibilist reports about their experience at face value, and thus grant that such expe-rience has genuinely incompatibilist content, there is still an important respect inwhich the experience is accurate, assuming determinism. Thus, I grant (at leastfor the sake of argument) an incompatibilist answer to the descriptive questionthat I outlined in Chapter 1, while defending compatibilism on the prescriptivequestion. As I explain in the next section (and in Chapter 4), the strategy I adoptdoes not require defending the idea that people?s experience of modal freedom isincompatibilist, as is suggested by the results of the studies reported in Chapter 2.Instead, I argue that it turns out to be dialectically advantageous for compatibiliststo grant that this is the case, at least arguendo. Thus, commitment to incompat-ibilist experience is simply taken hypothetically, for the sake of argument, in thepresent chapter and in Chapter 4. (In Chapter 5, I again take the existence of in-compatibilist experience seriously, and I there develop a deflationary etiology forsuch experience.)3.2 Agentive ExperienceTo forestall any confusion, let me begin by clarifying some terminology.Granting that we actually experience being free to do otherwise, experience-incompatibilists think that this experience is inaccurate if determinism is true.Call such experiences libertarian or incompatibilist. Experience-compatibilists67think the opposite: the experience might be accurate, even assuming determinism.Call such experiences compatibilist. Finally, call the question whether our expe-rience of being free to do otherwise is compatible with determinism (in the wayjust outlined) the experience-compatibilism question.I take an experience to be any representational mental state with phenomenalcharacter, where phenomenal character is what-it?s-like (or what it feels like) to bein that mental state. An experience?s phenomenology is just its phenomenal char-acter. The satisfaction conditions for an experience are its accuracy or veridicalityconditions. For any experience, its content yields a veridicality condition: the con-tent specifies how the world must be in order for the experience to be veridical. Ifa visual experience has the content squareness, where this property is attributed toa particular object, then that experience is veridical only if the object in questionis square. Moreover, I assume a close tie between content and phenomenology, sothat an experience?s phenomenology shares a veridicality condition with its con-tent. Call this phenomenal content. A visual experience of seeing a red apple willhave the phenomenal character reddishness, and thus the content that a certainobject one sees?the apple?is red. Such content is constitutively determined bythe phenomenal character,2 and it is veridical only if the apple actually is red.The property of being able to act otherwise is the property of being free in aparticular way: it is to possess a specific?not just a general?ability or freedom,as I explained in Chapter 1. To recap briefly: General abilities are uncontrover-sially compatibilist. I might possess the general ability to raise my hand an hourfrom now, without having the specific ability to exercise it just then, perhaps be-cause I will be asleep. Determinism is compatible with my retaining these sorts ofunexercised abilities. Only specific abilities are at issue in the question whetherfreedom is compatible with determinism. In Chapter 1, I characterized this no-tion as follows. One has the specific ability to do something only if (i) one has a2 I remain neutral about whether representational properties determine phenomenal properties,or vice versa.68general ability to do it, (ii) one has an opportunity to do it, and (iii) holding fixedone?s motivations at the time (including the exact strengths of one?s motivations),one can exercise one?s general ability to act in that way at that time. If this is trueof more than one available option at a time, then one has the specific ability to dootherwise.What about agentive experience? For a start, one might wonder whether thereis such a phenomenon. Even if there is, it is a further question whether high-levelproperties like being able to do otherwise feature in it. One might think that evenif people believe that they are free to do otherwise, they do not experience suchfreedom. Yet, as we shall see, appeals made by philosophers to experience implythat such properties do feature in agentive experience. I will deal with these twoissues in turn.First, certain disorders of agency speak against the suggestion that agentive ex-periences in general are reducible to beliefs about agency (Cf. Bayne 2008: 185?87; 2011: 360). A patient with anarchic hand syndrome may discover her hand?doing? things?perhaps against her wishes. She may even describe her handas ?having a will of its own.? It is plausible that such a patient?s disavowal of herhand?s ?actions? may be due to the fact that she does not experience its movementsas issuing from her own agency. She might come to believe that these movementsare, in a sense, her own actions (who else?s could they be?). Yet this judgmentwill hardly affect her experience. If that is right, then agentive experiences are notbeliefs about agency.3Even granting agentive experiences, it is a further question whether high-levelproperties like being able to do otherwise feature in them. On a conservativeview, the contents of agentive experiences contrast with those of other perceptualmodalities, like vision (Cf. Bayne 2008: 189?90). On this view, agentive expe-3 Bayne claims that such cases show that agentive experiences are not influenced by cognitivestates (2011: 360). However, just because my experience of seeing a stick in the water seeming tobe bent is not altered by my explicit belief that it is straight does not show that my visual experienceis not shaped by my background beliefs. Likewise, I propose, for agentive experiences. I discussthis issue further in section 3.4.69riences merely represent one as acting, controlling one?s action, and acting witheffort (Cf. Horgan forthcoming; Bayne 2008: 190?92). By contrast, a more liberalview takes the contents of agentive experiences to be relatively rich. For exam-ple, they additionally represent one as being the source of one?s actions, and one?sactions as being free?including in the sense that one is free to do otherwise.4At any rate, the way in which many people describe their agentive experienceindicates that they experience being free to do otherwise, and such beliefs drivea good deal of philosophical theorizing about free will. Indeed, we have alreadyseen that many philosophers take it as a basic insight that being free to do other-wise is central to the notion of human freedom, and our possessing such freedomis confirmed by our phenomenology of deliberation. This goes for both compati-bilists and incompatibilists, as noted at the opening section of this chapter (as wellas in chapters 1 and 2).5Finally, the results of the experiments that I reported in Chapter 2 indicate thatit is not just philosophers who report experiencing being free to do otherwise?ordinary people report such experiences as well. In these experiments, every effortwas made to ensure that participants attended to their relevant experience (if any).Over 91% of participants reported experiencing being free to choose either way,where it was left open for them to interpret such freedom however they wished.These results support the claim that participants had an experience of being freeto do otherwise.Furthermore, participants in these studies went on to judge their experienceas incompatibilist when the notion of determinism was explained to them. Fol-lowing a series of comprehension checks, participants were asked whether theexperience they had earlier reported was consistent with determinism. Across arange of conditions, participants said their experience was inconsistent. Whether4 See Bayne (2008) for discussion of this issue. For analogous debates about whether high-level properties are represented in perceptual experience, see Siegel (2006) and Bayne (2009).5 In addition to the authors mentioned in section 3.1, many other philosophers grant that thereare identifiable experiences of being free to do otherwise. See for instance Hume (1960/1739),Reid (1788), C. A. Campbell (1951), Lehrer (1960), Ginet (1997), Nahmias et al. (2004), andHolton (2006).70participants merely imagined making a choice or actually made a choice, whetherthe choice was morally salient or not, or whether the choice was present-focusedor retrospective, participants reported an experience of being free to do otherwisethat is incompatible with determinism. Thus, apparently, people not only experi-ence being free to do otherwise, but they also report that their experience in thisregard is inaccurate if determinism is true. This offers at least prima facie evi-dence in favor of experience-incompatibilism, evidence that must be addressed bycompatibilists.6For compatibilists, such results are problematic for at least two reasons. First,if a compatibilist theory of freedom fails to capture our experience of being freeto do otherwise, then it is not clear whether the theory is explaining the rightphenomenon. It seems plausible that the freedom the theory explains should bethe one that we experience possessing. Second, if all our experiences of beingfree to do otherwise are illusory, assuming determinism, then this leaves us withno way of distinguishing?just in terms of the accuracy conditions of our phe-nomenology?illusory experiences from experiences we normally think are ac-curate. Imagine that you wake at night and consider switching on the light. Asyou lie there, you experience being free to switch on the light, or to refrain fromdoing so?for instance, by intentionally remaining motionless. Surely, we wantto distinguish this case from one in which, unbeknownst to you, you have beenparalyzed by a drug, yet still you experience being free to switch on the light orto refrain from doing so. If all your experiences of being free to do otherwise areillusory because determinism is true, then there is no compatibilist way to makesense of the idea that your experience in the first case is accurate in exactly theway it is not in the second. Granting that we experience being incompatibilisti-cally free, we are left with the verdict that the content of our phenomenology in6 As we shall see, the claim that these participants? experiences were incompatibilist is ten-dentious in the present context, since Horgan?s claim is precisely that people often misinterprettheir experiences of freedom. By contrast, I will argue that even if these participants? experiencesare incompatibilist in a certain respect, they might nevertheless be accurate in another respect,assuming determinism.71each case makes any such experience utterly illusory, assuming determinism. Onemight avoid this verdict if the experience had compatibilist content?for instance,if it were best described as an experience of conditional freedom. Yet, as we haveseen, this appears to go against the evidence. Moreover, for the reasons given amoment ago, compatibilists cannot insist that the experience is theoretically unim-portant. Compatibilists need an alternative strategy.Terry Horgan proposes just such a strategy. Horgan agrees that the relevant ex-perience is theoretically important, and he agrees that many people have a strongtendency to think that their experience is incompatibilist. However, he argues thateven when people judge their experience as incompatibilist, actually it is compati-bilist: people tend to misinterpret their actual experience. By spelling out how thishappens, Horgan provides an error theory for incompatibilist judgments about ex-perience. If Horgan is right, then the participant responses in Chapter 1 may wellbe in error.Horgan?s compatibilist strategy is attractive, and it has clear theoretical advan-tages. However, it also incurs significant theoretical costs. In section 3.3, I sketchHorgan?s view, then in sections 3.4 and 3.5, I outline the costs that it incurs. Afterconsidering Horgan?s position, I sketch an alternative way to be a compatibilistabout the experience of freedom, which avoids these costs. My position also hasdistinct advantages over Horgan?s. For instance, I argue that even if we take peo-ple?s incompatibilist reports about their experience of freedom at face value, andthereby grant both that introspection reliably latches onto the content of such ex-perience and that such content is rich enough to be incompatibilist, there is stillan important respect in which the experience is compatibilist: it is veridical evenif determinism is true.3.3 Horgan?s ViewHow does Horgan defend experience-compatibilism? According to Horgan, peo-ple?s incompatibilist reports about their experience of freedom are in error. Now,72incompatibilist experience might, of course, be mistaken in the sense of beingnon-veridical. However, Horgan suggests another way in which people might bemistaken?they might misinterpret the relevant experience. Horgan concedes thatpeople often judge their experience of being free to do otherwise as incompati-bilist. For instance, he says,When one attends introspectively to one?s agentive phenomenology,with its . . . [representational] . . . aspects of freedom . . . and when onesimultaneously asks reflectively whether the veridicality of this phe-nomenology is compatible with causal determinism . . . , one feelssome tendency to judge that the answer to such compatibility ques-tions is No. (Forthcoming)7Horgan thinks that while introspection is reliable in some domains, introspectivejudgments about whether our agentive experience is compatible with determin-ism are uniquely unreliable (see e.g., 2012; 2011). He begins by distinguishingbetween two sorts of introspection: (1) attentive introspection, which involves?paying attention to certain aspects of one?s current experience?; and (2) judg-mental introspection, which involves ?forming a judgment about the nature ofone?s current experience? (2011: 84). The kind of content on which we atten-tively introspect is ?presentational content,? which is ?. . . the kind that accrues tophenomenology directly?apart from whether or not one has the capacity to artic-ulate this content linguistically and understand what one is thus articulating, andapart from whether or not one has the kind of sophisticated conceptual repertoirethat would be required to understand such an articulation? (forthcoming). Thisis the sort of content about which we make judgments when we judgmentallyintrospect.87 Horgan deals with several types of agentive experience. For the sake of simplicity, I fo-cus solely on Horgan?s treatment of what he calls the ?freedom? aspect of experience, which heunderstands as the experience of being free to do otherwise.8 Presentational content is roughly what I am calling phenomenal content. As we will see,though, it is not equivalent to phenomenal content, since there may be sorts of phenomenal content73In judgmental introspection, we attend to certain aspects of our experi-ence, and then form a judgment about them. Thus, ?Judgmental introspection. . . deploys attentive introspection, while also generating a judgment about whatis being attended to? (2011: 84). There is no appearance/reality gap when weattentively introspect. Yet, in judgmentally introspecting on our experience, wecan, as it turns out, go wrong: we may be subject to what Horgan calls a label-ing fallacy (2012: 408?409). For instance, we might make a performance error inapplying the ordinary judgmental concept ?red? to our experience of redness: wemight mistakenly apply the ordinary concept ?green.? In Horgan?s parlance, wemight ?mislabel? the phenomenology. Presumably, this hardly (if) ever happens.So, while attentive introspection is infallible, judgmental introspection about colorexperiences is not quite infallible, although it nearly is.9The point for now is that Horgan thinks judgments about our agentive experi-ence are not like this. It is not merely that our judgments are fallible; they are notreliable at all?at least when it comes to our experience of freedom and whether itis compatible with determinism. For a start, answering this compatibility questionfar outstrips what attentive introspection is capable of. The question can only beanswered by means of judgmental introspection. Yet this process is uniquely ill-qualified for the task. Horgan thinks that when we try to answer the compatibilityquestion about our experience of being free to do otherwise by judgmentally in-trospecting, we find that such introspection is unable to provide a reliable answer,even though the question is about the character of our introspectively availableexperience of freedom.10 It is not simply that we are subject to the occasional la-other than what is ?presented? in phenomenology. Even so, what Horgan calls ?presentationalcontent? is an important aspect (or sort) of phenomenal content. I leave aside this wrinkle since itdoes not affect what I have to say.9 Horgan (2012; Horgan and Kriegel 2008) argues that there are also cases in which we areimmune even from labeling fallacies when we make judgments about our experience?e.g., whenwe judge that ?this experience has this feature.? Such judgments are infallible. These cases do notconcern me here.10 This is despite the fact that, for Horgan, experiences of freedom have intrinsic, determinatesatisfaction conditions that are compatibilist. However, such compatibility is a ?non-manifest?74beling fallacy. Rather, we cannot reliably tell what the answer to the compatibilityquestion is just by judgmentally introspecting.Horgan thinks there are good reasons for this, which I will outline in a mo-ment. Nevertheless, he admits that there are ?sophisticated philosophers? whothink that there is what he calls a ?read-offable? incompatibilist answer to thecompatibility question about determinism and the experience of being free to dootherwise, ?since they have said so to me in philosophical discussion. And I con-fess to experiencing some temptation to think so myself, as I suspect you thereader do too?a temptation that needs explaining? (2012: 416). Horgan?s expla-nation has two parts.For instance, if we are asked whether Alice is free to do otherwise than steala piece of candy?when she is in possession of her faculties, is not being orderedto steal at gunpoint, is not subject to an irresistible addiction, and so on?we areliable to answer, ?Yes.? By contrast, if we are asked whether Fred is free to do oth-erwise than lie in bed all morning, when?as it turns out?he is securely strappeddown and is prevented from moving, we are liable to answer, ?No.? By contrastwith such cases, when it comes to knowing whether to apply the notion of beingfree to do otherwise to our experience while assuming the truth of determinism,we go beyond the limits our competence. Here, we are reasoning about whetherthe satisfaction conditions of our phenomenology are met under the assumptionof a general hypothesis about the world, and we are liable to make mistakes. In-stead, we should proceed by inference to the best explanation: seek a hypothesisthat yields an answer to the compatibility question, while also explaining various?data,? such as the fact that we are normally competent in applying the notionof being free to do otherwise to concrete cases and without ever considering thethesis of determinism. Horgan thinks it reasonable to assume that such intuitivejudgments result from our competence in applying the relevant notion, so thatsuch judgments are ?normally true? (forthcoming).feature of the experiences. (Horgan allows that such experiences may have wide satisfaction con-ditions as well.)75Compatibilism accommodates these judgments easily, by enabling them tocome out true even when we assume determinism. Incompatibilism, by contrast,requires that a more stringent condition be met?namely, that indeterminism (at aminimum) be true. So, Horgan thinks we should prefer compatibilism to incom-patibilism. As he puts it, other things being equal ?one hypothesis is better thananother if it accommodates the attributional practices of competent users of therelevant concept better than the other? (forthcoming).This is meant to show that any answer to the question whether our experienceof freedom is compatibilist or not will go beyond the capabilities of judgmentalintrospection, and thus any such answer will be unreliable. Of course, Horgan alsoneeds to explain why we have any tendency in the first place to judge our experi-ence as incompatibilist. It is one thing to open up scope for error in our judgments,as Horgan does, but it is another to explain away this judgmental tendency. To thisend, Horgan offers a two-part debunking explanation of incompatibilist judgmentsabout experience. First, Horgan suggests a way in which we might introspectivelyconfabulate. Second, he tells a contextualist story about the application conditionsof the notion of freedom, which also applies to judgments about our experience offreedom. I will take these in turn.First, Horgan suggests that if we think we can tell by introspection that our ex-perience is incompatible with determinism, this may reflect a form of ?introspec-tive confabulation.? After all, it is one thing for me to know (A) by introspection:(A) My experience does not present my behavior as determined by my priorstates.Yet it is another thing for me to know (B) simply by introspecting on my phe-nomenology:(B) My experience presents my behavior as not determined by my prior states.Horgan thinks that we can ascertain whether (A) is true by introspecting. How-ever, (B) is distinct from (A), and we cannot ascertain whether (B) is true by intro-spection. Even if (B) were true, we could not know this merely by judgmentally76introspecting on experience. When we judge that our experience is incompati-bilist, and thereby assert (B), either we are mistakenly inferring (B) from (A), orconflating (A) and (B).11Second, Horgan has a contextualist story to tell about the application condi-tions of the notion of freedom, which also applies to judgments about our experi-ence of freedom:I maintain that many concepts that figure importantly in philosophicalproblems are governed by implicit, contextually variable, semanticparameters?and that some forms of philosophical puzzlement ariselargely because (i) posing a philosophical problem can tend to shiftthe implicit parameters toward settings under which the claims madeusing a given concept are more ?demanding? in their truth conditionsthan the claims that would normally be made using that concept, and(ii) one tends not to notice this shift of the ?score in the languagegame? when one is contemplating the philosophical problem. . . . Imaintain that the very posing of the question whether human freedomis compatible with . . . determinism tends to alter the contextually op-erative settings on certain implicit semantic parameters that governthe concept freedom?and tends to drive those parameter settings sohigh that, in the newly created context, no item of behavior that is. . . determined counts as free. (2007: 21?22)These contextual parameters do not apply to phenomenal content. After all, itis plausible that non-human animals also have agentive phenomenology, despitetheir mental content not being governed by contextual semantic parameters. Evenso, Horgan thinks that when we introspect on our experience of freedom, whileasking ourselves whether its content is compatible with determinism, our judg-ment gets ?infected? by scorekeeping confusions, just as happens when we ask11 Alternatively, we might just be over-interpreting the character of what we introspectivelyattend to.77compatibility questions about the judgmental content of ordinary notions like?freedom.?3.4 Problems with Horgan?s ViewEven if we grant Horgan?s hypothesis about introspective confabulation, moreneeds to be said about how this sort of mistake occurs. For example, ShaunNichols notes that generally people do not make this sort of mistake when it comesto headaches: ?the phenomenology of headaches doesn?t present us with a setof deterministic headache-causes, but we don?t leap to indeterminist conclusionsthere? (2012: 296). In other words, we do not mistakenly infer from the claim thatour experience does not present our headache as determined the further claim thatwe experience our headache as not determined. Thus, Horgan needs to say howthe phenomenology of deliberation is relevantly different from that of headaches.This requirement is a theoretical cost of Horgan?s view, which the alternative posi-tion that I outline in section 3.6 (and develop further in Chapter 4) does not incur,since my view does not claim that incompatibilist judgments about experienceare mistaken. Of course, I need to say something about the source of incompati-bilist experiences, whereas Horgan does not. Yet my position has resources in thisregard, as I outline later in this section.Note too that Horgan?s proposal is backward-looking, since it focuses on ourintrospective access to the causes of our decisions. By contrast, deliberation andaction-planning are importantly forward-looking. When we face decisions, ourprimary interest as deliberating agents is not the causal antecedents of our deci-sion, but rather the alternatives with which we are presented, and our sense of be-ing free to decide between them. These are some of the aspects of agentive expe-rience that libertarians most often cite as incompatibilist, and they are also of con-cern to compatibilists. When we focus on the future, the content of our experienceis not that our behavior is not determined by our prior states and the laws of na-ture, since for one thing our experience does not concern the laws of nature. Moreplausibly, our experience has a content, P, that is in fact incompatibilist, where P78is something like openness to the future. Notably, even some compatibilists grantthat our experience of such openness is non-veridical if determinism is true. Forinstance, Keith Lehrer has claimed that when it comes to the way in which thefuture feels open, the incompatibilist ?accurately describes what I find by intro-specting, and I cannot believe that others do not find the same? (1960: 150). Such aparadigmatic compatibilist as David Hume (1960/1739) agrees with this sentimentwhen he writes, ?There is a false . . . experience . . . of the liberty of indifference?(Bk. II, Part III, ?II). This sense of openness to the future has been characterizedby the (semi-)compatibilist John Fischer as like a ?Garden of Forking Paths.? Thismetaphor portrays our options as arrayed before us like branches in a path, eachone seemingly a realizable extension of the actual present into the future (Fischer1994: 190). Fischer thinks that being free to do otherwise is incompatible withdeterminism, since for him determinism has the consequence that there is onlyone possible extension of our actual present into the future. For Horgan, deter-minism does not have this consequence. Although Horgan thinks that experiencesof freedom are aptly described by metaphors like ?Garden of Forking Paths,? andhe agrees that libertarian descriptions of agentive experience are phenomenologi-cally apt, he still thinks that it remains open whether the satisfaction conditions ofsuch experience are compatibilist. However, nothing in his position speaks to thisview of what the content of our agentive experience amounts to. This is anothertheoretical cost of his view, since there is at least prima facie reason to think thatwhat it amounts to entails indeterminism. By contrast, my position avoids thiscost by granting (at least for argument?s sake) that our experience of freedom haslibertarian content, yet is normally veridical even if determinism is true.There is also a further worry for Horgan?s position: judgments about experi-ence can feed back into the experiences themselves. It is widely known that whatone judges about a situation can affect one?s perception of the situation, via cog-nitive penetration. Roughly, cognitive penetration occurs when the phenomenalcharacter of one?s experience is altered by one?s cognitive states?for instance,by one?s background beliefs. There is ample evidence that this actually occurs79in cases of visual perception.12 Something similar may occur in agentive experi-ence. For instance, one?s background beliefs may influence the character of one?sexperience of freedom. Indeed, Horgan appears to recognize this possibility, andhe notes that the distinction between phenomenal and judgmental content is notalways sharp: ?it may well be that the two kinds of content can interpenetrateto a substantial extent? (forthcoming). Thus, even if one?s phenomenal contentis initially compatibilist (as Horgan claims it is), that does not mean that it re-mains compatibilist. The incompatibilist judgment might shape the ?presenta-tional? content.Whether this concern has bite depends on whether cognitive penetration ac-tually occurs in agentive experience. Although there is no definitive evidence ofthis, certain considerations suggest that it is not just possible, but likely. First, tosay that experience is cognitively penetrable is just to say that it is theory-laden,which is widely accepted.13 Indeed, it would be somewhat surprising if agentiveexperience turned out to be exceptional in not being theory-laden, or subject totop-down processing.14 Second, there is evidence that beliefs about determinismand free will have measurable effects on both cheating and punishing behavior inexperimental settings. Priming participants to believe that determinism is true ap-pears to result in their cheating more (Vohs and Schooler 2008), whereas primingthem instead to believe that neural mechanism is true results in their punishingothers less than when they retain the belief that people have whatever sort of free-dom is (for them) undermined by neural mechanism?presumably, libertarian freewill (Shariff et al. forthcoming). Such evidence, taken together with the fact thatexperience often tends to be theory-laden, suggests that background beliefs?for12 See e.g., Delk and Fillenbaum (1965) and Levin and Banaji (2006) for classic experimentsthat support the thesis of cognitive penetration. See also MacPherson (2012) for recent discussionof this phenomenon.13 See e.g., Hanson (1958) for an early discussion of the theory-ladenness of perception and itsbearing on science.14 See Palmer (1999, Ch. 9) for a detailed discussion of top-down processing in everyday per-ceptual experience. The claim here is that a good deal of early-stage perceptual information issubject to subsequent top-down processing, whereby experience is influenced by concepts, be-liefs, and expectations.80instance, about the falsity of determinism or the obtaining of an indeterminist orlibertarian freedom?might well influence people?s agentive experiences of free-dom.Of course, if an explicit incompatibility judgment is what affects our experi-ence, then one might object that before we make such a judgment our experiencewill not have been altered in this way. Yet this objection is mistaken. Let us grantthat experiences of being free to do otherwise have a content, P, that is in fact in-compatibilist, and which results from cognitive penetration. Even so, such contentdoes not have to be caused by any explicit belief. It might instead be caused by abackground assumption implicitly held about P. This is the standard view of whathappens in visual experience (Cf. Delk and Fillenbaum 1965).Conversely, someone might think that if, so to speak, the judgmental tail iswagging the experiential dog, then that is far less worrisome to compatibiliststhan if the relevant experience were inherently incompatibilist. Compatibilistscan simply grant that these incorrect judgments result in incorrect experience, yetinsist that the experience can be altered to be veridical once the relevant judgmentsare corrected. However, it is not clear that such correction will always (or evenever) be possible. After all, my visual experience of a stick in the water seeming tobe bent is not altered by my forming the explicit belief that it is straight. Likewise,as we saw in section 3.2, the fact that a patient with anarchic hand syndrome formsthe explicit belief that her hand?s movements are, in a sense, her own actions doesnot affect her alienated experience of these actions not being her own (Cf. Bayne2008: 185?87; 2011: 360).Finally, one might think that even if experiences of being free to do otherwiseare theory-laden, and as a result many people end up with (perhaps unalterably) in-compatibilist experience, still people?s experience beforehand was compatibilist.This may be so. Yet agents for whom this is the case will still have agentive expe-rience that is genuinely non-veridical if determinism is true. To grant that much,for the experience-compatibilist, seems to give the game away.81However, it bears repeating that this worry has bite only if it turns out thatagentive experience is in fact theory laden. Yet given the relative likelihood of thisbeing the case, it would be theoretically advantageous for a compatibilist viewabout agentive experience to be immune to the worry. Horgan?s position is notimmune. By contrast, the alternative compatibilist position that I sketch in section3.6 (and develop more fully in Chapter 4) is immune. Indeed, the possibilitythat agentive experience is theory laden serves to bolster my view, by positinga possible explanation for the source of people?s incompatibilist experiences ofbeing free to do otherwise.3.5 Problems with Horgan?s ContextualismThere are also problems with Horgan?s contextualist proposal, the most importantof which is that global worries about whether we are free?even in the sense ofbeing able to do otherwise?can arise when contextual parameters are normal. Ifthat is right, then Horgan?s claim that such worries arise only when we raise theparameters beyond their normal settings and explicitly ask the compatibility ques-tion is false. Rather, it seems that our competence in applying the relevant notionof freedom is such that it might enable us to answer the compatibility question,and to do so reliably. That, in turn, would undermine Horgan?s claim that incom-patibilist answers to the compatibility question are the result of a scorekeepingconfusionOn the sort of contextualism that Horgan seems to adopt, ?can? (or ?is ableto?) may be used with varying degrees of stringency. According to David Lewis(1986):To say that something can happen means that its happening is com-possible with certain facts. Which facts? That is determined, butsometimes not determined well enough, by context. An ape can?tspeak a human language?say, Finnish?but I can. Facts about theanatomy and operation of the ape?s larynx and nervous system are82not compossible with his speaking Finnish. But don?t take me alongto Helsinki as your interpreter: I can?t speak Finnish. My speakingFinnish is compossible with the facts considered so far, but not withfurther facts about my lack of training. What I can do, relevant to oneset of facts, I cannot do, relative to another, more inclusive set. When-ever the context leaves it open which facts are to count as relevant, itis possible to equivocate about whether I can speak Finnish. (77)Lewis?s view is contextualist since the meaning of ?can? does not, by itself, de-termine which facts are relevant; the additionally relevant facts are determined bycontext. While all uses of ?can? share a common semantic element?they expresscompatibility with certain facts?the precise meaning of a particular use dependson something else, which Lewis calls ?context,? and which we can take to bewhatever facts determine the precise meaning of a particular use of ?can.? So, ?Scan A? means that S?s A-ing is compatible with certain facts, where the relevantfacts depend on the stringency with which ?can? is used. The sense in which de-terminism makes it impossible for someone to do anything other than what shedoes is this: given the actual history and laws, it is not physically possible for herto do anything else. Obviously, a similar contextualist line can be run for claimsabout being able to do otherwise.Horgan maintains that we go beyond the limits of our competence when itcomes to applying notions like being able to do otherwise, at least while assumingdeterminism. Here, we are reasoning about whether the application conditions ofthis notion are met under the assumption of a general metaphysical hypothesis,and we are liable to make mistakes. We do best, Horgan thinks, to proceed insteadby inference to the best explanation; and that, he thinks, favors compatibilism.Still, global worries about whether we are free?even in the sense of beingable to do otherwise?can arise even when contextual parameters are not limit-case, as they are when we explicitly ask the compatibility question. Such worriescan arise even when contextual parameters are normal, as they are when we ask83whether someone is free while making only whatever assumptions are appropriateto everyday situations.Consider, for instance, the sort of case made famous by Harry Frankfurt(1969). In Frankfurt?s original case, Black?a neurosurgeon?wants Jones tochoose A. Black can intervene to control Jones?s brain processes should Jonesbe about to choose B. Yet Black prefers not to intervene unnecessarily. Instead,he waits to see how Jones will choose on his own. Jones is unaware of Black?spresence. Frankfurt claims that Jones lacks alternative possibilities in the case.However, if Jones chooses A on his own, he is apparently responsible for hischoice even though he has no alternative, given that Black is ready to intervene tocontrol Jones?s brain processes.Now assume that Jones?s choosing A is choosing to kill Smith. A natural re-sponse to this case is to ask whether it is reasonable to expect that Jones havedone something else instead, given that the conditions in which he found himselfruled out any alternative?and this despite the fact that these conditions play norole in why Jones does kill Smith (Cf. Widerker 2006). If we think it reasonableto expect that Jones not have killed Smith, then it seems we have located a con-flict in our thinking about how to apply the notion of freedom. On one hand, ifwe consider the case just by focusing on the intervener, without ever consideringdeterminism, we might want to grant?given that Black did not intervene?thatJones freely killed Smith. After all, he killed Smith on his own. On the otherhand, it is not clear whether it is reasonable to expect that Jones have done some-thing else instead. Recall, he was unable to do otherwise. Did Jones freely killSmith? Perhaps we do not know. Have we illicitly raised the contextual parame-ters governing application of the relevant notion of freedom? It is not clear that wehave. Once we point out that determinism is meant to function in the same wayas Black, by blocking the availability of alternatives and thus blocking the abil-ity to do otherwise, we have generated a global worry according to the ordinarystandards governing application of the notion of freedom.15 The standards are15 This is the standard way to understand how Frankfurt-cases work. Nevertheless, one might84ordinary since they do not invoke determinism, yet if Black functions in the sameway as determinism (see footnote 15) then all we have to do in order to generate aglobal worry is to ask someone to imagine that there is always a figure like Blacklurking in the background whenever anyone deliberates about doing anything.If global worries about whether we are free arise from our competence in ap-plying the relevant notion of freedom, even when the parameter settings are nor-mal, then Horgan?s contextualist move proves doubtful.16 Even when we applyordinary standards, which are internal to our everyday judgmental practices anddo not explicitly invoke determinism, it seems that we can generate worries aboutwhether people are free, even in the sense of being able to do otherwise. Thus,it is not clear that any scorekeeping confusion does occur when we raise the pa-rameters and explicitly ask the compatibility question about agentive experience.We may simply be exhibiting our competence in applying the notion of freedomin that context as well.I suggest that people?s incompatibilist judgments about their experience areat least prima facie evidence of the actual nature of the phenomenal content withwhich people are presented when they attend to their agentive experience. Thus,we would need some positive reason, other than the hypotheses canvassed by Hor-worry whether Black really functions in the same way as determinism, since libertarians denythat Black does block alternative possibilities (even though they grant that determinism blockssuch alternatives). However, even if Black does not block all alternatives, he plausibly blocks thesorts of alternatives that would be required for the ability to do otherwise. After all, many haveargued that even if certain alternatives in a Frankfurt-case offer Jones a ?flicker of freedom? (Cf.Fischer 1994, Ch. 7), they are not robust enough to underwrite the sort of freedom to do otherwisethat might plausibly be required for moral responsibility. The central point is that while alternativepossibilities (of some sort, i.e., without begging the question either way regarding the compatibilityissue) are necessary for the ability to do otherwise, they are nevertheless not sufficient for suchan ability. Conversely, a compatibilist might insist that determinism allows for a compatibilistability to do otherwise, even if Black does not. That may be right. However, most compatibilistsaccept that determinism does block alternative possibilities, and thus the ability to do otherwise.Although I happen to be sympathetic to the idea that it does not, it should be noted that there iswide agreement that compatibilist accounts of such an ability are subject to fatal criticisms (Cf.Lehrer 1968; see also Clarke 2009 for why more recent compatibilist accounts of the ability to dootherwise fail).16 In Chapter 5, I suggest another sort of contextualist move that is more plausible, and whichcan deal with Frankfurt-cases.85gan, to think that people systematically misinterpret the nature of such experience.Of course, if Horgan were right that judgmental introspection is unable to providea reliable answer to the experience-compatibilism question, then we would indeedneed an error theory to account for why people make the judgments that they do.In that case, Horgan?s contextualist hypothesis would be preferable to the alter-native hypothesis proposed here, namely that such judgments are accurate andcompetent (as they simply could not be). Yet, given the considerations that countagainst Horgan?s view, his contextualist error theory looks doubtful. That is, it isunclear whether Horgan?s contextualist hypothesis fares better than the alternativehypothesis that agents are competent and in normal contexts.3.6 An Alternative Compatibilist ProposalIn this section, I give a broad outline of an approach that I will explore more fullyin the next chapter. This approach takes takes people?s incompatibilist reportsabout their experience of freedom at face value. The trick for compatibilists, Isuggest, is to argue that such experience has two sorts of content, and thus two as-sociated veridicality conditions. Even granting (at least for the sake of argument)that people?s introspective reports about their experience are incompatibilist andreliably latch onto the relevant phenomenology, there might still a respect in whichthe experience is veridical, assuming determinism.Recall that for any experience, its phenomenal content yields a veridicalitycondition: the content specifies how the world must be for the experience to beveridical. If the experience has two sorts of content, then it has two associatedveridicality conditions. Now take agentive experience. Perhaps our experience ofbeing free to do otherwise has two sorts of content. According to one of these, wewould have to be libertarian agents for our experience to be veridical. Yet maybethis very experience also has another sort of content, which is sometimes veridicalunder determinism.Consider an analogy. On what we might call a pre-Newtonian view of col-ors, we experience colors as primitive properties of objects, spread out over their86surfaces.17 When we see a red apple, what is presented to us in phenomenologyis that a certain object, the apple, has a certain simple property, redness, spreadout over its surface. This property seems irreducible: the apple?s redness does notseem, at least in phenomenology, to consist in any more fundamental property?say, a microphysical or dispositional property of the apple, or some unspecifiedproperty of the apple that plays a causal role in generating our visual experience.The apple just seems primitively red. David Chalmers (2006) calls this perfectcontent.18 Of course, as Newton and Galileo first saw, such a view of colorsis false. Physics tells us that apples are not red (or green) in anything like theway we experience them as being. For physicists like Newton and Galileo, aswell as for philosophers like John Locke, the result of this discovery was coun-terintuitive: there are no colors. Thus, all our experiences of colors are non-veridical. This not only leaves us with no way of distinguishing red from green,but with no way of distinguishing?just in terms of the veridicality conditions ofour phenomenology?illusory (or hallucinatory) color experiences from experi-ences we normally think are veridical. As I will outline in a moment, Chalmersproposes a novel way of escaping this unsatisfactory situation.Now consider agentive experience. Let us grant, at least for argument?s sake,that when we experience being free to do otherwise our phenomenology presentsus (to ourselves) as possessing a certain property: an indeterministic ability. Like-wise, let us grant that this property does not seem, at least in phenomenology, tobe any more fundamental property, such as a microphysical or dispositional prop-erty. It just seems?experientially?that we are able to do otherwise, and in a waythat requires the falsity of determinism. Call this libertarian content. If the phe-nomenal content of our agentive experience is libertarian, this has the result that17 Or spread throughout a volume (e.g., wine), etc.18 Chalmers also calls this content Edenic?it is the content of experiential representations ofthe primitive properties instantiated in ?Eden? (2006: 66). In the ?Garden of Eden,? Chalmerswrites, ?We had unmediated contact with the world. We were directly acquainted with objects inthe world and with their properties. Objects were simply presented to us without causal mediation,and properties were revealed to us in all their true intrinsic glory? (48). My ?pre-Newtonian world?is Eden for colors.87all our experiences of being able to do otherwise are illusory, at least assumingdeterminism. Thus, it leaves compatibilists with no way of distinguishing?justin terms of the veridicality conditions of our phenomenology?illusory agentiveexperiences from experiences we normally think are veridical. Recall our exam-ple from earlier: you wake at night and consider switching on the light. As youlie there, you experience yourself as being free to switch on the light, or to refrainfrom doing so by continuing to lie there motionless. What we wanted was a wayof distinguishing such a case?just in terms of the veridicality conditions of thephenomenology?from one in which, unbeknownst to you, you have been para-lyzed by a drug, yet nevertheless you experience yourself as being free to switchon the light or refrain from doing so. Yet if all your experiences of being free todo otherwise are illusory because determinism is true, compatibilists cannot makesense of the idea that your experience in the first case is veridical in exactly theway it is not in the second, since the phenomenology is the same. This, too, isunsatisfactory.Before I outline my compatibilist proposal for agentive experience, let me firstoutline Chalmers?s proposal for color experiences. When it comes to such experi-ences, Chalmers argues that in addition to perfect content there is also another sortof phenomenal content that makes color experiences veridical, at least in the rightkinds of cases. This sort of content allows us to differentiate illusory experiencesof seeing red from experiences we normally think are veridical. Chalmers callsthis second sort of content ordinary content. Crucially, perfect content serves asa regulative ideal in picking out the ordinary content. What does this mean? Fora given experience of seeing red to be perfectly veridical, we would have to livein a world in which colors actually are primitive properties of objects, spread outover their surfaces. We would have to live in a world of pre-Newtonian colors.The best we can do in our world is to have the relevant experience be causedby whatever actual properties play the role that the relevant primitive propertieswould play in a pre-Newtonian world. Even though no property can play this roleperfectly, some property (or properties) may be able to play it well enough?i.e.,88by being, as a matter of fact, the normal cause of experiences of seeing red. Thiscondition constitutes the ordinary phenomenal content of experiences of seeingred. Such experiences are veridical once they are caused by whatever propertiesof objects normally cause them (under normal conditions). Of course, ordinarycontent does not yield, by itself, an adequate account of the phenomenal contentof experiences of seeing colors, since it does not capture how things seem to usphenomenologically. As a result, Chalmers suggests that color experiences shouldbe thought of as having two sorts of content: perfect and ordinary content. Eachsort of content is useful for a different purpose. Perfect content captures our colorphenomenology, while also serving to pick out the ordinary content by being itsregulative ideal. Ordinary content allows us to make the kinds of distinctions wewant to make between illusory experiences of seeing colors and experiences wenormally think are veridical.Might not a similar story be told for experiences of being free to do other-wise? Perhaps such experiences also have two sorts of phenomenal content, andtwo associated veridicality conditions. First, they have libertarian content, whichcaptures how things are presented to us phenomenologically. Second, they haveanother sort of content as well, which allows us to distinguish cases of illusoryexperiences of freedom from experiences we normally think are veridical. Callthis compatibilist content. Analogously with the color case,19 libertarian contentserves as a regulative ideal in picking out the compatibilist content. What doesthis mean? For an experience of being free to do otherwise to be veridical ac-cording to the standard of libertarian veridicality, which is the standard associatedwith libertarian content, we would need to possess exactly the libertarian freedomwe experience possessing. The best we can do in a deterministic world, however,is to have our experience undergirded by whatever properties actually play therole in such a world that libertarian properties would play in a libertarian world.Even though no property can play this role perfectly, some property (or properties)may play it well enough, by being, as a matter of fact, the normal undergirding19 Note that this analogy need not be airtight; it is only meant to be illustrative.89of our experience of being free to do otherwise. This condition constitutes thecompatibilist phenomenal content of experiences of freedom. Such experiencesare veridical once they are undergirded by whatever properties normally undergirdthem (under normal conditions). Of course, compatibilist content is not, by itself,an adequate account of the phenomenal content of experiences of being free to dootherwise, since it fails to capture how things seem to us phenomenologically: ourexperience is, we are granting arguendo, as of our possessing an indeterministicfreedom. Thus, my proposal is that experiences of freedom have two sorts of phe-nomenal content: libertarian and compatibilist. Each sort of content is useful fora different purpose. Libertarian content captures our phenomenology, while alsoserving to pick out the compatibilist content by being its regulative ideal. More-over, compatibilist content enables us to make the distinctions we want to makebetween illusory experiences of freedom, and experiences we normally think areveridical.3.7 ConclusionThe proposal I have sketched needs to be worked out in greater detail, which is atask I undertake in the next chapter. Yet even in rough outline it exhibits some at-tractive features. First, on the assumption that there are agentive experiences, andspecifically experiences of being free to do otherwise, even if such experienceshave libertarian content there is still wiggle-room for compatibilists to argue thatthese very experiences are veridical, assuming determinism. This enables compat-ibilists to make the sorts of distinctions that they need to make between illusoryexperiences and veridical experiences, while assuming determinism. Second, myview opens the way for a compatibilist theory of freedom to capture the abilitythat most people apparently experience having, even if such experience has lib-ertarian content. Finally, my proposal blocks the libertarian move of trying tojustify belief in libertarianism on the basis of people?s experiencing being indeter-ministically free. On my view, there is no more reason to endorse libertarianism90on the basis of our experience than there is to endorse compatibilism. Indeed,my view has a further attractive feature, since it concedes that libertarianism is atleast descriptively right about the content of phenomenology. Thus, my proposalgoes further than Horgan?s in addressing incompatibilist concerns while nonethe-less remaining prescriptively compatibilist. By all means, compatibilists may nowgrant, our agentive experience has content that is non-veridical under the assump-tion of determinism, just as libertarians claim. Yet that does not pose any problemfor compatibilism.In closing this chapter, I want to note the following. The view that I havesketched here (and which I develop further in Chapter 4) differs from Horgan?s ina crucial respect. My view can grant that people?s experiences are descriptivelyincompatibilist (as the evidence adduced in Chapter 2 suggests it is), and thus arein error in the sense of being (in part) non-veridical if determinism is true. Bycontrast, Horgan?s view says that people?s experiences are not descriptively com-patibilist, and so they are not even partly non-veridical under the assumption ofdeterminism. The error, Horgan thinks, occurs not in the experience?s being non-veridical, but rather in the judgments that people make about the experience. So,in positing a gap between the descriptive and prescriptive questions, I am com-mitted to the possibility of something?s being mistaken or in error if prescriptivetheory says that how we should think about modal freedom differs from how wedo think about it. Yet that does not mean that I am committed to an error theoryin the sense that Horgan is.91Chapter 4Against an Argument forLibertarianismLibertarians often claim that our experience of freedom is indeterministic. Thisclaim then functions as a step in an argument in favor of libertarianism, the viewthat freedom requires indeterminism and we are free. Since, all else being equal,we should take our experience at face value, libertarians argue, we should believelibertarianism. I argue that compatibilists, who think that freedom is consistentwith determinism, can do better than their usual responses to this argument. Ibegin by conceding that our experience of freedom is in a sense libertarian, butI argue that it is also in another sense compatibilist. Thus, even if libertariandescriptions of the experience are correct, there is still a sense in which that veryexperience might be veridical, assuming determinism.4.1 IntroductionThis chapter responds to a certain argument in favor of libertarianism, the viewthat freedom requires indeterminism and we are free:92Argument L(1) The content of our experience of freedom is (presumptively) veridical.1(2) Our experience of freedom has libertarian content.(3) If our experience of freedom has libertarian content, then our experience isveridical only if libertarianism is true.(4) So, libertarianism is (presumptively) true.2The libertarian content is this. For a possible action, A, that one is consideringperforming, one?s experience is as of being able to decide to A, and as of beingable to decide, in an unconditional sense,3 to refrain from A-ing.For compatibilists, who maintain that freedom is consistent with determinism,the obvious responses to Argument L are to reject (1) or to reject (2). This chaptermakes a case for rejecting (3), and also for the claim that (1) is neither true norfalse since it presupposes that there is a unique content to experiences of freedom.4.2 Compatibilist StrategiesSome compatibilists think that premise (1) of Argument L is false.4 These com-patibilists grant premise (2), and thus admit that the experience of freedom has1 More will be said later about veridicality, but for now let us loosely define the term ?veridical-ity? as: accurate in a relevant sense, user some specification of how the world might be. Differentways of spelling out what exactly this means will be sketched later, but this broader, univocal senseof ?veridicality? will remain constant throughout the chapter.2 This argument appears, for instance, in Reid (1788), C. A. Campbell (1951: 463), andO?Connor (1995: 196?197).3 According to the compatibilist ?conditional analysis? of the freedom to do otherwise, anagent, S, is free to do otherwise iff the following is true: If S wanted (or tried, etc.) to do otherwise,then she would do so. That is, something about S?s beliefs, desires, and so on, would have to bedifferent in order for S to do otherwise. Libertarians think that S is free to do otherwise given heractual beliefs, desires, etc. This is what is meant by saying that S is free to do otherwise ?in anunconditional sense?; it is possible that S will do otherwise given her actual mental states. Each ofthe compatibilist?s and incompatibilist?s claims usually require the laws of nature to be the actuallaws.4 For example, Lehrer concedes that when it comes to experiences of freedom, the libertarian?accurately describes what I find by introspecting, and I cannot believe that others do not find the93libertarian content, but they insist that it is non-veridical. However, compatibilistswho adopt this strategy concede what arguably they should not concede: that ourexperience of freedom is non-veridical if determinism is true. Adopting this strat-egy also requires showing that we lack libertarian freedom, rather than merelyshowing that there is a variety of freedom that is compatible with determinism. Itis not clear that compatibilists need to take on this extra argumentative burden.Most compatibilists prefer to reject (2).5 These compatibilists insist that ourexperience of freedom is veridical, but they claim that it has exhaustively com-patibilist content. The experience is simply of having a conditional freedom: aslong as we are free from constraint, coercion, and so forth, our experience is asof being free to decide to A, or to refrain from A-ing, if we want (or try) to doso.6 By rejecting (2), though, compatibilists enter into intractable disputes withlibertarians7 about the nature of the presentational content of our experience offreedom?about whether the content is of having a compatibilist (conditional)freedom or instead a libertarian (unconditional) freedom. It is exceedingly diffi-cult to know how to adjudicate such disputes, given that they turn on competingintrospective claims. Is one side mistaken about what it introspects?With this in mind, another compatibilist strategy is to develop an error the-ory for libertarian judgments about experiences of freedom. According to thisstrategy, even if people judge their experience as incompatibilist, actually it iscompatibilist: people misinterpret their experience.8 Dialectically, this is moresame? (1960: 150). So Lehrer concedes (2). Yet he thinks that our experiences of freedom arenot veridical, and thus he disables Argument L at step (1). Similarly, Hume thinks that ?There isa false . . . experience . . . of the liberty of indifference? (1960/1739: Bk. II, Part III, ?II), by whichhe means that our experience of freedom is libertarian in nature, yet is non-veridical.5 See, for instance, John Stuart Mill (1865: 285), Adolf Gru?nbaum (1952: 672), and TerryHorgan (e.g., 2012; 2011), the latter of whose views I discussed in Chapter 3, and I discuss againin a moment.6 See footnote 3 above regarding the compatibilist ?conditional analysis? of freedom.7 And, more generally, with incompatibilists, who think that freedom is incompatible withdeterminism.8 Horgan (e.g., 2012; 2011) makes this compatibilist move. See Chapter 3 for details.94helpful than just banging one?s fist on the table and insisting that the experienceis compatibilist. Even so, it commits one to arguing that the experience is exhaus-tively compatibilist. It would be useful if compatibilists could avoid shoulderingthis burden.This paper presents a case for rejecting (3). The trick for compatibilists is toargue that our experience of freedom has two sorts of phenomenal content, andthus two associated veridicality conditions. The experience might be veridicalif the satisfaction conditions of just one of these types of content are met. Ifone type of phenomenal content is libertarian, yet an experience of freedom hasanother type of phenomenal content that is compatibilist, the experience mightbe veridical even if only the latter content is satisfied. Libertarianism need notbe true. Thus, Argument L would be blocked at premise (3). However, sincelibertarianism plausibly implies the view that libertarian phenomenal content isveridical, (1) and (2) jointly entail (3). Thus, (3) cannot be denied without denying(1) or (2). In the end, this chapter makes the case that (1) is neither true nor falsesince it presupposes that there is a unique phenomenal content to experiences offreedom.4.3 An Analogy with ColorAn experience that has libertarian content might be veridical if determinism istrue. This is because such an experience might have more than one distinct typeof phenomenal content. The experience might be veridical if the satisfaction con-ditions of just one of these types of content are met. Here, ?phenomenal content?is content that is tied in a particular way to an experience?s phenomenology: itis constitutively determined by the experience?s phenomenal character.9 We can9 This is not to beg any questions about content determination, an important issue in the phi-losophy of mind. In short, we can be neutral about whether content or phenomenology has ex-planatory priority. According to some philosophers, we should explain phenomenology in terms ofcontent: in virtue of redness being represented in experience, for instance, one has phenomenologyas of redness being instantiated. So, intentional properties constitutively determine phenomenal95see how this might work by first looking at how David Chalmers (2006) makes asimilar move in connection with the phenomenal content of color experience.Chalmers thinks that the view about phenomenal content that is most adequateto our phenomenology in visual color experiences is primitivism. On this view, weexperience colors as simple intrinsic properties of objects, spread out over theirsurfaces. As Chalmers puts it,When I have a phenomenally red experience of an object, the objectseems to be simply, primitively, red. The apparent redness does notseem to be a microphysical property, or a mental property, or a dispo-sition, or an unspecified property that plays an appropriate causal role.Rather, it seems to be a simple qualitative property, with a distinctivesensuous nature. (2006: 66)If this is right, then experiences of color have contents that attribute primitiveproperties. This is primitivist content.10 Chalmers thinks it natural to considersuch content to be phenomenal content, given that the properties presented in theexperience seem to be fully determined by the phenomenology.However, Chalmers concedes that, ?For all its virtues with respect to phe-nomenological adequacy, the . . . primitivist view has a familiar problem. Thereis good reason to believe that the relevant primitive properties are not instanti-ated in our world? (2006: 66). According to primitivism, then, none of our ex-periences of color is veridical. As a consequence, primitivism fails to provideproperties. (Proponents of such a view must then explain how content suffices for phenomenol-ogy; see e.g., Tye (1995: ?5.2), Lycan (1996), Rosenthal (1990).) By contrast, others maintain thatwe should explain content in terms of phenomenology: in virtue of the property of redness beingpresented in phenomenology, one?s experience has content that derives from this property. So,phenomenal properties constitutively determine intentional properties. (Defenders of this view areless numerous; they include Horgan and Tienson (2002), and Kriegel (2002).) The reason we canbe neutral on the question of explanatory priority is that even if one?s experience has its content invirtue of what is presented in phenomenology, it might still be true that one has such phenomenol-ogy in virtue of first having content (Cf. Chalmers (2006: 51)). Presumably, one?s having thatcontent would then be explained by an intentionality-first theory, plus a story about how mentalstates get content. So, we can leave it open whether content determines phenomenology, or viceversa, while still making use of the notion of phenomenal content.10 See e.g., John Campbell (1997) for a canonical statement of this sort of view.96us with any way of distinguishing?just in terms of the veridicality conditionsof our phenomenology?illusory or hallucinatory color experiences from experi-ences that are accurate.In addition to primitivist content, Chalmers argues that there is another typeof phenomenal content that makes color experiences accurate, at least in the rightkinds of cases. This second content enables us to differentiate illusory or hallu-cinatory experiences from experiences we know to be accurate. This is ordinarycontent, which has its own associated veridicality condition: it is satisfied iff therelevant object has whatever property (or set of properties) normally causes phe-nomenally red experiences.11Ordinary content is not adequate to our phenomenology, since it does not re-flect the phenomenal character of our color experience.12 As a result, while thissecond content reflects our judgments about veridicality, it fails the important testof phenomenal adequacy, which primitivism passes.Chalmers?s idea is to combine these two views in a way that captures both thetruth-conditional virtues of ordinary content, and the phenomenological virtues of11 The central idea behind this sort of view is that the attribution of color concepts is justified,and our experience of colors is veridical, despite the evident fact that primitive color propertiesare not instantiated in our world. See, for instance, Johnston (1992) for an influential argument tothe conclusion that part of what a philosophical theory of color should do is to make sense of thesituations and practices in which we apply color concepts. Johnston grants that in an importantsense, our world is not colored, since the driving belief behind primitivism, which is that the in-trinsic nature of a color is fully revealed by visual experience as of a given colored object, is mostlikely false, or at least incompatible with other core beliefs we have about color. Johnston callsthis belief ?Revelation? (1992: 223). Even so, he argues that in another important sense, our worldis colored, since there are (usually) properties instantiated that make true ?enough? of our beliefsabout color. Johnston?s view is that colors are best conceived as dispositional properties. Chalmersdenies this claim (2006: 56?58), while nevertheless adopting something close to Johnston?s strat-egy. In place of Johnston?s dispositionalism, Chalmers endorses what he calls a ?quasi-Fregean?view (2002: 135) about how color terms refer, and about how the content of color experiences isveridical (2004, 2006) despite its not being veridical in the way that it would be were primitivecolor properties actually instantiated in our world.12 This is because, as Chalmers puts it (in the passage quoted above), a phenomenally redexperience is as of a ?simple qualitative property, with a distinctive sensuous nature,? and not as of?a microphysical property, or a mental property, or a disposition, or an unspecified property thatplays an appropriate causal role? (2006: 66).97primitivist content.13 His proposal is this. For color experiences to be perfectlyveridical, objects would have to instantiate primitive color properties. Even if anexperience is not veridical in this way, however, it might nevertheless be imper-fectly veridical: it might be veridical according to our ordinary standards of accu-racy.14 These are the standards according to which we differentiate (in everydaylife) veridical from non-veridical experiences of colors?as when we see, ratherthan merely hallucinate, a red apple. According to Chalmers, there is no conflicthere, once we keep in mind that the two notions of veridicality are associatedwith distinct conditions of veridicality. Imperfect veridicality is associated withan ordinary veridicality condition, while perfect veridicality is associated with aprimitivist condition. The result is that color experiences have more than one typeof phenomenal content, depending on the associated notion of veridicality, and anexperience is veridical as long as one of these is satisfied.The most fundamental type of content, Chalmers argues, is primitivist content.This is because primitivist content determines ordinary content via a ?matching?relation, which works as follows. For a color experience to be perfectly veridi-cal, we would have to live in ?Eden,? which is a world in which primitive colorproperties are instantiated by objects. The best that we can do in our world is tohave certain properties ?match? the primitive properties attributed by primitivistcontent, by playing the role that these properties would play in Eden. While noproperty can play this role perfectly, some property (or properties) may be able toplay it well enough, by being the normal cause of phenomenally red experiences.So, ordinary phenomenal content is grounded in primitivist phenomenal content.In Chalmers?s terms, primitivist content serves as a ?regulative ideal? in determin-ing the ordinary content. That is, the primitivist content sets an ideal standard forthe veridicality of phenomenal content, and the ordinary phenomenal content is13 This is why Johnston?s view (1992; see footnote 11 above, is an important precursor toChalmers?s strategy.14 As Chalmers notes (2006), it does not follow that merely because an experience is imperfectlyveridical, it is not ?really? veridical. If anything, it seems more plausible that imperfect veridicalityis what our ordinary term ?veridicality? picks out.98a condition that relates us to whatever properties in fact come closest to meetingthis standard.A similar story can be told for experiences of freedom. Even if the presen-tational phenomenal content of such an experience is libertarian, and therefore isnon-veridical if determinism is true, there is still a second type of phenomenalcontent that might be veridical. This second content is compatibilist.15Let us frame this strategy in another way. It is an empirical discovery thatprimitive color properties are not instantiated in our world. As a result of thisdiscovery, we are left with two options. First, we can conclude that the contentof color experiences is non-veridical. The result of this would be that none ofour experiences of color is veridical. Second, we can rescue the intuition thatthere is still a sense in which our color experiences are veridical, by introducingordinary content. Similarly, the claim of the present chapter is that the correctresponse to Argument L is to say that it is an empirical question whether (1) and(2) are jointly true. At best, Argument L establishes a defeasible presumption infavor of libertarianism. If it turns out that libertarianism is empirically defeated(e.g., if indeterminism turns out to be false), then even if we grant, arguendo, thatexperiences of freedom have libertarian content, we can still rescue the intuitionthat there is a sense in which experiences of freedom are veridical, by appeal tocompatibilist content.4.4 Libertarian ContentAssume that there is something it is like to be an agent, and that agentive expe-riences have rich contents. In perceptual experience, rich contents attribute notonly low-level properties like redness or squareness, but also high-level propertieslike being an apple. Likewise, rich contents for agentive experience attribute high-level properties like being free to do otherwise, in addition to low-level propertieslike being an action.15 Pereboom (2011: 29?40) uses Chalmers?s theory in developing a account of phenomenalconcepts. The view presented here is indebted to Pereboom?s use of Chalmers?s theory.99One might doubt whether there are experiences of freedom. Moreover, even ifone grants that there are such experiences, one might doubt whether they can havecontent that is rich enough to be non-veridical if determinism is true.16 However,let us grant for the sake of argument that this idea makes sense. Assume thatpremise (2) of Argument L is true: there are experiences of freedom that havelibertarian phenomenal content.This phenomenal content has at least two parts. For some action, A, the rele-vant phenomenal content is that (i) one is able to decide to A, and (ii) whether onedecides to A is not determined. Here, (i) is agentive. Yet (i) is not enough for thecontent to be libertarian. We need (ii) for that.How could anyone?s experience have as content that her deciding to A is un-determined by her prior states together with the laws of nature?17 After all, ourexperience presumably does not concern such laws. Furthermore, even if some-one?s experience is not that her deciding to A is determined by her prior states(together with the laws), it would clearly be a mistake to conclude from this thather experience is thereby that her deciding to A is not determined by those states.For present purposes, however, let us assume that content (ii) is of one?s feeling acertain unconditional openness to the future. The future feels open in a way thatwould require indeterminism for the feeling to be accurate. It feels ?as if? one isfree to decide to A or, in an unconditional sense, to refrain from A-ing.We are assuming for the sake of argument that experiences of freedom havelibertarian content of the sort just described. This is what is premise (2) of Argu-ment L says. By analogy with primitivism about color experiences, it may seemthat an experience with libertarian phenomenal content is veridical iff libertari-anism is true. Call a world in which libertarianism is true an ?Agentive Eden.?Perhaps the actual world is an Agentive Eden, although very likely it is not. Leavethat question aside. What concerns us here is only whether experiences of freedom16 After all, the richer the content is, the more demanding the veridicality condition will be.17 The standard characterization of determinism is that whether one decides to A is entailed bya description of one?s prior states together with a statement of the laws.100with libertarian content might be consistent with determinism, in the sense of be-ing veridical even if determinism is true.18 This question can be answered withoutcoming to any verdict about whether libertarianism is true or instead determinismis true.19On the face of it, a libertarian experience of freedom is veridical only if welive in an Agentive Eden, and so it would seem to be obviously non-veridical ina deterministic world. However, the proposal to be outlined in the next sectionsays that this is not obvious. An experience of freedom with libertarian contentmight be veridical, even assuming determinism (and thus despite the fact that thelibertarian content is not veridical).4.5 Compatibilist ContentBy analogy with the color case, the proposal of this paper is that there is a secondsort of phenomenal content to experiences of freedom that is compatibilist. Thiscontent is a condition that a property must satisfy in order to be the property thatis attributed by the experience.20 The property attributed by the experience is, ofcourse, the freedom to do otherwise. What condition might work as the secondcontent for such an experience?18 Note that when we ask whether someone?s experience as of being able to do otherwise iscompatible with determinism, we are asking whether her experience is veridical, assuming deter-minism. We are not asking whether the experience itself is compatible, even if it has libertariancontent. After all, presumably any experience whatever, including a libertarian one, is potentiallycompatible with determinism, since we could be determined to have that very experience (Cf.Mele 1995: 133?37, 246?49).19 Obviously, one cannot determine whether a given content is veridical without specifyingan evaluation context, which is either deterministic or not. Even so, one does not have to becommitted regarding either (a) whether any experience actually has that content, or (b) whetherthe specified evaluation context is actually true. One is merely assuming both the content and theevaluation context in order to assess whether that content could be veridical in that context, in thesame way that one can assess whether two claims are compatible without being committed to thetruth of either claim.20 According to Chalmers?s ?quasi-Fregean? view, the phenomenal content of an experience isthe mode of presentation of a property presented in that experience, not the property itself. This iswhat Chalmers calls an ?epistemic intension? (2002: 135). For experiences, such an intension is?a condition that a property must satisfy in order to be the property attributed by the experience?(2006: 59).101For color, the second content is whatever property (or set of properties) ordi-narily causes phenomenally red experiences. In the agentive case, however, wecannot say that the second content is whatever property (or set of properties) ordi-narily causes experiences of being free to do otherwise, since presumably no onethinks that the meaning of ?being free to do otherwise? is being such that it causesan experience of being free to do otherwise. More plausibly, the second content isthe following condition: that there is instantiated whatever relevant property (orset of properties) is ordinarily instantiated when one experiences being free to dootherwise. This content, which is compatibilist, is veridical iff this condition ismet.There is good reason to think that this condition really does pick out a genuinesecond content for experiences of freedom. Consider that it is at least somewhatplausible to think that ?freedom? is a natural-kind term that refers to whateverrelevant processes are at work in choices or decisions that we ordinarily call ?free,?or that feel free to us. Consequently, we might be free even if determinism istrue.21 It follows from this view that so long as the relevant processes actuallyconstitute a natural kind, we are free.22Another way to state this sort of compatibilist position is as follows. We of-ten begin theorizing about freedom by pointing to paradigm cases?for example,cases in which an agent acts while free from coercion, constraint, and so forth. Theparadigm-case view says that we should stop there: simply point to actual choicesor decisions that we call ?free,? or that feel free to us, and those are paradigm casesof freedom.2321 Heller (1996) has argued that ?freedom? (or ?free will?) is plausibly a kind term comparable tonatural-kind terms. Heller thinks that in just the way that cats might turn out to be non-biologicalrobots, so our free actions may turn out to be determined.22 Cf. Heller (1996: 336). If the choices we call ?free? do not constitute a kind, then there areno free choices.23 The paradigm-case view is suggested by Flew (1959), among others. For criticism of theview, see Danto (1959), van Inwagen (1983: ch. 4), and Double (1996: ch. 2; 1997). As Hellerpoints out (1996: n.7, 336), the paradigm-case view is actually stronger than the kind view aboutfreedom, since it guarantees that there are free choices, whereas the kind view does not (seefootnote 22).102The natural-kind view is not unproblematic. One problem is that ?freedom?seems to behave differently from standard natural-kind terms like ?water? (cf. Bal-aguer 2010: n.5, 22).24 ?Water? refers to the aqueous stuff that flows in our riversand falls from our sky as rain, whatever that stuff happens to be. If we were to dis-cover that our aqueous stuff is actually XYZ rather than H2O, then ?water? wouldrefer to XYZ (and would have done so all along). One thing that presumably wewould not say if we made such a discovery is that water does not exist. The prob-lem for the natural-kind view about freedom is that ?freedom? appears to workdifferently from ?water? in this regard. If we were to discover that the processesat work in choices that feel free to us include our being remotely controlled byMartians, it seems that we would not conclude from this that ?freedom? refers toour being remotely controlled by Martians. If we made such a discovery, it wouldbe a discovery that we lacked freedom. So the natural-kind view about freedomseems mistaken.This problem might be remedied. Imagine that what we experience as water is,as it turns out, a hologram accompanied by remotely-induced sensations. There isno aqueous stuff at all. Nevertheless, it might be true that ?water? refers to a naturalkind just so long as no hoax of this sort is going on. The same goes for freedom.As long as nothing like remote control by Martians is going on, ?freedom? refersto a natural kind.Even granting that Martian-control scenarios do not threaten a compatibilistnatural-kind view about freedom, incompatibilist arguments put pressure on sucha view by suggesting that libertarian conditions on freedom should be includedamong the constraints that must be satisfied for there to be a proper referent for?freedom.? In order to provide a non-question begging defence of the claim that?freedom? picks out a compatibilist natural kind, the appeal of such incompatibilistarguments must be explained away.2524 Mark Balaguer talks about ?free will.? ?Freedom? picks out the same thing as ?free will.?25 I owe this point to Gunnar Bjo?rnsson, in correspondence.103Whatever intuitive plausibility the natural-kind view about freedom has, theanalogous view for experiences of freedom is more plausible. This is because theanalogous view is part of an overall position that already accommodates an incom-patibilist perspective. By granting libertarian phenomenal content, the claim thatthere is a second, compatibilist content for experiences of freedom does not comeunder the same sort of pressure from incompatibilist arguments as the natural-kindview about freedom does. The analogous view says that one type of phenomenalcontent is libertarian, and is non-veridical under the assumption of determinism.Still, such experiences have a second type of phenomenal content that is compati-bilist. This content is veridical under normal conditions, even assuming determin-ism.What are normal conditions? First, they are conditions in which the relevantexperience is not obviously illusory. Consider an example. Someone has beenhypnotized. A performing hypnotist has primed a subject to pour a glass of waterover her own head as soon as she hears the hypnotist cough. The subject is un-aware of this. The hypnotist tells his audience what will happen. Then he gives hissubject a glass of water. He tells her that if she drinks it in less than five seconds,she will win a prize. He also reminds her that she should not feel obliged to drinkit. The subject agrees. She takes the glass, the hypnotist coughs, and the subjectpours the water over her own head.When the subject accepts the glass of water, presumably she experiences be-ing free to do various things: to drink it, or to refrain from drinking it (e.g., byrefusing it, or by pouring it over her own head, or . . . etc.). Yet the hypnotist isskilled: whenever he succeeds in hypnotizing a subject, the subject never fails topour the water over her own head. In this case, too, the subject is unable to do oth-erwise. If so, then her experience of being free to do otherwise is non-veridical,and obviously so.2626 This experience is illusory on both contents?libertarian and compatibilist. The point is thatunless we introduce compatibilist content, then if indeterminism is false we have no way of distin-guishing (just in terms of the veridicality conditions of the experience) non-veridical experiences104Call this a locally abnormal situation. Experiences of freedom that are notillusory in anything like this way are candidates for those that occur under normalconditions.Additionally, we might want to rule out globally abnormal situations, such aseveryone?s being remotely controlled by Martians. Thus, we might add a require-ment that there are certain constraints that must be satisfied?for instance, ournot being controlled by Martians?in order for a situation in which we experiencebeing free to do otherwise to count as normal.In normal situations, then, one?s experience of freedom is veridical iff a certaincondition is met. This condition is that one instantiate whatever relevant property(or set of properties) is ordinarily instantiated when one experiences being free todo otherwise. This is compatibilist content.4.6 The Two-Stage ViewHow are the two types of phenomenal content related? Among them, the mostfundamental type of content is libertarian. This is because we are assuming thatlibertarian content is the content that most closely reflects our presentational phe-nomenology in experiences of freedom: it accurately reflects (we are assuming)what is presented in our phenomenology.27Analogously with the color case, presentational content determines the secondcontent via a matching relation. For an experience of being free to do otherwiseto be perfectly veridical, we would have to live in an Agentive Eden. The best thatwe can do if determinism is true is to have certain properties match the libertarianproperties that are attributed by the presentational content, by playing the role thatthese properties would play in an Agentive Eden. No property can play this rolelike this from experiences we normally judge to be veridical.27 If premise (2) of Argument L were false, and experiences of freedom actually had exhaus-tively compatibilist content, then our presentational content would be compatibilist.105perfectly. Yet some property (or set of properties) may be able to play it wellenough, by being the property (or set of properties) that is ordinarily instantiatedwhen one experiences being free to do otherwise. Thus, compatibilist contentis grounded in libertarian content. Or, to use Chalmers?s phrase, presentationalcontent acts as a ?regulative ideal? in determining the second content. That is,the libertarian phenomenal content sets an ideal standard for veridicality, and thesecond, compatibilist content is a condition that relates us to whatever propertiesin fact come closest to meeting the ideal standard. Once the second content issatisfied, the experience is imperfectly veridical, even if determinism is true.Despite granting premise (2) of Argument L, which says that the relevant pre-sentational content is libertarian, our experience of freedom might be veridicaleven if it turns out that determinism is true.28 Thus, Argument L is disabled atits third premise. The experience can be veridical, it can have libertarian con-28 It might be objected that if there are two phenomenal contents to our experience of freedom,then the claim ?our experience of freedom is veridical? is ill-defined, since veridicality is judgedrelative to a content and context, yet this claim only gives the context, not the content. However,the present chapter adopts Chalmers?s pluralism about representational contents, which amountsto the claim that there are multiple content relations (i.e., relations that associate experiences withcontents). If one relation associates a given experience (such an experience of freedom) withone sort of content, nevertheless another relation might very well associate that experience withanother sort of content. In this way, an experience of freedom might be associated with more thanone sort of content via different content relations. Of course, as Chalmers notes (e.g., 2006: 51?52), not all of these contents can be phenomenal contents, since if a content varies independently ofthe phenomenal character of the relevant experience, then that content is not phenomenal content.So, even if it is definitional of an experience that it has phenomenal character, still the contents ofan experience do not have to be exhausted by the content associated with its phenomenal character.However, here we are focusing just on an experience?s phenomenal content relations, and thus theclaim is that a given experience?for instance, an experience of freedom?might well have morethan one phenomenal content relation and, as a result, more than one sort of phenomenal content.As Chalmers puts it, ?For ease of usage, I will speak of the phenomenal content of an experience,but we should leave open the possibility that there is more than one phenomenal content relation,so that a given experience can be associated with phenomenal contents of more than one sort?(2006: 52). Thus, the claim of the present chapter is that (for ease of usage) an experience offreedom that is associated with libertarian phenomenal content via one content relation, mightwell be associated with another, compatibilist phenomenal content via another content relation,such that even if (for ease of usage) it is non-veridical when assessed according to its libertariancontent (e.g., if determinism is true), still it might be veridical when assessed according to itscompatibilist content.106tent, yet libertarianism can be false. Since libertarianism plausibly implies theview that libertarian phenomenal content is veridical, however, premises (1) and(2) of Argument L jointly entail (3). Thus, we cannot deny (3) without denying(1) or (2). This chapter has argued that (1) is neither true nor false since it pre-supposes that there is a unique phenomenal content to experiences of freedom,namely libertarian content. Yet experiences of freedom plausibly also have thesort of compatibilist content just defended. If we discover that libertarianism isfalse because determinism is true, then even granting (arguendo) that experiencesof freedom have libertarian content, there is still a sense in which experiences withlibertarian content are veridical (despite their veridicality not being in virtue of thelibertarian content).4.7 ConclusionThat is the case for rejecting premise (3) of Argument L, and thus Argument Litself.The unsoundness of Argument L undermines a central motivation for libertar-ianism, since it removes any presumption in favor of libertarianism based solelyon our experience of freedom. Showing (3) to be false also secures new groundfor compatibilism. Indeed, this strategy has theoretical advantages over rival com-patibilist responses to Argument L.First, responses that grant libertarian phenomenal content by granting premise(2) of Argument L, yet which deny the presumed veridicality of the experienceby denying (1), concede what they should not concede: that our experience offreedom is entirely non-veridical if determinism is true. The compatibilist viewdefended here makes the experience veridical, despite granting that it has libertar-ian content. The verdict about premise (1) is that it is neither true nor false, sinceit presupposes a unique phenomenal content to experiences of freedom?namely,libertarian content. The central claim of this chapter is that this presupposition isfalse.Second, responses that deny premise (2), and which thereby insist that the107relevant experience has exhaustively compatibilist presentational content, cannotavoid entering into intractable disputes with libertarians about the nature of thepresentational content of experiences of freedom. By accepting, at least for ar-gument?s sake, that the presentational content is libertarian, the strategy proposedhere avoids such disputes. Let such content be libertarian. That is no threat tocompatibilism.108Chapter 5Indeterminism, Experience, andCompatibilismRecent evidence shows that (1) people tend to believe that they possess indeter-minist freedom, and (2) people also experience possessing such freedom. Somealso argue that (3) belief in indeterminist freedom has its source in people?s expe-rience. Shaun Nichols denies (3), despite endorsing (1) and (2). Nichols providesan alternative account of the source of indeterminist beliefs. I argue that Nichols?account has significant shortcomings, and that belief in indeterminist freedom has(in part) its source in indeterminist experience. I explain how this works by appealto the phenomenon of prospection, which is the mental simulation of future pos-sibilities. Crucially, prospection can be experienced. Further, because of the wayin which prospection models choice, it is easy both to experience and to believethat one?s choice is indeterministic. Even so, belief in indeterminist freedom isnot justified.5.1 IntroductionIn this chapter, I address the question where the libertarian content of agentiveexperiences of modal freedom comes from. As we saw in chapters 3 and 4, even if109people actually have experiences of being free to do otherwise, it remains unclearwhether such experiences could possibly have content that is rich enough to benon-veridical if determinism is true. In chapters 1?4, we also saw that libertarianstend to think that people?s experience of deliberating and choosing actually haslibertarian content. Moreover, libertarians think that this sort of experience offreedom leads people to believe that human choice is indeterministic. (The resultsof the studies I reported in Chapter 2 indicate that libertarians may be right aboutthis.)In previous chapters, I assumed for the sake of argument that the relevant liber-tarian content was of one?s feeling a certain unconditional openness to the future:the content is of the future being open in a way that would require indeterminismfor the experience to be accurate. Where might this sort of experiential contentcome from? I also assumed that people?s beliefs about freedom plausibly havetheir source in experience, and thus libertarian beliefs can be explained by appealto libertarian experience.Shaun Nichols (2012) argues that belief in libertarian freedom cannot have itssource in experience. Nichols argues instead that people believe that they possessindeterminist or libertarian freedom since they think that the psychological factorsthat are introspectively accessible do not determine their choice. Since peoplebelieve that they have introspective access to all the relevant factors that influencetheir choice, they infer that their choice is not determined. But this inference is notwarranted, since people do not actually have access to all the factors that influencetheir choice.As I will argue in the present chapter, a shortcoming of Nichols? view is itsbackward-looking focus: it relies on the idea that people?s indeterminist beliefsderive from introspection on the causes of decisions. By contrast, I will argue thatpeople?s indeterminist beliefs derive from their experience of navigating into thefuture, rather than from introspection on the causes of decisions.110A number of recent authors have embraced a forward-looking model of agen-tive experience. They think that people?s agentive experience enables them?minimally?to distinguish self-generated actions from both involuntary bodilymovements and the externally caused movements of objects. Tim Bayne (2011,2008) develops a view of this sort. Yet Bayne?s view is unable to explain people?sexperience of indeterminist freedom, and so it cannot explain how people?s beliefin indeterminist agency derives from their experience as deliberating agents. Thisis because Bayne?s model focuses on perceptual experience, and it is difficult tosee how such experience could have as content that one is free to do otherwise.That would require a comparison of two or more distinct representations?thealternative possibilities themselves?in the mind, and that is not a perceptual op-eration.I explain people?s experience of indeterminist freedom, and thus the source oftheir belief that they possess such freedom, by appeal to prospection, which is themental simulation of future possibilities for the purpose of guiding action. Cru-cially, prospection can be experienced, and because of the way in which prospec-tion models choice, it is easy for deliberating agents to experience (and to think of)their alternatives as ones they can ?get to? in the sense that libertarians maintainmight be provided for by indeterminism. Even so, belief in indeterminist freedomis not thereby justified.In this way, my view provides a deflationary explanation of a major motivationfor libertarianism?namely, the practical experience of deliberation and choice?while nevertheless granting that it is natural for agents to believe that their choiceis indeterministic. At bottom, people?s libertarian beliefs derive from their experi-ence of navigating into the future, rather than from introspection on the causes oftheir choices. Thus, people?s indeterminist beliefs have their source in the expe-rience of deliberating and choosing, exactly as libertarians maintain, and againstNichols? claim.1115.2 Do Indeterminist Beliefs Have Their Source inExperience?Many people apparently believe that human choice is (at a minimum) undeter-mined. They believe that when it comes to human choice and decision-making,the world is, as John Martin Fischer puts it (borrowing a phrase from Jorge LuisBorges), a ?Garden of Forking Paths.? In Fischer?s words:We naturally think of the future as open. We think of the future ascontaining various paths that branch off one past; although we knowwe will travel along just one of these paths, we take it that some ofthe other paths are (at least sometimes) genuinely accessible to us. Indeliberating and deciding on a course of action, we intuitively thinkof ourselves (at least sometimes) as determining which path to take,among various paths we could take. (1994: 190)If determinism is true, however, there is only one physically possible path into thefuture. In other words, assuming determinism, there is just one physically possibleway for any agent to extend the present into the future, given the laws and the priorstates of the world (Cf. Fischer 1994: 88). Thus, the way in which deliberatingagents believe that they can choose among alternatives seems incompatible withdeterminism. If agents believe they are free in this way, they believe that theypossess indeterminist freedom.Libertarianism is the philosophical formalization of this view: freedom is in-compatible with determinism, and we are free. A central element in libertarianismis the belief that indeterminism is usefully implicated in decision-making: inde-terminism is necessary for free choice. Since libertarians believe that we are free,they consequently believe that indeterminism is true.Libertarianism is not widely held as a philosophical view. Yet it appears to bestrongly implicated in ordinary thinking about human agency, given that partic-112ipants in a number of experimental studies have tended to regard human choiceas indeterministic. In one experiment (Nichols and Knobe 2007), participantswere given descriptions of a deterministic universe (A) and another universe (B)in which everything is determined except human choices. Participants were asked?Which of these universes is most like ours?? More than 90% of respondentssaid that the indeterministic universe, i.e., universe (B), is most like our own(2007: 669). Further evidence in support of the view that people think of humanchoice as indeterministic comes from experimental studies conducted by Nichols(2004). In one experiment, children were asked whether an agent was able to dootherwise than she did, even if everything stayed exactly the same right up untilthe moment she made her choice. A sample question went as follows:Joan is in an ice cream store and wants some ice cream. She choosesvanilla . . . If everything in the world was the same right up until shechose vanilla, did Joan have to choose vanilla? (2004: 486?7)The children were old enough to understand counterfactual conditionals and weregiven comprehension checks to ensure they understood what was being asked.They were also given a contrast case involving a pot of boiling water, and wereasked whether, with everything staying the same right up until the water boiled, itwas possible that the water would not have boiled. The results showed a signifi-cant difference between the agentive case and the non-agentive case involving theboiling water. Participants tended to agree in the agentive case that the agent couldhave done otherwise, yet they tended to deny that the water could not have boiled.Nichols interprets these results as indicating that people regard human choice asindeterministic.Where do indeterminist beliefs come from? Libertarians often cite experienceas the source. They think that the way in which people experience their choicesleads them to believe that choice is indeterministic. The results of the studies re-ported in Chapter 2 indicate that libertarians may be right about this. Recall thatin those experiments, participants were asked to decide between two options (for113instance, two charities) and were asked whether, as they faced their decision, theyexperienced being able to choose either option. Participants were free to interpretsuch an ability however they wished. Most participants reported experiencing be-ing able to choose either way. Determinism was then explained to the participants.Following comprehension checks, participants were asked whether the experiencethey had earlier reported having was consistent with determinism. Most partici-pants tended to judge their experience as incompatibilist, whether the decisionwas present-focused or retrospective, imagined or actual, or morally salient ormorally neutral. The only case in which participants did not give an incompati-bilist response was when they were asked whether their experience of ignoranceof the future was consistent with determinism. This suggests that libertarians areright in claiming that people?s belief in indeterminist freedom has its source inexperience.Yet according to Nichols (2012), belief in indeterminist freedom cannot derivefrom experience. Nichols thinks that some beliefs?such as the belief that it iscurrently raining outside?may derive directly from raw experience?i.e., fromexperience not shaped by beliefs.1 However, Nichols insists that indeterminism istoo sophisticated a notion to grasp simply on the basis of experience: ?. . . as anexplanation for the belief in indeterminism, the appeal to experience is too anemicto be convincing? (2012: 293).2Nichols suggests instead that indeterminist beliefs might derive from experi-ence shaped by beliefs. After all, it is well known that background beliefs canalter perception in certain ways, via the phenomenon of cognitive penetration.Roughly, cognitive penetration occurs when the phenomenal character of one?sexperience is altered by one?s cognitive states?for instance, by one?s backgroundbeliefs or thoughts. If two subjects report different visual experiences from oneanother when looking at the same patch of color under the same conditions, thethesis of cognitive penetration says that these perceptual differences are to be ex-1 I explain what I mean by ?shaped by beliefs? in the next paragraph.2 It is my aim in this chapter to challenge this claim.114plained by differences in the subjects? cognitive states, which in turn alter theirrespective experiences. There is ample evidence that this occurs in cases of visualexperience (e.g., Delk and Fillenbaum 1965; Levin and Banaji 2006). If cognitivepenetration occurs even in cases of visual experience, then in the absence of anyreason to think that it cannot occur in cases of agentive experience, it is at leasta plausible hypothesis that background beliefs may influence the character of ex-periences of freedom as well. As a result, even if experiences of being free tochoose otherwise are initially compatibilist in character, that does not mean thatthey remain compatibilist. Incompatibilist beliefs might shape the experience andmake it incompatibilist.A problem with this explanation of indeterminist beliefs is that it is circular:it explains the relevant belief by appeal to experience, which is in turn shaped byindeterminist beliefs.3 As Nichols puts it, ?If experience is supposed to providea noncircular explanation for our belief in indeterminism (or our theoretical resis-tance to determinism), then it has to be in virtue of experience that is not guidedby an indeterminist belief? (2012: 294).4To avoid this problem, we might reconsider the possibility that Nicholsrejects?namely, that indeterminist beliefs in fact have their source in experi-ence.5 For instance, Richard Holton maintains that, ?Our experience tells us thatour choice is not determined by our beliefs and desires, or by any other psycho-logical states?intentions, emotions etc.?to which we have access. Those could3 Nichols? proposal that experiences of freedom are shaped by beliefs is also problematic sincethere is no positive evidence of cognitive penetration occurring in these experiences, howeverplausible this hypothesis might be.4 The possibility remains that explaining the belief as deriving from experience is circular inan unproblematic way: it is a positive feedback loop. Experience by itself may be too ?anemic?to be the source of indeterminist beliefs. But experience might influence beliefs in a minor way atfirst, and then beliefs might penetrate experience, and then experience in turn support belief, andso on, in a sort of ?ratchet? or ?bootstrapping? account. However, Nichols does not explore thispossibility, and I leave it aside in what follows. Nichols thinks that ?The idea of indeterminism. . . is presumably much too complex to be directly given by raw experience? (294). It is thisstronger claim that I wish to challenge.5 This move is also motivated by the fact that there is no positive evidence that experiences offreedom are shaped by beliefs. See footnote 3.115be the same, and yet we could choose differently? (2006: 15). As a result of suchexperience, people form indeterminist beliefs.Terry Horgan (e.g., forthcoming) agrees that people form indeterminist beliefson the basis of experience. Yet he denies that the experience is indeterminist. Aswe saw in Chapter 3, Horgan concedes that people might think that they can telljust by introspection that their experience is indeterminist. However, this reflectsa form of introspective error or confabulation. As Horgan puts matters, it is onething to know (A) by introspection:(A) My experience does not present my choice as determined by my prior states.But it is another thing to know (B) by introspection:(B) My experience presents my choice as not determined by my prior states.Horgan agrees that it might be possible to know whether (A) is true by introspect-ing. Yet it is not possible to know whether (B) is true by introspecting. When oneasserts (B), thus judging one?s experience as indeterministic, either one is mistak-enly inferring (B) from (A) or conflating (A) and (B). Either way, the result is afallaciously formed belief in indeterminism.Nichols (2012) thinks that even this etiology for people?s belief in indeter-minist freedom cannot be correct, since it fails as an adequate explanation of therelevant belief. We are not introspectively aware of the causes of our headacheseither, Nichols argues, yet we do not infer indeterminism from that:It would be a kind of scope fallacy to move from ?I don?t experi-ence my actions as determined? to ?I experience my actions as notdetermined.? Now, people surely do commit fallacies. But notice wedon?t seem to commit the scope fallacy when it comes to headaches.That is, the phenomenology of headaches doesn?t present us with a setof deterministic headache-causes, but we don?t leap to indeterministconclusions there. (2012: 296)116Nichols proposes an alternative explanation for the source of belief in indetermin-ist freedom. First, he outlines how scientists judge whether a system is indetermin-istic. Researchers control for inputs to the system, and if it has different outputsgiven the same inputs, then one is warranted in concluding that the system is in-deterministic (2012: 297). Next, Nichols presents an argument for indeterminismadapted from William of Ockham (2012: 298?99):(1) The factors that are introspectively accessible do not determine my choice.(2) I have introspective access to all the (proximal) factors that influence mychoice.(3) Therefore, my choice is not determined.Here, (2) is the crucial premise. One might, of course, think that this premise isfalse. Indeed, Nichols cites evidence showing that people do not assume that theyhave access to everything in their minds. However, (2) is actually a more limitedclaim: ?All that is required is a kind of default (but defeasible) assumption thatthe causal influences on decisions [emphasis added] are introspectively available.And there is evidence that people do have such a default assumption? (2012: 299).So, people tend to believe (2) as stated. Furthermore, it seems that decisions arespecial in this regard. People do not have the same default assumption when itcomes to whether all the proximal factors that influence their urges, for example,are introspectively accessible. Presumably, the same goes for headaches.Why do people make the default assumption for decisions, but not for urgesor headaches? Nichols thinks that it is because of a certain bias: ?Our atten-tion is drawn to factors that are present, rather than to the possibility that thereare hidden factors? (2012: 301). Normally, Nichols maintains, people take them-selves to have a good understanding of how mechanisms with discrete, accessiblecausal parts work?for instance, locks or zippers. (The relevant causal parts are?present.?) Yet when people are asked to explain how locks or zippers work, of-ten they find themselves at a loss. As a result, people ?downgrade? their level of117presumed understanding. By contrast, people do not think that they have a goodunderstanding of how objects like flash drives work, given their lack of easy ac-cess to the causal parts of such devices. (The causal parts are ?hidden.?) Nicholsthinks that when people introspect on decisions, they find that they have access todiscrete mental states (beliefs, thoughts, desires, etc.) that causally influence thesedecisions. As a result, they take decisions to be like locks or zippers. When theyare presented with evidence from psychology detailing the actual, unconsciouscauses of their decisions, they may be brought to downgrade their level of under-standing. That is, they may be brought to reject (2). People?s default setting is tothink of decisions as like zippers, but of headaches as like flash drives. Yet justas people can be brought to downgrade their presumed understanding of zippers,they can also be brought to downgrade their understanding of decisions.This explains why people tend to believe premise (2)?in other words, theclaim that we have introspective access to all the proximal factors that influenceour choice?as long as this claim about access is restricted to choices or decisions.It also shows that belief in indeterminism is not warranted, at least if it results fromanything like an inference from (1) and (2) to (3). This is because people?s defaultassumption that they have introspective access to all the factors that influence theirdecisions is mistaken.A major shortcoming of Nichols? proposal is its backward-looking focus: itfocuses on individuals? access to the causes of their decisions. This is true of bothHorgan?s and Holton?s accounts as well. By contrast, deliberation is centrally aforward-looking phenomenon. When agents look to the future as deliberators, twoaspects of agency become especially salient, which Holton, Horgan, and Nicholsall fail adequately to address: (i) the experience of having alternative possibilitiesfor action, and (ii) the experience of being able to decide between such alterna-tives. These are some of the aspects of experience that libertarians most often citeas indeterministic.66 See e.g., Searle 1984: 95; C. A. Campbell 1951: 463; Ginet 1997: 89; O?Connor 1995: 196?197.118Against Nichols? claim, I maintain that people?s indeterminist beliefs derivefrom their experience of navigating into the future, rather than from introspectionon the causes of their decisions. I also maintain that such beliefs do not derivefrom a mistaken judgment about experience, as Horgan claims. Further, althoughsuch beliefs partly derive, as Holton thinks, from a feeling that one?s choice isnot determined, this is due in the first instance to the forward-looking character ofdeliberative experience. Nevertheless, while I concede that it is natural for agentsto arrive at indeterminist beliefs on the basis of such experiences, these beliefs arenot justified.5.3 A Forward-Looking ModelA number of authors have recently embraced a forward-looking model of agen-tive experience.7 According to these authors, our experience as agents enablesus?minimally?to distinguish self-generated actions from both involuntary bod-ily movements and the externally caused movements of objects. One prominentview of this sort is developed by Tim Bayne (e.g., 2011).On Bayne?s view, agentive experiences are produced by a dedicated percep-tual system, which informs us about aspects of our own agency. This systemincludes forward models of action control. The forward models receive a copy ofthe agent?s motor commands, which are used to predict the sensory consequencesof the agent?s bodily movements. The copy of the motor commands is also sent toa comparator, so-called because it compares the predicted sensory consequencesof the agent?s movements with sensory feedback.8 When the comparator iden-tifies a match between prediction and feedback, it identifies the changes as self-7 See, e.g., Bayne 2008, Synofzik et al. 2008, Bayne and Pacherie 2007, Blakemore and Frith2003.8 According to Bayne, the comparators lie between the standard perceptual systems and themotor system, since they take both perceptual representations and motor representations as inputs.In this way, the states generated by the comparators have the functional role of perception. That isto say, like other perceptions, the function of the sense of agency is to generate representation ofsome domain and make these available to the agent?s cognitive systems in an experiential format(2011: 358?59).119generated. When there is no match (or a weak match), the changes are identifiedas externally caused.In this way, the model explains how agents are able to distinguish?experientially?self-generated actions from (i) the externally caused movementsof objects in the environment and (ii) their own involuntary bodily movements.This is as far as Bayne extends his model. Yet one might think that it canexplain more. In a suggestive passage on the experience of freedom, Horganwrites:Some survival-important features of a creature?s ambient environmentwill be ones that are susceptible to causal influence by suitable bodilymotions by the creature itself, motions that can be internally gener-ated by the creature?s inner motion-control mechanisms. (Consider abear, for instance. In an appropriately fortunate ambient environmentin which there is a bush nearby with edible berries on it, there are po-tential bodily motions available to the bear that would have the effectof transferring some of those berries from the bush itself to the bear?sstomach. For such potential bodily motions, the anticipatory-freedomphenomenology of ?I can? (vis-a`-vis those potential bodily motions)will be beneficial to the bear, as will the ongoing free-agency phe-nomenology experienced by the bear during feeding.) Other survival-important features of a creature?s ambient environment will involveevent-causal goings on that are not susceptible to causal influence bythe creature?s potential bodily motions, but that need attending to (andresponding to) if the creature is to survive and flourish. (For instance,if a bear sees a huge boulder rolling down the mountainside in hisdirection, this ought to be registered by the bear as a state-causal se-quence that (i) cannot be influenced by certain bodily motions, and(ii) will be big trouble if the bear?s body remains where it currentlyis.) (Forthcoming)The suggestion here is that something like Bayne?s model might explain experi-120ences of having a general ability to do various things. Over time, an agent learnsto identify, experientially, the kinds of options that exist for her as alternatives foraction at a time, and these are picked out against a backdrop of phenomena that sheexperiences herself as powerless to affect (e.g., eating those berries vs. stoppingthat boulder rolling down the mountainside). This provides her with experien-tial representations both of alternative possibilities and of a general ability to dootherwise, and in such a way that each of these representations is uncontrover-sially compatible with determinism. The agent not only experiences (i) volun-tary actions differently from externally caused movements, but also experientiallyidentifies (ii) the kinds of alternatives that are amenable to her control at a time.Even so, such a model is unable to explain libertarian experience, or to explainhow indeterminist belief might derive from an indeterminist experience of choos-ing and deliberating. More would need to be said about how experiences of (ii)are generated. Bayne?s model invokes only ?systems that are concerned with mo-tor control and production? (2008: 198), and thus it is not clear whether his modelcould possibly account for more than experiences of (i). In short, ?the low-levelcontents of agentive self-awareness,? such as the experience of self-generated ac-tions or of externally-caused movements, might be ?generated exclusively by low-level comparator systems? (198). By contrast, ?the high-level contents of agentiveself-awareness (such as the kinds [emphasis added] of actions that one takes one-self to be carrying out),? might have their etiology in another system, or in thecomparator system operating together with some other system (198?99).Nevertheless, Bayne thinks that we have a robust experience of freedom:?there is clearly an intuitive sense in which we can?and often do?experienceourselves as acting freely,? although ?[t]he difficulties that confront us in at-tempting to articulate what it is like to experience oneself as a free agent are. . . particularly imposing? (2008: 195?6). In particular, Bayne is unsure whetherit is possible for anyone to experience libertarian freedom:[P]erhaps experiential systems are incapable of inserting a negationquite where it needs to be inserted in order to represent libertar-121ian freedom. In order to represent an action as free in a libertariansense one must not only represent it as undetermined by one?s priorpsychological properties but also as undetermined by one?s physicalproperties?or indeed any physical properties. And it is not obviousthat experiential systems have that kind of representational power.(2008: 196)9Whatever the attractions of Bayne?s comparator view might be, such a model can-not explain indeterminist experience, and it cannot explain either how anyone?sbelief in indeterminist freedom might derive from such agentive experience. Thereason for this, I think, is that Bayne?s model focuses on perceptual experience,and it is difficult to see how such experience could have as content that one is freeto do otherwise, in any sense. To borrow Nichols? phrase, perceptual experienceis too ?anemic? to represent the possibilities required for such an experience. Thatwould seem to require a comparison of two or more distinct representations?thealternatives themselves?in the mind, and that is clearly not a perceptual opera-tion.10In order to account for experiences of freedom, I turn to recent work done onprospection, which is the mental simulation of future possibilities. By embracinga forward-looking account of prospection, I will show how we can explain howpeople experience indeterminist freedom, and how they acquire a belief in suchfreedom. Even so, libertarian beliefs are not warranted.9 This echoes the point made by Horgan and Nichols: even if we do not experience our choicesas determined by our own prior states, it seems implausible that we also experience them as notdetermined by such states. Note also that Bayne?s comment here seems at odds with his own view,since it slips into the backward-looking framework that considers only the causes of decisions,while his comparator model is importantly forward-looking.10 As a consequence, it is not obvious how any development of Bayne?s view could account forthe sort of freedom at issue in free-will debates. Experiences both of general abilities and of thekinds of actions one can perform at a time are obviously compatibilist. No one thinks that generalabilities are incompatibilist, and thus the experience of having general abilities could be accurateassuming determinism. So, even a development of Bayne?s view along the lines suggested in thepassage from Horgan would fall short of being an account of anything that is a matter of contentionregarding freedom.1225.4 Etiology of Indeterminist ExperienceAlthough Bayne is correct to focus on the forward-looking aspects of agency, hisview does not go far enough. Bayne?s view does not explain the sort of experiencethat might generate a belief in indeterminism. Still, unlike Nichols, I maintain thatexperience is exactly the right place to look for an explanation of people?s beliefin indeterminist freedom. Nonetheless, such experience is different from the sorton which Bayne focuses. The sort of experience that causes indeterminist beliefsresults from the phenomenon of prospection,11 which is the mental simulation offuture possibilities for the purpose of guiding action. These simulations functionas effective competitors to perceptual experience, so that they are actually experi-ential.As I will show, the way in which prospection models choice and alternativepossibilities for action makes it easy for deliberating agents both to experienceand to believe that their freedom to do otherwise is indeterministic.5.4.1 Prospection is Forward-LookingLet us get a clearer picture of prospection. According to Martin Seligman, PeterRailton, Roy Baumeister, and Chandra Sripada (henceforth ?SRBS?), ?Prospec-tion . . . is guidance . . . by present, evaluative representations of possible futurestates. These representations can be understood minimally as ?If X, then Y? con-ditionals, and the process of prospection can be understood as the generation andevaluation of these conditionals? (2013: 119).12 Central to the phenomenon ofprospection is its focus on how agents ?navigate? into the future, as SRBS put it,rather than on how they are ?driven by the past.? Thus, prospection nicely cap-tures the forward-looking aspect of deliberation and agency. According to thisframework,11 The term ?prospection? is due to Gilbert (2006), Gilbert and Wilson (2007), and Buckner andCarroll (2007).12 As we shall see, a useful way to think about the hypotheticals generated in prospection is interms of causal modeling, as developed by Pearl (e.g., 2000) and Woodward (2003), among others.123. . . people and intelligent animals draw on experience to update abranching array of evaluative prospects that fan out before them.Action is then selected in light of their needs and goals. . . . Theseprospects can include not only possibilities that have occurred be-fore but also possibilities that have never occurred?and these newpossibilities often play a decisive role in the selection of action.(2013: 119)According to SRBS, in order to regulate an organism?s interactions with its am-bient environment, the brain must construct representational models of that envi-ronment. 13 The most efficient models will be of the form, ?if in circumstanceC and state S, then behavior B has outcome O with probability p? (2013: 124).Like the comparator models of agentive experience developed by Bayne, these?feedforward/feedback? models take the following form:(1) expectation ?? observation ?? discrepancy detection ?? discrepancy-reducing change in expectation ?? expectation ?? . . . 14As SRBS put it,Expectation is pivotal in schema (1) because it transforms experienceinto experimentation?continuously generating a ?test probe? so that13 SRBS argue that this is how a systems theorist would approach building an organism like us(Cf. Conant and Ashby 1970; Eykhoff 1994). They also note that this sort of approach is centralto the prescient learning theory of Miller, Galanter, and Pribam (1960), as well as to both adaptivecontrol theory (A?stro?m and Murray 2008; Carver and Scheier 1990) and Bayesian epistemology(Earman 1992).14 SRBS cite as empirical support for the phenomenon of prospection the near-optimal forag-ing behavior of various species (Dugatkin 2004). According to SRBS, ?Foraging mammals havesystems of neurons whose firing rates and sequences correlate with differences in: the identity ofstimuli, their intensity, the magnitude of specific positive vs. negative hedonic rewards or food val-ues, the relative value of a stimulus (e.g., deprivation vs. satiation), the absolute value of a stimulus(e.g., physiological need), the probability or expectation of a given outcome, the occurrence of abetter- or worse-than-expected predicted error, and the absolute risk and expected value of givenactions? (2013: 125; Cf. Craig 2009; Grabenhorst and Rolls 2011; Kringelbach and Berridge,2009; Preuschoff et al. 2006; Quartz 2009; Rolls et al. 2008; Schultz 2002; Singer et al. 2009;Tobler et al. 2006). SRBS also cite recent evidence for the neural implementation of evaluativeprospection during experiments with rats in T-mazes (Ainge et al. 2012; Derdikman and Moser2010; Gupta et al. 2010).124the next experience always involves an implicit question and suppliesan answer, which can then function as an error-reducing ?learningsignal.? (2013: 124)In this way, agents generate and use mental simulations of future alternative pos-sibilities, often by drawing on their past experience, for the purpose of enablingthem to navigate into the future and select appropriate actions. The expectationsgenerated by these simulations are tested against observed results in order to at-tenuate future expectations and actions. In this way, agents exercise ?teleological?control over their decisions and actions:15We call such accounts ?teleological,? meaning explanation by selec-tion in light of values and goals. . . . A good prospector must knowmore than the physical landscape?what is to be found where, withwhat probability?but also at what cost in effort and risk and withwhat possible gain. The prospecting organism must construct an eval-uative landscape of possible acts and outcomes. The organism thenacts through this evaluative representation, electing action in light oftheir prospects. And the success or failure of an act in living up toits prospect will lead not simply to satisfaction or frustration, but tomaintaining or revising the evaluative representation that will guidethe next act. To be sure, learning and memory necessarily reflect pastexperience. But at any given moment, an organism?s ability to im-prove its chances for survival and reproduction lies in the future, notthe past. So learning and memory, too, should be designed for action.(2013: 120)1615 SRBS allow that many sorts of everyday action do not require prospection, of either a con-scious or an unconscious variety. When actions ?can be successfully repeated without need forevaluation of alternatives? (2013: 125), they move from being under teleological control to beingunder habitual control.16 While SRBS maintain that motivation is often teleological in the way that they outline intheir ?feedforward/feedback? model, SRBS concede that drive-like, non-goal-directed motivationsometimes occurs: ?Addiction and salt deprivation, for example, can produce wanting without lik-125For SRBS, this forward-looking, desire-focused framework makes agency intelli-gible:The driven-by-the-past framework makes agency and choice diffi-cult to understand?individuals are responders rather than navigators.. . . If instead we see the individual as using past experience as infor-mation, as continually forming and evaluating a range of future possi-bilities, and as electing action from among these possibilities in lightof what she likes and values, then we can see that active agency isa natural part of the causal structure of action. Motivation for suchaction is not determined by fixed drives or past conditioning, but iselicited by the evaluative process itself through the normal workingof desire. (2013: 127)In generating and using prospections in this way, agents are ?drawn? toward thefuture by their evaluative representations, rather than ?driven? by the past.5.4.2 Prospection Can Be ExperiencedAccording to SRBS, prospection typically occurs unconsciously and is unavail-able to introspection, since it would be inefficient for agents consciously to keeptrack of all the prospections that they generate. Indeed, ?even when individu-als engage in conscious prospection, their intuitive sense of the value of alterna-tives may be underwritten by unconscious simulation? (2013: 126; Cf. Railton inpress). Yet prospection can become conscious. One reason SRBS think it may being . . . Certain physiological demands, natural or artificial, can produce ?driven? motivation evenin the face of profound distaste and resistance, but this is atypical indeed. Ordinary action, eveneating a meal when hungry, does not work this way?for hunger makes eating attractive, not dis-tastefully compulsive? (2013: 127). Yet, normally, motivation depends crucially on desire, not ondrive: ?Philosophers since Aristotle have emphasized that desire is not a blind urge but rather rep-resents its object as an ?apparent good? . . . or under a ?desirability characterization? . . . an attractiveprospect that can elicit motivation to seek it??liking? a representation gives rise to ?wanting? itsobject? (2013: 126). Cf. Aristotle ca. 330 BC/1999: 1113a15; Anscombe 1957: viii; Railton2002; Berridge 2004.126useful for prospection to become conscious is that this enables agents to engage inshared prospection. Even though conscious prospection may be less efficient thanunconscious prospection, conscious prospection might make for more effectiveprospection.Whatever the benefits of conscious prospection might be, SRBS hypothesizethat affect is central to how prospection becomes conscious, whenever it does.According to their story, when the process of prospection encounters ?incommen-surable dimensions and conflicting values and perspectives? (2013: 131), agents?engagement in explicit, conscious comparison of these elements is facilitated bythe brain?s ?common metric? of affect:Affect is the brain?s common currency for value, and conscious, sub-jective affect would permit the possible futures to be brought into theopen for explicit comparison with each other. . . . conscious subjectiveaffect attached to prospections would enable them to compete effec-tively with ongoing experience. (2013: 131)In other words, whenever agents have conflicting, incommensurable thoughtsabout what to do, their options feed into ?an experientially rich and detailedworkspace,? so that they can ?use their intelligence and imagination to best ef-fect.? In such cases, ?it can be best to act in awareness [emphasis added] of. . . conflicting thoughts? (2013: 131). As a result, forward-looking prospection isconsciously experienced by the agent. Nevertheless, this experience is not percep-tual, in contrast to Bayne?s model.5.4.3 Prospected Choices Are Free VariablesIt is useful to think about the hypotheticals generated in prospection as carry-ing causal information about what would happen under certain variations in thevalues of exogenous variables in a causal model. A causal model is a representa-tion that encodes hypothetical relationships between variables, where the variables127represent causal relata. According to causal modeling, when considering whethersomething is a cause, we ask, ?What if things had been different?? and by answer-ing this question we identify factors whose manipulation would produce changesin the outcome being explained. If this (cause) variable were altered in these ways,this (effect) variable would be altered in these ways. The main restriction on whatcounts as a variable is that it must represent particular events in such a way thatthey can be set to different values by interventions (Cf. Woodward 2003: 11?14).Thus, a variable that represents the event of my choosing dessert might take thevalue ?1? if I choose cake, or ?0? if I refrain from having dessert. An interventionin a model is an exogenous change to the value of a variable: we consider whathappens in the model by tweaking just this variable?s value. (By contrast, an en-dogenous change to the value of a variable occurs because of the values taken byother variables within the model.) In this way, interventions are ?surgical,? in thesense that the usual causes of a variable, or of a variable?s taking a given value,are are ignored or suspended. When we causally model a situation, we ?carve off?the situation from the rest of the world, and from its causal antecedents. We al-low the variables whose antecedent causes we have thereby ignored to vary freelyacross a range of values, where this range of values has the following pragmaticrestriction: none of the values should correspond to possibilities that we considertoo remote (Hitchcock 2001: 286). The hypotheticals specifying the relations thathold between the variables in any causal model are stated as structural equations,which are asymmetrical in the following way: the values of the variables on theleft hand side of the ?=? are determined by the equations on the right hand side.Assume, for instance, that you will choose dessert only if I choose dessertfirst or a friend joins us. Take the variable representing the event of my choosingdessert to be C, the variable representing the event of a friend?s joining us to beF , and the variable representing the event of your choosing dessert to be D. Ofthese, C and F are exogenous, while D is endogenous. Let C take the value ?1? ifI choose dessert; otherwise D takes the value ?2.? Likewise, let F take the value?1? if a friend joins us; otherwise, F takes the value ?2.? Finally, let D take the128value ?1? if [C=1 or F=1]; otherwise, ?2.? Now assume that I choose dessert andno one joins us. Thus, the structural equations that specify this extremely simplemodel are:C= 1 or 2F= 1 or 2D= 1 if [C=1 or F=1]; else 2Here, the first two equations state the possible values that the exogenous two vari-ables in the model, C and F , may take. The third equation encodes four hypothet-ical conditionals, two for each possible value of C and F . This equation says thatyou choose dessert only if I choose first or a friend joins us. These equations com-prise the model, which may also take a graphical form indicating the dependencyrelations obtaining between the variables by means of ?directed edges? connectingthe variables in the graph:XXXXXXXXXz:CFDJenann Ismael (2013) thinks that agents mentally construct models of this sortwhen deliberating about what to do.17 Agents carve off the event of their makinga choice from its causal antecedents, and treat it as an exogenous variable in amodel, in order to assess the downstream effects of this variable?s varying freelyacross a range of values. In this way, agents capture causal information relevant toaction-planning via the hypotheticals comprising the model. These are the same17 The models I describe here are far simpler than those we presumably construct when delib-erating about what to do. For one thing, my examples only involve variables with discreet values,whereas many models we construct in deliberating will have probabilistic values, as both SRBSand Ismael note. I have used simple models for the sake of clarity.129hypotheticals that SRBS claim our brain constructs in order to regulate interactionwith the environment.According to Ismael, a model, M, of a set of prospected options for any choiceis ?narrow-scope,? in the sense that it focuses only on a segment of the world?namely, the event of one?s making a choice and the prospected consequences ofchoosing in various different ways. In M, the event of one?s choice features asa free or exogenous variable, which we will call C. The variables that representthe prospected consequences are ?downstream? of C, in the sense that the valuesthey take are determined by the value C takes, together with the values of any othervariables internal to the model (holding all other inputs and background conditionsfixed). The downstream variables are endogenous and represent various possibleconsequences of one?s choice. The only way in which one can influence the valuesthese variables take is by determining the value of the exogenous variable, C.Imagine (once again) that C represents the event of my choosing dessert. As-sume that C can take only two possible values: C=1 if I choose cake; C=2 if Idecline dessert. Here, C is exogenous, and it is allowed to range across two val-ues. Downstream of C, there is an endogenous variable, S, which represents theevent of my falling asleep that night. Assume that the value S takes is determinedsolely by the value that C takes (holding all background conditions fixed). I knowthat if I choose cake it will (as always) keep me awake. If I decline dessert, I willfall asleep early. I have an important talk to give tomorrow morning, and I know Iwill perform at my best only if I go to sleep early. The event of my giving the talkis represented by L. Once C takes a value, the value L takes will be determinedsolely by S?s value, and thus L is endogenous. The way in which C differs from Sor L, in terms of how it functions within the model, is that C is allowed to rangeacross more than one value in a way that does not depend on the values takenby any ?upstream? variables. In this way, C is ?carved off? from its antecedentcauses.In M, the set of prospected options is narrow-scope, given that it focuses onlyon a segment of the world: one?s choice. This contrasts with ?wide-scope? mod-130els, which carve off larger chunks of the world. The narrow-scope model, M,might be embedded in a wider-scope model, W, in which the formerly exogenousvariable C functions as a newly endogenous variable, the values of which are de-termined by the values of the exogenous variables of the wider model, togetherwith the values of this model?s other upstream endogenous variables (whose val-ues would also ultimately depend on the values of the exogenous variables in thewide model).To see how this works, consider a ball-and-socket joint in a robotic arm usedin a factory assembly line. Recall that we select models on pragmatic grounds,according to the sort of question we are asking. If we are interested in how thisjoint works (perhaps we are engineers, designing a better joint), we will create avirtual separation of the joint from its environment and model it. The ?frame? weput around this isolable causal structure has, at its boundaries, exogenous vari-ables, which we allow to vary freely across a range of values. These variables?taking different combinations of values determines the values of the endogenousvariables. When we insert the joint back into its environment?the robotic arm?we model the entire arm as a causal structure. As a result, the formerly exogenousvariables that were allowed to vary freely become endogenous: the range of valuesthat they can take is constrained by the exogenous variables at the boundaries ofthe more encompassing model. We will select this wider-scope model if we wantto assess the efficacy of the robotic arm in an assembly line. Likewise, the event ofan agent?s choice might be embedded as an endogenous variable in a wider-scopemodel of social psychology.The central point is that prospection treats the event of choice as exogenous,which effectively requires us to ignore its causal antecedents. This is importantfor two reasons. First, as I explain in the next subsection, it makes it natural foragents to experience (and also to think of) their prospected alternatives as onesthat they can ?get to.? Second, it shows how radically forward-looking delibera-tion actually is. In generating prospections, agents not only pay less attention tothe causal antecedents of their decisions, and more attention to what SRBS call131a ?branching array of evaluative prospects that fan out before them? (2013: 119),but mentally they ignore the antecedent causes. Nevertheless, prospected choicescan additionally feature as endogenous variables in wider-scope models servingother purposes (like those of social psychology), where they feature as endoge-nous variables and thus have their causal antecedents ?reattached.?5.4.4 Indeterminist Experience and JudgmentAlthough SRBS weigh in briefly on the topic of free will, they do not addressthe sort of empirical evidence that I discussed earlier. This evidence indicatesthat people tend to experience possessing and to believe that they possess an in-determinist freedom to choose among alternatives. Even so, I claim that SRBS?saccount of prospection, together with a causal-modeling account of how the hypo-theticals generated in prospection should plausibly be modeled, can explain howpeople get to be indeterminists.The experience of being free to do otherwise may seem indeterminist for tworeasons. First, we experience an openness to the future that appears to requireindeterminism for the experience to be accurate. Second, we experience choiceas not being sufficiently caused by anything prior to it, and thus as not havingantecedent deterministic causes. On the basis of such experiences, we tend tohave indeterminist beliefs, which are subsequently revealed when we explicitlyentertain the notion of determinism.First, our prospected experience suggests an openness to the future that re-quires indeterminism to be accurate. Our experience when we engage in consciousprospection is that there is more than one way that we can extend the present intothe future, depending on our choice. This makes it easy to experience (and tothink of) our prospected alternatives as ones that we can ?get to? in the sense thatlibertarians are tying to pick out, and which they claim might be provided forby indeterminism. After all, if determinism is true then there is only one physi-cally possible extension of the present into the future. In that case, it would seemthat our experience is inaccurate, and thus apparently the experience is incom-132patibilist. We need not be aware of this. Our experience might be, so to speak,implicitly incompatibilist: it might have a content, P, that is in fact incompatiblewith determinism. As a result, were we to entertain the thesis of determinism, wewould explicitly judge our experience as incompatibilist.In the language of causal modeling, we can put the point this way. Wheneverwe prospect future possibilities for action in the course of deliberating about whatto do, the variable representing the event of our making a choice is exogenous,meaning that it is permitted to vary freely over a range of values. As a result,in our practical, deliberative experience we treat the event of making a choice asan exogenous variable in a narrow-scope model of deliberation. Yet, when weconsider the same event while assuming determinism, we treat it as an endoge-nous variable in a wide-scope model. Here the variable is permitted to take justone value.18 This creates an apparent conflict between treating one and the samevariable as exogenous and endogenous. This conflict makes the experience seemindeterminist.Second, we experience our choice as not determined by anything prior to it.This is in direct denial of the claims made by Nichols, Horgan, and Bayne thateven if we do not experience our choice as determined by our prior states, it isimplausible that we experience it as not determined by such states. Why are theywrong about this? Recall that prospection models choice as an exogenous vari-able, which is carved off from its antecedent causes and allowed to vary freelyacross a range of values. If we experience choice in this way, we experience it asnot having antecedent sufficient causes. A choice that is experienced as not hav-ing antecedent sufficient causes is, a fortiori, experienced as not having antecedentdeterministic causes.If that is right, then we are able to explain people?s belief in indeterministfreedom as being due to their agentive experience. People believe that the futureis open in a way that requires indeterminism, since that is what they experiencein prospection. People also believe that their choice is not sufficiently caused by18 Obviously, this value may be unknown, or even unknowable, prior to actually deciding.133anything prior to it, and thus implicitly it does not have antecedent determinis-tic causes. Of course, people?s belief that their choice is not sufficiently causedby anything prior to it might be defeated once they learn that it does have suchcauses?perhaps even deterministic ones. Still, people?s initial tendency will beto believe that it does not. Moreover, the experience itself will presumably remainunaltered.The result of all this is that, contra Nichols? claim that experience is too ?ane-mic? to be a plausible source of people?s belief in indeterminism, indeterministbeliefs might well come from experience. Yet, contra Bayne?s suggestion that in-determinist beliefs might come from perceptual agentive experience, such beliefsmore plausibly come from the experience of prospection.5.5 Belief in Indeterminist Freedom is Not JustifiedEven if people have genuinely indeterminist experiences of freedom, it certainlydoes not follow that these experiences are accurate, or even that a belief in inde-terminist freedom formed on the basis of such an experience is justified. First,even though it might easily seem that the future is open in a way that would re-quire indeterminism for the experience to be accurate, all that is actually going onis that the agent is considering what outcomes she can cause, depending on whatchoices she makes, which requires letting the event of her choice range acrossmore than one value. In this way, she can prospect the possible downstream ef-fects of her choosing in different ways. Second, even if the agent introspects thather choice lacks deterministic causes, it does not follow that it actually lacks suchcauses, since the causes might not be introspectible. Third, the conditionals gen-erated in prospection are subjective and relative, and so they cannot support theview that the future is indeterministically open. This is partly due to the epistemicnature of the possibilities that such conditionals capture, and epistemic possibilityis obviously compatible with determinism.Even so, that cannot be the whole story. We want to be able to say that agents134are free to choose among alternative possibilities?that is, that their freedom is amatter of fact, not something about their epistemic access to facts.In closing, I want to suggest how experiences of indeterminist freedom are infact compatible with determinism, in the sense of being accurate even if deter-minism is true. Further, I want to suggest that we are justified in believing that weare free to do otherwise, and that such freedom is consistent with determinism.This is because the claim that we are free to do otherwise should be assessed foraccuracy according to the narrow-scope model that we use in deliberating aboutwhat to do. In that context, it is true. When the claim is instead assessed accordingto the wide-scope model of determinism, it is false. Even so, there is no conflictbetween these models.Recall: we select models on pragmatic grounds. In a situation of choice, themodel that we naturally select treats the event of our making a choice as an ex-ogenous variable. This occurs automatically. The apparent threat from deter-minism comes from the thought that wide-scope models in some way overridenarrow-scope models. As Ismael puts it, ?To get the purported conflict with freewill going we are invited to see action in the context of wider embedding models. . . The worry about physical determinism is the most extreme version of this sortof model? (2013: 229?230).Ismael argues that the rules governing narrow-scope models do not derive fromthe rules governing the wider-scope models in which they are embedded. In factit is the other way round: we ?start with the basic building blocks with a greatdeal of freedom of movement and build up more complex systems by restrictingtheir relative motion? (2013: 227). That is to say, the rules governing the wholeare derived from the rules governing the parts, not vice versa. Further, the rulesgoverning the parts provide agents with richer causal information than the rulesgoverning the whole. This is because ?[v]ariables that were allowed to vary freelyin the original model are constrained by the values of variables in the embeddingmodel and so we just lose information about what would happen if they were135allowed to vary freely? (2013: 218). As a result, the wide-scope model of deter-minism does not override the narrow-scope model used in prospection.The correct model to adopt when deliberating about what to do is the narrow-scope model, since it is more useful relative to the pragmatic purposes of makinga choice. This model is compatible with wider-scope models. We develop widermodels by taking narrower models and adding constraints on the values that theirexogenous variables can take. Thus, we arrive at the maximally wide-scope modelof determinism by taking models of things like prospected choices and restrictingthe value that each variable in such a model can take to just one. However, as Is-mael puts it, ?there is no more conflict between these models than there is betweenthe view of a building from close-up and the view from a very great distance?(2013: 230).Ismael thinks that her naturalistic picture of causation and deliberation is?faithful to the experience of agency? (2013: 232). She is right about this inmore ways than she explores. Even if people have indeterminist experiences offreedom (a possibility that Ismael does not consider), these experiences are com-patible with determinism, in the sense of being accurate even if determinism istrue. This is because the accuracy of such an experience in a given context of de-liberation is to be judged according to the narrow-scope model appropriate to thatcontext. Moreover, if people form a belief in indeterminism on the basis of suchan experience, not only is this belief not justified, but the opposing belief may bejustified?namely, the belief that freedom is consistent with determinism. In thenext two subsections, I sketch in more detail how this proposal is meant to work.5.5.1 Radical CompatibilismConsider the feeling that an agent, S, might have about her openness to the futureand her freedom to do otherwise: ?It feels like I can A at t, or refrain from A-ingat t by B-ing instead.? In order for this experience to be accurate, the followingwould have to be true for S: (ATDO) ?I can A at t, or refrain from A-ing at t by B-136ing instead.?19 On the usual way of assessing the truth of such a claim, traditionalcompatibilists insist that we may consider possible worlds with differences in theevents antecedent to t or in the laws of nature. Incompatibilists insist that weshould only consider worlds with the same antecedent events and laws as theactual world.On the present way of looking at matters, however, the relevant hypotheticalsare not assessed according to the usual semantics for counterfactual condition-als.20 Rather, they are assessed according to the structural equations compris-ing the relevant causal models. In a deliberative context, (ATDO) may be truewhen assessed according to the model relevant to that context, i.e., a narrow-scopemodel in which the event of choice features as an exogenous variable. Yet when(ATDO) is assessed according to the wide-scope model of determinism, it is false.In such a model, there is just one value that the variable representing the event ofchoosing can take. When trying to decide whether to A, that is not a useful modelto adopt.Imagine that S judges (ATDO) as false when she explicitly assumes the truthof determinism, defined as the thesis that there is only one physically possiblefuture, given the past and the laws. So S judges her experience of being free todo otherwise as incompatibilist (i.e., as inaccurate if determinism is true). Thus,presumably her experience had a content that, when considered in relation to de-terminism, is judged as inconsistent with that thesis.Perhaps a traditional compatibilist will tell us that S is making a mistake injudging that (ATDO)?and so her experience of being free to do otherwise?isfalse under the assumption of determinism. Somehow, S is assessing the truth of(ATDO) by considering worlds with the same antecedent events and laws as theactual world, and this is not what she should do. Yet, if there is something inthe content of S?s experience of being free to do otherwise that encourages her to19 Here, ?ATDO? stands for ?Ability To Do Otherwise.?20 Problems have been suggested for possible-worlds analyses of counterfactuals (e.g., Barker2011, Fine 2012).137do this, then the traditional compatibilist?s claim will be idle: S?s experience isinaccurate, and that is that.On the present proposal, however, nothing about the content of S?s experi-ence is altered, yet that very experience is accurate, even assuming determinism.This is in stark contrast to the traditional compatibilist?s proposal. Let us call thethe present view radical compatibilism. By all means, says the radical compat-ibilist, (ATDO) is false when assessed according to the structural equations thatencode the hypotheticals comprising the maximally wide-scope model of physi-cal determinism. (Traditional compatibilists will not concede this.) Nevertheless,this verdict is consistent with judging (ATDO) as true when assessed accordingto the narrow-scope model of S?s situation of choice. So, the truth of (ATDO) isconsistent with determinism, even when there is something implicit in the contentof a person?s experience that encourages her to judge it as incompatibilist.If one?s belief about being free to do otherwise derives from this sort of expe-rience, then what amounts to an implicitly libertarian belief about freedom turnsout to be consistent with determinism. Only when one makes an explicit incom-patibility judgment is one?s belief false. What might appear to be a libertarianbelief is not incompatibilist, when properly understood.5.5.2 Contextualism RevisitedThe view described in the previous subsection is a contextualist proposal: in adeliberative context, the claim that the agent is free to do otherwise is true, since itis assessed according to the narrow-scope model of a situation of choice; but thisvery claim is false in a context in which it is assessed according to the maximallywide-scope model of physical determinism. Recall that in Chapter 3, I criticizedHorgan?s contextualist proposal. Thus, it may be thought that there is a primafacie tension between my endorsing contextualism here, yet my criticizing it there.However, this is not so.First, notice that the two forms of contextualism are distinct. On the sort ofcontextualist view endorsed by Kratzer (1977), which I sketched in Chapter 1,138?can? is treated as an existential quantifier over possible worlds restricted by an?in-view-of? clause. Thus, for instance, it may be true that ?In view of the rules ofchess, you cannot move the pawn three spaces ahead,? while it may be false that?In view of your physical constitution, you cannot move the pawn three spacesahead.? Likewise, as we saw in chapters 1 and 3, Lewis (1986) gives a treatmentof ?can? in terms of restricted possibility: to say that ?S can A? is to say that S?sA-ing is compossible with certain facts, where the relevant facts depend on thestringency with which ?can? is used. This is the sort of view that Horgan adoptswhen he maintains that claims about S?s freedom to do otherwise are governed byimplicit, contextually variable, semantic parameters. On this view, simply askingthe compatibility question about freedom and determinism drives the semantic pa-rameters far beyond their normal settings to a maximally strict setting, accordingto which any claim that S is free to do otherwise is false.21 However, the falsityof such a claim is established according to a restricted possibility metric, such asLewis?s or Kratzer?s, which uses a possible-worlds apparatus to assess the truthof the claim. This is not how the sort of contextualism I outlined in the previoussubsection works. Although we arrive at the wide-scope model of determinismby restricting the possibilities available in narrower-scope models, this restrictionis not implemented by means of quantification over possible worlds. Rather, it isimplemented by embedding narrow-scope causal models in wider-scope models,and thus limiting the values that the previously exogenous variables in the narrow-scope model can take, once they become endogenous variables in a wider-scopemodel. Moreover, the truth of the claim that S is free to do otherwise issues fromthe model (and the hypotheticals that it encodes), which is selected on pragmaticgrounds. The claim is not assessed according to any standard possible-worldssemantics.Further, I argued in Chapter 3 that it was a problem for Horgan?s contextualismthat worries about whether we are free to do otherwise arise even when contextual21 It is again worth noting that the traditional compatibilist view sketched in the previous sub-section will not grant this.139parameters are normal. If that is right, then Horgan?s claim that such worries ariseonly when we raise the parameters beyond their normal settings and explicitly askthe compatibility question is false. Even when we apply ordinary standards and donot explicitly invoke determinism, it seems we can generate worries about whetherpeople are free, even in the sense of being able to do otherwise. Thus, it is notclear whether any scorekeeping confusion occurs when we raise the parametersand explicitly ask the compatibility question about experiences of being free todo otherwise and determinism. We may simply be exhibiting our competence inapplying the notion of freedom in that context as well.In Chapter 3, I demonstrated this point by considering Frankfurt-cases, inwhich Black?a neurosurgeon?wants Jones to choose A. Black can interveneto control Jones?s brain processes should Jones be about to choose B. Yet Blackprefers not to intervene unless he has to. Instead, he waits to see how Jones willchoose on his own. Jones is unaware of Black?s presence. Frankfurt claims thatJones lacks alternative possibilities in the case. Nevertheless, if Jones chooses Aon his own, then apparently he freely makes his choice even though he has no al-ternative. Despite this, it seems that we can ask whether it is reasonable to expectthat Jones have done something else instead, given that the conditions in whichhe found himself ruled out any alternative. If we think it reasonable to expectthat Jones not have A-ed (where A-ing is killing Smith), then we have located aconflict in our thinking about how to apply the notion of freedom. If we considerthe case just by focusing on the intervener, without considering determinism, wemight want to grant?given that Black did not intervene?that Jones freely killedSmith. After all, he killed Smith on his own. Yet it is not clear whether it is rea-sonable to expect that Jones have done something else. Recall, he was unable todo otherwise. Did Jones freely kill Smith? Perhaps not. Have we illicitly raisedthe contextual parameters governing application of the relevant notion of free-dom? It is not clear that we have. Once we point out that determinism is meantto function in the same way as Black in the case, by blocking the availability ofJones?s alternative possibilities and thus blocking his ability to do otherwise, we140have generated a worry about Jones?s freedom according to the ordinary standardsgoverning application of this notion.Notice that this way of understanding Frankfurt-cases relies on the idea thatBlack functions in the same way as determinism in the case. As I pointed out inChapter 3, libertarians might deny this claim. They might insist that Black doesnot block alternatives, even though they will grant that determinism does blockalternatives.22 However, I noted in Chapter 3 that even if Black does not block allalternatives, he plausibly blocks the sorts of alternatives that would be required forthe ability to do otherwise. Conversely, I noted that a compatibilist about modalfreedom might insist that determinism does allow for a compatibilist ability to dootherwise, even if Black does not, and so Black does not function in the sameway as determinism. The problem is that most compatibilists actually grant thatdeterminism does block alternatives, and thus blocks the ability to do otherwise.What I want to show is that the causal modeling framework on which thecontextualist view that I dubbed ?radical compatibilism? depends enables us toshow that Jones has relevant alternatives in a Frankfurt-case, and therefore is freeto do otherwise. So: Black functions differently from determinism in the case. Asa result, when the claim that Jones is free to do otherwise is assessed according tothe hypotheticals comprising the relevant model (which is the relevant context ofevaluation for the case), it is true that Jones is free to do otherwise, even thoughwhen the same claim is assessed according to the maximally wide-scope model ofdeterminism, it is false. The upshot is that a Frankfurt-case does not generate anyworry about modal freedom when judged according to the relevant causal model,whereas on Horgan?s contextualist proposal, it does generate such a worry. Thus,22 The idea here is that if the prior sign by which Black knows that Jones is about to B (e.g., aneurological pattern) is a deterministic predictor of Jones?s A-ing, then it begs the question againstthe libertarian by assuming determinism in the case. Yet if it is an indeterministic predictor, thenBlack does not know for sure that Jones will A, and so Jones retains alternative possibilities in thecase.141the form of contextualism that I endorse is free from the criticism that I levelledat Horgan?s brand of contextualism in Chapter 3.How does this work? In a Frankfurt-case, we assume that Black has access toa prior sign (PS=2) that predicts whether Jones is about to decide to B (JD=2). Ifso, then Black intervenes (BL=1) to ensure that Jones decides to A (JD=1), suchthat Jones A-s (JO=1). Otherwise, PS=1, Jones decides to A on his own (JD=1),Black sits idly by (BL=2), and Jones A-s (JO=1). Thus, the model looks like this(variables, equations, graph):VariablesPS= 1 if prior sign occurs that Jones is about to decide to A; PS=2 if prior signoccurs that Jones is about to decide to BBL=1 if Black intervenes to ensure that Jones decides to A; BL=2 if Black sits idlyby without interveningJD=1 if if Jones decides to A; JD=2 if Jones decides to BJO=1 if Jones does A; JO=2 if Jones does BEquationsPS= 1 or 2BL= 1 if PS=2; else 2JD= 1 if (PS=1 or BL=1); else 2JO= 1 if JD=1; else 2Actual Case: PS=1; BL=2; JD=1; JO=1The corresponding graph is:142-@@@@@R-6PS JDBLJOIn the actual case, PS=1 (rather than PS=2) is an actual cause of JO=1 (rather thanJO=2).23 Yet if (counterfactually) PS=2, then it would be true that BL=1, JD=1,and JO=1.In the causal model, we intervene on the value of Jones?s decision (JD) in or-der to see what happens, while holding fixed other direct causes of JO. That is howwe test whether a variable is a difference-maker. In this case, there are no otherdirect causes of JO, so when we intervene on JD we not only ignore PS but alsoBL as a direct cause of JD. Ignoring or suspending both these variables as inputsto JD is necessary in order to assess whether JD=1 is an actual cause of JO=1, i.e.,in order to assess whether Jones?s decision considered on its own (while ignoringBlack) is the relevant causal variable in the situation. Note that here JD is allowedto range freely over two values, which represent Jones?s deciding either to A or toB, and so the causal model allows us a way of saying that Jones?s doing what hedoes rather than doing something else is the difference-maker in what happens ina Frankfurt-case. This is because when we focus on Jones?s decision, we screenoff or ignore the variable representing Black?s possible intervention. When we doso, notice that Jones is (in a relevant sense) free to do otherwise. What the model-ing does is illustrate our intuition that Jones is the difference-maker in the actualcase, even though there is another potential cause that would ensure the sameoutcome. More importantly for present purposes, since the variable representingJones choice (JD) ranges over two values, he is free to do otherwise.23 I am explicitly combining the interventionist model with a contrastive account of causation,as I suggest in Deery (forthcoming). The model I adopt is from co-authored work with EddyNahmias (unpublished).143This way of understanding a Frankfurt-case reveals that Black does not func-tion in the same way as determinism. This is because Black?s choice to interveneis a variable in the model rather than (as determinism would be) part of the back-ground conditions that are being held fixed. Thus, Frankfurt-cases do not generateglobal worries about freedom when judged according to the relevant causal model,whereas on the sort of contextualism that Horgan endorses, they do generate suchworries. As a result, the form of contextualism I endorse in the present chapter isfree from the sort of criticism that I made of Horgan?s contextualist proposal inChapter 3.5.6 ConclusionThe view that I have outlined accomplishes two things. First, it explains whysome people believe that being free to do otherwise is incompatible with deter-minism: they do so because of their experience as deliberating agents, due to thephenomenon of prospection. Second, it explains how these apparently incompati-bilist beliefs and experiences are compatibilist. The experience of having alterna-tive pathways into the future, and so of being free to do otherwise, is compatiblewith determinism. Furthermore, any implicitly incompatibilist belief that has itssource in such an experience is also compatibilist, when considered correctly.That is to say, even though the claim that one is free to do otherwise is correctlyjudged as false when assessed according to the maximally wide-scope model ofdeterminism, this is consistent in a relevant way with the judgment that the claimis true when assessed according to a model appropriate to a deliberative context.This type of contextualism is free from the problem that undermines Horgan?scontextualist proposal, namely, that his proposal generates global worries aboutfreedom even in when the context is ordinary.Of course, the view described in this chapter falls short of providing a com-prehensive compatibilist theory of the freedom to do otherwise. That remains asa task for another day. It remains open what properties and capacities such free-dom consists in. Nonetheless, on the view that I have outlined here, our beliefs144and experience regarding such freedom?whatever it ends up actually consistingin?are perfectly compatible with determinism.145Chapter 6ConclusionIn this chapter, I review the dissertation and show how its various strands hangtogether.6.1 What the Dissertation AchievesThis dissertation addresses the question whether experiences of freedom are con-sistent in a certain way with determinism. The thesis of determinism says that astatement of the non-relational facts of the world at a time, together with a state-ment of the laws of nature, entails all other facts about the world at other times.So, determinism says that, at any given time, there is exactly one physically pos-sible future. Throughout the dissertation, I take ?free will? to mean the freedom tochoose among alternative possibilities for action. In doing so, I rely on a distinc-tion between moral freedom and modal freedom. We have moral freedom whenwe control our actions in the strongest manner necessary for being morally re-sponsible for them, whereas we have modal freedom when we are free (or able,in the relevant sense) to perform a given action at a time, or to refrain from per-forming that action at that time. In the dissertation, I focus exclusively on modalfreedom (i.e., the ability to do otherwise), while leaving it entirely open whethersuch freedom is required for moral freedom.146I take modal freedom to have two aspects: (a) an alternative possibility aspect,and (b) an ability aspect. The alternative possibility aspect is necessary, but notsufficient, for modal freedom. The ability aspect involves the power of an agentto originate changes in the environment, or to ?make things happen.? This is anintentional, not merely a simple ability. You have a simple ability to roll a 6 ona die as long as (for instance) your arm is working sufficiently well to roll a die.Yet a simple ability is not an intentional ability: you cannot ever intentionally rolla 6. I am interested in intentional ability. An account of intentional ability willinvolve simple ability, plus a story about intentional action, where for an eventto be an intentional action requires that it be non-deviantly caused by certain ofone?s mental states (such as belief and desires). So, modal freedom is, first of all,an intentional ability: it is the ability to perform intentional actions.Modal freedom is also a specific, not just a general ability. I am generallyable to do many things that I cannot do right now (even things that normally Ican intentionally do), because I lack the opportunity to do them. For example, Icannot play the guitar right now, even though I am well able to play the guitar.This is because I do not have a guitar to hand now.In Chapter 1, I characterize specific ability as follows. An agent has a specificability to perform a given action at a time only if (i) she has a general ability toperform it at that time; (ii) she has an opportunity to perform it (which, minimally,requires that she have a certain possibility for action available at the time); and (iii)holding fixed the past and the laws (including the agent?s motivations regardingher opportunities as they are), the agent can exercise her general ability to performthat action at that time. If an agent has a specific ability in this sense regardingtwo distinct actions, A and B, at a time, then she has the specific ability to dootherwise. That is modal freedom: it is the specific ability to intentionally performan action at a time, or to refrain from performing that action at that time.My focus is on the compatibility question about modal freedom: Is such free-dom compatible with determinism? To begin with, though, I focus on the ?nat-ural? compatibility question, which is a descriptive question about people?s pre-147theoretic belief-tendencies about their own and others? agency. Here, I rely on adistinction between three sorts of question about free will: descriptive, substan-tive, and prescriptive. The descriptive question is the question that I have justoutlined: it asks about people?s pre-theoretic belief-tendencies. The substantivequestion asks whether people are in fact agents of the sort that they tend to believethey are, in a given respect. Finally, the prescriptive question asks how best to the-orize about freedom, given how we have answered the descriptive or substantivequestions.There are two ways to address the descriptive question: (a) ask people fortheir intuitions or beliefs about modal freedom (a good deal of experimental phi-losophy does exactly this), or (b) ask people about their experiences of modalfreedom. I focus on the second method, concerning people?s experiences of free-dom. One advantage of this approach is that it is plausible that experiences offreedom partly drive people?s beliefs (or intuitions) about freedom. However, fo-cusing on experiences requires slightly reframing the compatibility question, andredefining compatibilism and incompatibilism. Those whom I call experience-compatibilists about modal freedom think that experiences of such freedom aresometimes veridical (or accurate) even if determinism is true, whereas experience-incompatibilists think that such experiences are non-veridical (or inaccurate) ifdeterminism is true.In the dissertation, I present new evidence supporting experience-incompatibilism on the descriptive question: people?s experiences (and thus be-liefs) tend to be incompatibilist about modal freedom. By contrast, I leave thesubstantive aside as a project for another day. Finally, in chapters 3?5, I defendcompatibilism on the prescriptive question: we should be compatibilists, I argue,about experiences of (and beliefs about) modal freedom.1486.2 New Evidence on the Descriptive QuestionIn Chapter 2, I report the results of studies I conducted with Matt Bedke and ShaunNichols (Deery et al. 2013). These results support experience-incompatibilism onthe descriptive question about experiences of modal freedom.In the experiments that my colleagues and I ran, we found that participantsdescribed their experience of modal freedom as inconsistent with determinismwhether the decision was (i) present-focused or retrospective, (ii) imagined oractual, or (iii) morally salient or morally neutral. The only case in which par-ticipants did not report incompatibilist experience was when the question wasexplicitly about whether experiencing ignorance of the future is compatible withdeterminism?i.e., when the question was about epistemic openness, which is ob-viously compatible with determinism.These results support experience-incompatibilism on the descriptive question.They also show that existing compatibilist accounts of freedom are inadequate topeople?s reported experience. So, if we want to be compatibilists, it turns out thatwe have got work to do on the prescriptive question.In chapters 3?5, I turn to the prescriptive question. I begin by assuming (atleast for the sake of argument) that there are experiences of modal freedom withlibertarian content. First, I defend this hypothetical claim against an error theoryfor incompatibilist reports about the experience of freedom, and then I argue thateven assuming that there are such experiences, they do not threaten experience-compatibilism. Finally, in Chapter 5, I develop an etiological story about thesource of libertarian experiences.6.3 Against a Compatibilist Error TheoryChapter 2 describes results that support experience-incompatibilism on the de-scriptive question. But do they support incompatibilism? In Chapter 3, I consideran error theory for incompatibilist judgments about experiences of freedom?aview that says such judgments are somehow mistaken.149For instance, Terry Horgan (e.g., 2012; 2011) agrees that people often thinkthat their experience is incompatibilist. Yet he argues that even when people judgetheir experience as incompatibilist, actually it is compatibilist: people misinterprettheir experience. By spelling out how this happens, Horgan develops an errortheory for incompatibilist judgments about experience. If Horgan?s error theory isright, then it threatens to undermine the results of the experiments that I report inChapter 2.First, Horgan thinks that when we pay attention to certain aspects of our cur-rent experience, there is no significant appearance-reality gap. However, when wemake a judgment about what we are paying attention to, it turns out that we makemistakes, especially when we try to make sophisticated judgments like judgingwhether our agentive experience is compatible with a general hypothesis aboutthe world, such as the thesis of determinism.Even so, Horgan concedes that many people (including many philosophers)actually judge their experience of freedom as incompatibilist. Horgan must ex-plain why people make such judgments. He does so in two ways. First, he sug-gests a way in which people introspectively confabulate. Second, he tells a con-textualist story about the application conditions of the notion of freedom, whichalso applies to judgments about experiences of freedom.In this review, I focus on the first explanation. Horgan suggests that if wethink that we can tell by introspection that our experience is incompatibilist, weare mistaken. It is one thing, Horgan thinks, for us to claim to know the followingby introspection: (A) My experience does not present my behavior as determinedby my prior states. But it is another thing entirely to claim to know the followingjust by introspecting on phenomenology: (B) My experience presents my behav-ior as not determined by my prior states. The first claim is unproblematic. Yet wecannot ascertain the truth of the second claim by introspection alone. When wemake the second claim, i.e., when we judge that our experience presents our be-havior as not determined by our prior states?and thereby judge our experience as150incompatibilist?either we are mistakenly inferring this claim from the innocuousfirst claim, or else we are conflating the two claims.Even if we grant Horgan?s hypothesis about introspective confabulation, Imaintain that more needs to be said about how such a mistake occurs. Noticethat people do not make this sort of mistake when it comes to headaches. Wedo not mistakenly infer from the claim that our experience does not present ourheadache as determined the further claim that we experience our headache as notdetermined. Thus, Horgan needs to say how the experience of deliberation is rel-evantly different from that of headaches. This requirement is a theoretical cost ofHorgan?s view, which the alternative position that I develop in chapters 3 and 4does not incur. Indeed, while Horgan?s view is attractive in many respects, itincurs a number of theoretical costs that do not need to be incurred by compati-bilists.A central theoretical cost of Horgan?s view is that it requires maintaining thatthe content of experiences of freedom is exhaustively compatibilist. This claimenters compatibilists into intractable disputes with libertarians about the contentof experiences of freedom. By contrast, it is a theoretical advantage of my viewthat it does not claim that experiences of freedom have exhaustively compatibilistcontent. This enables me to avoid entering into intractable disputes with liber-tarians. On my view, even if we take people?s incompatibilist reports about theirexperience of freedom at face value, and thereby grant both that introspection re-liably latches onto the content of such experience and that such content is richenough to be incompatibilist, there is still an important respect in which the ex-perience is compatibilist: it is veridical even if determinism is true. I sketch thisview at the end of Chapter 3 and develop it at length in Chapter 4.6.4 An Alternative Compatibilist ProposalOn my view, experiences of freedom have two sorts of phenomenal content. Ifone type of content is libertarian, yet an experience of freedom also has anothersort of content that is compatibilist, the experience might be veridical if the latter151content is satisfied. David Chalmers (2006) makes a similar move in connectionwith the phenomenal content of color experience.Chalmers thinks that the view about phenomenal content that is most adequateto the phenomenology of visual color experiences is primitivism. According to theprimitivist view, phenomenology presents colors to us as simple intrinsic proper-ties of objects, spread out over their surfaces. As a result, experiences of colorhave contents that attribute primitive properties. A problem with this view is thatthere is good reason to think that the relevant primitive properties are not instan-tiated in our world. Thus, according to primitivism, none of our experiences ofcolor is veridical.Chalmers argues that there is also another type of phenomenal content thatmakes color experiences accurate, at least in the right kinds of cases. This isordinary content, which has its own veridicality condition: that the relevant objecthave whatever property (or set of properties) normally causes phenomenally red(etc.) experiences. Here, the phenomenal content is a ?condition on extension.?However, there is also a problem for this view: ordinary content is not adequate tothe phenomenology of color experience, since it does not reflect the phenomenalcharacter of such experience.Chalmers?s idea is to combine these views in a way that captures the truth-conditional virtues of ordinary content and the phenomenological virtues of prim-itivist content. I do something similar for experiences of freedom.I begin by assuming (at least for argument?s sake) that the libertarian con-tent of an experience of modal freedom is of one?s feeling a certain unconditionalopenness to the future. The future feels open in a way that would require indeter-minism for the feeling to be accurate. It feels ?as if? one is free to decide to Aor, in an unconditional (i.e., libertarian) sense, to refrain from A-ing. Of course, itmay seem that an experience with such libertarian phenomenal content is veridi-cal only if libertarianism is true. However, my view shows that experiences offreedom with libertarian content might still be consistent with determinism, in the152sense of being veridical even if determinism is true (and thus, even if libertarian-ism is false).My proposal (by analogy with Chalmers?s move) is that there is a second sortof phenomenal content to experiences of freedom that is compatibilist. This con-tent is a condition that a property must satisfy in order to be the property thatis attributed by the experience. The property attributed by the experience is, ofcourse, the freedom to do otherwise. What condition might work as the secondphenomenal content for such an experience?For color, the second phenomenal content Chalmers?s proposes is the follow-ing condition: whatever property (or set of properties) ordinarily causes phenom-enally red (etc.) experiences. In the agentive case, I think we cannot say that thesecond phenomenal content is whatever property (or set of properties) ordinarilycauses experiences of being free to do otherwise, since presumably no one thinksthat the meaning of ?being free to do otherwise? is being such that it causes anexperience of being free to do otherwise. More plausibly, the second content isthe following condition: That there is instantiated whatever relevant property (orset of properties) is ordinarily instantiated when one experiences being free to dootherwise.There is good reason to think that this condition picks out a genuine secondphenomenal content for experiences of freedom. In the dissertation, I explainthe plausibility of this content by analogy with a natural-kind view about modalfreedom. This view says that ?freedom? (or ?free act?) is a natural-kind term thatrefers to whatever relevant processes are at work in choices or decisions that weordinarily call ?free,? or that feel free to us, whatever those processes turn out tobe (as long as they constitute a relevant kind).I also explain how libertarian and compatibilist phenomenal content are re-lated in the following ways. For an experience of being free to do otherwise tobe perfectly veridical, we would have to live in a world where libertarianism istrue. The best that we can do if determinism is true, by contrast, is to have certainproperties ?match? the libertarian properties that are attributed by the presenta-153tional content, by playing the role that libertarian properties would play in such aworld. So, the libertarian phenomenal content sets an ideal standard for veridical-ity, and the second, compatibilist content is a condition that relates us to whateverproperties come closest to meeting this ideal standard. Once the second content issatisfied, the experience is imperfectly veridical, even if determinism is true.Thus, despite granting that our experience of freedom has libertarian content,this experience might be veridical under the assumption of determinism.6.5 The Source of Libertarian ContentIn Chapter 5, I address the question where libertarian content might come from.Libertarians often think that people?s experiences of deliberating and choosinghave libertarian content, and that such experiences lead people to believe thattheir choice is indeterministic. The studies that I report in Chapter 2 indicate thatlibertarians may be right about this. However, Shaun Nichols (2012) argues thatbelief in libertarian freedom cannot have its source in experience. Nichols arguesthat raw experience is too ?anemic? to have content that requires indeterminismto be true. Nichols instead proposes an alternative etiology for belief in libertarianfreedom.I argue that a major shortcoming of Nichols? view is its backward-lookingfocus: it relies on the idea that people?s indeterminist beliefs derive from intro-spection on the causes of decisions. Yet, deliberation and choice are cruciallyforward-looking. When agents look to the future in the course of their deliber-ating about what to do, two aspects of agency become especially salient, whichNichols does not adequately address: (i) the experience of having alternative pos-sibilities for action, and (ii) the experience of being free to choose between suchalternatives. These are some of the aspects of experience that libertarians mostoften cite as indeterministic.By contrast, Tim Bayne (2011) embraces a forward-looking model of agen-tive experience. Bayne thinks that agentive experience enables us to distinguishour self-generated actions from both involuntary bodily movements and the exter-154nally caused movements of objects in the environment. Yet Bayne?s view cannotexplain people?s experience of indeterminist freedom, and so it cannot explainhow people?s belief in libertarian freedom derives from agentive experience. Thisis because Bayne?s view focuses on perceptual experience, and it is difficult to seehow perceptual experience could have as content that one is free to do otherwise.That would require a comparison of two or more distinct representations?thealternative possibilities themselves?in the mind, and that is not a perceptual op-eration.I explain people?s experience of libertarian freedom, and thereby the sourceof people?s belief that they possess such freedom, by appeal to the phenomenonof prospection, which is the mental simulation of future possibilities for the pur-pose of guiding action (Cf. Seligman et al. 2013). Crucially, prospection canbe experienced, and because of the way in which the hypotheticals generated inprospection should be modeled in an interventionist framework, it turns out to beeasy for deliberating agents both to experience their choice as indeterministic, andto believe that their choice is indeterministic (and thus libertarian), even thoughthe modeling itself is consistent with determinism. Prospection treats the eventof one?s making a choice as an exogenous variable in a model of prospected out-comes, and (as I outline in Chapter 5) this requires ignoring its causal antecedents.This way of modeling choices is important for two reasons.First, it makes it easy for an agent to experience (and to think of) herprospected alternatives as ones that she can ?get to? in the sense that libertarianswant to pick out, and which they claim might be provided for by indeterminism.That is, it might easily seem that the future is open in a way that would requireindeterminism for the experience to be accurate. Even so, all that is going on isthat the agent is considering what outcomes she can cause, depending on whichchoices she makes, which requires letting the event of her choice range acrossmore than one value in order to prospect the downstream effects of her choosingin different ways. Modeling her own choice in this way is no different in kind155from modeling how contingent events, such as different paths of an approachinghurricane, would cause various outcomes, such as damage to different cities.Second, prospection models the event of choice as exogenous, and thus ascarved off from its antecedent causes and allowed to vary freely across a range ofvalues. If an agent experiences deliberation in this way, then she very likely expe-riences her choice as not having antecedent sufficient causes. And a choice that isexperienced as not having antecedent sufficient causes is, a fortiori, experiencedas not having antecedent deterministic causes.Even so, belief in libertarian freedom is not justified by such experiences. Forone thing, the hypotheticals generated in prospection are subjective and relative,and so they cannot support the view that the future is indeterministically open.This is partly due to the epistemic nature of the possibilities that such hypotheti-cals capture, and epistemic possibility is compatible with determinism. Moreover,even if one introspects that one?s choice seems to lack deterministic causes, it doesnot follow from this that it actually lacks such causes, since many of the causesmight not be introspectible. Finally, modeling choice in this way illuminates howthe process of deliberation itself is a causal difference-maker?prospection, per-haps experienced as indeterministic, is an important causal contributor to whathappens in the world, even if the world turns out to be deterministic.In this way, my view provides a deflationary explanation of a major motiva-tion for libertarianism?namely, the experience of deliberation and choice?whilenevertheless granting that it is natural for agents to form the belief that their choiceis indeterministic. At bottom, libertarian beliefs derive from agents? experiencesof navigating into the future, rather than from their introspection on the causes ofchoices. Thus, people?s indeterminist beliefs have their source in their practicalexperience of deliberating and choosing.In Chapter 5, I also show how the sort of contextualism about the freedom todo otherwise that follows from my application of causal modeling to questionsabout modal freedom is distinct from the sort of contextualist proposal endorsedby Horgan (which I criticize in Chapter 3). I also show how my version of con-156textualism is immune to the specific criticism that I direct at Horgan?s view. Inshort, consideration of Frankfurt-cases shows how worries about whether agentsare free arise for Horgan?s view even when contextual parameters are normal,whereas they do not arise for my view.6.6 ConclusionIn the dissertation, I first present new empirical findings that support experience-incompatibilism on the descriptive question about modal freedom. Next, I defendthese findings against an error theory for incompatibilist judgments about experi-ences of freedom developed by Terry Horgan. I then propose a compatibilist storyabout the veridicality of experiences of modal freedom that have libertarian con-tent, and thus I defend experience-compatibilism on the prescriptive question?which asks how we ought to think about experiences of modal freedom. Finally,I develop a deflationary etiological story about the source of libertarian agentiveexperiences and beliefs.What is at stake in all this is our self-image as agents who are free to navigateamong alternative pathways into the future. The compatibilist view that I developin the dissertation preserves this self-image, even assuming that determinism istrue.157Bibliography[1] Ainge, J.A., M. Tamosiunaite, F. Worgotter, and P. A. Dudchencko. 2012.?Hippocampal Place Cells Encode Intended Destination, and Not aDiscriminative Stimulus, in a Conditional T-maze Task,? Hippocampus,22: 534?543.[2] Anscombe, E. 1981/1971. ?Causality and Determination,? in Metaphysicsand the Philosophy of Mind: Collected Philosophical Papers of G.E.M.Anscombe. Minneapolis: University of Minnesota Press. 133?47.[3] Anscombe, G. E. M. 1957. Intention. Oxford: Blackwell.[4] Aristotle. ca. 330 BC/1999. Nicomachean Ethics, trans. T. Irwin.Indianapolis: Hackett.[5] Armstrong, D. 1983. What Is a Law of Nature? Cambridge: CambridgeUniversity Press.[6] Armstrong, D. 1978. A Theory of Universals. Cambridge: CambridgeUniversity Press.[7] A?stro?m, K. J. and R. M.Murray. 2008. Feedback Systems: An Introductionfor Scientists and Engineers. Princeton: Princeton University Press.[8] Austin, J. L. 1956. ?Ifs and Cans,? in Philosophical Papers, Third Edition.Oxford: Clarendon Press 1979: 205?32. Originally published inProceedings of the British Academy, 42.[9] Ayer, A. J. 1954. ?Freedom and Necessity,? in Philosophical Essays.London and Basingstoke: Macmillan. 271?284.158[10] Balaguer, M. 2010. Free Will as an Open Scientific Problem. Cambridge,MA: MIT Press.[11] Balaguer, M. 2004. ?A Coherent, Naturalistic, and Plausible Formulationof Libertarian Free Will,? Nou?s, 38(3): 379?406.[12] Barker, S. 2011. ?Can Counterfactuals Really Be about Possible Worlds??Nou?s, 45(3): 557?576.[13] Bayne, T. 2011. ?The Sense of Agency,? in F. Macpherson (ed.), TheSenses. Oxford: Oxford University Press. 355?74.[14] Bayne, T. 2009. ?Perception and the Reach of Phenomenal Content,?Philosophical Quarterly, 59(236): 385?404.[15] Bayne, T. 2008. ?The Phenomenology of Agency,? Philosophy Compass,3(1): 182?202.[16] Bayne, T. and E. Pacherie. 2007. ?Narrators and Comparators: TheArchitecture of Agentive Self-Awareness,? Synthese, 159(3): 475?491.[17] Bealer, G. 1998. ?Intuition and the Autonomy of Philosophy,? inRethinking Intuition: The Psychology of Intuition and Its Role inPhilosophical Inquiry, M. R. Depaul and W. Ramsey (eds.), Lanham,Maryland: Rowman and Littlefield Publishers, Inc. 201?239.[18] Bedke, M. S. 2008. ?Ethical Intuitions: What They Are, What They AreNot, and How They Justify,? American Philosophical Quarterly, 45(3):253?270.[19] Bennett, J. 1984. ?Counterfactuals and Temporal Direction,? ThePhilosophical Review, 93: 57?91.[20] Bernstein, M. 2002. ?Fatalism,? in R. Kane (ed.), The Oxford Handbook ofFree Will. Oxford: Oxford University Press. 65?81.[21] Berofsky, B. 2002. ?Ifs, Cans, and Free Will: The Issues,? in R. Kane(ed.), The Oxford Handbook of Free Will. New York: Oxford UniversityPress. 181?201.[22] Berridge, K. 2004. ?Motivation Concepts in Behavioral Neuroscience,?Physiology and Behavior, 81: 179?209.159[23] Bird, A. 2005. ?The Dispositionalist Conception of Laws,? Foundations ofScience, 10(4): 353?370.[24] Blakemore, S.-J. and C. Frith. 2003. ?Self-Awareness and Action,?Current Opinion in Neurobiology, Special Issue 13(2): 219?224.[25] Brogaard, B. 2013. ?The Phenomenal Use of ?Look?,? PhilosophyCompass.[26] Buckner, R. and D. Carroll. 2007. ?Self-Projection and the Brain,? Trendsin Cognitive Science, 11: 49?57.[27] Buhrmester, M.D., T. Kwang, and S.D. Gosling. 2011. ?Amazon?sMechanical Turk: A New Source of Inexpensive, Yet High-Quality, Data?Perspectives on Psychological Science, 6: 3-5.[28] Campbell, C. A. 1951. ?Is ?Freewill? a Pseudo-Problem?? Mind, 60(240):441?65.[29] Campbell, John. 1997. ?The Simple View of Colour,? in A. Byrne andD. Hilbert (eds.), Readings on Color. MIT Press. 257?268.[30] Campbell, Joseph. 2005. ?Compatibilist Alternatives,? Canadian Journalof Philosophy, 35:387?406.[31] Campbell, Joseph., M. O?Rourke, and D. Shier. 2004. ?Freedom andDeterminism: A Framework,? in J. Campbell, M. O?Rourke, and D. Shier(eds.), Freedom and Determinism. Cambridge, MA: MIT Press, BradfordBooks. 1?18.[32] Carroll, J. W. 2011. ?Laws of Nature,? in E. N. Zalta (ed.), The StanfordEncyclopedia of Philosophy.[33] Carver, C. S. and M. F Scheier. 1990. ?Origins and Functions of Positiveand Negative Affect: A Control-Process View,? Psychological Review, 97:19?35.[34] Chalmers, D. 2006. ?Perception and the Fall from Eden,? in T. Gendlerand J. Hawthorne (eds.), Perceptual Experience. Oxford: OxfordUniversity Press. 49?125.160[35] Chalmers, D. 2004. ?The Representational Character of Experience,? inB. Leiter (ed.), The Future for Philosophy. Oxford: Oxford UniversityPress. 153?181.[36] Chalmers. D. 2002. ?On Sense and Intension,? Philosophical Perspectives,16. Atascadero: Ridgeview. 135?182.[37] Chisholm, R. 1957. Perceiving: A Philosophical Study. Ithaca: CornellUniversity Press.[38] Clarke, R. 2009. ?Dispositions, Abilities to Act, and Free Will: The NewDispositionalism,? Mind, 118: 323?51.[39] Conant, R. C. and W. R. Ashby. 1970. ?Every Good Regulator of a SystemMust Be a Model of That System,? International Journal of SystemsScience, 1: 89?97.[40] Craig, A. D. 2009. ?How Do You Feel?Now? The Anterior Insula andHuman Awareness,? Nature Reviews Neuroscience, 10: 59?70.[41] Danto, A. C. 1959. ?The Paradigm Case Argument and the Free-WillProblem,? Ethics, 69(2): 120?124.[42] Deery, O. Forthcoming. ?Absences and Late Preemption,? Theoria.[43] Deery, O., M. Bedke, and S. Nichols. 2013. ?Phenomenal Abilities:Incompatibilism and the Experience of Agency,? in David Shoemaker(ed.), Oxford Studies in Agency and Responsibility. New York: OxfordUniversity Press. 126?150.[44] Deery, O., T. Davis, and J. Carey. In preparation. ?The Free-Will IntuitionsScale and the Question of Natural Compatibilism.?[45] Delk, J. L. and S. Fillenbaum. 1965. ?Differences in Perceived Color as aFunction of Characteristic Color,? The American Journal of Psychology,78(2): 290?93.[46] Dennett, D. 1978. ?On Giving Libertarians What They Say They Want,? inBrainstorms. Cambridge, MA: MIT Press.[47] Derdikman, D. and E. I. Moser. 2010. ?A Manifold of Spatial Maps in theBrain,? Trends in Cognitive Sciences, 14: 561?569.161[48] Double, R. 1997. ?Misdirection in the Free Will Problem,? AmericanPhilosophical Quarterly, 34(3): 357?366.[49] Double, R. 1996. Metaphilosophy and Free Will. New York: OxfordUniversity Press.[50] Dretske, F. 1977. ?Laws of Nature,? Philosophy of Science, 44: 248?268.[51] Dugatkin, L. A. 2004. Principles of Animal Behavior. New York: W. W.Norton.[52] Earman, J. 2004. ?Determinism: What We Have Learned and What WeStill Don?t Know,? in J. Campbell, M. O?Rourke, and D. Shier (eds.),Freedom and Determinism. Cambridge, MA: MIT Press, Bradford Books.21?46.[53] Earman, J. 1992. Bayes or Bust? Cambridge, MA: MIT Press.[54] Earman, J. 1986. A Primer on Determinism. Dordrecht: D. ReidelPublishing Company.[55] Earman, J. 1984. ?Laws of Nature: The Empiricist Challenge,? in R.Bogdan (ed.), D. M. Armstrong. Dordrecht: D. Reidel PublishingCompany.[56] Eykhoff, P. 1994. ?Every Good Regulator of a System Must Be a Model ofThat System,? Modeling, Identification, and Control, 15: 135?139.[57] Fara, M. 2008. ?Masked Abilities and Compatibilism,? Mind,117(468): 843?865.[58] Fine, K. 2012. ?A Difficulty for the Possible Worlds Analysis ofCounterfactuals,? Synthese, 189(1): 29?57.[59] Fischer, J. M. 2013. ?My Compatibilism,? in P. Russell and O. Deery(eds.). 296?317.[60] Fischer, J. M. 2008. ?Freedom, Foreknowledge, and Frankfurt: A Reply toVihvelin,? Canadian Journal of Philosophy, 38(3): 327?42.[61] Fischer, J. M. 2007. ?Compatibilism,? in J. M. Fischer, R. Kane,D. Pereboom and M. Vargas, Four Views on Free Will. Oxford: BlackwellPublishing. 44?84.162[62] Fischer, J. M. 1994. The Metaphysics of Free Will. Oxford: BlackwellPublishing.[63] Fischer, J. M. and M. Ravizza. 1998. Responsibility and Control: ATheory of Moral Responsibility. Cambridge: Cambridge University Press.[64] Flew, A. 1959. ?Divine Omniscience and Human Freedom,? in A. Flew,and A. McIntyre (eds.), New Essays in Philosophical Theology. London:New York: Macmillan.[65] Frankfurt, H. 1969. ?Alternate Possibilities and Moral Responsibility,? TheJournal of Philosophy, 66(23): 829?39.[66] Gilbert, D. 2006. Stumbling on Happiness. New York: Knopf.[67] Gilbert, D. and T. Wilson. 2007. ?Miswanting: Some Problems in theForecasting of Future Affective States,? in S. Lichtenstein and P. Slovic(eds.), The Construction of Preferences. Cambridge: CambridgeUniversity Press. 550?563.[68] Gill, M. 1980. ?Aristotle?s Theory of Causal Action in Physics III, 31,?Phronesis 25(2): 129?147.[69] Ginet, C. 1997. ?Freedom, Responsibility, and Agency,? in The Journal ofEthics, 1(1): 85?98.[70] Grabenhorst, F. and E. T. Rolls. 2011. ?Value, Pleasure, and Choice in theVentral Prefrontal Cortex,? Trends in Cognitive Sciences, 15: 56?67.[71] Gru?nbaum, A. 1952. ?Causality and the Science of Human Behavior,?American Scientist, 40(4): 665?676.[72] Gupta, A. S., M. A. A. van der Meer, D. S. Touretzky, and A. D. Redish.2010. ?Hippocampal Replay is Not a Simple Function of Experience,?Neuron, 65: 695?705.[73] Hanson, R. N. 1958. Patterns of Discovery. Cambridge: CambridgeUniversity Press.[74] Heller, M. 1996. ?The Mad Scientist Meets the Robot Cats:Compatibilism, Kinds, and Counterexamples,? Philosophy andPhenomenological Research, 56(2): 333-37.163[75] Hitchcock, C. R. 2001. ?The Intransitivity of Causation Revealed inEquations and Graphs,? Journal of Philosophy, 98: 273?299.[76] Hobart, R. E. 1934. ?Free Will as Involving Determination andInconceivable without It,? Mind, 43(169): 1?27.[77] Hobbs, J. Unpublished. What Can Ought Imply? Ph.D. thesis, CornellUniversity.[78] Hoefer, C. 2010. ?Causal Determinism,? in E. N. Zalta (ed.), StanfordEncyclopedia of Philosophy.[79] Holton, R. 2010. ?Disentangling the Will,? in R. Baumeister, A. Mele andK. Vohs (eds.), Free will and Consciousness: How Might They Work?New York: Oxford University Press. 82?100.[80] Holton, R. 2006. ?The Act of Choice,? Philosophers? Imprint, 6: 1?15.[81] Honderich, T. 1993. How Free are You? The Determinism Problem.Oxford: Oxford University Press.[82] Honore?, A. M. 1964. ?Can and Can?t,? Mind, 73: 463?479.[83] Horgan, T. Forthcoming. ?Causal Compatibilism about AgentivePhenomenology,? in M. Sabates, D. Sosa, and T. Horgan (eds.), Festschriftfor J. Kim.[84] Horgan, T. 2012. ?Introspection about Phenomenal Consciousness:Running the Gamut from Infallibility to Impotence,? in D. Smithies andD. Stoljar (eds.), Introspection and Consciousness. Oxford: OxfordUniversity Press. 403?422.[85] Horgan, T. 2011. ?The Phenomenology of Agency and Freedom: Lessonsfrom Introspection and Lessons from Its Limits,? Humana.Mente,15: 77?97.[86] Horgan, T. 2007. ?Agentive Phenomenal Intentionality and the Limits ofIntrospection,? Psyche, 13(2): 1?29.[87] Horgan, T and U. Kriegel. 2008. ?Phenomenal Intentionality Meets theExtended Mind,? The Monist, 91: 353?380.164[88] Horgan, T. and J. Tienson, 2002. ?The Intentionality of Phenomenologyand the Phenomenology of Intentionality,? in D. Chalmers (ed.),Philosophy of Mind: Classical and Contemporary Readings. Oxford:Oxford University Press. 520?33.[89] Hume, D. 1960/1739. A Treatise on Human Nature, L. A. Selby-Bigge(ed.). Oxford: Clarendon Press.[90] Ismael, J. 2013. ?Causation, Free Will, and Naturalism,? in H. Kincaid,J. Ladyman, and D. Ross (eds.), Scientific Metaphysics. New York:Oxford University Press. 208?235.[91] Johnston, M. 1992. ?How to Speak of the Colors,? Philosophical Studies,68(3): 221?263.[92] Jenkins, C. S. I. Forthcoming. ??Intuition,? Intuition, Concepts and the APriori,? in A. Booth and D. Rowbottoma (eds.), Intuitions. OxfordUniversity Press.[93] Kapitan, T. 1986. ?Deliberation and the Presumption of OpenAlternatives,? The Philosophical Quarterly, 36(143): 230?251.[94] Kratzer, A. 1977. ?What ?Must? and ?Can? Must and Can Mean,?Linguistics and Philosophy, 1(3): 337?355.[95] Kriegel, U. Forthcoming. ?A Hesitant Defense of Introspection,?Philosophical Studies.[96] Kriegel, U., 2002. ?Phenomenal Content,? Erkenntnis, 57: 175?198.[97] Kringelbach, M. and K. Berridge. 2009. ?Towards a FunctionalNeuroanatomy of Pleasure and Happiness,? Trends in Cognitive Science,13: 479?487.[98] Lehrer, K. 1960. ?Can We Know that We Have Free Will byIntrospection?? The Journal of Philosophy, 57(5): 145?57.[99] Lehrer, K. 1968. ?Cans Without Ifs,? Analysis, 29(1): 29?32.[100] Lehrer, K. 1976. ??Can? in Theory and Practice: A Possible WorldAnalysis,? in Brand and Walton (eds.), Action Theory. Dordrecht: D.Reidel. 241?270.165[101] Levin, D. T. and M. R. Banaji. 2006. ?Distortions in the PerceivedLightness of Faces: The Role of Race Categories,? Journal ofExperimental Psychology: General, 135(4): 501?12.[102] Lewis, D. 1994. ?Humean Supervenience Debugged,? Mind,103: 473?490[103] Lewis, D. 1986. ?The Paradoxes of Time Travel,? in Philosophical Papers,Vol. II. Oxford: Oxford University Press.[104] Lewis, D. 1983. ?New Work for a Theory of Universals,? AustralasianJournal of Philosophy, 61(4): 343?377.[105] Lewis, D. 1981. ?Are We Free to Break the Laws?? Theoria, 47: 112?121.[106] Lewis, D. 1973. Counterfactuals. Harvard: Harvard University Press.[107] Lycan, W. 1996. Consciousness and Experience. Cambridge, MA: MITPress.[108] MacPherson, F. 2012. ?Cognitive Penetration of Colour Experience:Rethinking the Issue in Light of an Indirect Mechanism,? Philosophy andPhenomenological Research, 84(1): 24?62.[109] Mele, A. 2006. Free Will and Luck. New York: Oxford University Press.[110] Mele, A. 2003. ?Agents? Abilities,? Nou?s, 37: 447?470.[111] Mele, A. 1995. Autonomous Agents. New York: Oxford University Press.[112] Mill, J. S. 1865. An Examination of William Hamilton?s Philosophy.Boston: William V. Spencer.[113] Miller, G., E. Galanter, and K. Pribam. 1960. Plans and the Structure ofBehavior. N.Y.: Holt, Rhinehart, and Winston.[114] Monroe, A. E. and B. F. Malle. 2010. ?From Uncaused Will to ConsciousChoice: The Need to Study, Not Speculate about People?s Folk Concept ofFree Will,? Review of Philosophy and Psychology, 9: 211?224.[115] Moore, G. E. 1912. ?Free Will,? from Ethics. Oxford: Oxford UniversityPress. 122?37.166[116] Nahmias, E., S. Morris, T. Nadelhoffer and J. Turner. 2004. ?ThePhenomenology of Free Will,? Journal of Consciousness Studies,11(7?8): 162?79.[117] Nahmias, E. and D. Murray. 2011. ?Experimental Philosophy on FreeWill: An Error Theory for Incompatibilist Intuitions,? in New Waves inPhilosophy of Action, Aguilar, J., A. Buckareff, and K. Frankish (eds.).Palgrave Macmillan. 189-216.[118] Nelkin, D. 2004. ?Deliberative Alternatives,? Philosophical Topics,32: 215?240.[119] Nichols, S. 2012. ?The Indeterminist Intuition: Sources and Status,? TheMonist, 95(2): 290?307.[120] Nichols, S. 2008. ?How Can Psychology Contribute to the Free WillDebate?? in J. Baer, J. C. Kaufman and R. F. Baumeister (eds.), Are WeFree? Psychology and Free Will. New York: Oxford University Press.10?31.[121] Nichols, S. 2004. ?The Folk Psychology of Free Will: Fits and Starts,?Mind and Language, 19(5): 473?502.[122] Nichols, S. and Joshua Knobe. 2007. ?Moral Responsibility andDeterminism: The Cognitive Science of Folk Intuitions,? Nou?s, 41(4):663?685.[123] O?Connor, T. 1995. ?Agent Causation,? in T. O?Connor (ed.), Agents,Causes, and Events: Essays on Indeterminism and Free Will. New York:Oxford University Press. 173?200.[124] Palmer, S. E. 1999. Vision Science: Photons to Phenomenology.Cambridge, MA: MIT Press.[125] Paul, L. A. 2010. ?A New Role for Experimental Work in Metaphysics,?Review of Philosophical Psychology, 1(3): 461?476.[126] Parfit, D. 1987. ?Divided Minds and the Nature of Persons,? inC. Blakemore and S. Greenfield (eds.), Mindwaves: Thoughts onIntelligence, Identity and Consciousness. Oxford: Blackwell Press.167[127] Parfit, D. 1971. ?Personal Identity,? The Philosophical Review,80(1): 3?27.[128] Pearl, J. 2000. Causality. Oxford: Oxford University Press.[129] Pereboom, D. 2011. Consciousness and the Prospects of Physicalism.New York: Oxford University Press.[130] Pereboom, D. 2008. ?A Compatibilist Account of the EpistemicConditions on Rational Deliberation,? The Journal of Ethics,12(3): 287?306.[131] Perry, J. 2004. ?Compatibilist Options,? in J. Campbell, M. O?Rourke, andD. Shier (eds.), Freedom and Determinism. MIT Press. 231?254.[132] Preuschoff, K., P. Bossaerts, and S. Quartz. 2006. ?Neural Differentiationof Expected Reward and Risk in Human Subcortical Structures,? Neuron,51: 381?390.[133] Quartz, S. R. 2009. ?Reason, Emotion, and Decision-Making: Risk andReward Computation with Feeling,? Trends in Cognitive Sciences, 13:209?215.[134] Railton, P. In Press. ?The Affective Dog and Its Rational Tale,? Ethics.[135] Railton, P. 2002. ?Kant Meets Aristotle Where Reason Meets Appetite,? inC. U. Moulines and K.-G. Niebergall (eds.), Argument Und Analyse.Paderborn, Germany: Mentis.[136] Reid, T. 1788. Essays on the Powers of the Human Mind. London: Tegg,Griffin & Cumming.[137] Rolls, E. T., J. Tromans and S. M. Stringer. 2008. ?Spatial SceneRepresentations Formed by Self-Organizing Learning in a HippocampalExtension of the Ventral Visual System,? European Journal ofNeuroscience, 28: 2116?2127.[138] Rosenthal, D. 1990. ?A Theory of Consciousness,? in N. Block, O.Flanagan, and G. Gu?zeldere (eds.), The Nature of Consciousness.Cambridge, MA: MIT Press. 773-788.168[139] Russell, P. and O. Deery (eds.). 2013. The Philosophy of Free Will:Essential Readings from the Contemporary Debates. New York: OxfordUniversity Press.[140] Saunders, J. T. 1968. ?The Temptations of Powerlessness,? AmericanPhilosophical Quarterly, 5: 100?108.[141] Searle, J. 1984. Minds, Brains, and Science. Cambridge, MA: Harvard UP.[142] Seligman, M., P. Railton, R. Baumeister, and C. Sripada. 2013.?Navigating into the Future or Driven by the Past: Prospection as anOrganizing Principle of Mind,? Perspectives on Psychological Science,8(2): 119?141.[143] Shariff, A. F., J. D. Greene, and J. W. Schooler. Forthcoming. ?His BrainMade Him Do It: Encouraging a Mechanistic Worldview ReducesPunishment.?[144] Siegel, S. 2006. ?Which Properties are Represented in Perception?? inT. Gendler Szabo and J. Hawthorne (eds.), Perceptual Experience. Oxford:Oxford University Press. 481?503.[145] Singer, T., H. D. Critchley, and K. Preuschoff. 2009. ?A Common Role ofInsula in Feelings, Empathy and Uncertainty,? Trends in CognitiveSciences, 13: 334?340.[146] Smart, J. J. C. 1961. ?Free Will, Praise and Blame,? Mind, 70(2): 91?306.[147] Smith, M. 2004. ?Rational Capacities,? in Ethics and the A Priori:Selected Essays on Moral Psychology and Meta-Ethics by M. Smith. NewYork: Cambridge University Press. 114?135.[148] Schultz, W. 2002. ?Multiple Reward Signals in the Brain,? Nature ReviewsNeuroscience, 1: 199?207.[149] Strawson, G. 1994. ?The Impossibility of Moral Responsibility,?Philosophical Studies, 75(1?2): 5?24.[150] Symons, J. 2008. ?Intuition and Philosophical Methodology,? Axiomathes,18(1): 67?89.169[151] Synofzik, M., G. Vosgerau and A. Newen. 2008. ?Beyond the ComparatorModel: A Multi-Factorial Two-Step Account of Agency,? Consciousnessand Cognition, 17(1): 219?239.[152] Taylor, R. 1983. Metaphysics, 3rd edition. Englewood Cliffs, N.J.:Prentice Hall.[153] Tobler, D. N., J. P. Dougherty, R. J. Dolan, W. Schultz. 2006. ?RewardValue Coding Distinct from Risk Attitude-Related Uncertainty Coding inHuman Reward Systems,? Journal of Neurophysiology, 97: 1621?1632.[154] Tooley, M. 1977. ?The Nature of Laws,? Canadian Journal of Philosophy,7: 667?698.[155] Turner, J. and E. Nahmias. 2006. ?Are the Folk Agent-Causationists??Mind and Language, 21(5): 597?609.[156] Tye, M. 1995. Ten Problems of Consciousness: A Representational Theoryof the Phenomenal Mind. Cambridge, MA: MIT Press.[157] van Inwagen, P. 2008. ?How To Think About the Problem of Free Will,?The Journal of Ethics, 12(3): 327?341.[158] van Inwagen, P. 2000. ?Free Will Remains a Mystery,? PhilosophicalPerspectives, 14: 1?20.[159] van Inwagen, P. 1983. An Essay on Free Will. Oxford: Oxford UniversityPress.[160] Vargas, M. 2013. ?How to Solve the Problem of Free Will,? in P. Russelland O. Deery (eds.). 400?416.[161] Velleman, J. D. 2006. ?So It Goes,? The Amherst Lecture in Philosophy.[162] Vihvelin, K. 2011. ?Arguments for Incompatibilism,? in E. N. Zalta (ed.),The Stanford Encyclopedia of Philosophy.[163] Vihvelin, K. 2004. ?Free Will Demystified: A Dispositional Account,?Philosophical Topics, 32(1?2): 427?450.[164] Vohs, K. D., and J. W. Schooler. 2008. ?The Value of Believing in FreeWill: Encouraging a Belief in Determinism Increases Cheating,?Psychological Science, 19: 49?54.170[165] Widerker, D. 2006. ?Blameworthiness and Frankfurt?s Argument Againstthe Principle of Alternative Possibilities,? in D. Widerker and M.McKenna (eds.), Moral Responsibility and Alternative Possibilities.Aldershot: Ashgate Publishing. 53?73.[166] Woodward, J. 2003. Making Things Happen: A Theory of CausalExplanation. New York: Oxford University Press.171Appendix ASurvey Materials for Chapter 2STUDY 1CONDITION 1[Page 1]Thank you for consenting to participate in the study.On the next page, you will read a brief passage. Then we will ask you somequestions.Click ?Next? below to begin.[Go to Page 2][Page 2]Please read the following passage, and answer the questions that follow as bestyou can:Imagine that you are sledding down a snowy path on a mountainside. Your sledhas a steering mechanism that allows you to control the direction of the sled.172Below you is a fork in the path with snow built up in the middle, and you cantell that, if you don?t direct your sled one way or the other, the contours of themountain will channel you and your sled either to the left or to the right.Consider how things seem to you as you approach the fork in the path. In partic-ular, consider what it?s like to decide which way the sled will go.Please indicate your level of agreement with the following statement:When deciding which way the sled will go, it feels like I can either go to the leftor go to the right.1. Disagree completely2. Strongly disagree3. Disagree4. Neither agree nor disagree5. Agree6. Strongly agree7. Agree completely1[If 1?4, then go to Page 3; if 5?7, then go to Page 5][Page 3]Still considering how things seem to you as you approach the fork in the path,please indicate your level of agreement with the following statement:When deciding which way the sled will go, it feels like I cannot either go to theleft or go to the right.[If 1?4, then go to Page 4; if 5?7, then go to Page 12]1 This 7-point Likert scale was used whenever participants were asked to indicate their level ofagreement with a statement. It will be omitted from the materials presented below, and should betaken for granted.173[Page 4]Still considering how things seem to you as you approach the fork in the path,please indicate your level of agreement with the following statement:When deciding which way the sled will go, it is unclear whether or not the fol-lowing is true: it feels like I can either go to the left or go to the right.[Go to Page 12][Page 5]We will now describe to you the notion of causal completeness and then ask youhow it relates to the sledding situation.According to causal completeness, everything that happens is fully caused by whathappened before it. This is true from the very beginning of the universe, so whathappened in the beginning of the universe fully caused what happened next, andso on right up until the present. Causal completeness holds that everything is fullycaused in this way, including people?s decisions.To understand this idea, let?s start with an example of a real event. On May 18,1980, Mount St. Helens erupted into a fiery volcano. The eruption was triggeredby a 5.1 earthquake. Scientists agree that this earthquake (along with other factors)fully caused the eruption.According to causal completeness, if we could somehow replay the entire pastright up until St. Helens erupted on May 18, 1980, then St. Helens would onceagain erupt at that time. Another way to put this is to say that all the events leadingup to the eruption made it so that the eruption had to happen.The causal completeness view maintains that this is true for everything in the uni-verse, not just volcanoes and earthquakes. So, to take another real example, onAugust 22, 2008, Barack Obama decided to have Joe Biden as his Vice Presiden-tial running mate in the U.S. Presidential election. There were obviously manyfactors that led up to this decision?including Obamas feelings about Biden, his174feelings about Hillary Clinton, his beliefs about how each would help his can-didacy, his beliefs about how each would actually function as Vice President,and so on. According to causal completeness, if we replayed the past right upuntil Obama?s decision?including everything that was going through Obama?smind?then Obama would once again make exactly the same decision. That is,all the events leading up to Obama?s decision (including everything that was goingthrough Obama?s mind), made it so that it had to happen that Obama would pickBiden.Causal completeness also maintains that this goes for everything in the future. Soin the future, every time you make a decision, what you ultimately decide will befully caused by everything that happens leading up to the decision. That is, giveneverything that happens before any future decision of yours, that future decisionhas to happen the way that it does.Just to be sure that you understand the idea of causal completeness, please indicatewhether the following is True or False.According to causal completeness, St. Helens would have erupted on May 18,1980 even if there had been no earthquake.True / False[If true, then go to Page 6; if false, then go to Page 7][Page 6]The correct answer is False.Recall that causal completeness says that what happens is fully caused by whathappened before it. So causal completeness doesn?t say that the eruption of St.Helens was inevitable no matter what. Rather, causal completeness would saythat the eruption happened because of the earthquake.Given this clarification, please indicate whether the following statement is True orFalse.175According to causal completeness, if there hadn?t been an earthquake, St. Helensstill would have erupted on May 18, 1980.True / False[If true, then go to Page 12; if false, then go to Page 7][Page 7]Just one more question to make sure you understand causal completeness. Pleaseindicate whether the following statement is True or False.According to causal completeness, if a week from now Barack Obama decides tohave soda with dinner, all the events leading up to that decision will make it thecase that he has to decide to have a soda with dinner.True / False[If true, then go to Page 8; if false, then go to Page 12][Page 8]The correct answer is True.Recall that causal completeness says that what happens in the future is fullycaused by what happened in the past. There is no exception for human decisions.According to causal completeness, when you make a decision in the future, thatdecision will be fully caused by everything leading up to it. So causal complete-ness says that if you make a decision in the future?for example, if a week fromnow you decide to have soda with dinner?that decision has to happen in the par-ticular way that it does, given everything that leads up to it (including everythingthat is going through your mind right before the decision).Given this clarification, please indicate whether the following statement is True orFalse.According to causal completeness, if everything that happens leading up toObama?s future decision?for example, his decision a week from now to have176soda with dinner?stays exactly the same, then that decision does not have tohappen.True / False[If true, then go to Page 12; if false, then go to Page 9][Page 9]Now, recall the sledding situation. You previously agreed with the following state-ment:When deciding which way the sled will go, it feels like I can either go to the leftor go to the right.Considering this previous statement and your understanding of causal complete-ness, please indicate your level of agreement with the following:Even though it felt like I could either go to the left or go to the right, if causalcompleteness is true there is something mistaken about how that decision felt tome.[Go to Page 10][Page 10]Please briefly explain why you answered the last question as you did. (This ques-tion is optional.)[Go to Page 11][Page 11]Thank you for completing the main portion of the survey. If you have any com-ments about the questions, please enter them below. Otherwise, please help usout by entering some demographic information in the few questions that followbefore finalizing your survey.[Go to page Page 12]177[Page 12][The following demographic information was gathered: participants? sex, age, thenumber of philosophy classes they had taken, whether participants had taken aclass in which free will was a topic, participants? ethnic background, religiousaffiliation, and level of religiosity.]178STUDY 1CONDITION 2[This condition used the same introduction page and training section, Pages 5?8,as Study 1, Condition 1.]Page 2Please read the following passage, and answer the questions that follow as bestyou can:Imagine that, many years ago, you were sledding down a snowy path on a moun-tainside. Your sled had a steering mechanism that allowed you to control thedirection of the sled. Below you was a fork in the path with snow built up in themiddle, and you could tell that, if you didn?t direct your sled one way or the other,the contours of the mountain would channel you and your sled either to the left orto the right. In the end, you decided to go left and you went left.Still imagining that you made this decision many years ago, and assuming no extrafacts not described in the passage above, please indicate your level of agreementwith the following statement:I could have gone right instead of left.[If 1?4, then go to Page 3; if 5?7, then go to Page 5][Page 3]Still imagining that you made this decision many years ago, and assuming no extrafacts not described in the passage above, please indicate your level of agreementwith the following statement:I could not have gone right instead of left.[If 1?4, then go to Page 4; if 5?7, then go to Page 12]179[Page 4]Still imagining that you made this decision many years ago, and assuming no extrafacts not described in the passage above, please indicate your level of agreementwith the following statement:It is unclear whether or not the following is true: I could have gone right insteadof left.[Go to Page 12][Page 9]Now, recall the sledding situation. You previously agreed with the following state-ment:I could have gone right instead of left.Considering this previous statement and your understanding of causal complete-ness, please indicate your level of agreement with the following:Even though I said I could have gone right instead of left, if causal completenessis true there is something mistaken about what I said.[Go to Page 10][Page 10]Please briefly explain why you answered the last question as you did. (This ques-tion is optional.)[Go to Page 11][Page 11]Thank you for completing the main portion of the survey. If you have any com-ments about the questions, please enter them below. Otherwise, please help us180out by entering some demographic information in the few questions that followbefore finalizing your survey.[Go to page Page 12]181STUDY 2[This study used the same introduction page and training section, Pages 5?8, asStudy 1, Condition 1.]CONDITION 1[Page 2]Please read the following passage, and answer the questions that follow as bestyou can:Imagine that you have $0.50 to donate. You have two options:Donate to a foundation that protects the endangered tree castanea dentata.ORDonate to a foundation that protects the endangered tree ulmus dentata.These are your only two options.Now, consider how things seem to you as you face your decision. In particular,consider what it?s like to decide which option to choose.Please indicate your level of agreement with the following statement:When deciding which option to choose, it feels like I can either choose to donateto castanea dentata or choose to donate to ulmus dentata.[If 1?4, then go to Page 3; if 5?7, then go to Page 5][Page 3]Still considering how things seem to you as you face your decision, please indicateyour level of agreement with the following statement:When deciding which option to choose, it feels like I cannot either choose todonate to castanea dentata or choose to donate to ulmus dentata.[If 1?4, then go to Page 4; if 5?7, then go to Page 12]182[Page 4]Still considering how things seem to you as you face your decision, please indicateyour level of agreement with the following statement:When deciding which option to choose, it is unclear whether or not the followingis true: it feels like I can either choose to donate to castanea dentata or choose todonate to ulmus dentata.[Go to Page 12][Page 9]Now, recall the donation situation. You previously agreed with the following state-ment:When deciding which option to choose, it feels like I can either choose to donateto castanea dentata or choose to donate to ulmus dentata.Considering this previous statement and your understanding of causal complete-ness, please indicate your level of agreement with the following:Even though it felt like I could either choose to donate to castanea dentata orchoose to donate to ulmus dentata, if causal completeness is true then I couldn?treally have chosen differently than I did.[Go to Page 10]183STUDY 2CONDITION 2[Page 2]Please read the following passage, and answer the questions that follow as bestyou can:You have $0.50 to donate. We, the researchers, will actually donate this moneyfor you whichever way you decide. You have two options:Donate to a foundation that protects the endangered tree castanea dentata.ORDonate to a foundation that protects the endangered tree ulmus dentata.These are your only two options. Each option is currently available for you tochoose at the bottom of this page. But don?t decide just yet.First, consider how things seem to you as you face your decision. In particular,consider what it?s like to decide which option to choose.Please indicate your level of agreement with the following statement:When deciding which option to choose, it feels like I can either choose to donateto castanea dentata or choose to donate to ulmus dentata.HERE ARE YOUR TWO OPTIONS:After you have answered the question above, please choose a charity:The endangered tree castanea dentata / The endangered tree ulmus dentata[If 1?4, then go to Page 3; if 5?7, then go to Page 5]184[Page 3]Considering how things seemed to you as you faced your decision, please indicateyour level of agreement with the following statement:When deciding which option to choose, it felt like I couldn?t either choose todonate to castanea dentata or choose to donate to ulmus dentata.[If 1?4, then go to Page 4; if 5?7, then go to Page 12][Page 4]Still considering how things seemed to you as you faced your decision, pleaseindicate your level of agreement with the following statement:When deciding which option to choose, it was unclear whether or not the follow-ing was true: it felt like I could either choose to donate to castanea dentata orchoose to donate to ulmus dentata.[Go to Page 12][Page 9]Now, recall the donation situation. You previously agreed with the following state-ment:When deciding which option to choose, it feels like I can either choose to donateto castanea dentata or choose to donate to ulmus dentata.Considering this previous statement and your understanding of causal complete-ness, please indicate your level of agreement with the following:Even though it felt like I could either choose to donate to castanea dentata orchoose to donate to ulmus dentata, if causal completeness is true then I couldn?treally have chosen differently than I did.[Go to Page 10]185STUDY 2CONDITION 3[Page 2]Please read the following passage, and answer the questions that follow as bestyou can:You have $0.50 to donate. We, the researchers, will actually donate this moneyfor you whichever way you decide. You have two options:Donate to a foundation that protects the endangered tree castanea dentata.ORDonate to the Childhood Cancer Foundation.These are your only two options. Each option is currently available for you tochoose at the bottom of this page. But don?t decide just yet.First, consider how things seem to you as you face your decision. In particular,consider what it?s like to decide which option to choose.Please indicate your level of agreement with the following statement:When deciding which option to choose, it feels like I can either choose to donateto the endangered tree castanea dentata or choose to donate to the ChildhoodCancer Foundation.HERE ARE YOUR TWO OPTIONS:After you have answered the question above, please choose a charity:The endangered tree castanea dentata / The Childhood Cancer Foundation[If 1?4, then go to Page 3; if 5?7, then go to Page 5]186[Page 3]Considering how things seemed to you as you faced your decision, please indicateyour level of agreement with the following statement:When deciding which option to choose, it felt like I couldn?t either choose to do-nate to the endangered tree castanea dentata or choose to donate to the ChildhoodCancer Foundation.[If 1?4, then go to Page 4; if 5?7, then go to Page 12][Page 4]Still considering how things seemed to you as you faced your decision, pleaseindicate your level of agreement with the following statement:When deciding which option to choose, it was unclear whether or not the follow-ing was true: it felt like I could either choose to donate to the endangered treecastanea dentata or choose to donate to the Childhood Cancer Foundation.[Go to Page 12][Page 9]Now, recall the donation situation. You previously agreed with the following state-ment:When deciding which option to choose, it feels like I can either choose to donate tothe endangered tree castanea dentata or choose to donate to the Childhood CancerFoundation.Considering this previous statement and your understanding of causal complete-ness, please indicate your level of agreement with the following:Even though it felt like I could either choose to donate to the endangered treecastanea dentata or choose to donate to the Childhood Cancer Foundation, if187causal completeness is true then I couldn?t really have chosen differently than Idid.[Go to Page 10]188STUDY 3[This study used the same introduction page and training section, Pages 5?8, asStudy 1, Condition 1.]CONDITION 1[Page 2]Please read the following passage, and answer the questions that follow as bestyou can:At the bottom of this page, there are two buttons, labelled H and V. Each optionis currently available for you to choose. In a moment, we?ll ask you to choosejust one of them. For this survey, only one of the buttons will give you an extra$0.05 (as bonus payment on MTurk) if you choose it. But we won?t tell you whichbutton it is?you?ll have to make a choice and find out.But don?t decide just yet.First, consider how things seem to you as you face your decision. In particular,consider what it?s like to decide which option to choose.Please indicate your level of agreement with the following statement:When deciding which option to choose, it feels like I can either choose H orchoose V.HERE ARE YOUR TWO OPTIONS:After you have answered the question above, please choose a button.Once you have you chosen, we?ll ask you a few more questions before telling youwhether you picked the bonus button.H / V[If 1?4, then go to Page 3; if 5?7, then go to Page 5]189[Page 3]Considering how things seem to you as you face your decision, please indicateyour level of agreement with the following statement:When deciding which option to choose, it feels like I can either choose H orchoose V.[If 1?4, then go to Page 4; if 5?7, then go to Page 12][Page 4]Still considering how things seem to you as you face your decision, please indicateyour level of agreement with the following statement:?When deciding which option to choose, it is unclear whether or not the followingis true: it feels like I can either choose H or choose V.[Go to Page 12][Page 9]Now, recall the button-choosing situation. You previously agreed with the follow-ing statement:When deciding which option to choose, it feels like I can either choose H orchoose V.Considering this previous statement about how things felt to you before yourchoice and your understanding of causal completeness, please indicate your levelof agreement with the following:?If causal completeness is true, then I couldn?t really have chosen differently thanI did.[Go to Page 10]190STUDY 3CONDITION 2[Page 2]Please read the following passage, and answer the questions that follow as bestyou can:At the bottom of this page, there are two buttons, labelled H and V. Each optionis currently available for you to choose. In a moment, we?ll ask you to choosejust one of them. For this survey, only one of the buttons will give you an extra$0.05 (as bonus payment on MTurk) if you choose it. But we won?t tell you whichbutton it is?you?ll have to make a choice and find out.But don?t decide just yet.First, consider how things seem to you as you face your decision. In particular,consider what it?s like to wonder which option you?ll choose.Please indicate your level of agreement with the following statement:When wondering which option I?ll choose, it feels like I don?t know for surebefore I select a button which button is the bonus button.HERE ARE YOUR TWO OPTIONS:After you have answered the question above, please choose a button.Once you have you chosen, we?ll ask you a few more questions before telling youwhether you picked the bonus button.H / V[If 1?4, then go to Page 3; if 5?7, then go to Page 5]191[Page 3]Considering how things seem to you as you face your decision, please indicateyour level of agreement with the following statement:When wondering which option I?ll choose, it feels like I do know for sure beforeI select a button which button is the bonus button.[If 1?4, then go to Page 4; if 5?7, then go to Page 12][Page 4]Still considering how things seem to you as you face your decision, please indicateyour level of agreement with the following statement:When wondering which option I?ll choose, it is unclear whether or not the follow-ing is true: it feels like I don?t know for sure before I select a button which buttonis the bonus button.[Go to Page 12][Page 9]Now, recall the button-choosing situation. You previously agreed with the follow-ing statement:When wondering which option I?ll choose, it feels like I don?t know for surebefore I select a button which button is the bonus button.Considering this previous statement about how things felt to you before yourchoice and your understanding of causal completeness, please indicate your levelof agreement with the following:If causal completeness is true, then I knew for sure before I selected a buttonwhich button was the bonus button.[Go to Page 10]192

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0074120/manifest

Comment

Related Items