Hard Luck is a challenging, provocative, and engaging book that wrestles with a number of key issues in metaphysics, epistemology, and moral psychology.  In it, Levy advances a sophisticated argument that there is no such thing as free will.   However, rather than presenting another argument about the incompatibility of free will and determinism Levy’s claim is that “it is not ontology that rules out free will, it is luck.” (2) While luck objections to free will are not new, Levy provides a novel and detailed account while engaging a great deal of the contemporary literature.

Levy’s book begins by advancing and defending an account of luck.  Building on the work of Pritchard and Coffman, Levy lays out an account of luck that is a function of chance, significance, and a lack of control.   Significance is understood as something that matters, where a subject caring about it is sufficient to make it matter.  Levy gives a modal account of chance that allows for both indeterministic and deterministic worlds to exhibit chance.  It suffices for an occurring event to be chancy, that in a large enough number of nearby worlds it does not occur.  This account of chance is admittedly vague, but Levy argues that fact should not bother us since, luck is itself a vague concept.  Finally, for an event to be lucky for an agent, its occurrence must not be under agent’s the direct control.  In all, we are given the following account of chancy luck:

          An event or state of affairs occurring in the actual world is chancy lucky for an agent if (i) that event or state of affairs is significant for that agent; (ii) the agent lacks direct control over that event or state of affairs; and (iii) that event or state of affairs fails to occur in many nearby worlds; the proportion of nearby worlds that is large enough for the event to be chancy lucky is inverse to the significance of the event for the agent.  (36)

Levy, however, claims that chancy luck fails to account for all the luck that there is.  Drawing from cases of epistemic luck and Nagel’s account of constitutive luck, Levy claims that there is also non-chancy luck.  For instance, while it may not be a matter of chance that I have the psychological traits and dispositions that I do (since in most nearby worlds my genetic makeup and culture do not differ significantly), it is still a matter of luck (whether good or bad luck) that I have those traits and dispositions.  After all, those traits and dispositions vary greatly among people, are significant, and they are no more ‘up to me’ than matters of chance even if they are themselves not matters of chance.  So, Levy gives the following account of non-chancy luck:

An event or state of affairs occurring in the actual world that affects an agent’s psychological traits or dispositions is non-chancy lucky for an agent if (i) that event or state of affairs is significant for the agent; (ii) the agent lacks direct control over that event or state of affairs; (iii) events or states of affairs of that kind vary across the relevant reference group, and (iv) in a large enough proportion of cases that event or state of affairs fails to occur or be instantiated in the reference group in the way which it occurred or was instantiated in the actual case.  (36)

With an account of luck in hand, Levy goes on to argue that both libertarian and compatibilist accounts of free will leave an unacceptable degree of luck in human actions — a degree that precludes moral responsibility.  Levy claims that libertarian accounts of free will must be able to provide contrastive explanations — explanations for why the agent acted in the way that she did as opposed to some alternative way — and that they are unable to do so.  Levy examines both event-causal and agent-causal versions of libertarianism and argues that both fail on this count.  The failure for these libertarian accounts to offer satisfactory contrastive explanations can be seen using van Inwagen’s ‘replay argument’.  If, as libertarians claim, agent’s choices are undetermined, then rewinding and replaying the universe will have them sometimes making distinct choices given the very same reasons, dispositions, and so forth.  However, Levy claims that if this is so then nothing explains why they chose as they did.  After all, everything that you may point to as a reason is also present in the counterfactual cases where a different choice is made.

It may be thought that agent-causal libertarian accounts have something to add since they claim that the agent herself is there to provide the ‘final push’ to choice.  Free will, however, is a rational power; the variety of free will required for moral responsibility must be reason responsive, it cannot be blind.  Levy argues that agent-causal libertarian view cannot have the agent both exercising direct control over its choice and making its choice based on reasons.  While the agent has reasons to choose any number of choices, the link between those reasons must be indeterministic (given libertarianism).  Since the agent’s reasons do not compel her choice, we can, and must, ask why she chose as she did.  Since she could have chosen differently given those very same reasons, those reasons fail to contrastively explain her choice.  While agent-causal libertarians may retreat to talking about the agent’s weighing of reasons, such a move only pushes the question back a level and we are left wondering why the agent weighed her reasons as she did, since in other nearby worlds she had the same reasons yet weighed them differently.  So, according to Levy, most versions of libertarianism face a dilemma: either fail to give a contrastive explanation of the actions of agents or fail to have that explanation be grounded in reasons. (I say ‘most’ since at the conclusion of chapter 3 Levy lays out several versions of libertarianism that avoid these particular luck problems.  However, these versions of libertarianism are quite close to compatibilist accounts, and, according to Levy they fail to avoid problems with luck altogether.)

Levy extends his argument to compatibilist accounts of free will, arguing that they too face a luck problem, even if it is less significant than the problem for libertarianism.  While compatibilists can give a contrastive explanation as to why an agent acted in a certain way (as opposed to other ways), Levy argues that these explanations are themselves infected with luck (either constitutive luck or present luck, and often both).   When the agent’s constitution (or ‘endowment’ as Levy calls it) settles the agent’s action, the agent is subject to constitutive luck.  The agent is subject to present luck, when chancy factors like one’s mood, what considerations come to mind, and so forth, settle the agent’s action.  Levy examines a number of compatibilist attempts at giving ownership conditions for one’s actions, but finds all lacking.  So, compatibilists too face a dilemma: either have it that the constitution of agents is decisive with respect to an action or allow for it to be modified over time.  If the former, then the compatibilist account falls victim to constitutive luck, and if the latter then the account falls victim to a combination of present luck and constitutive luck.  Since “we cannot undo the effects of luck with more luck” (89) compatibilist accounts also fail to avoid the luck problem.

Since free will is a rational power, Levy argues that there is yet another way that we fail to have relevant control over our actions — by failing to have relevant control over our reasons, and thus our beliefs.  Put differently, according to Levy we typically fail to meet the epistemic requirements of free will and moral responsibility.   Even when agents can directly bring about states of affairs, they do so freely only if they know that this is so, know how it is that this is so, and properly appreciate (or are culpable for not appreciating) the significance of bringing that state of affairs about (the agent must appreciate the reasons to bring it about).  (112-3) This epistemic condition presents a significant obstacle, since whether agents are non-culpably ignorant of the relevant details is frequently itself a matter of luck (often chancy luck or constitutive luck).  Culpable ignorance is rare, then, because it requires origination in a benighting act (an act where an agent “knowingly and freely passes up an opportunity to improve her epistemic position” (117)) and such acts are themselves rare.   

If all blameworthy actions must trace back to some benighting act to meet the epistemic condition on responsibility, all blameworthy action must originate in akratic actions.  In chapter 6, Levy argues, contrary to the received view, that agents are not responsible for akratic actions, and so agents cannot begin a series of actions that they are morally responsible for on the basis of some akratic action.  Levy examines a number of accounts of akratic actions, dividing them between judgment-based and desire-based accounts, and finds all of them incapable of grounding moral responsibility.  Judgment-based accounts, according to Levy, fail to have it that the subject knows or sufficiently appreciates the significance of her action, and thus the agent fails to meet the epistemic conditions for moral responsibility.  Desire-based accounts are left with a dilemma: either akratic actions are a matter of compulsion or luck, or they fail to meet the epistemic requirement for moral responsibility.  So, Levy argues, we have no good reason to believe that agents can be morally responsible for akratic actions.

In the final two chapters, Levy gives his assessment of accounts of moral responsibility that simply bite the bullet on constitutive luck.  According to such theories, agents can be responsible for their actions irrespective of their history, irrespective of how they came to be the way that they currently are.  Levy’s attention is aimed at ‘quality of the will’ theories in particular — those that claim that agents are morally responsible for those actions that express who the agent is (158).  Levy argues that variants on Frankfurt cases can demonstrate that who the agent is (i.e. what capacities and endowments the agent has) can also be manipulated such that there is not enough there (inside the skin) to hang moral responsibility on.  Rather, even the capacities of an agent rely on factors outside of the ‘inner citadel’.  Levy also argues that such accounts of moral responsibility also fail on empirical and moral grounds.  They fail empirically by resting on the false claims about our implicit judgments and how they are expressed.  They fail on moral grounds by treating individuals unfairly; in fact they only worsen the pre-existing inequalities by further rewarding the lucky and further penalizing the unlucky. 

Hard Luck will be of great interest, and a must read, for those with any interest in the philosophical debates concerning free will.  Levy advances a number of the central debates here, and his arguments deserve careful attention.  Those with interests in epistemology, moral psychology, and cognitive psychology will also find plenty of interest here.

I am indebted to Sean Rech for many helpful conversations on these matters.

© 2014 Jonathan Matheson

Jonathan Matheson, University of North Florida