Monthly Archives: June 2013

Notes on Thomas Nagel’s “Moral Luck”


In his paper ‘Moral Luck’ Thomas Nagel argues against Kant’s idea that the moral will can and must be taken as the proper object of our moral judgments independently of the judgments we make about the consequences that issue from the actions of that will.  Against Nagel, I will defend Kant’s idea and claim that that the proper object of our moral judgments is the will and not the outcome of our willed actions.

Both Kant and Nagel agree that the outcome of one’s actions is largely beyond the control of the acting agent.   That is to say that a man may do everything that is within his power to do an action A and still do not-A.  A fireman may try with all his might to save a child dangling from a 22-story window but the result of his efforts may be that the child falls 22 stories to her death as the fireman’s grip gives out from sheer exhaustion.  In such a case, Nagel argues,  the fireman is morally blameworthy in a way that he would not have been had the child, for instance, weighed but a few pounds lighter such that the fireman’s grip on her would not have given way when it did.  In this second case, the fireman through no additional effort or power of his own, would be judged a hero.  And Nagel calls this ‘moral luck’. 

Moral luck, for Nagel, obtains (in the one measure) when circumstance conspires with the intentions of a good will so that the actual outcome of one’s action is in harmony with its intended outcome.  Luck enters the picture precisely because no agent is or could be in complete control of the circumstances in which and through which she acts.  Whether she does good or does bad is therefore, not something she can claim responsibility for.  Chance has the final say over what she does and thus over her moral assessment as an agent in toto.

Nagel explicates his theory of moral luck, or luck in relation to good action, by presenting it on analogy with epistemic luck, or luck in relation to true belief.   This analogy is helpful , for granting its validity, it allows us to apply a defense of the power of knowledge – call such a power the intellect – to a defense of the power of action – call it the will – such that if one defense  is successful, so is the other.  

Now, in epistemology, it is often held that on the justified true belief conception of knowledge, knowledge is impossible because no subject is in the position to know whether or not her beliefs are in agreement with the object that they are about.  Since no singular belief satisfies the JTB conception of knowledge, no singular act of believing is or could be an act of knowing.   From this, it is thought to extend generally such that there can in principle be no knowing subject, or stated differently, no subject with a power to know.  This can be named the ‘argument from fallibility’, it states that:

                A power vulnerable to malfunction is not a functional power.

In its epistemic form the principle states that an intellect vulnerable to false belief is not a knowing intellect.  In its moral form the principle states that a will susceptible to the production of wrong actions is not a moral will.  In its biological form the principle states that a behavioural trait susceptible to failure is not an adaptive trait.  I argue that this principle from infallibility, in all its abovementioned forms is false.  We will start with the last.

The fish has a behavioural trait which allows it to move itself through water. This trait is known typically as the ability to swim.  The fish is a good swimmer even if it becomes entangled in a fisherman’s net or if it is lifted from the water by an eagle and carried into the sky.  That the fish is afforded the opportunity to express its ability to swim is subject to luck but that the fish is able to swim is not subject to luck for luck is something that happens, and the ability to swim is not something that happens to the fish.  Rather, the power to swim is something that constitutes the fish as a fish.  A fish that has no such power is not a fish.  Thus, the fish is a good swimmer, even when it is prevented from swimming.

In the epistemic case, the human agent enjoys, through its intellect, the power to gain knowledge about its world and its own place in this world. Specifically, the intellect functions such that false beliefs once shown to be false will be corrected and the gap between the thought and the world repaired. The human intellect has this power even if circumstance dictates that now one believes falsely that-p.  That circumstance conspires to allow the agent to believe truly that p obtains or does not obtain is subject to luck but that the agent is able to know through this process of belief correction is itself not subject to luck.  As with the fish’s power to swim, the human’s power to know is not something that happens to the human – chance has no say in the matter – but something which constitutes the human as human.  A human that has no intellect is not a human.  Thus, the human has a power of knowledge even when it is prevented from knowing.

Now, in the moral case, the structure is no different. The good will is such that it acts according to principles productive of what it is right to do.  If, acting from the good will, an agent brings it about that something terrible occurs, then, though luck plays a role in the outcome of the action and the state of affairs that obtains, luck has no say in whether or not the acting agent’s will is good or bad.  The will is not something that happens to one, but rather it is constitutive of what that person is.  The morality of the agent cannot be assessed apart from the morality of the will.  Thus, she who acts from a good will who does wrong is no less moral than she who acts from a good will who does right.  And she who acts from a bad will who does right is no less morally blameworthy than she who acts from a bad will and does wrong.

Let us consider some further examples:

A driver who hits a pedestrian while driving drunk or while speeding, is guilty in a way that a train conductor whose train kills a man who falls from the station platform onto the tracks is not.  In the one case, the driver is responsible for the jeopardy that the pedestrian unknowingly steps into when he steps into the road; in the other case, nothing the conductor does causes any threat to the life of the poor man who falls in front of his train. The disparity holds even though it can be noted that in neither case did the driver or conductor have any control over the presence of the victim.  If bad luck enters the picture in either case, it is not the bad luck of the drivers but of the victims and in no sense is it moral luck.

But if we remove the presence of the victims in either case, then it does appear that the drunk driver is ‘lucky’ in a way that the train conductor is not for though the drunk driver does wrong, nothing bad or tragic comes of it. In a sense, we might say, that ‘he gets away with it’. But why should we think that this is moral luck?  He may be lucky in the sense that he escapes legal culpability and the punishment and shame he might otherwise have acquired but morally speaking, the man has done wrong; he is immoral. A reckless driver, especially a flagrant and habitual one, whose driving never causes any deaths, is no less morally blameworthy than one whose driving does result in someone’s death. The blameworthiness is directed at the will not the consequences of the action.

Now, on another measure, Nagel thinks that the will itself, specifically the kind of will fortune bequeaths to us, is subject to luck. ‘What if one is born with a wicked will?’ we might ask.  Clearly in such a case the formation of this will is not under one’s antecedent control in the same way that the drunk driver’s choice to drive recklessly is. Therefore it is tempting to think that the agent is not responsible for the acts that issue from such a will. But we should resist this temptation.  For succumbing to it obscures the fact that it is the will that is the object of our moral judgments.  We should not look beyond it for an agent of whom we can say that the will is a mere attribute. And a bad will deserves all the blame we will heap on it. That the will and not the person is the proper object of moral, as opposed to say legal, judgment should explain why there is no moral difference between the drunk driver who kills a pedestrian and the drunk driver who does not. Similarly, a mother whose intentional negligence results in the harm of her child is no more morally blameworthy than a mother whose intentional negligence causes her child no such harm. They are both expressing an immoral will.  

All of this is to say that there is no such thing as moral luck. For luck is something that happens to one, while morality is no such thing.

I’d like to close by highlighting a tension that I do believe Nagel brings out very well in his paper and that is the following: whether or not one is acting from a good or bad will is often obscure until one has acted and the consequences of that action can be evaluated in broad daylight.

Gauguin leaves his family to become a painter.  How can we assess the morality of such an action unless we wait and see whether or not history deems his art to be worthy of such a sacrifice? 

Here is my answer:

Though the will may be revealed by the outcome of the action, the will is not determined by the outcome. The will is, instead, visible in the response the agent has to her initial action. If I feel truly guilty for my action after the fact, then I blame myself for having done wrong. But a will that blames itself for its past failings is a good will. Just as an intellect that corrects its past judgments when it sees that they have erred is, ipso facto, a power of knowledge. We judge a will properly if we judge it not by only one of its particular expressions but by its disposition to respond to its expressions in a way that manifests its teleological function.

Gauguin was not morally lucky that history has appraised him as a worthy painter. If his decision to quit his job as a merchant and abandon his family was the result of flippancy or boredom then the beauty of his art does not render his action in any way morally justified.  But if his decision was made from all the earnestness and necessity befitting of a good and honest heart then he is not morally blameworthy for that decision even if his art is rubbish.  It is not the beauty of his art that would morally justify his decision but rather that he decided as he did because he believed that it was the right thing to do, not the easy thing, not the pleasant thing, but the right thing.  That we are in no position to see into the morality of such a decision from our present point of view does not entail that there is no fact of the matter whether Gauguin acted morally or not when he picked up the paintbrush.  We however, can judge our own will by examining whether or not we strive towards the good, (a striving that includes correcting ourselves for our own past transgressions). A good will like a good intellect seeks not constant perfection but constant improvement.

Brandom on Hegelian Magnanimity in the face of Reductive Naturalism

The story is familiar. The esteem enlightenment Reason feels for itself as it explains and disenchants more and more of the world turns soon enough into dispair and confusion as Reason turns on itself to explain its own normative discourse in non-normative causal terms.  Such a picture is what Brandom calls here genealogy and it is associated with the skepticism and eliminativism we read in Schopenhauer,  Nietzsche and in contemporary philosophy and science today in thinkers like the Churchlands and Metzinger.  Following Hegel’s metaphor, Brandom characterizes this form of hermeneutics as the metaphysics of the “valet” who, as the saying goes, sees the man never as the hero but only as the man behind the hero’s façade. As far as genealogy is concerned rational discourse – the once noble pursuit of truth and or knowledge, the game of giving and asking for reasons – is the heroic façade masking the plebian reality of senseless causal processes done up to look like dialogue and normative order. The fMRI scan of the so-called ‘reason-responsive’ brain is the valet’s view of the hero as the ordinary man behind the scenes, perspiring and defecating like all the rest of us. The problem with this view is however, as Brandom points out in his lecture and elsewhere, that the genealogical account is itself only intelligible insofar as it is part of the normative discourse it wants to destroy. Reductionist metaphysics is a move in a language game and in a rational tradition and cannot if it is to remain intelligible overturn the source and succour of its own intelligibility. The hero pays the valet’s wages.

There is of course a compelling argument put forward by Brandom to render this defense of Reason more than just convincing. But I want to focus the current thought on what I see as a flaw or lacuna in Brandom’s system.

For Brandom, the problem of the purely genealogical account of Reason which explains thoughts, actions and norms as nothing other than the mere epiphenomena of so much pulpy neuronal palpitations and firings is that such an account fails to recognize that the semantic content of our concepts and propositions does not come ready-made into the world ex nihilo. Instead, Brandom insists that semantics gains its meanings and conceptual content only through our linguistic and conceptual practices and thus, makes no sense outside of that context.

I agree wholeheartedly with Brandom on this point. However, the way Brandom tells the story, our normative discourse is an essentially social phenomenon and a necessarily social one. He takes this to be a Hegelian point and maintains with the latter that Geist is not a brainbound entity or process but a property, if you will, of the social world. Again, not much to contest here. But I want to call into question the notion that we could ever understand the individual society as a locus of Geist without first availing ourselves of an understanding of the Self as a more fundamental form of the individual. To be sure, the Person is a decidedly social entity but persoonhood, it seems to me, supervenes on selfhood just as my name supervenes on my “I”. The Person is something I become or rather take on. Now, it will be argued that the reverse is no less true: the child only becomes an I or a Self by being treated as a person with a name and as a member of the family-sized society. But this picture would have it that what the family-sized society welcomes into its hearth as a child is i) not bound by conceptual norms and ii) not an entity who has a “world in view” to quote McDowell. That is to say, that Brandom’s story seems to have it that selfhood just is personhood and to me this account is clearly an impoverished view of what it is to be a self-conscious human individual. What I think we need is a story that narrates the Person as an expansion or fulfillment of the Self and the We as an expansion and fulfillment of the I. Otherwise, Brandom is, in his own way, playing the valet.

A few more thoughts on ‘can,’ ‘ought,’ and death

1) FAP: S can do A only if S can do not-A

Why did we affirm FAP?  Because we need a principle that accounts for the logical structure of abilities and dispositions which grant us a use of ‘can’ that predicates over a range of counterfactually obtaining states (and within a range of possible worlds).  Without FAP we cannot speak of dispositions and abilities and without the latter categories, we lose the vantage point from which to reason counterfactually in toto about any thing as it is part of the concept of a unified object, whether agent or non-agent, that it persists and endures through a range of changing states and remains self-identical as it sheds layer upon layer of predicative skin.   FAP gives us the moral agent in tact and as such FAP gives us the venue without which it would be impossible to do moral philosophy at all.

2) K: Ought implies can

Why did we argue that the appropriate conjunct of K is FAP instead of FAO? Because K + FAO implies ‘ought implies opportunity’ but as moral agents are material beings and as such, are sensitive to external conditions, an agent does not have the power to, in every case, see to it that she has the opportunity to discharge her obligation.  Further, if ‘ought implies opportunity’ then it is impossible for an agent to do other than she ought as we saw earlier that the concept of opportunity is logically dependent on the concept of ability and from this taken with the principle of FN – if S ought to A and can A and has the opportunity to do A then S does A – it would make it necessary that no agent ever did other than she ought have and thus, that no wrong had ever been done in the history of human action.  This would render the concept of failure void and with it we would lose the ability to evaluate actual states against ideal states, nor would be able to speak of moral education and improvement. But if moral philosophy is to have a subject-matter than this cannot be the case.  Hence, we affirm the alternative conjunction ‘K + FAP’ and judge to our satisfaction that ‘ought implies ability’.

2)b) There is a further argument in favour of K + FAP: the view that ‘ought implies telos’. Here, we take telos to mean ‘essential purpose’ and see that on this view, ‘S ought to do A’ implies that it is essential to the concept of S that it do A.  If S does not-A then either S is not a S or S has been impeded from or did not have the opportunity to do A.   But if the ability to do A holds for S, then S is an S.

3) If doing A is a telos of S then when S does not-A some failure has occurred. What remains is to distinguish what we can call, following Bernard Williams, intrinsic failure from extrinsic failure. Intrinsic failure describes a scenario when the impediment to doing what S ought to do stems from S herself so that we can render S responsible for not doing what she is obligated to do, or what was her telos. Extrinsic failure describes a situation in which the failure to do A has a source beyond the agent who has failed to do as she should. But to say that the source is intrinsic to S is to say that S did something or failed to do something which resulted in her lacking the opportunity to do as she should. In other words, if S is responsible for not doing A then she was under obligation to see to it that she have the opportunity to do A. If, on the other hand, she is not responsible for failing to do her duty, then we do not cite another obligation or telos of S’s to explain why she failed to do A; instead we cite an external constraint that prevented her from doing what she ought to do or what it was her purpose to do.

4) Ascriptions of blame and moral responsibility will tell us then what we judge the source of a wrong to be. They will also tell us what we judge an agent’s abilities and obligations to be, and thus, tell us what particular speciations of a general genus S falls under or is thought to fall under.  For example, if Jones is responsible for killing Smith, then Jones is a rational being, an adult human, and able to act on her reasons and intentions.  If the source of the wrong is in Jones then she is morally responsible for killing Smith. Now, as we said above, if the source of the wrong is in Jones then it has to be attributable to Jones as Jones or agent qua agent and not qua material body.  A brain tumour is not an intrinsic property but a belief is.  Thus, if S is blameworthy for failing to do A then she was obligated to see to it that she have the opportunity to do A.  But is she blameworthy for failing to see to it that she have the opportunity to do A?  Do we not embark on an infinite regress of obligations if we anchor ascriptions of moral responsibility and blameworthiness to further obligations and abilities?   Though we do embark on a regress it is not infinite.  The regress of obligations bottoms out in a telos. The telos is an ability and an obligation that one discharges and expresses just by having it, or, just by being the kind of thing that has it. If S loses its telos then it is no longer a S. It is because by reference to a telos that we can explain an agent’s actions that we can give an “infinite explanation” (Rödl, citing Hegel) of any and all of its actions, including its failures. In short, a telos names an ability that provides its own opportunity. This is fruitful indeed, for it is appropriate that the only thing to end an infinite regress is an infinite explanation. This is what teleological judgment allows us to provide, and it is nothing short of a grounding condition for our moral judgments about what S does and what happens to her for it is nothing short of a grounding condition for our judgments about what S is.

5) Now, we said that understanding what an agent’s telos is will let us understand what her obligations and abilities qua agent are.  What are the obligations and abilities basic to all forms of agency? Another way to ask this question is to ask what is the content of agency that we can derive purely from its form?   If we want to say, as I think is uncontroversial, that agency requires the ability to act on reasons and for reasons then an essential element of agency will be first and foremost, the ability to consider reasons and to make judgments.   In other words , an agent is a thinking thing. Though it bears repeating for our purposes what the requirements of thought as such would need to be in order to be possible, I cannot provide a transcendental deduction here. Suffice it to say that one can think only if one can make inferences and that one can only make inferences if one can make inferences about something and that this requires the ability to make temporal judgments about a unified object as it changes in time.  If our subject cannot do this then it ipso facto cannot act for reasons since it cannot entertain thoughts. Now, we said also that an agent has the ability to act for reasons in addition to her ability to consider reasons. This implies that she is able not only to differentiate thoughts but also to evaluate them and judge of one that it is better than another. This means that she is able to compare thoughts to an ideal and to derive inferential and motivational force from that thought which most approximates that ideal.   This means that an agent must have an intuition of the Good in addition to an intuition of time. To have an intuition of the Good is to have a preference for some states over others and to feel obligated to do as she prefers. Now, if our agent is not bound by external conditions then she is divine, as she, in accordance with principle FN (S necessarily does A when it ought to, can, and has the opportunity to), has the power to do what she ought since she is free of extrinsic intervention. But since it is the case that states obtain that are not in line with our preferences, we can infer our own finitude and receptivity. Hence we learn that it is not of the concept of agency in general but of finite human agency that S be sensitive to extrinsic conditioning. Here we discover the teleological structure of human agency at the limit point of external conditioning.  If the agent loses the ability to think, to prefer, to act on preferences and to have the content of her judgments partially constituted by external states that obtain through her receptive faculties, then she is no longer a human agent.

6) We have seen that not all states of human agency are preferable and that it is not within an agent’s power to guarantee that all of its states will be those which it finds Good.  Is agency itself therefore Good? Agency itself is judgeable as good or bad only if it can be judged against an ideal and that ideal, seems to be, divine agency, which as we saw, was the power to bring about states in perfect accord with one’s preferences. Now, if divine agency is impossible then the only alternative to agency is death, which is the null state and the absence of all judgment, including possibility. But if divine agency is possible then the alternative to its actualization is the polar opposite ideal, which would be one wherein an agent has no power at all to bring about states that are in line with its own preferences.  This is certainly a picture of hell.  Given that it is possible for an agent qua finite agent to find itself in a hellish or near-hellish state, it may seem rational for it to reject its own existence as an agent and instead opt for death. Even though, strictly speaking, when the agent opts for death, she opts for the Good (since this is the only way – representing A as preferable to B – she can opt for anything) the intended outcome is thought to be pure void, i.e. an impossible state.

7) On what grounds does the finite human agent infer death as a genuine possibility? On what grounds does she choose death as a genuine possibility? Is there something in us, some telos, that is not thought, that is not life? I phrase the question this way because the speculative possibility of suicide seems to fly in the face of FN if we refuse to see suicide as opting for the Good. Here, the suicide chooses the void state as if it were the better state of a disjunction but there is a difference between choosing death because it represents the Good and choosing death because it represents the void. The former seems to represent death as being akin to a dreamless sleep-like state; the latter devests it of any representation at all. Kant’s prohibition against suicide only makes sense if one is already a proponent of life and agency, but if one longs for life without agency one can opt for animal or vegetative consciousness with for example, drugs and certain meditative states, even if only temporarily. And if one opts for nothingness, although one may not be entitled to do so when judged as an agent what if one is judged as a mere entity? Insofar as consciousness is not synonymous with agency, it seems we as finite agents have the right to opt out of our own agency since agency is afterall, something we have been thrown into, by luck, by fate, by society and so on. But this means finite agents are the only entities whose telos is optional, that is, who have no obligation to be what they are. Is this because we are also void?

More to think about…

Notes on Haji’s “Deontic Morality and Control” ch. 8

In chapter 8 of “Deontic Morality and Control” Haji presents, perhaps here more forcefully and persuasively than in any of the previous chapters of his book, a case for his asymmetry thesis. 

Moral responsibility requires dual-control (ability + opportunity) over S’ s doing A whereas Deontic Anchors require only singular control (ability) – in an indeterministic universe – to do A.
Here, Haji adjusts the asymmetry to make room for a further principle that he introduces, namely, the “authenticity requirement”.

To motivate the thought, Haji places the hypothetical agent (who we’ve been calling ‘S’) in a situation of absolute manipulation at the hands of the ‘Psychohacker’ Max. Max grants her the ability to act in accord with her own will and according to reasons which she endorses but so arranges the case that S acts precisely as Max wants her to act. Under such constraints, S is not, Haji believes, morally responsible for acting as she does. Since S satisfies all the conditions for free will and responsibility stipulated under compatibilism and modest forms of Libertarianism,  Haji thinks Psychohacker reveals the need for a further condition that must obtain in order to secure freedom and responsibility.  This condition has it that in order for an agent to be morally responsible for an act A, it must have stemmed from principles and a will that are “authentically” her own. Under Psychohacker,  her will and rational principles are inauthentic since they perfectly mirror Max’s and thus express nothing at all about what S (as S) morally stands for. Haji’s thinking here is also sophisticated in that it is not solely because S is Max’s “marionette” that she is not morally responsible for her actions but because she has unwillingly become Max’s marionette.  That is, had S chosen to be manipulated by Max in this way, her actions would then spring from and express her own authenticity as a rational, moral agent and as such she would be responsible for doing as Max has her do.  What this tells us is that an agent is morally responsible for doing A/not-A at t only if she at t0 has an obligation to see to it that she has the opportunity at t to do A/not-A. Haji doesn’t put it this way in this chapter (perhaps because he is afraid it will lead to an infinite regress, though it does not) but his authenticity condition harmonizes nicely with our reasoning here.

Now, though S, under Psychohacker is not morally responsible for doing A, Haji maintains that her act of doing A is open to assessment of rightness, wrongness and obligatoriness.  Since it is prima facie plausible that S can still do wrong and can still do what she ought not to given that, as she acts, she does so using principles and reasons that she willingly endorses as her own and so has proximal control over, Haji takes the Psychohacker scenario to exonerate his asymmetry thesis.  He concludes that this makes sense because deontic ascriptions describe acts while ascriptions of moral responsibility describe agents. (In fact he says that the former are “act-directed” while the latter are “agent-directed” though how obligatoriness can describe an act and not be binding for any agent under any circumstances as would be the case under Haji’s determinism, I do not understand). 

An alternative and I think better way to parse the asymmetry is to say the following:

Deontic anchors apply generally while ascriptions of moral responsibility and blameworthiness apply particularly.  The sentence is awkward but the thought is not. Insofar as I am a moral agent I am bound by duties that apply categorically across the entire landscape of agency. However,  insofar as I am this agent and not another I am responsible for fulfilling my categorical duties to the extent that I have had the opportunity to do so.  And if I have the ability and obligation to see to it that I have the opportunity to discharge a further duty then I am responsible for having or failing to have this opportunity. 

In the Psychohacker scenario,  S is under no obligation to see to it that she have the opportunity to resist the covert hacking of her mind by Max and so, she as this S will not be responsible for the wrongs she performs under Max’s control.  However, we need not say that the post-hacked S is devoid of blame or responsibility as long as we stipulate that this S is in a fundamental sense, a different agent. Historiscity of reasoning seems therefore to capture the spirit of Haji’s argument from authenticity. 

But if this is true, then would an originary case of psychohacking, imagine a psycho-hacked baby at birth, for example – who does not have a history of reasoning – be responsible for what it does later in life?  If not, then we need to be able to say why such originary intervention or manipulation differs in essence from cultural indoctrination and socialization under which the baby will later be held accountable for its actions. But this is for another time…

Nevertheless,  the asymmetry thesis holds and Haji has given an intriguing, convincing argument.

Notes on Haji’s “Deontic Morality and Control” ch. 7

In chapter 7 of “Deontic Morality and Control” Ishtiyaque Haji develops the notion of the Contrastive Explanation Condition (CEC) to show why he thinks Kane’s Indeterministic Libertarianism fails to underwrite any conception of moral responsibility and free will “worth wanting”.

CEC states that for an action A to be within an agent S’ s free and responsible control S must be able to cite a causally efficacious reason for her choosing to do A over B that would not play the same causal role if, in a counterfactual world, S had chosen to do B instead of A.

Kane’s account of free will does not seem to satisfy this criterion. The reason being that because Kane wants to insert indeterminism between deliberation and execution (rather than, say, between nonoccurent and occurent belief formation like Mele does) the Kaneian agent will have no way to give a causally contrastive account of why she chose to do A rather than B. This follows from the fact that on Kane’s account, the input that ‘flips the switch’ and triggers the action is pure chance, all the weighting and mental effort from S’ s antecedent reasoning not being by itself enough to tip the scales one way or another.

Haji’s critcism is that this view of agency succumbs to the luck objection.

The luck objection, recall, states that if luck or chance causes S to do A rather than not-A then S is not morally responsible for doing A.

I’d like to rehearse a possible Kaneian response to Haji here: the Kaneian may object by pointing out that the agent’s responsibility is, by the time a course of action is finalized, already represented not in either of the alternatives taken on its own but rather in the disjunction itself. The agent is, in other words, responsible insofar as it has come to this garden of forking pathways and not another.

In this sense Haji’s Deweyian point that an agent expresses herself objectively in her actions would bear emendation: it is not what an agent chooses to do that represents her moral character so much as it is the live options she perceives as open to her, when taken together as a totality, that display what she morally stands for.  By the time luck enters the picture the agent’s moral character – up until the moment of action – is already in full disclosure: S is the kind of person who has equally forceful reasons for doing A and B.

This need not be as implausible as it may seem. For it is surely only very rarely that we find ourselves with truly equal reasons for doing one action or another and it is, on Kane’s account, only in these cases that a causal space is opened up for luck take the wheel.  To be sure, it is not that all of an agent’s actions and decisions will demand the kind of soul-searching deliberation that Jones exhibits as he makes up his mind whether or not to quit smoking.  The reason Jones cannot give a contrastive account for deciding to smoke rather than to quit is because there is no extant account available to him at t; he is thus entitled, given his reasons, to keep smoking and to quit smoking.  Another way to grasp the thought here is to say that chance only ‘has the chance’ to determine what S does at t because and insofar as S has no obligation at t to choose A over B.

We can put this yet another way: S only ought to do A if A is what S ought to do insofar as she is under obligation to be SA and not SB. But in a self-forming action (SFA) such as Jones’s, S is compelled by two alternative and mutually exclusive obligations: to be SA or to be SB. And this cannot be decided based on S’s past up until time t.

As Kane says, the justification for an SFA is grounded in the future not in the past.

So, even though for Jones it is true that the choice to smoke or not to smoke is, given his character, a disjunction with equally good and forceful reasons on either side, this is not going to be the case for most other smokers who find themselves entertaining the thought that maybe, just maybe they should quit.

Smith for instance, might consider it virtually inconceivable to quit smoking and thus, it is not the case that he has the opportunity, as Jones does, to do A (continue to smoke) and not-A (quit smoking). Rather, he simply is not deliberating about his addiction here. When Smith, after entertaining the possibility of kicking the habit, proceeds to put his Gambier pipe back in his mouth, he is not deciding between the same A or not-A that Jones is.   He is not, contrary to how it may seem, renewing his original decision to keep smoking.  Smith is not taking the side of the disjunction that Jones rejected.  Smith and Jones deliberate over two separate disjunctions. Since quitting is not a live option for Smith neither is “reaffirming his decision to keep smoking”.

That is, in Smith’s mouth the proposition “I should quit smoking” is Moore-paradoxical; it has the form:

p; but I don’t believe that p

In such cases, p is not an assertion. It is not a judgment. As such it cannot be an action.  Unlike Jones, what Smith is deciding here is not “to smoke or not to smoke” but what his reasons (or justifications) for smoking should be.  Given this, Smith at this moment, might be blameworthy for giving disingenuous reasons for his smoking when he knows that there is really one reason that explains his inability to quit, viz., his addiction, but this act and the agent’s connected blameworthiness should be adjudicated separately from the fact that he may or may not be guilty for getting himself addicted in the first place.  Definitely, Smith may be responsible and blameworthy for smoking to begin with. But this is to be decided by examining Smith’s reasons for smoking at the time before he became addicted. In contrast, we need to be able to judge Smith’s moral worth at t, independently of what he may be guilty of at t0. At time t, he is blameworthy for pretending that he smokes, say, “because life is not worth living without my Gambier pipe” when he should admit to himself instead that he smokes because he is powerless to stop.

If Smith really has no contrastive explanation for why he chose to justify his smoking by citing the enjoyment he derives from it rather than choosing to justify it by citing his addiction then regardless of whether or not chance intervenes and he chooses A over B, we can judge Smith’s moral character by the fact that these two options, for him, have equal rational persuasion, (and for the record, it sounds like he has a lousy moral character).    This should only feel counterintuitive if we forget that such a disjunction as Smith’s is highly unlikely for most rational smokers and even for those pesky rationalizing ones.   In most cases, it is perfectly obvious why one does not quit smoking: because one is addicted.  Thus, it is only for the select few that the kind of puzzle Smith finds himself in is awakened to agentive deliberation.

All of this is to say that even if indeterminacy enters the picture where Kane invites it in – between deliberation and action – we can hold an agent responsible for their actions insofar as these actions express a disjunctive bind that the agent has ‘given herself’ (in the Kantian sense of giving oneself the rational law), a bind that is itself a function of that agent’s reasons, beliefs, efforts and moral character.

As for the action itself – Jones’s decision to fight the addiction, and Smith’s decision to rationalize it – these can be contrastively explained by future reasons made available to Jones and Smith once indeterminacy has had its say.   If we ask Jones and Smith after the fact, “why did you choose A rather than B?” they can say first, “because, at the time, I was under no obligation to do one and not the other and as such, I take responsibility for both,” and secondly, “because, now that I have acted, look at all the good things that came about (or now I see were reasonably likely to come about) given my choosing A over B.”

Now, I want to highlight one thing about this account that I think is highly implausible: the idea that this kind of scenario paints an accurate picture of what happens the majority of the time that we are acting.  The situations in which an agent will have equally forceful reasons for doing this as she has for doing that seem extremely few and far between, even over an entire lifetime.  Most of the time, it is our reasons, beliefs and past history that determine what it is we must and will do and whatever input chance has is negligible precisely because these, call them ‘rational inputs’, are so forceful.

Returning to the example from Kane’s paper:

If it is true that the businesswoman really has no better reason for choosing to help the assault victim than she does for choosing to continue on to the career-defining conference, then it is a matter of luck that she resolves to do the former.  But if this is the case, then she is no more or less blameworthy or responsible than she would have been had she chosen to walk on by.  The alternative is to say that indeterminacy plays a role in her action here but that it is likely not going to figure in her decision but in her opportunity to act on her decision or perhaps to come to a decision in time to act.  That is, the businesswoman may find herself continuing on to the conference in spite of herself, in spite of, that is, her certainty that she is obligated to help the victim.  That is, she may do B even though she knows she should do A.  Here the agent is blameworthy in a way that she cannot be in Kane’s original

Notes on Kane’s “Responsibility, Luck and Chance”

In “Responsibility, Luck and Chance” Robert Kane offers a penetrating, insightful critique of the Luck Principle (LP) which states that: if an action is undetermined at time t then its happening or not happening at t would be a matter of chance or luck and would therefore not be a free act or one that an agent could be responsible for.

Kane’s insight is that the rhetorical power of LP trades on an ambiguity regarding the source of the indeterminism involved in the execution of the action. Kane argues persuasively that if the agent herself (or more pointedly her own will) is the source of the indeterminism that precipitates her choosing to do A rather than not-A then it follows without a hitch that she acted freely and is responsible for so acting. This is brought out intuitively when we consider such a case of internal indeterminacy against a case wherein the indeterminacy is external to the agent’s will.

Where the indeterminism is internal, Kane asks us to ponder the businesswoman who notices an assault taking place on the street while she is on the way to a career-defining conference on which her professional success depends. She must deliberate, if only for a moment, what to do: to help or not to help the assault victim. Here, Kane speculates that the agent’s feeling of being torn between two conflicting choices expresses a conflict internal to her own will: as she deliberates what to do she is prevented from continuing on to the conference by the ‘noise’ and distortion clouding her judgment from the force of the compulsion she feels to help the victim and vice versa. She wants to help the victim but she cannot ignore her commitment to her career which, bearing its own significant moral and/or non-moral motivational weight, is something she also wants to do. If a brain scientist tells her after the fact that she chose to help the victim due to an indeterministic process in the brain and as such that she did not act freely, Kane insists that the businesswoman should balk at this suggestion since it is precisely an expression of her own character that her choice was indeterministically caused in this way.

Contrast this case with one in which the indeterminacy is external:

The husband means to smash his wife’s favourite glass table but fails to do so due to insufficient momentum in his fist when it impacts the tabletop. Assuming he otherwise has the strength required to break the table, his failure to do what he intends is due to indeterministic processes in the neural pathways controlling the force with which he hammers down his fist. Unlike the businesswoman, he is not in this case, torn between the disjunction: to smash or not to smash.

Nevertheless, Kane maintains that the husband is not off the hook here just because the table did not break. He is still responsible for attempting to smash the table and he still acted freely in doing so.

The essential difference between the two cases is not in the nature of their outcomes; it is not relevant what the consequences of either the businesswoman’s or the husband’s actions turn out to be. What is relevant is the fact that it is only in the businesswoman’s case that her action is a “self-forming action” or SFA (Kane, p. 230) while the husband’s is not. In an SFA, an agent’s failing to do A is not the result of accident or a lack of volition or mistake but rather a direct result of her succeeding in doing not-A. In an SFA, an agent expresses herself, not in spite of being internally torn, but precisely in virtue of this self-conflict.

Though the final decision may be caused by an indeterministic process akin to the rolling of the dice, the fact that the faces on the dice are a function of the agent’s own character makes the final decision itself a function of that same character. The freedom is represented in the fact that it was these dice that were rolled rather than some other. Thus, the businesswoman is free whether she helps the victim or hurries on to the meeting.

Both choices express her character which is itself a function of her previous SFAs.

The version of the Luck Principle (LP*) that would be corrosive to SFA-relative free will would have to stipulate that the indeterminism that causes, for example, the businesswoman to help the victim rather than to continue on her way to the conference has nothing (or next to nothing) to do with her reasoning at the same time in favour of going to the conference. That is, those who argue that chance and freedom cannot live in the same world must show that the faces on the ‘dice of action’ are largely shaped at random and do not reflect to any relevant degree the agent’s past choices, convictions, beliefs etc. But LP* cannot show this and neither does our best empirical science.

However, I would like to comment briefly on the above conclusion:

What if we devise a version of the Luck Principle, call it LP**, that stipulates:

if an agent’s original SFA is caused by a process that is more or less indeterministic (or beyond the agent’s proximal control) and as such is the result of luck or chance it follows that it cannot be a grounding source of free will.
Does this spell trouble for all of Kane’s work?

I think the answer has to be a cryptic “yes and no”.

“No”, because we’ve been working with an implicit definition of ‘freedom’ that views any act as free insofar as it stems from and expresses one’s past SFAs (in the husband’s case) or is itself a ‘live’ SFA (as in the businesswoman’s case) and something like an original SFA would seem to satisfy these criteria. A child, Jack, may decide to do his homework to please his mother instead of playing outside and either of these options would be free choices on Jack’s part even though it might be the case that he has these desires – to please his mother and to play – simply in virtue of being a socialized human being and not the result of any past deliberations of self-formation. That is to say, he need not have chosen these desires in order for them to be his own.
But it seems to me that LP** (the view that our choices are at bottom the result of luck and thus jeopardize our capacity for freedom and responsibility) might require an affirmative answer when an agent is born or thrown into the world with a nature or character – i.e. ‘faces of the dice’ – that expresses an inclination towards what we would want to call immorality or ‘evil’. Jack’s twin brother, Zack, may, it is conceivable, find himself torn between two different desires, neither of which developed out of previous SFAs. He may find himself, like his brother, deliberating about whether to do his homework or to play outside but may find that he chooses the latter because he takes pleasure in the knowledge that, by playing outside, he is causing his mother anguish. Zack’s counterfactual counterpart might instead decide, all things considered, that he should go inside and do his homework with Jack, but not because he wants to please his mother, but because he is afraid of punishment. We need not assume that Zack is a victim of abuse nor of poor parenthood to make sense of his reasoning. The child simply rolls the dice of action and expresses his own character in doing so and it seems he is responsible for his wickedness in the first instance and cowardice in the second even though it is not grounded in any previously decided SFA.

All this is to say the following: unless we are willing to live with an infinite regress of SFA’s built upon previous SFAs – which, given our human finitude, beggars comprehension – it seems we must bite the bullet and ground the chain of freedom in a ‘raw’ or original nature that an agent is responsible for but did not form self-consciously. Because the ingredients that go into the original SFA are an expression of contingent circumstances (e.g. genes, environment, blind chance) it would appear that Jack’s good will is clearly the result of good luck while Zack’s wickedness is clearly the result of bad luck. Even if we exonerate Zack in this case (given his young age) and instead hold his mother and his society culpable in forming his character, we cannot regress forever. At some point in human prehistory, when the first self-forming action was performed, it seems a matter of luck that a Jack resulted instead of a Zack. That nature itself produces circumstances congenial to the development of goodwilled persons may be an example – the example – of Kantian beauty. But this is a thought for another day.

One more thing I think is worth mentioning about Kane’s paper is that his account of free action is sophisticated enough to survive in both and indeterministic and deterministic universe.
This, in my opinion, is a sign of a moral philosopher who is on the right track…

Getting clear on ‘can’

In this entry I argue that the following principle has paramount importance in moral theory:

FAP: S can do A at t only if S can do not-A at t.

In taking up the defense of such a principle I hope to bring out the fact that there are, broadly speaking, two senses of the word ‘can’ that need concern the moral philosopher and that understanding how these two senses are applied in various cases (and misapplied) will shed equal light on some of our most belabored ethical puzzles and on our deepest moral intuitions. FAP is, I argue, the key to distinguishing the two senses of ‘can’ we need to hold distinct in order to do moral philosophy.

The first thing to notice is that ‘can’ plays a pivotal role in how we conceptualize the lion’s share of our moral principles. It is part and parcel for example, of our concept of obligation that ‘ought implies can’. The same goes for our judgments of moral responsibility and blameworthiness. It is not clear, however, just what function ‘can’ plays in these principles nor whether it plays the same function in each case. ‘Can’ is an ambiguous term. In this essay, I want to suggest a method for disambiguating these senses of ‘can’ in a way that preserves ascriptions of both moral responsibility and deontic anchors. It is my contention that developing an understanding of FAP is crucial if we are to understand how to adjudicate our moral discourse.

In defending FAP, I place myself in opposition to those who argue that Frankfurt-type cases (henceforth FTCs) demonstrate it to be a patent falsehood that for an agent to be able to do some action A, she must also have it within her power to do otherwise. Specifically, it has been held by some philosophers (see for example, Harry G. Frankfurt, Moral Responsibility and Alternate Possibilities, Michael J. Zimmerman, The Concept of Moral Obligation, and Ishtiyaque Haji, Deontic Morality and Control) that FTCs render FAP false in virtue of the fact that the presence of a counterfactual intervener ensures that the agent has exactly one course of action open to her despite her acting on the supposition that she has, at the time of action, some number of available alternatives. The thought is that if S does some action A, at time t, believing all the while that she could have done not-A, her doing A and only A is, nonetheless, fully determined insofar as any genuinely accessible world wherein she would have recourse to some possible alternative is ruled out tout court by the positing of the counterfactual intervener. We might say that FTCs render S’s choice at t ‘saturated’ by A for the fact that any garden of forking pathways available to S at t – n will, at t, collapse into one and only one available future: S does A.

Now, assuming for our purposes here, that FTCs are actually possible – (there would be good reason to think that they are once we understand that and how actions can be overdetermined not because the beliefs that cause them supervene on physical states but because its is likely that the same action can be caused by a variety of beliefs in the form of reasons. See Eric Marcus Rational Causation, and Crawford Elder Familiar Objects and their Shadows) – it is clear that, wherever FTCs apply, in some commonsense idea of ‘can’, the sentence ‘S can do other than A’ is undeniably false. This is the notion of ‘can’ that is modally rigid, for it applies only to actually obtaining states of affairs insofar as they do obtain or to the extent that it is known beyond any doubt that they will obtain. Following Haji’s terminology, we can call this sense of ‘can’ ‘opportunity’.

To be sure, it is clear from the logical structure of FTCs that FAP cannot survive if we take it to mean that

S has the opportunity to do A only if S has the opportunity to do not-A

But there is another sense of ‘can’, one which I would argue is more fundamental than opportunity and one that can withstand the acidity of FTCs.

In his thought experiment of the fallen plate J.J.C. Smart articulates a notion of ‘can’ that is applicable even in cases where circumstance fixes it that one and only one of a range of possible states of affairs obtains. If I drop a plate on the floor and fortuitously find that it does not break I am perfectly entitled to say to myself that it ‘could have’ broken. When I say this, I express a judgment about the plate’s disposition or ability to break, and it is in this sense that the alternative possibility, that of the plate’s breaking, is represented, even if, as in the case of FTCs, this ability never finds opportunity for expression.

‘Can’, in this more supple sense, is a modal term that predicates over a range of counterfactually obtaining states. It is, for this reason, invulnerable to negation by the sort of counterfactual intervention that is represented in Frankfurt examples. We do not want to say that the plate is unbreakable, not like a tin plate is, or a diamond is, even in the presence of Frankfurt’s resourceful intervener.

If we read FAP as implying this second sense of ‘can’, a sense which we will call ‘ability,’ then we secure a principle that can ground our understanding of the differences and connections between moral concepts such as obligation, responsibility and blameworthiness by linking them to the parallel differences and connections that hold between ability and opportunity. For the remainder of this piece, I will be using ‘can’ to mean ability, unless otherwise stated.

We said earlier that ability is more fundamental than opportunity. What this means is that it is more conceptually basic than its rigid sibling and as such, ability acts as a grounding concept for opportunity. To put it simply, opportunity only makes sense in the context of an ability. A plate has the opportunity to break only insofar as it is able to break. An agent has the opportunity to perform an action only insofar as she has the ability to perform that action or, formally,

OA: S has the opportunity to do A only if S can do A.

It makes no sense to say that an agent has the opportunity to perform brain surgery if the agent cannot perform brain surgery just as it makes no sense to say that the jello has the opportunity to shatter when jello cannot do this. Thus, the converse of OA, call it

AO: S can do A only if S has the opportunity to do A

is false. From these reflections we demonstrate that while opportunity implies ability, ability does not imply opportunity.

Consider the fact that the opportunity analog to FAP

FAO: S has the opportunity to do A only if S has the opportunity to do not-A

is certainly false (at least in FTCs). It is made doubtful by the fact that S does A at t but does not do not-A at t. If FAO were true then FTCs would not have the requisite force to bind the agent to one particular action since it is the job of the counterfactual intervener to bar S from having the opportunity to do otherwise than she does. Another way to put this is to say that FAO, were it true, would render the difference between ability and opportunity void. But accepting this comes at the high cost of denying any sense to the notion of ‘can’ that refers to disposition. And we have already seen that if the latter goes, so does opportunity go with it. It would seem less costly to simply affirm the fact that while ‘S does A at t’ does not entail ‘S has the opportunity to do not-A at t’ (which is refuted by the fact that S does A at t but does not do not-A at t) ‘S does A at t’ does entail that ‘S can do not-A at t’, holding fast to a notion of ‘can’ that predicates over a range of counterfactual alternatives and as such, is FTC-proof.

Now that we have made explicit the conceptual differences between these two sense of ‘can’ as they relate to non-moral cases, we can employ them in our study of our major moral concepts. First, we will take, what many believe to be axiomatic, the Kantian ‘ought implies can’ principle which states that

K: S ought to do A only if S can do A

In line with our preliminary findings regarding ‘can’, we may take K to either mean ‘ought implies opportunity’ or that ‘ought implies ability’. If we assume the former then the agent has no obligation to perform an action if she does not have the opportunity to perform that action. Under the surveillance of the Frankfurt intervener, this renders it impossible for S to do anything she ought not to as FTCs stipulate that she never have the opportunity to do other than she does. If oughts corresponded to opportunities than there would be no room for failure and no possibility of moral infraction. If, on the other hand, we take K to refer to ability we do not saddle ourselves with such an absurdity. For it may seem perhaps harsh to hold S under obligation to perform an action that she does not have the opportunity to perform, but we need only stipulate further that in such cases, she is also obliged to see to it that she have the opportunity required to dispatch her duty.
Furthermore, if we add to K the following principle

FN: if S ought to do A and S can do A and S has the opportunity to do A then necessarily S does A.

this gives us a rule for discerning where failure occurs in the process if S does not-A.

Since ought implies can, then if S ought to do A then S can do A. Therefore, we get a disjunction:

either S does not-A because she lacks the opportunity to do A or S does not-A because she lacks the obligation.

The third option, S does not-A because she lacks the ability to do A is ruled out of hand by FAP, since if S lacks the ability to do A then necessarily ~’S does A’. But from FAP we get from ‘S does A’ also S can do not-A Thus, if S does not-A, necessarily S lacks either the opportunity or the obligation but never the ability to do A.

How do we decide whether the lack is on the side of the ought or on the side of the opportunity?
If S lacks the obligation to do A then no wrong has been committed when S does not-A.

If S lacks the opportunity to do A it is not the case however, that no wrong has been committed by S’s doing not-A. Here S fails to fulfill her obligation and commits a wrong. (From K and its partner OW: S ought to do A only if it is wrong for her to do not-A).

Is S blameworthy for doing not-A in every case that S ought to do A?

Here is my answer:

No. S is not blameworthy for doing not-A in every case that S ought to do A.

Here is my reason:

S is blameworthy for failing to do A at t1 only if S has, at t0, an obligation to see to it that she have the opportunity to do A at t1.

This renders:

S is blameworthy for doing not-A at t1 only if S has, at t0, the ability to see to it that she has at t1, the opportunity to do A.
Blameworthiness and moral responsibility occur only when the doing of a wrong at t1 is traceable to the doing of a wrong at t0.

If at t0, no wrong was committed then at t1, S is not blameworthy nor morally responsible for doing not-A.