Free Will Zombies

Here're the first couple pages of a paper I presented at Manuel Vargas' excellent USF-hosted conference a year and a half ago; the rest may be found here.  But this is enough with a little exposition to make my case.

***

"Libertarians were understandably jubilant when the invention of the I-detector was announced.  This device detects in people the presence or absence of an indeterministic process that—as it turned out—always and only attaches to ordinary capacities for reason in any exercise of choice, and thus “effectively establishes a powerful correlation indicating the presence or absence of libertarian free will” (so said the Surgeon General of Bioethics on a label eventually affixed to the device).  In other words when the detector shows that indeterminism is absent by producing no reading, effective reasoning capacity is absent too (as empirically demonstrated), so it maps the familiar territory of the presence or absence of basic-desert-bearing responsibility in libertarian terms.  Compatibilists in particular grumbled that this demonstrated strict correlation of indeterminism with reasoned choice did not yield a plausible mechanism for how it works, but libertarians just snapped back that Humean causation and laws hadn’t troubled them much before—why should similar rigid correlations be worrisome now?  Admittedly there were still many conferences headlining the back-and-forth between event-causal libertarians and their agent-causal cohorts because the I-detector data was epistemically compatible with both, but not many conferences were funded for hard incompatibilists or compatibilists or free will skeptics, though they still gathered together often for informal symposia (in the Platonic sense).

One product of this discovery was a re-energized emphasis on retribution in both personal and public policy matters.  Critics complained that even the granted truth of libertarianism was conceptually insufficient to justify retributive practices, but advocates shoved I-data back in their faces as a plausible basis for a claim to moral ultimacy as a necessary condition for separating moral wheat from chaff.

So things continued in this way, and the I-detector proved increasingly indispensible for matters of justice.  Then something unexpected happened:  evolution threw a wrench in the gears with a mutation.  Over time a population appeared who were outwardly and (as far as self-reports can tell us anything) inwardly just like people with libertarian free will, but the I-detector yielded null results when trained on them.  Of course initially these individuals were prejudicially thought to lack ordinary capacities for reasoning and treated accordingly.  However, it eventually became apparent that that was not the case:  double-blind studies proved that these individuals were indistinguishable in reasoning capacities from those with libertarian free will.  They pondered choices in the same kinds of way, sometimes choosing well, sometimes ill and in comparable proportion to libertarians.  Naturally a derisive term emerged to describe them.  They were ‘free will zombies.’

Of course the I-detector became useless when attempting to assign libertarian-style ultimate responsibility in any practical sense to free will zombies, especially when it became a tactic for some more unsavory zombies to feign insanity in order to escape retributive blame with the assist of the I-detector.  But this was also something of a philosophical windfall for those few struggling compatibilists and hard incompatibilists and free will skeptics left (usually employed as baristas rather than academics since such non-zombie ‘free will flat-earthers’ chose to deny obvious truth ironically of their own free will).  They questioned whether in cases where libertarians and zombies behaved identically by all standards of assessment except for different I-detector readings, would it be the case that retributive treatment (based primarily on incompatibilist grounds) is fairly applied to zombies, since they lack what had become an accepted feature of praise- and blameworthiness?  Would two systems of justice (perhaps modeled on the juvenile/adult systems) be necessary for equivalently rational defendants who were precisely distinguishable in just one way—the presence or absence of indeterministic I-properties?  Or, were distinguishable zombies really people worthy of retributive concern?  And perhaps more pertinently some whispers grew into a shout:  was the system of retributive justice based on libertarianism somehow faulty from the get-go? "

***

As the paper further lays out, this is a possible-world-crunching-together of a best-case epistemic foundation of libertarianism (given what we know) supplied by the I-detector (analogized with cancer/smoking epistemic grounds available even circa 1964) along with a stipulated clean-case of emergently indiscernibly thinking-and-acting and thus only I-contrastively  free will "zombies" to see what we might conclude about agency and responsibility.

The issues I think play out thus:

(i) Are there any purely metaphysical grounds to elevate the status of agency of Ls over Zombies aside from ad hoc screening strategies related to the I-detector that would move Zombies into the same category as Ls or advocate agnosticism about Zombie agency on some similar basis (I mean by focus on the device in terms of operation, reliability, finks, masks, etc.)?  I mention these latter strategies--and not in the paper though maybe I should--because I think they are only diversionary to avert one's moral vision about the salient issues of Ls versus zombies

(ii) Assuming then that the way Ls and Zombies work as revealed by the I-detector won't do the job of establishing some differential of agency, then what's left to distinguish them?  There are axiologically stipulative ways to do so.  I-fingered Ls are thus just on the one hand taken to be real, ultimacy-and-desert-bearing responsible moral entities, and Zombies are something less.  Or, alternately, Zombies are sufficiently non-ultimacy-and-yet-adequately-desert-bearing moral entities, and Ls are perhaps something more--though not morally so.  But this means that one assigns values to what are taken to be morally relevant features and denies that other features--even ones empirically grounded, whether features of reason or I-properties--are necessary or sufficient to assigning responsibility.

(iii) (i) and (ii) involve two different kinds of factors playing into opposing strategies of argument that finally might all well end in Fischer's familiar dialectical stalemate.  But the thought-experiment is about a possible world that either somehow tolerates such stalemates theoretically, and results in Buridan-ass inaction as favoring neither position as justified, or places a higher value on having some meta-value of practicability--something that one can justifiably do in the face of thus-assigned lower-order conflicts.  The ultimate meta-value of doing something rather than nothing is what the Zombies tale pushes:  should dialectical stalemate of positions on a comparable level of metaphysics with attached values be paralyzing for recommending action, or should we mount up the stairs of inquiry to ask what would constitute a satisfactory place for recommending sufficient values for action based on salient features that are common to dialectically opposing views assessed at some lower level?  My story attempts to push that meta-value envelope:  should we treat Zombies the same or differently from demonstrably indeterministically choosing Ls?  I can't imagine any non-prejudicial grounds for treating them differently.  (And as my paper argues, differential treatment in that possible world that favors less punitive treatment for Zombies (ala juveniles in our own current systems of justice) would no doubt evoke loud cries from Ls as otherwise completely empirically equivalent save the I-detector difference which stands behind L-retributivism; Ls might thus argue unequal harsher treatment on insufficient grounds--absent some distinct account of how I-properties indisputably add-in to responsibility.  I would note as well that thus self-interested Ls in that world would probably not endorse the Consequent Argument as setting them apart from Zombies--which tends to show as well that the CA itself is not purely a logical/metaphysical argument, but as a putative thesis about determinism involves assumed values prejudiced against determinism, which the Ls in my Zombies world might well under duress renounce. )

(iv) If one puts the push and shove of higher-order pragmatic values against the ones axiologically and metaphysically dialectically stymied in the Zombies world, I think second-order utility of practice axiologically defeats any sort of position at the lower level.  Of course that assessment itself is a pragmatic one.  In addition, if higher-order pragmatism is right (most workable) it might even challenge equally higher-order concerns about constitutive luck, ala Levy, and Waller's global skepticism of the conceptual sufficiency of what constitutes desert and the like,  though I have a lot more thinking to do about all that.  But my argument here is that despite the world-bound evolutionary-source-disparities of Ls and Zombies with regard to significant moral action, there is a sufficiently shared empirical ground in the uses of reason with respect to action by both parties at the zombie-world level to hold at least some of both camps responsible for what they do.  Therefore I'd argue that pragmatically assessed, the Zombies tale takes L-based retributive justice off the table as a preferred option thus higher-ordered assessed.

I think (iv) is right (again by pragmatic standards of truth and argumentative assessment), and even advances something like Vargas' perspective of FW/MR revisionism, though this is not by any means offered as representative of his own approach either by strategy or result.  So Manuel, please accept this as an amicus brief.

Comments

[David Duffy]

So do you see this as a reductio for p-zombies? Or is it, as you hint, a model for real world interindividual differences, where one set of people (adolescents) might lack insight that they are missing a particular ability necessary for self-control?

[V. Alan White]

Hi David and thanks for the question.

My story really only attempts to focus on and enliven the sort of dialectical stalemate between incompatibilists and compatibilists that we often see in the literature by having two almost identical beings co-habit one world with only one epistemically justified distinction: one group involves indeterminism in making choices (the Ls) and the other doesn't (the Zombies; dubbed by the Ls originally as a disparagement). What I'm looking for is some factor that would favor treating either party as really something more or less free or responsible. Frankly, I don't see any that simply flows out of the narrative. As I suggest, treating the Zombies more favorably (like adolescents) would probably result in claims of reverse-discrimination by Ls who think and act in exactly the same ways. The empirically founded indeterminist I-properties seem incapable of supporting (say) retributivism for only Ls. Given all that, I think some sort of pragmatism about what matters for both Ls and Zombies--reasons, identification with acts, expressions of deep selves or whatever--is the only viable solution. As I say in a footnote in the paper, Neil has told me that he thinks this is an argument for compatibilism, and I agree in part. It argues for moral compatibilism (a version of the second disjunct of (1') in my previous post), but doesn't commit to any favored version of FWx, even compatibilist FWb. On such grounds Ls and Zombies may be treated identically morally, even if they metaphysically work in different ways.

Hope that helps. Ok, I plead guilty to deliberately using zombie-talk to engage the reader, but please note that this is *not* p-zombies in Chalmers' sense.

[Paul Torek]

Well, I liked it! Although I don't think most compatibilists would be troubled by indeterminism, as long as it doesn't disrupt the link between reasons and acts too badly. Hobart might be an exception, but he just went a little overboard.

I'm not sure why it's "pragmatic" rather than "axiological". It seems to me that it's *morally* wrong to treat Zombies and Ls differently in the justice system.

[Angra Mainyu]

Hi Alan,

The I-detector assumes there is a way to empirically establish that a process is indeterministic. I'm not sure about that. But assuming there is a way and the I-detector works as explained, in light of the fact that Zombies and Ls make similarly rational decisions, seem to behave in the same way, etc., I think a more probable interpretation of the data would be that the indeterministic process detected in the Ls not a part of their specifically decision-making process, but a [not fully determined, of course] consequence of one of the brain events involved in making some decisions - an event that does not happen in Zombies.

Perhaps, a difference might exist in, say, regulating brain temperature, or something like that - maybe something favored the spread of the mutation? -, or maybe another process (deterministic or not) is happening in the brains of Zombies but not Ls. But in any case, I would be inclined to conclude that their choices are probably not being affected by the process detected by the I-detector.

That said, the issue I just raised seems to fall - as not "purely metaphysical" - among the issues you intend to set aside in (i), but it seems salient to me, since it gives me the impression that Ls and Zombies probably make decisions in the same manner.

Tentatively, I guess perhaps one may just stipulate that the indeterministic process that is detected is one of the brain processes involved in coming up with a choice, while Zombies actually reach the conclusion through another, deterministic process (though that would seem to require big changes in brain structure; maybe instead of evolution, genetic engineers introduced many changes at once?).
However, in that case, I'm not sure why there would not be other differences. I would expect Zombies' choices - all other things equal - to better reflect their previous character, values, etc., and as a result be more rational more often, while the Ls would have the problem of some randomness getting in the way of their reasoning. But maybe not all other things are equal. Maybe one may also stipulate that the brains of Zombies have another, deterministic but unreliable (from a perspective of rationality) process that affects their choices, rendering them no more rational than Ls, overall.

In that case, I see no good reason to treat them differently, unless perhaps one can detect when the unreliable process affected the Zombies (that might reduce culpability), or when the randomness hits the Ls badly (but for that, one would need more info about the sort of randomness involved, how it hits the Ls, etc.).

Of course, libertarians may disagree and hold that indeterminism in the brain processes leading up to a choice does not need to be a case of randomness reducing rationality, but I don't see a way around that, at least not as long as all decisions are influenced by the indeterministic process (rather than, say, only those in which a person might rationally and deterministically choose to make an indeterministic choice).

Side note: I'm not implying determinism is required for rationality, MR, etc., I think indeterminism may or may not be a problem depending on the specifics of how it happens, the extent of the effects, etc.

[V. Alan White]

Thanks so much Paul--that means a lot, believe me.

I'd hoped to make myself a bit clearer in the OP than I managed to be in the paper about the pragmatic stance. What I hope the Zombies tale tries to convey is that it seems unavailing to anchor any particular value that favors Ls or Zombies unless and until there are explanations of how I-properties work in delivering dual-ability control of L choice, which would at least revive arguments about ultimacy of choice that might make Zombies look like metaphysically second-class citizens, perhaps like androids or the like. But despite valiant efforts like Peter's here a couple months ago, we epistemically seem to live in something like the Zombie world where we might adhere to differing values that line up with compatibilisam or incompatibilism, but no one seems to have an argumentative upper hand. So--maybe we should take a step up from that struggle, seize on some reasonable and workable standards of moral responsibility that can be shared by compatibilists and incompatibilists, and go from there. By invoking pragmatism we take a different track than just stipulating the "right" values as being reflected in one of the two accounts. Given the moral equality of Ls and Zombies--and unless someone has a better insight than I have, they are equal in the Zombie world--a pragmatic approach to values seems warranted.

[V. Alan White]

Hi Angra--thanks so much for your insightful comment.

I agree with your analysis about what well could be going on with the I-detector and L-brains, because the best I can do in the story is show a strong correlation in the smoking-causes-cancer sense: the constant conjunction of I-properties and rational choice--and the absence of such in non-rational pre-Zombie non-Ls--is the best epistemic stance I can take to bolster belief in L free will in that world. Your account would certainly fit in that large explanatory hole.

I also very much like your argument that Zombies might be better situated to have their choice express their pasts and character even granted that Ls necessarily I-choose. One thing I set up in the story is that it is not known whether Zombies work deterministically--just that they lack I-properties in doing pretty much the same things that rational Ls do. That helps focus attention (I think) on pragmatic assessments of the two types of beings in terms of believably being moral equals.

And thanks so much again!

[Eddy Nahmias]

Hi Alan, I like your thought experiment. Why not use it as a reductio of Kane-style libertarianism? I remember a question (alas, not the questioner) for Kane at a conference that I love using with my students. Imagine that future neuro-physicists discover we don't have any indeterministic processes occurring in the right place at the right time to give us L-FW, but they also discover a way to introduce an indeterministic process into our brains in just the right place. On Kane's (Balaguer's, Nozick's, maybe James') view, these scientists could now, for the first time in history, make humans free and responsible by surgically implanting the new "Liberator 2200"(TM) into people's brains. Alas, no one can tell who has the Liberator in their brain and who doesn't. Seems like a reductio to me (or at least a tough bullet to bite for these theorists, or a possibility they would have to deny, etc.). I read your thought experiment in the same light.

[V. Alan White]

Eddy--

Now that's really interesting! There is one possible asymmetry. My account attempts to establish an epistemic ground for believing in the existence of L free will by correlation, which leaves explanatory space for event- or agent-causation consistent with I-property events as constitutive of any resultant decisions or actions. In your account, as I understand it, the decisions are already constituted by whatever processes obtain, and something like I-properties are added on "in the right place". But this puts Liberator 2200 subjects into a logical bind: either the "right place" is detectable or it is not. If a certain class of decisions are rendered L-2200 detectable by reduction to 50/50 chance, then the L-2200 is thus demonstrated as effective, but also undermining control as detected to be a merely chance influence. If the L-2200 does not affect the relative frequency of types of decisions, then it is epiphenomenal and inconsequential and undetectable (somewhat as Angra suggests above), and leads back in the direction of my Zombie tale at least epistemically as an inquiry comparing what we might think about Ls and I-detector-contrastive Zombies. As my paper says a bit more directly, this puts the onus on defenders of libertarianism to show how indeterminism contributes to our sense that it lends control over decisions that bolsters confidence that ultimacy-bound retributivism is justified in such cases given what we know about Ls in the Zombie world (which mirrors what we know about agents in our own world--we might be Ls; we might be Zombies). The L-2200 case moves like something in the opposite direction of that, and in the first footnote of my original paper I even posit that the story might move from people being D-detected (deterministic) and thus assessed by compatibilist responsibility criteria to a later evolution that produces "Angels" that are exactly like D-beings but fail the D-detector test. Perhaps putting both stories against one another best advances the pragmatist agenda as opposed to either one presented as a reductio of libertarianism: reductio is usually an absurdity against some established case, and the pair of antithetical stories--emerging Zombies against Ls and emerging Angels against Ds--moves above the question of reductio to pose a perspective of pragmatism that is meta-theoretic. Maybe I should tell both tales explicitly to better make the point? That's intriguing.

Excellent comment to make me think harder about all this, and thank you.