In Consciousness Explained, Daniel Dennett sketches an argument to the effect that we could not be brains in a vat, since combinatorial explosion rapidly outgrows any possible system’s capacity to provide the appropriate deceptions.

In this note I shall discuss an objection to Dennett’s argument which seems to spring naturally to mind for many philosophers, but whose only published formulation (as far as I know) is Stjernberg (1993). In his poignant note, Stjernberg professes to be a ‘vat operator’, offering some disheartening news-ex officio, as it were-about how the deception is, after all, accomplished. But I shall argue that his reasoning fails to discredit Dennett’s argument. The only deception we suffer from Stjernberg concerns his job description.

Against Dennett, Stjernberg objects that there is no need to provide the brain with genuine exploratory powers-an illusion of freedom will do, and conveniently reduces the burden of calculating the proper experiences. Hence, the combinatorial explosion can be avoided, since only experiences appropriate to a single line of ‘action’ have to be provided and there will be no need to cater to unactualized choices. Moreover, Dennett is not entitled to the "facts about the world and about the human mind (or brain)" (p. 182) on which he relies.

The first objection claims that the task facing the deceivers is less forbidding than Dennett makes it look. The second pre-empts the first by suggesting that Dennett may not have (even an overstated) case. I shall discuss this latter and more general objection first.

Why does Stjernberg deny Dennett the right to use facts about the brain? Because he thinks that rather than facts, they may be vat-scientific fabrications designed to instil a false sense of comfort in deluded subjects. As often happens, however, this attempt to deprive the anti-skeptic of his premises drains the skeptic’s own position of interest. For part of what makes the brain-in-a-vat argument attractive is that it employs "facts about the brain" in order to explain how we may come to be so deceived. In this respect, the brain-in-a-vat argument is more than a modern-day version of Descartes’ evil demon; it is a substantial improvement on it. It is skepticism wedded to gradualism: the Quinean realization that epistemologist and skeptic alike must conduct their enterprise within the fabric of largely empirical beliefs they are concerned with, since there is no independent vantage point from which to judge it.

By employing and accommodating more beliefs about the world than Descartes’ demon argument, then, the brain-in-a-vat argument gains in plausibility and intelligibility. The skeptic pays for these virtues, however, by committing herself to the legitimacy of empirical assumptions about brains. Without such assumptions, the brain-in-a-vat argument cannot even be coherently stated. And what the skeptic is granted, Dennett must surely be allowed too.

So it seems that Dennett does have a case. How does it fare in view of Stjernberg’s suggestion that the explosion may be avoided by controlling the subject’s ‘choices’? Let us first be clear about what sort of control is at issue here. Stjernberg writes (p. 183): "You can go on believing that if you had chosen to move your finger in some other way, you would have had other sensations. But there is no need for me to make any extra provisions for such contingencies-you never move your limbs (‘limbs’?!) in any other way than the way in which you do move them." This might perhaps suggest that the vat operator is a ‘Laplacean’ predictor of choices rather than a ‘Cartesian’ instigator of them. But such a predictor would face a combinatorial explosion just as bad as that confronting the original operator, considered by Dennett, who tries to keep pace with the choices in real time. [1] For the sake of (the skeptic's) argument, let us therefore concentrate on the claim that the vat scientist can evade combinatorial explosion by imposing choices on the brain and hence limit the number of ‘input’ sequences to be calculated.

Incidentally, a similar suggestion is offered by Dennett himself: "if the evil scientists can force the brain in the vat to have a particular set of exploratory intentions, they can solve the combinatorial explosion problem by preparing only the anticipated material" (1991, p. 10; Dennett’s italics). Nonetheless, the suggestion seems misguided.

For there is something quite puzzling about the very contrast invoked by the suggestion. What is meant, in this context, by saying that our exploratory capacities are not ‘genuine’, that our choices are ‘illusory’ or ‘forced’? Not, presumably, merely that they are caused. Nor, clearly, just that they belong to an envatted brain, whose exploratory activities are by hypothesis all illusory, in the trivial sense that no muscles are moved by signals coming from its efferent (or ‘output’) nerves. The idea appears to be rather that they are interfered with in some way that goes beyond what is implied by their envatted condition. At best, they are made at the mercy of the operators. At worst, they are inflicted willy-nilly on the ‘choosing’ brain (although in ways imperceptible to it).

This picture does not seem to make good sense, however. For suppose you are an envatted brain in these direst of straits. Then you still deliberate and choose on the basis of your beliefs and desires, and given these, your choices are generally rational. This is a feature of your psychology which, you may be pleased to notice, sticks with you whatever the hypothesized scientists are doing at the moment. If you want a glass of water and believe doing certain things will lead to the satisfaction of this desire, you will very likely decide to do those things (in your virtual world)-you will not find yourself suddenly and perplexingly having chosen to go for the milk bottle instead. Nor will your desires and beliefs themselves generally emerge in a wanton or inexplicable way. In brief, your beliefs, desires, choices, deliberations and ‘actions’ are generally marked by rationality, coherence, and responsiveness to your ‘environment’. But now what more could be asked to make your choices genuine? Presumably, nothing. To have one’s choices ‘forced’ in ways that (a) are unnoticeable and (b) do not impeach their rationality, is not to have them forced in any sense that threatens their status as genuine choices.

The dilemma then is this. Either we think of the vat operators’ interference as something that really governs the brain’s choices and threatens their rationality-but such interference would be painfully noticed by the subject, in conflict with the postulated indistinguishability between envatted and embodied first-person experiences. Or else the interference leaves the rationality of the brain’s choices untouched, in which case it does not undermine their status as choices and hardly merits the name ‘interference’. In the latter case, the ‘interference machinery’ really becomes just an inert appendix to the brain’s processes of deliberation, a roundabout but harmless causal detour. But the latter case is the only option that is consistent with the premises of the thought experiment.

The genuineness of the choices lies in their rationality, which springs from the overall coherence of the subject’s psychology. The vat scientists cannot-must not, according to the terms on which they were employed-undo this coherence. Therefore, they cannot make the choices illusory in the pertinent sense. [2]

If this conclusion is correct, the distinction invoked by Stjernberg just isn’t there, and so cannot be exploited to reduce the vat operators’ plight one bit (or byte). But then Dennett’s argument is unperturbed (so far). This is not, of course, to say that it is sound. I believe that, on a correct construal, it is, but I shall not defend this view here. [*]

References:

Dennett, Daniel (1991) Consciousness Explained, Little, Brown,

Dennett, Daniel (1984) Elbow Room: The Varieties of Free Will Worth Wanting, Oxford University Press.

Stjernberg, Fredrik (1993) "A Note From Your Local Vat Operator", Analysis 53, pp. 182-3.

[1] Note that the counterfactual mentioned by Stjernberg is true on any variant here: either because if you had chosen to move your finger in a different way, the predictor would have predicted this (and hence prepared other sensations), or because if you had chosen to move it differently, this would have been because the instigator had arranged for this to happen (and hence prepared other sensations).

[2] This parallels the familiar compatibilist argument that determinism cannot threaten the genuineness of our choices and deliberations, since this consists in the patent (though not perfect or exceptionless) rationality of the causal processes which make up these choices and deliberations. Since one of the foremost advocates of this line is Dennett himself (see Dennett 1984), it seems injudicious and unnecessary for him to concede the distinction between forced and genuine choices in the context of brains in vats. But perhaps Dennett means to concede it only for the mini-exercise in vat technology he describes on pp. 4-5 in Consciousness Explained: convincing a subject reduced to a brain in a vat that she is lying on the beach listening to music, blind and paralyzed but for one finger, which she is free&emdash;or so she thinks&emdash;to wiggle at will. In this case&emdash;where only the crudest (virtual) exploratory options are possible anyway&emdash;it may perhaps make sense to imagine that one particular choice to wiggle is somehow forced on the subject. My claim is that the distinction cannot be generalized to the case where the deception is full-blown and the subject has a full range of experiences and (virtual) activities.

[*] Thanks to Fredrik Stjernberg and Daniel Dennett for comments on an earlier draft of this note.