Saturday, December 03, 2005

Is "desire" a natural kind term?

Here by the invitation of the Socratic Shadow himself (aka Richard Chappell), I'm Neil the Ethical Werewolf. Werewolves are typical Humean beings, driven by their desires, and I'm going to be doing lots of desire-related blogging in the days to come. Today I'm going to attack a view about what desire is, and make some suggestive arguments about how to better understand desire.

In his new book, Three Faces of Desire, Tim Schroeder claims that "desire" is a natural kind term. He offers a supervenience base for desire. Drawing on psychological and neurobiological results, he claims that “To be a desire is to be a representational capacity contributing to a reward or punishment signal.” To be a reward or punishment signal is to be a signal with the power to strengthen or weaken an organism's mental dispositions in processes of reinforcement learning. If I keep getting something that I desire (i.e., if I keep being rewarded) after engaging in some behavior, the disposition to engage in that behavior will be strengthened. If I keep getting something I have an aversion to -- if I keep being punished -- my disposition will be weakened.

This doesn't merely work for behavioral dispositions, but certain perceptual capacities as well. When I pass my finger over the Braille letters in an elevator, it's hard for me to get any precise tactile sensation of how the dots are arranged. It strikes me as surprising that anyone can get such a feeling by touching them – as one must in order to understand the dots as a word. Blind people who have learned Braille, however, can do so. This is because correct sensory discrimination has been repeatedly rewarded in the process of learning Braille. The desire to understand a particular word was satisfied, and perhaps a teacher praised the blind person. So the mental dispositions required for fine-grained perception of Braille letters were reinforced.

So now you're probably thinking, "All this mental-disposition-changing reward stuff amounts to an interesting fact about desire. But why does Schroeder think that desiring B is nothing more than representing B in such a way that it contributes to the strengthening of mental dispositions when B comes about? How does this have anything to do with being motivated to do A when one believes that doing A will bring about B? Or maybe being pleased when discovers that B will happen? Aren't these the things that really make something a desire?"

Well, here's where the empirical evidence gets really interesting. The brain structure that realizes the reward system -- the VTA/SNpc -- is causally upstream from the brain centers that realize pleasure and desire-belief motivation. Schroeder writes that “The neural basis for reward is the normal cause of pleasure and an important cause of motivation.” I won't go into the evidence for this here (it's gonna be a pretty long post as is). So just take it as a premise of Schroeder's argument for the time being. Actually, if you like you can give Schroeder the following, more useful premise: "In every case of human pleasure and motivation that desire intuitively ought to explain, the reward system does the explanatory work."

Suppose we see "desire" as a natural kind term like "water," and regard desire's ability to cause motivation and pleasure as similar to water's having wetness, drinkability, and clarity. In other words, these are merely parts of the stereotype of the thing, not its actual essence. Being a representational capacity contributing to a reward or punishment signal is the hidden essence that explains how desire motivates us and causes us to be pleased, just as being H2O is the hidden essence that explain's water's wetness, drinkability, and clarity.

Now let me launch my counterexamples to Schroeder's view. When considering each of these cases, assume Schroeder's premise -- that in humans, the neural basis for reward explains pleasure and motivation. It's just like the Twin Earth cases, where you accept the empirical claim that the stuff in our lakes and rivers is actually H2O. My first case will attack the claim that reward systems are necessary for desire. The second will attack the claim that reward systems are sufficient.

First, suppose there are creatures on some other planet who are much like us, except that their mental dispositions cannot be strengthened or weakened by reward signals. Perhaps these creatures – call them the Unconditionable – spring from the womb fully formed and able to do all the things that we humans require a reward system to learn. In particular cases, their phenomenology and behavior is identical to ours, minus whatever immediate phenomenological and behavioral effects require a reward system. The Unconditionable have a particular conative mental state which plays a major role in explaining their behavior. When they believe that doing A will bring about B, and they have this conative state towards B, they do A. If they fail to attain B, they feel the same emotions of disappointment and frustration that we do. But if they succeed, they feel as excited and happy as we would. Does their lack of a reinforcement learning system make the conative state that motivates their actions not count as desire?

Second, imagine the Creatures of Habit, whose mental lives are largely unlike ours, but who have the same kinds of reward systems that we do. They regularly engage in habitual, unintentional tics. They have representational states picking out certain states of affairs, and when they happen to tic before one of these states is produced, ticcing behavior of that kind is reinforced. Their reward and punishment systems strengthen and weaken neural connections contributing to these behaviors and other things, changing their mental dispositions just as our reward systems change our mental dispositions. But when they represent B as a reward and believe that they can bring about B by getting A, they never do A (unless A coincidentally happens to be a tic that they have learned). Intentional action that is motivated by a combination of belief and some conative state is entirely foreign to them. Do these creatures have desires, in virtue of the fact that they represent certain states of affairs as rewards?

It seems to me that the Unconditionable have desires, while the Creatures of Habit do not. Schroeder’s theory gives counterintuitive results in both cases. (Do you think so too? Post your intuitions in the comments! I wish I had time to make up cool buttons like the ones you get when you take an online quiz. One would say "I'm Tim Schroeder!" and the other would say "I'm Neil Sinhababu!" and I haven't decided on the other two. Unfortunately, I'm lazy.)

I haven't given sufficient arguments for this conclusion yet, but let me give you the conclusion I'm pushing towards: It's wrong to think of “desire” as a natural kind term like “water”, and see its ability to cause behavior and pleasure as analogous to the clarity and wetness of water, which are merely part of a stereotype that helps us identify it. Desire’s ability to cause behavior and pleasure aren't merely part of a stereotype which allows us to identify desire. They're essential to desire itself. Far from being necessarily constituted by a hidden essence that explains its outward properties, desire wears its essence on its sleeve.

(For the record, I don't have Putnam's intuition and I actually think that there's a possible world where water isn't H2O. Guess I'm just not the kind of werewolf who likes natural kind terms.)


  1. I think our definition of "desire" or anything like that is somthign that doesnt create tight definitions. What this means is that there woud be no common cause or nature to "desire" since anything that simulates it is also it.

    So I might say -I desired the ice cream but I am not thinking of it in the terms that would be required to tightly define it (eg neurons and whatever else).

    One question in your example might be are your theoretical examles likely in the real world? Maybe "almost all the time" is enough.

  2. I agree with your intuitions, but have some qualms about your conclusion that desire isn't a natural kind. Before I get to this, though, here's another point.

    First, the necessary condition part of your argument reminds me of the chauvinism objection raised in discussions of functionalism.

    The specific version of the chauvinism objection I have in mind goes like this. "Psychofunctionalists" hold that mental states like belief and desire are defined by a certain theory, an a posteriori psychological theory discovered through cognitive science. Well, imagine that cognitive science shows that human beings consistently update their beliefs in a way that involves committing the Monte Carlo fallacy: the more times the board spins red, the more strongly human beings believe that it's gotta come up black next time.

    Now, imagine a species psychologically (and even physically, if you like) a great deal like human beings, except for the following single exception: they aren't susceptible to the Monte Carlo fallacy. Then, the empirical psychological theory which is true of us will not be true of this species. But then, since psychofunctionalists say that the theory which is true of us is what defines mental states like belief, it would seem to follow according to psychofunctionalists that these creatures would not have any beliefs. According to the objection, this is illegitimate. It is a kind of "chauvinism": it involves denying that creatures trivially different from us lack beliefs simply because of that trivial difference (akin to denying equal rights to a group of people because they have a different skin color than we do). So, you might want to check out Shoemaker (especially) or Block or Jackson on chauvinism.

    Now, on natural kinds. What is a natural kind property? One answer is that it is the sort of property which can figure in natural laws and ground inductions. So for instance, being grue isn't a natural kind property since there aren't any laws about being grue (though there may be laws about being green or about being blue) and being grue can't legitimately figure in inductions (just because all observed emeralds are grue you can't safely infer that all other emeralds are grue too). If this is how we understand natural kinds, I doubt that your argument establishes that desires can't be natural kinds -- in doesn't show that there can't be psychological laws about desire or that desire cannot figure in inductions.

    So, here would be my alternative take on where things stand. Some natural kinds (like water, heat, and the vast majority) have hidden essences, while maybe other natural kidns have unhidden essences. The natural kind/unnatural kind distinction is metaphysical, while the hidden essence/unhidden essence is epistemological, and so I see no reason why they shouldn't cross cut. Your argument might show that desire doesn't have a hidden essence, but from this point alone I don't think it follows that desire isn't a natural kind.

    Let me acknowledge, though, that this line I'm taking here is controversial. So maybe my point is just this: talk of natural kinds is fairly loaded, and for the sake of the argument you're making you don't need to embroil yourself in all of the controversies involved, and so should take care to avoid doing so.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)