Tuesday, December 06, 2005

What desire is

[By Neil Sinhababu]

My last post was about what desire isn't, necessarily. (Namely, a representational capacity contributing to a reward or punishment signal.) Now I'm going to tell you what necessarily is. I've got a conceptual analysis that offers two conditions on something's being a desire, individually necessary and jointly sufficient. Here goes:

1> If someone desires B, and she occurrently believes that doing some action A can bring about B, she will do A.

2> If someone desires B, at least one of the following four relations between her beliefs regarding B and her hedonic state holds:
a. Increases in the subjective probability of B will cause her some pleasure.
b. Decreases in the subjective probability of B will cause her some displeasure.
c. Vividly imagining B coming about will cause her some pleasure.
d. Vividly imagining a state of affairs incompatible with B will cause her some displeasure.


So that's the analysis. The necessity of 2 is more controversial than the necessity of 1, so let me offer the case that inspires me to include it. It's a case I blogged about some time back:

On another planet, there exist the Neutrals -- intelligent creatures who are exactly like us, except that they are psychologically incapable of ever experiencing pleasure or displeasure. They engage in many motions similar to ours. Like a human, a Neutral would move quickly and suddenly towards his baby if he saw that the baby was about to crawl into a busy street. But while a human father might have an unpleasant experience of fear just as he began to move, a Neutral would not. Though the Neutral's attention would be intensely focused on the baby as he began to move and he would have lots of visual and auditory sensations, he would feel nothing unpleasant at all. Even if, in the future, he imagined what could've happened if he hadn't seen the child in time, he wouldn't feel the unpleasantness of horror in imagining. To an observer, Neutrals are indistinguishable from normal human beings. When you do things to one of them that would make a person laugh or cry, they show the outward behaviors of laughter and crying. But they don't feel the pleasure of laughter or the pain of sadness that we usually do when crying.

There are two questions I like to ask after giving this example: First, do any of the Neutrals' motions count as actions? (This becomes important for assessing questions about the relation between desire and motivation.) Second, (and more importantly for the topic of this post) do any of the Neutrals' mental states count as desires? I'd say that the Neutrals act, but the states motivating them aren't desires. Certainly, the Neutrals have some conative mental state. But I don't want to call this mental state a desire, since desire has to feel a certain way, and this way involves feelings of pleasure and displeasure. There's disagreement on this question, though (see the linked comments). Perhaps the word "desire" expresses two concepts -- one that applies to any old conative state, and one that only applies to states with a certain affective character. In that case, it's the latter concept that I'm analyzing.

12 comments:

  1. 1> is going to need some serious tweaking, because there are many envisionable scenarios in which S desires B and has an occurrent belief that doing A will bring about B, but will not do A.

    E.g.:

    A may be impossible to perform per se, or impossible for S to perform, either at that time or at any time.

    Or S may merely believe that A is impossible in one of these senses.

    Or S may have conflicting desires that come into play, and on balance may decide not to do A, even though she can still be said to desire B.

    Or S may be an inveterate procrastinator, or neurotically incapable of making a decision to action, or have some other psychological impediment to doing A.

    I find this reminiscent of attempted behaviorist analyses of mental states, which, it is my impression, often failed due to an inability to precisely and exhaustively specify such qualifications in a way that left the analysis with any explanatory force.

    ReplyDelete
  2. Oops; in my first sentence above I should've said "can bring about B," not "will bring about B," but this doesn't affect my point.

    ReplyDelete
  3. Yeah Neil, I have some serious concerns, including hte one that Mason cites about (1). Here's a counterexample to (1): I desire to be rich, and I believe that if I kill you and take all your money, I will be rich. But I do not kill you and take all your money.

    As for (2)... I'm not sure how to give an argument that speaks to the point, but I can definitely say that I don't at all share your intuition. Why think that desires have to feel a certain way? I have no reluctance in attributing desires to the Neutrals at all. I know this is just a report of an intuition and not an argument... do you think there's an argument in favor of your view?

    I guess the relevant test case is sort of a compliment to the Neutrals. Let's see, we'll call them... hmm. No English word comes to mind, so I'll make up a new one: the Listful. The Listful are phenomenologically much like we are, and they often have states, schmesires that feel just like our desires feel, and usually toward the same kinds of things. Most Listful children schmesire candy, and most listful adults schmesire the well-being of their children, etc. And when the thing that they schmesire comes about, it usually makes them happy. Likewise, thinking about things they schmesire feels good. But the Listful are not motivated by their schmesires; they are no more likely to pursue something they schmesire than something they don't schmesire, or even something they schmesire the negation of. If the object of the schmesire comes about, it makes them happy, but they don't take any steps to increase the chances of this being the case.

    The Listful don't satisfy Neil's (1), so they don't desire, on his view. The question is whether something like (1) and something like (2) are BOTH necessary for desire, and (here's the brute intuition again) it seems that once we've separated the phenomenological from the motivational components of desire, we should admit that the neutrals do have desires.

    ReplyDelete
  4. Yeah, 1 needs a lot of help. So how about this:

    1*> If someone occurrently desires B, and she occurrently believes that she can bring about B by doing A, she is motivated to begin doing A.

    Then I'm going to say that the action one engages in will be the one that one has the most motivation to do. This should settle Jonathan's example.

    What to make of the procrastination cases? Well, there's a few things -- (Hume and) I have an explanation of akrasia on which vividly perceived or imagined objects raise the desires for them to temporarily higher levels of intensity. Future objects usually don't do this, so they are put aside for trivial desires of the moment whose objects are right there near us. I'd want to give similar explanations of other psychological impediments, but I'd have to know more about how they work to give an account.

    Jonathan, I'll give a fuller account of the theoretical motivation behind 2 in just a little bit... Just want to start by saying that I think the Listful pretty clearly lack desires. The Neutrals are, I accept, the harder case.

    ReplyDelete
  5. Actually, change 1* to "motivated to A" -- that'll get motivations not to act into the picture more easily.

    ReplyDelete
  6. Here are 3 points
    -------------------------
    A.
    Here is another version of the objection to (1) already on the board.

    You desire B (a glass of water), and believe that by doing A (asking the waitress for a glass) you will bring about B (you will get your glass of water). But, before you do A (ask the waitress), I shoot you and kill you dead. In this scenario, you satisfy the antecedent of (1) but not the consequent, and so you don't meet condition (1). This entails that you never really desired B after all (you didn't really desire a glass of water). But that's crazy.

    Here is a Davidsonian take on the problem you're encountering here. For any given type of mental state, there is nothing that *always* causes or is *always* caused by mental states of that type. Or in other words, there aren't any strict psychological laws like (1) purports to be. You could try replacing (1) with:

    (1'):If someone desires B, and she occurrently believes that doing some action A can bring about B, then *ceteris paribus* she will do A.

    (1') might be able to get you around the other objections to (1) presently on the board. The worry here, though, is that the ceteris paribus clause will be doing a lot of work, and so Mason's worry about "precisely and exhaustively" specifying qualifications may have some force.
    -----------------------------
    B.
    What exactly is the relation between desire and pleasure and pain, as specified by 2? Is it that desire is part of the *cause* (i.e., it combines with subjective probabilities to cause pleasure or pain in 2a or 2b, while it combines with vividly imagining to cause pleasure or pain in 2c or 2d), or is it that desire is partly *constituted* by (and thus not the cause of) pleasure and pain?

    If you take the cause option, then your account is straightforwardly functionalistic. According to you, desire is defined in terms of its causal interactions with belief (including subjective probabilities), action, vividly imagining, pleasure, and pain. This view is functionalistic even if, for instance, you think that pleasure and pain cannot be given a functionalist analysis (you might deny functionalism about qualia).

    On the other hand, if you think that desire is partly constituted by pleasure and pain, then your account is functionalistic only if you think pleasure and pain can be given a functionalist analysis.

    Either way, there is at least a functionalist element in your proposal. But now, a further question. Functionalists typically define mental states in terms of what they typically cause *and* what they are typically caused by. But your analysis doesn't seem to address what desires are caused by. Do you think there is no necessary constraint on what a state must be typically caused by for it to be a desire? (I don't have an objection lurking here, it's just a question.) You at least think that a desire must cause action in certain situations for it to count as a desire (this is what 1 says); do you think there is nothing a state must typically be caused by for it to count as a desire?
    ---------------------------------------
    C.
    Let me go along with the spirit of 2 and suppose that desires must "feel a certain way", as Jonathan puts it. Still, I don't find it intuitive why they should feel precisely the way specified by 2. Suppose I have a putative desire for a glass of water which meets condition 1. However, the subjective probability I attach to getting a glass of water is in no way correlated with any pleasure or pain I feel; that probability can go up or down, and I feel no pleasure or pain. What's more, when I vividly imagine getting a glass of water, I am left cold, feeling no pleasure or pain. However, when I vividly imagine it being impossible to get a glass of water, I feel some pain. Again then, coming to believe that I can't get a glass of water causes me no pain, but vividly imagining that I can't get a glass of water causes me some pain. Then I satisfy condition 2d and so satisfy 2. And so I count as desiring a glass of water.

    The objection is that even when I get into the thought that desires need to feel a certain way, I don't find it intuitive that this counts as a perfectly acceptable way for desires to feel. Even if the 4 conditions of 2 are getting at something right, then, I doubt that 2 as it stands is acceptable.

    ReplyDelete
  7. This is supposed to be an intuition test, Jonathan, so your disavowal of any intuition that the neutrals lack desires has me worried. But let me see if I can say some stuff to get you into my way of thinking.

    There have been lots of people throughout the history of philosophy -- Kant being one of the most notable examples -- who claimed that humans were capable of multiple kinds of motivational states. One kind was desire/passion, which was associated with pleasure. (The terms of this association were different from 2, since 2 doesn't send you into psychological hedonism, while Kant thought that psychological hedonism was true of our inclinations, though of course moral motivation worked differently. But I hope the parallels are clear.) Then there was another kind of motivation, the kind that came from reason.

    The case of the Neutrals is designed to separate one of these kinds of motivational states from the other. I don't think it's a linguistic accident that we often talk about people who seem to lack any emotion as acting "dispassionately", and the Neutrals are the ultimate case of that. Perhaps some people call "passion" what I call "desire", but I think there's some ordinary concept or other that's being analyzed here.

    (Justin, I'm heading off to German class now, but I'll get back to you soon.)

    ReplyDelete
  8. A. Yeah, I'm thinking about doing the "ceteris paribus" thing, or perhaps it should be called the "normal conditions" thing for 1. (I wonder why they call it ceteris paribus, anyway. It's not like anything else is being made equal in cases like this.)

    B. I don't think desires are constituted by the pleasure / displeasure, so I'm going the way where they're merely causes of it. I don't want to make it a necessary truth about desire that desires are caused in any particular way. Really, I think that the ways desires are caused are many and weird. (Hormones, appetite, reinforcement learning, etc.)

    C. Yeah, 2 is weird like that. Still, it does seem to me that in your case, any hedonic experience or other can make the mental state a desire, and the agent does desire the water. Can you tell me anything about how you think this affective component should work? I'm guessing it's not just a matter of saying, okay, we need two of the four things instead of just one.

    ReplyDelete
  9. Hi, Neil.

    I think a different kind of objection to (1)centers on the rationality of the agent. If an agent is seriously means-ends irrational, then even if an agent (most) desires that P and believes that the best means to take to get P is to do A, then agent still might fail to do A. Does this strike you as conceptually possible? If so (as it does to me), then it suggests that connection between desire and motivation is more indirect than you seem to think - rationality is an extra component that links desire and motivation. Note that even a Humean about motivation, like me, can accept this (see, e.g., Smith in various papers). Perhaps this relates to the Neutrals case somehow, but it's late, and I don't immediately see how at present.

    Neat dissertation topic. Hope all is well for you.

    ReplyDelete
  10. I take it the idea you're trying to get at in (1/*) is that we are disposed to [be motivated to] act in ways we believe will satisfy our desires. I'm not sure we can get much more specific than that though, for the sorts of reasons others have pointed out. Most notably: desires can clash or be overwhelmed, or we might believe that some alternative action A+ would be a better way to achieve the desire. But perhaps you can factor these in?

    ReplyDelete
  11. Good to hear from you, Todd! I'd have to get into the details of cases to figure out what to say about irrationality. Some cases, like akrasia, are cases where I'll say that the desire we act on is in fact stronger than the desire that governs our deliberation. Maybe you have in mind a case like depression, where people don't act on their desires. I'm wondering if the thing to do here is to present my analysis as an analysis of occurrent desire, and say that depression prevents one's desires from becoming occurrent. (One can still be aware that one has them, in the sense that a sleeping person desires things, but they don't do the things desires do, in terms of causing pleasure or motivation.)

    Here are three things I'm not trying to do: 1) analyze the concept of rationality, 2) say what the norms apply to action, 3) bring in a third causal factor, rationality, which is separate from desire and belief, to explain action.

    Richard, that is the basic idea. I suppose that if sticking with disposition-talk can successfully hide me from counterexamples, I might end up chickening out and going there. But right now it seems dishonorably imprecise to me.

    ReplyDelete
  12. A set of neural inputs, A, that is desired by a neural entity relative to another set of neural inputs for that entity, B, is a set of neural inputs such that neural outputs associated with A by the neural entity will have a greater chance of being triggered then the neural outputs associated by the entity with set B.

    What we experience as disappointment and frustration is the process of our mind/brain destroying the associations we previously formed between our neural inputs and outputs. This happens when we do not achieve the neural inputs we thought optimally desirable, or more specifically, when our expectation of our future sensory input changes in an undesirable way. Therefore we change behaviors that produces relatively more undesirable outcomes relatively more often then behaviors that produce more desirable outcomes.

    To use an example, a monkey that is presented with a button that when pressed removes a clear plastic divider separating the monkey from a desirable food may be confused and disappointed when the button is disconnected from the plastic divider. When this happens, its mind begins to loosen its association of button pressing and food. It may then begin to act erratically and experiment with other neural outputs, such as making noise and throwing stuff.

    To use your example of a car nearly hitting a baby, the father would experience a negative feeling disassociating whatever impulses lead to the near miss (id est, what lead him to over estimate the safety of his carriage parking location), as well as this sudden interruption in his current thought process which stops the positive association his current activity is based on, focusing him on the important task of baby rescue.

    Opposite of this is contentment which preserves the connections we previously made. Excited contentment is a process by which we further integrate already existing connections into larger and more complex patterns.

    To return to our monkey example, the monkey may begin to fantasize about all the females he can impress with his button food.

    In the baby carriage example, the fathers initial contentment with the (precarious) position of the carriage allows him to divert his attention to other issues (until the car enters the picture).

    In your examples, the neutrals would not be able to act normally because their minds would not operate: emotional qualia are inherent in the ability to think (I suppose a robot could be programmed to laugh and cry on cue to some script, but eventually the ruse would be exposed as a simple program rather then a sentience).

    In the case of the fully formed Unconditionals, they would be unable to experience disappointment or excited contentment (although there exists a case that they have achieved some sort of serene nirvana) since they are unchanging neural entities (unless you are supposing these are some sort of tortured souls trapped in a body they cannot control).

    In the case of the creatures of habit, they would experience a qualia similar to disappointment when they perform a tic associated with desirable inputs and do not observe those inputs. The association the creature of habit had previously created would be destroyed.

    Occasionally our genes may work against our minds and make some part of our mind unable to assert control against our minds interest. In this case I would say that our genes force the creation of a separate less able neural entity relative to our usual neural entity that does not have access to faculties that would predict less immediate consequences of actions. In this sense it is not possible to assert that we desired the undesirable consequences of “our” actions in that while an “individual” would usually be able to predict the results of the action, and thus be expressing a desire for that outcome by committing the action, that “person” was unable to do so at the time, however I assert the existence of separate entities at the separate times and am uncomfortable identifying the entity in the body with singular pronouns. (This is not to preclude responsibility and subjection to negative consequences since they are determinates of the extent to which a mind will be able to force itself over impulses.)

    ReplyDelete

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.)

Note: only a member of this blog may post a comment.