Recall our discussion of How to Interpret the Turing Test.
One of the options construed passing the Turing Test as good evidence that an entity was intelligent, but not yet a guaranteed proof. This option allowed that it might be possible for an unintelligent entity to pass the test. That just seems to be unlikely: the more likely and reasonable explanation would be that the system really was intelligent.
A more radical option thought that passing the Turing Test suffices for being intelligent. According to this view, being able to respond to questions in the sophisticated ways demanded by the Turing Test is all there is to being intelligent. This is an example of what philosophers call behaviorism about intelligence.
We’re now going to spend some time unpacking what this view says, what can be said in favor of it, and what against it.
“Behaviorism” has been used to label a couple of different programs and theories.
There were research programs in psychology in the early 1900s associated with the psychologists Watson and Skinner. In our texts, those views get called “scientific” or “psychological” or “methodological behaviorism.” Kim discusses these views at pp. 61-2, and 82-6; but we won’t talk about them in class.
Then there were philosophical theories about the mind/body relation, starting around 1930. Philosophers associated with these views include Carnap, Hempel, and Ryle. We call these views logical or analytical behaviorism, and they will be the views we focus on. Their basic idea is that talk about mental states (like pain, beliefs, and intelligence) can be defined or translated into talk about what behavior the subject exhibits or is disposed to exhibit. One upshot of this is that there are no mental facts “over and above” facts about people’s behavior and behavioral dispositions.
Earlier in the class, we introduced this contrast between conceptual (or analytic) definitions and scientific (or theoretical or empirical) definitions/identities. Unlike Smart’s identity theory, the logical behaviorists are aiming to give conceptual/analytic definitions of our mental concepts. If their definitions are correct, then we shouldn’t be able to “discover” nor even coherently imagine that they’re false.
A third kind of view emerged later in philosophy. This goes by various names; Kim calls it “ontological behaviorism.” This keeps the logical behaviorist’s idea that there are no mental facts “over and above” behavior and behavioral dispositions. But it gives up the idea that mental talk can be defined or translated into talk about them. Kim discusses this later form of behaviorism at pp. 78-80; but we won’t be talking about it further in our class. (It improves on the second kind of view in some ways, but it hasn’t played as important a role in the history of philosophy of mind: shaping other theories and arguments, and so on.)
In our class meeting, I asked whether it made sense for there to be a joke that was really funny, even though nobody ever found it amusing. It never ever made anybody laugh. It really was funny, it just always appeared unfunny, to everyone. That seems hard to make sense of, doesn’t it? With funniness, we’re more inclined to say that being funny and seeming funny have to be more closely connected. Of course, sometimes things might seem funny to some people, or in some contexts, and not to others. As a result, some would deny that there are any facts about funniness. But if we admit that there are facts about funniness — if any things ever really are funny — then maybe there can be some disconnect between its being funny and whether it seems funny to every audience. Maybe some audiences are just giggly, and it wasn’t really that funny. Or maybe it really was funny, but the audience was in angry mood for independent reasons and so didn’t appreciate it. We might make sense of that. What seems hard to make sense of is that there could be a total disconnect between whether something really is funny and whether it ever seemed funny to anyone. A complete, perfect, permanent illusion of funniness (or of unfunniness) isn’t easy to imagine.
When it comes to things like rocks and trees, on the other hand, it seems like we can make sense of complete, perfect, permanent illusions. Set aside the question of how we can tell whether we’re really seeing rocks or trees, or just are having the illusion of doing so. Those at least seem to us like two different possibilities that we can coherently imagine. So we can make sense of a kind of appearance/reality distinction with rocks and trees, that we can’t make sense of (or at least, have a harder time making sense of) with funniness.
Philosophers use the words “objective” and “realist” in several ways. But one main way they use these words is to talk about this kind of contrast. What we just said is the natural view about rocks and trees is a realist view about them, that says there are objective facts about the rocks and trees, that “transcend” or aren’t just logical consequences of what evidence we have, or how things appear to us.
A skeptic is someone who thinks there are such facts, but we’re not in a position to know what they are, or maybe even have reasonable beliefs about them. A non-skeptical realist thinks that there are such facts, and we can know or have reasonable beliefs about them. We’re not going to try to sort of their debate in this class. You can study that in an Epistemology class. One unfortunate bit of messiness in philosophical vocabulary is that sometimes people use “realist” in a narrower sense, to mean what I just called “non-skeptical realists.”
On the other hand, the natural view about funniness is a non-realist (or irrealist or anti-realist) view. It says that the facts about funniness are constrained by the facts about how funny things appear to us. There are limits to how far it makes sense for these to come apart.
More radical views would deny that there are any facts about funniness. These views might call themselves “non-factualist” or “non-cognitive” or “expressive” views about funniness.
A fourth strategy would be to deny that there are any facts about being funny, period; but allow that there are facts about being-funny-for-me, and so on. This fourth kind of view tends to have a lot in common with the second, non-realist view; but there are differences in the details of what they say.
This non-realist view, denying that it even makes sense for appearance and reality to be independent, is natural to have about topics like funniness and fashion. Philosophers have also advocated this kind of view for other topics, too, where it’s more controversial and sometimes really hard to believe. For example:
One way to understand philosophical behaviorism is as applying this kind of view generally to all facts about the mind. — Well, in the first place, to facts about other people’s minds. And then, if I want “pain” to mean the same thing when I talk about your pains and my own pains, I’ll be forced to apply the view to my own mind too.
That is, the behaviorist looks at the kinds of things that are our ordinary evidence for thinking other people have pains, beliefs, or any mental states. Our evidence is what those people do and say; how they behave. The behaviorist allows that sometimes that evidence can lead us astray. Sometimes it seems like your friend is angry but they’re just tired, or thinking about something else. But the behaviorist says it doesn’t make sense for the evidence and the reality to be in principle be fully independent. If the world were full of complete, perfect, permanent “illusions” of other people having mental lives of certain sorts, then for the behaviorist, it’d have to be true that they’re having those mental lives. There can’t be some further question beyond what all your behavioral evidence says.
This is why the behaviorist thinks if an entity passes the Turing Test — let’s say not just once, by accident, but is able to reliably and consistently pass it in a convincing way — then there can’t be some further question left open, beyond that evidence, about whether the entity really has thoughts, fears, plans, and so on.
For the behaviorist, the idea that everyone else behaves as though they have one kind of mind, but really their mental attitudes are completely different than they seem, or really they don’t have any mental life at all — these ideas are “pseudo-hypotheses,” things that only seem to make sense. If we think about it more carefully, we should see that these aren’t really things we can coherently imagine after all.
An example of an alleged pseudo-hypothesis we discussed in class was the idea that maybe everything doubled in size overnight, including all our measuring sticks, and the size of subatomic particles too, and at the same time the gravitational laws changed so that the change in size wouldn’t be detectable and so on. Of course this isn’t a hypothesis we have any reason to believe. But some philosophers would think that it’s an intelligible real alternative. One way reality could be is that there was no overnight doubling in size, another was is that there was. Other philosophers think this is absurd. They think there aren’t really two possibilities here. That only superficially seems to make sense. It’s not a distinction we can really coherently draw. Sometimes people say it’s a distinction without a difference.
We’ve mentioned several times in the class that we seem to have some special and privileged access to our own minds, that we don’t have to facts in our environment or to other people’s minds, and that they don’t have to ours.
Exactly how this special access should be spelled out is something philosophers disagree about. But there is a broad consensus that there’s something here that a theory of the mind should account for.
One especially strong way to understand this special access depicts your mind as a kind of private Cartesian theater. (See Kim pp. 63-66.) On this strong understanding, you can’t make mistakes about what’s going on in your mind: you’re infallible about it and the contents of your mind are “transparent” or unhidable. It’s not clear whether it’s right that we can’t make mistakes about our own mind. Philosophers also worry that on this picture, it might be too hard to achieve knowledge or even reasonable beliefs about other people’s mental states. Interestingly, an even more radical worry has arisen too. Some philosophers doubt whether on this picture, we could even get as far as understanding hypotheses about other people having the same mental states we do — never mind knowing when those hypotheses are true or false. These philosophers argue: how could anyone ever learn words that are supposed to be about other people’s inner mental states (as opposed to their outward behavior), if those inner mental states are the kinds of happenings on a private Cartesian theater that this picture says they are? Perhaps we could at least introduce private words for our own mental states — though it’s disputed whether we could even do that. But even if we could manage that much, it’s not clear we could ever teach other people to use those words with the same meaning.
Now the behaviorist goes to the opposite end of the spectrum. Intuitively, before we start doing philosophy, we think we know a lot, or at least have many reasonable beliefs about, other people’s mental lives. For instance, sometimes I know that my sister is angry. In order to explain this, the behaviorist wants to say the facts that make up my sister’s mental life are completely public facts about behavior, open to everyone’s view. Multiple people can observe those facts at the same time, and no one is immune to making mistakes about them, not even my sister. The behaviorist achieves this by identifying facts about mental lives with the behavior that constitutes our evidence for them.
As we’ll see, many philosophers who reject the private Cartesian theater think the behaviorist has gone too far. They’ve over-reacted, and now they have the problem of not giving any good account of what’s special about our access to our own minds. Perhaps the Cartesian picture of that was wrong, but we should still expect some good account of it.
The substance dualist thinks of the soul as a kind of ghostly engine, animating the body. But as we’ve seen, it’s mysterious how this ghostly engine could causally affect the body. Behaviorists are materialists; they postulate no mental substance, whose interactions with the physical world are mysterious and problematic. They only require there to be physical substances, that are disposed to move (behave) in certain physical ways.
In more detail, the behaviorist thinks that all there is to having some mental state is that the subject behaves, or is disposed to behave, certain ways in certain contexts.
What does it mean, to be disposed to behave a certain way?
First, consider the notion of a counterfactual claim. This is a claim about what would have happened if the world had been different in certain ways. For instance, a true counterfactual claim is:
If I had dropped this chalk out the second-story window, it would have fallen to the ground and shattered.
A false counterfactual claim is:
If I had dropped this chalk out the second-story window, it would have turned into an angel and flown to heaven.
A disposition is a related notion. (These can also be called “propensities” or “capacities” or “tendencies” or “powers”.) Dispositions include properties like being crushable, being fragile, and being soluble. Something can be crushable even though it is never crushed. It merely has to be such that it would be crushed if certain sorts of gentle forces were applied to it. Similarly, a glass can be fragile even though it never breaks. It merely has to be such that it would break, if it were struck in certain ways or if it were dropped from a modest height. And a sugar cube which never dissolves in water can still be soluble in water. It merely has to be such that it would dissolve, if it were placed in water. In general, having a disposition does not require that a thing actually undergo any changes; it only requires that the thing would undergo those changes if it were placed in the appropriate circumstances.
We call these events of being crushed, breaking, dissolving, and so on, the manifestations of the relevant dispositions. These manifestations are triggered in certain contexts: when a gentle force is applied, when it’s placed in water, and so on. Something can have a disposition even if it never happens to manifest it. Every sugar cube is soluble, but not ever sugar cube dissolves; because some sugar cubes are never placed in water.
Often, there will be some underlying facts about the object in virtue of which it has the dispositions it has. We call these underlying facts the categorical basis of the disposition. (“Categorical” here just means “non-dispositional.”) For instance, the sugar cube is disposed to dissolve in water, and the categorical basis of this disposition is a certain kind of molecular and crystalline structure, in virtue of which the sugar will dissolve when placed in water.
A disposition might have different categorical bases in different objects. For instance, fragility has one categorical basis in crystal wine glasses and a different categorical basis in soap bubbles. The underlying properties which make a wine glass likely to break when struck are different from the underlying properties which make a soap bubble liable to break when struck. Houses of cards will have yet further properties. But wine glasses, soap bubbles, and houses of cards still share a single disposition, fragility. They are all disposed to break when struck.
Some philosophers think it’s possible for there to be dispositions that have no categorical basis. Others think this isn’t possible; they think dispositions always have to “bottom out” in something non-dispositional. We don’t have to settle this dispute for our discussion.
So then: a behaviorist about pain will say that all there is to having pain is being disposed to behave in certain ways — wincing, groaning, saying “It hurts,” making doctor’s appointments, and so on. The kind of behaviorism we’re considering says that something like this is what the word or concept “pain” means.
Similarly, a behaviorist about belief will say that believing that Charlotte is south of Carrboro is a matter of having yet other behavioral dispositions. So too for mental states like wanting to greet your friend. (We’ll consider what the details of the dispositions might be below.)
And a behaviorist about intelligence will say that being intelligent consists in being disposed to give sophisticated, sensible answers to arbitrary questions. Since the behaviorist says that’s all there is to being intelligent, and since passing the Turing Test shows that an entity can do this, passing the Turing Test guarantees that the thing in question is genuinely intelligent. The behavior isn’t merely a sign or evidence that the thing is intelligent. Rather, behaving in the right ways is constitutes, or is all there is, to being intelligent.
This kind of behaviorism was one of the first worked-out materialist views about the mind. It’s a materialist view in that it doesn’t appeal to any souls or independent realm of mental properties; it tries to explain everything in terms of ordinary physical properties, like behavior and dispositions to behave. So if you wanted to reject dualism, you can see why behaviorism might seem initially attractive.
And for some mental states, this does seem like a plausible view. States like honesty and courage probably are well understood in terms of behavioral dispositions.
But when we think about other mental states, especially our own mental states, behaviorism doesn’t seem so plausible. In my own case, at least, there seems to be a real and important difference between really having a toothache, and just acting as if I have a toothache. When I really have a toothache, there seems to be more going on that just the external behavior. There’s also something pretty awful going on inside, which makes me behave in those ways.
When the behaviorist talks about people’s behavioral responses, what kind of thing do they have in mind? Some candidates are:
physiological responses (pulse rate, neural states, …)
bodily movements, physically described (your arm moves in such-and-such a trajectory, your vocal cords and mouth make so-and-so noises, …)
actions involving your body, described in mentally-loaded terms (greeting your friend, writing a check)
Whether you engage in these activities depends on your reasons/purposes. Someone else could have their body move in the same way, but be engaging in different activities, or not engaged in any intentional action at all.
mental actions: judging, guessing, planning
As Kim observes (pp. 67, 70-1), insofar as behaviorist is trying to define or analyze mental talk into physical talk, they can’t appeal to (iii) or (iv). And insofar as the behaviorist is trying to capture what everyone who understands concepts like “pain” and “belief” will know about, they can’t appeal to (i). Insofar as behaviorist is trying to give central role to what is available as public evidence for attributing mental states to other people, they also can’t appeal to (i) or (iv). So it seems like behaviorists should restrict themselves to (ii).
A few notes about the terminology Putnam uses in his article.
At some points, he talks about the intension of the word “pain” — here he just means the meaning of that word or concept.
At the beginning of the article, you may notice that he says that behaviorism is an alternative to both dualism and what he calls “materialism” (or “classical materialism”). Here he’s using “materialism” in a different sense that we’ve been doing. We said that “materialism” (and also “physicalism”, which we’re using the same way) could be explained broadly, as the view that there are no souls, or more narrowly, as the view that mental facts supervene on physical facts. Behaviorists do count as materialists in both those senses. But Putnam is using “materialism” in a third, yet narrower sense. This is for the view that mental states are identical to brain states, the kind of view we saw Smart defending. For us, that’s just one kind of materialist/physicalist view. But Putnam calls it “classical materialism.” Some other authors call it “physicalism” or “type physicalism.” Sorry that the terminology is so messy.
One central element of Putnam’s paper are his examples of the Super-Spartans and the X-Worlders. The Super-Spartans sometimes feel pains, but they suppress all the usual behavior that goes with pain: involuntary wincing, screaming, talking in distressed voice, and so on. They may still verbally admit that they have pains, but they’ll do so in “pleasant well-modulated voices.” If we manage to convert some of them to our own, non-Spartan ideology, then they’ll start to exhibit the same pain behavior we do, and they’ll say their pains feel the same way they did before. It’s just that now they behave differently when they’re in pain.
This example seems to make sense. It doesn’t yet pose a challenge to the behaviorist, because the behaviorist can just say the Super-Spartans always had the behavioral dispositions that are definitive of pain. They just managed to suppress those dispositions (or perhaps we should say, their ideology kept them from ever being in the right triggering conditions).
The X-Worlder example is where the behaviorists start to have trouble. Putnam imagines these as a more extreme case of Super-Spartans, where they won’t even verbally admit to having pain (even if they’re in agony). Also we should imagine it’s impossible to convert them to our ideology.
Putnam says: “If this last fantasy is not, in some disguised way, self-contradictory, then logical behaviorism is simple a mistake.” Because if pain were defined as a package of behavioral dispositions, then it shouldn’t even be coherently imaginable that we have the one without the other. But the X-worlders seem to sometimes have pain without having the behavioral dispositions.
Putnam considers some objections to his view. Most of the discussion is about whether the hypothesis that the X-worlders do really have pain would be testable, or whether it’s something that in principle there could be no empirical evidence for or against. If the latter, then the behaviorist might argue that it’s a “pseudo-hypothesis” or not a real way for the world to be. But Putnam argues that there are various tests that could make it reasonable for us to think the X-worlders do have pain. We just wouldn’t be able to tell it from their calm, untroubled behavior.
A second element of Putnam’s paper is his discussion on pp. 46-48 of how words like “polio” and “multiple sclerosis” work. To understand this, first let’s familiarize ourselves with some medical terminology.
a symptom is an observable condition or outward sign of something. In medicine, this is usually an abnormal condition that might be evidence of some kind of disorder, disease, injury, or so on. In discussions of the mind/body problem, a person’s outward behavior will be treated as analogous to medical symptoms.
a syndrome is a group of symptoms that can come as a package.
Calling something a “syndrome” leaves open whether or not it has some single underlying cause. For instance the “S” in AIDS stands for “Syndrome”, but we now know that AIDS is caused by the HIV virus. Similarly, we know that Down’s Syndrome is caused by having a third copy of part or all of chromosome 21. For other syndromes we don’t know yet whether they have a single cause. For still other syndromes we do know that they don’t have a single underlying cause. For instance toxic shock syndrome and premenstrual syndrome (PMS) don’t have single causes. You count as having these syndromes just in case you have the package of symptoms, no matter how they are caused. Sometimes it’s controversial whether some medical condition is “merely” a syndrome (package of symptoms) or whether there is an single underlying unifying causal explanation of the symptoms. This seems to be the case for schizophrenia.
I’ll say “mere syndromes” when we want to talk about a package of symptoms, rather than their shared underlying cause (if they even have one). “Toxic shock syndrome” and “PMS” are names for mere syndromes.
In Putnam’s discussion, what he wants to oppose is the idea that words like “polio” started out just referring to a syndrome / package of symptoms. And then when we discovered that the syndromes had a single cause, we changed the meaning of the word to refer to the underlying cause instead. So the word had two meanings. With the older, syndrome notion, “having polio” just meant having the package of symptoms. With the new underlying-cause notion, “polio” instead means the presence of a specific virus. And now it could happen that someone had the virus (so had polio) but had different symptoms.
That’s the picture Putnam opposes. Instead, his view is that “polio” all along meant whatever the cause of the package of symptoms turns out to be. It’s just that when we started using the word “polio”, we didn’t know what that was. When we learned that the symptoms were caused by such-and-such a virus, we learned more about the same disease, polio, we were always talking about. We all along meant for “polio” to pick out the underlying cause, whatever it turns out to be. (Putnam doesn’t address how we should think about words where we think there could be a single underlying cause, but later learn there isn’t one.)
Here is Putnam articulating his view:
We observe that, when a virus origin was discovered for polio, doctors said that certain cases in which all the symptoms of polio had been present, but in which the virus had been absent, had turned out not to be cases of polio at all. Similarly, if a virus should be discovered which normally (almost invariably) is the cause of the what we presently call “multiple sclerosis”… the lexicographer [would] reject the view that “multiple sclerosis” means “the simultaneous presence of such and such symptoms.” Rather he would say “multiple sclerosis” means “that disease which is normally responsible for some or all of the following symptoms…”
Here is Putnam articulating the view he opposes:
Some philosophers would prefer to say that “polio” used to mean “the simultaneous presence of such-and-such symptoms.” And they would say that the decision to accept the presence or absence of a virus as a criterion for the presence or absence of polio represented a change of meaning. But this runs strongly counter to our common sense…
Putnam thinks we should think of the concept of “pain” as working the way he says that “polio” works.:
[The behaviorist’s opponents] would want to argue that, although the meaning of “pain” may be explained by reference to overt behavior, what we mean by “pain” is not the presence of a cluster of responses, but rather the presence of an event or condition that normally causes those responses. … [They won’t accept] that “pain talk” is translatable into “response talk”…
To sum up: I believe that pains are not clusters of responses, but that they are (normally, in our experiences to date) the causes of certain clusters of responses.
As we just saw Putnam emphasizing, we ordinarily think that your behavior is caused by your mental states. The behavior is not itself the same thing as the mental state. It’s only a sign or piece of evidence for what mental state you’re in — in the same way that smoke is only a sign that there is a fire. This is why we think it’s possible for someone (like Putnam’s X-Worlders) to be in pain without engaging in pain behavior. Sometimes effects can be brought about by different causes.
The behaviorist likes it that his theory doesn’t appeal to any hidden states of some non-physical substance, to explain why people behave the way they do. But on his theory, mental states can’t be inner states of the brain, either. It’s not clear he can say mental states play any causal role at all in producing our behavior. .
If Putnam is right that mental states like pain aren’t clusters of behavioral responses or dispositions, but are instead inner states of our brain that cause our responses, this sounds like it’s something that Smart’s identity theory (what Putnam calls “classical materialism”) captures better.
As we said earlier, the behaviorist has an easier time than other philosophers at explaining how we can know about other people’s mental states. On their view, this is no more difficult in principle than knowing that a glass is fragile. But it’s unclear whether the behaviorist can give a satisfying account of how we know about our own mental states. We seem to have direct access to our own minds. We don’t need to observe our behavior to know how we’re feeling.
This is a problem for all sorts of mental states — I don’t need to observe my behavior to know whether I’m thinking about elephants right now, or whether I believe that Charlotte is south of Carrboro. But the problem seems especially acute when we consider qualitative states like pain and sensory experiences. Whether I’m in pain, or having the “reddish” kind of visual experience I normally have when I look at a ripe tomato, seems to a matter of what’s going on in me right now, which I can be directly and immediately aware of. It doesn’t seem to be a matter of how I would act in various hypothetical contexts.
How one behaves depends on all the mental states one has. There’s no such thing as the distinctive behavior associated with a single mental state in isolation. Any plausible connection between mental states and behavior will have to invoke many mental states in explaining the behavior.
Consider a behaviorist account of wanting a Caribbean vacation:
X wants a Caribbean vacation iff:
- X is disposed to read the Caribbean brochures first, if you give him a bunch of holiday brochures — really? that assumes that X is not already bored of Caribbean brochures.
- X is disposed to pack his bags when you give him a ticket to Jamaica — really? that depends on whether he believes the flight will be canceled, whether he can afford to miss work, etc.
It is difficult for the behaviorist to say what sorts of behavior one mental state will dispose a person towards, without making assumptions about what other mental states the person has. This raises a threat of regress or circularity: it seems impossible for the behaviorist to define any one mental state until he has already defined lots of other mental states. So the behaviorist’s account will never get off the ground.
We will return to this problem in upcoming classes, when we consider functionalist accounts of the mind. Like the behaviorist, the functionalist also thinks there are important and intimate connections between mental states and dispositions to behavior. However, the functionalist wants to improve on behaviorism in the same way the identity theorist does: by making a person’s mental states internal states which cause the person’s behavior. The functionalist also wants to allow that a person’s behavior is typically produced by several mental states working in conjunction. So there is no one-to-one correspondence between what one’s mental state is, and how one behaves. Like the behaviorist, then, the functionalist also faces a threat of regress or circularity: it will seem impossible for him to describe the connections between any one mental state and behavior, until he has already defined lots of other mental states. We will see how the functionalist responds to this circularity problem later.
Here is Kim discussing this objection (pp. 72-3):
[L]et us see how well the proposed definition of belief works as a behaviorist definition. Difficulties immediately come to mind. First, as we saw with Hempel’s “toothache” example, the definition presupposes that the person in question understands the question “Is it the case that p?” — and understands it as a request for an answer of a certain kind. (The definition as stated presupposes that the subject understands English, but this feature of the definition can be eliminated by modifying the antecedent, thus: “S is asked a question in a language S understands that is synonymous with the English sentence ‘Is it the case that p?’”) But understanding is a psychological concept, and if this is so, the proposed definition cannot be considered behavioristically acceptable (unless we have a prior behavioral definition of “understanding” a language). The same point applies to the consequent of the definition: In uttering the words “Yes, it is the case that p,” S must understand what these words mean and intend them to be understood by her hearer to have that meaning. It is clear that speech acts like saying something and uttering words with an intention to communicate carry substantial psychological presuppositions about the subject. If they are to count as “behavior,” it would seem that they must be classified as type (iii) or (iv) behavior, not as motions and noises.
A second difficulty (this too was noted in connection with Hempel’s example): When S is asked the question “Is it the case that p?” S responds in the desired way only if S wants to tell the truth. Thus, the condition “if S wants to tell the truth” must be added to the antecedent of the definition, but this again threatens its behavioral character. The belief that p leads to an utterance of a sentence expressing p only if we combine the belief with a certain desire, the desire to tell the truth. The point can be generalized: Often behavior or action issues from a complex of mental states, not from a single, isolated mental state. As a rule, beliefs alone do not produce any specific behavior unless they are combined with appropriate desires. Nor will desires: If you want to eat a ham sandwich, this will lead to your ham-sandwich-eating behavior only if you believe that what you are handed is a ham sandwich; if you believe that it is a beef-tongue sandwich, you may very well pass it up. If this is so, it seems not possible to define belief in behavioral terms without building desire into the definition, and if we try to define desire behaviorally, we find that that is not possible unless we build belief into its definition. This would indeed be a very small definitional circle.
Even if we do manage to identify single mental states that are associated with behavior, or we manage to earn the right to talk about clusters of mental states being associated with behavior, isn’t it a problem that different species may exhibit a mental state like pain in very different ways (see Kim pp. 77-78)?
This is analogous to an objection we saw when discussing identity theory. There, we considered whether there was some single physiological state that all species (even non-Earthling species, even merely possible species) will always be in when they have pain. With behaviorism, the objection is similar, except that now we’re talking about how creatures behaviorally express their pain, rather than what kind of physiological state they’re in. Is there any kind of outward behavior that all these creatures will be disposed to exhibit when they’re in pain? Couldn’t it be instead that different species express their pain very differently?
We’ll return to this and related issues again in upcoming classes.
Above we considered the question of what kind of “behavior” the behaviorist might be talking about. Remember that two of the options were:
bodily movements, physically described
actions involving your body, described in mentally-loaded terms (greeting your friend, writing a check)
Contrast raising your arm (mentally-loaded behavior) with your arm moving in such-and-such ways (physically-described behavior). Contrast intentionally saying that you want to go to the Caribbean (mentally-loaded behavior) with your mouth opening and certain noises coming out (physically-described behavior).
We asked a moment ago whether there’s some one specific kind of physically-described behavior which all pain-feeling organisms engage in (humans, cats, snakes, Martians, etc.)? It’s not clear there will be. Is there some one specific kind of physical noise which everyone who believes that Charlotte is south of Carrboro makes, when asked where Charlotte is located? Not clear there will be. Behaviorists could be accused of a kind of chauvinism, if they assumed that all creatures will express themselves in the same physical ways that we do.
Going in the other direction, just because you utter the sounds “I want to go to the Caribbean,” does that mean you want to go to the Caribbean? Perhaps you’re being deceptive. Or perhaps you didn’t even mean to be asserting anything: perhaps you made those noises as the result of a complicated hiccough.
The upshot of these observations seems to be that we can’t find any behavior described in purely physical terms which is invariably and definitionally associated with any mental state. For instance, the behavior associated with the desire to go to the Caribbean can’t just be making certain noises. It has to be making those noises intentionally and sincerely.
But if the behaviorist says this, then he’s no longer talking about purely physical behavior! He’s appealing to mentally-loaded behavior. And to understand that kind of behavior, we have to know what people’s reasons and intentions are. We have to know what their social customs are. And so on. Thus the behaviorist won’t have shown us how to understand mental talk just in terms of physical talk. He hasn’t succeeded in analyzing or defining facts about our mental life into purely physical facts about us. He’s covertly re-introducing some kinds of mental facts when he talks of mentally-loaded behavior.
Hence, the behaviorist faces a dilemma. Either he says:
But if he says that, then the claims he makes won’t be plausible. There is no single physical way in which all creatures who are in pain behave. Alternatively, the behaviorist can say:
But if the behaviorist says this, then he hasn’t succeeded in defining mental facts in terms of purely physical facts.
Here is Kim discussing this objection (p. 75):
Let us now turn to another issue. Suppose you want to greet someone. What behavior is entailed by this want? As we might say, greeting desires issue in greeting behavior. But what is greeting behavior? When you see Mary across the street and want to greet her, you might wave to her, cry out “Hi, Mary!” The entailment is defeasible since you would not greet her, even though you want to, if you also thought that by doing so you might cause her embarrassment. Be that as it may, saying that wanting to greet someone issues in a greeting does not say much about the observable physical behavior, because greeting is an action that includes a manifest psychological component (behavior of type [iii] distinguished earlier). Greeting Mary involves noticing and recognizing her, believing (or hoping) that she will notice your physical gesture and recognize it as expressing your intention to greet her, and so on. Greeting obviously will not count as behavior of kind (i) or (ii) — that is, a physiological response or bodily movement.
But does wanting to greet entail any bodily movements? If so, what bodily movements? There are innumerable ways of greeting: You can greet by waving your right hand, waving your left hand, or waving both; by saying “Hi!” or “How are you?” or “Hey, how’re you doing, Mary?”; by saying these things in French or Chinese (Mary is from France, and you and Mary are taking a Chinese class); by rushing up to Mary and shaking her hand or giving her a hug; and countless other ways. In fact, any physical gesture will do as long as it is socially recognized as a way of greeting.
And there is a flip side to this. As travel guidebooks routinely warn us, a gesture that is recognized as friendly and respectful in one culture may be taken as expressing scorn and disdain in another. Indeed, within our own culture the very same physical gesture could count as greeting someone, indicating your presence in a roll call, bidding at an auction, signaling for a left turn, and any number of other things. The factors that determine exactly what it is that you are doing when you produce a physical gesture include the customs, habits, and conventions that are in force as well as the particular circumstances at the time — a complex network of entrenched customs and practices, the agent’s beliefs and intentions, her social relationships to other agents involved, and numerous other factors.
Considerations like these make it seem exceedingly unlikely that anyone could ever produce correct behavioral definitions of mental terms linking every mental expression with an equivalent behavioral expression referring solely to pure physical behavior (“motions and noises”).