Philosophy 257 Behavior and Other Minds

Asst. Prof James Pryor Dept. of Philosophy

Criteria

What is a criterion? From the way Malcolm uses this term, we can distinguish the following features:

1. A criterion for p is something that gives you reason to believe that p.

2. It's part of understanding what it is for p to obtain that you recognize of p's criteria that they give you reason to believe that p.
   Some corollaries of this are:
   • You don't require any sort of inductive argument (or inference to the best explanation) to justifiably believe p on the basis of criteria for p.
   • It's knowable a priori of p's criteria that they give you reason to believe that p.

3. Knowing that a criterion for p is satisfied enables you to establish with certainty that p is true. (It's not clear how Malcolm is understanding "certainty." See Notes.)

4. There is a "logical gap" between the presence of p's criteria and p's being true. There are conceivable situations in which the criteria are present but p is false. (Hence Malcolm can't be understanding "certainty" to mean that you have conclusive evidence for p: that is, evidence which it's impossible to have when p is false.)
   This suggests:
   • A criterion for p only gives you a defeasible basis for believing p.

An Important Contrast

BE. Such-and-such behavior gives you reason to believe the agent is in pain.

• A criteria theorist says that BE is knowable a priori, because it's part of the meaning of "pain." (This view is often combined with a conception of the facts about other minds as "epistemically constrained." We'll discuss this further when we take up Dummett.)

• Some philosophers (Chisholm, BonJour, Price) agree that BE is knowable a priori, but they take BE to be a substantial philosophical claim, a deliverance of a certain sort of normative epistemology. (They don't conceive of facts about other minds as "epistemically constrained"; and they don't think that accepting BE is necessary for competence with the concept of pain.)

• The theorist who defends a "perceptual" account of the epistemology of other minds also says that BE is knowable a priori, but that's because he understands the "behavior" in question to be mentally-loaded behavior.

Why Should We Believe in Criteria?

1. If you know about others' pains by induction from their behavior, then you have to know what pain is "from your own case."

In fact, it's not so much the assumption that you acquire the concept of pain "from your own case" that the criteria theorist takes issue with. Rather, it's the following, more general claim:

2. It's possible for you to know what pain is without yet knowing what would justify the ascription of pain to others.

Well, what's supposed to be wrong with 2? One line of argument against 2 goes as follows. Suppose you accepted the following:

3. It's impossible to speak a private language; so your mastery of any concept has to be subject to external checks.

4. If you know what pain is, then it has to be possible for someone else to know that you know what pain is.

Many proponents of criteria believe that 2 and 4 are incompatible. They believe that no epistemology which permits 2 to be true can also respect 4. Rather, they argue:

5. If 4, then knowing what pain is has to involve a recognition that certain sorts of behavior count as good reason to believe that the agent is in pain.

Accepting the consequent of 5 is tantamount to accepting the existence of criteria for pain. (I'm ignoring the requirement that criteria confer "certainty.")

I granted 4 for the sake of argument. So the important question is: why should we believe 5?

We haven't yet seen any argument to show that 2 and 4 are incompatible. If it were possible to know that someone else were in pain by an argument from analogy, or by inference to the best explanation of their observed behavior, then perhaps it would be possible to understand what it is for someone to be in pain, without yet having recognized what sorts of behavior warrant the ascription of pain to others. So if the argument from analogy or inference to the best explanation were viable epistemologies of other minds, then 2 and 4 would seem not to be incompatible. If that were so, then we ought not to accept 5.

So if the criteria theorist want us to accept 5, he has to supply independent arguments that the argument from analogy and other epistemologies of other minds which permit 2 to be true won't work. So we haven't yet seen a convincing argument from the impossibility of private languages against the argument from analogy and similar epistemologies. Even if we accept the impossibility of private languages, the burden of proof remains on the criteria theorist. (Malcolm seems to overlook this.)

Doubts about Criteria

To see this, consider concepts like "lunacy" and "puncture," that assume certain causal connections (lunacy is madness with a lunar cause; a puncture is the insertion of a sharp object which causes deflation). The usefulness and perhaps even the existence of such concepts depends on our belief that there is a causal correlation of the sort the concept assumes; but the existence of the concept does not guarantee that the correlation is genuinely there, nor that, if it's there, it will continue.

Why shouldn't the same go for concepts that assume certain evidential connections between behavior and the mental life of others (as Malcolm says "pain" does)? If the evidential connections are already in place, then we're entitled to use concepts like Malcolm's "pain," which apply only when the evidential connections hold; but we can't establish that the evidential connections are there just by using concepts which assume they're there. We need an argument that concepts which make such assumptions do genuinely apply to things.