Sunday, March 11, 2007

Jottings on Category Mistakes

One of the complaints sometimes made against Atran's work in the cognitive science of religion is his terminology, namely, the way he uses terms like 'counterintuition', 'counterfactual', 'category mistake', 'pseudo-proposition,' etc. I made that complaint on my first reading of Atran. The point is not that the work itself is bad, but that the terms in which it is formulated are misleading, and sometimes even tendentious. The misuse of the term 'category mistake' is particularly egregious, because it's on the basis of classifying them as category mistakes that Atran makes the claim that counterintuitive propositions lack truth value. But since the sort of 'category' Atran has in mind is simply a sort of schematic expectation -- in the sense, for instance, that we expect inanimate objects not to talk -- it doesn't seem that there's any way to underwrite this reasoning. In response to some criticisms, Atran and Norenzayan argue (PDF, in the Authors' Response) this way:

The critique of our use of category mistake as “lacking truth conditions” follows in the same vein. From Aristotle and Kant to Gilbert Ryle and A. J. Ayer, a category mistake has been taken to occur when someone applies a concept that violates the necessary conditions of its application. Statements containing category mistakes lack truth conditions in that they can say nothing true or false. They are literally meaningless. Violations of necessary conditions may involve rupturing the logical, or analytic, structure of conceptual definitions (e.g., “the bachelor is married”), or sundering the synthetic a priori structure that the mind imposes on understanding the world (e.g., “the dead are resurrected”). Religious belief may involve both violations (e.g., “the dead are alive”), but we are most concerned with misapplication of concepts outside the commonsense ontological domain, or category, to which they meaningfully apply.


Category mistakes are tricky to pin down, but I'm inclined to think that this is a horrible defense. Suppose that a category mistake does certainly occur when someone applies a concept in a way that violates the necessary conditions of its (proper) application; it is not immediately inferrable, however, that statements containing category mistakes lack truth conditions. For instance, the following sentence has a category mistake and can be false:

I see all these colleges and libraries, but I don't see the University of Oxford.

The category mistake consists in thinking that the University of Oxford is something else to be seen besides its colleges and libraries. This is genuinely a mistaking of categories; it's false, because in seeing the colleges and libraries of the University of Oxford you are seeing the University of Oxford, in the only sense in which you can see it. The concept of the 'University of Oxford' is a concept in a different category than the concepts of its colleges and libraries; by treating it as if it were in the same category, the speaker makes a mistake and says something false about the University of Oxford. Another category mistake of the same sort:

I've visited Dallas, but I've never visited Texas.

But, of course, in visiting Dallas you do visit Texas; it's just that Texas is in a different category than Dallas. Anyone who says this has both committed a category mistake and said something false. So it's nonsense to think that because something is a category mistake it is meaningless, in the sense of lacking truth conditions. Since statements containing category mistakes are mistakes, they are very often false.

However, it is possible to say meaningless things due to a category mistake, e.g.:

The test and the chairs were equally hard.

The reason this is accounted as 'meaningless' is not that it can't be understood but that, assuming it is not a joke or a figure of speech, there's no way one could treat the the hardness of a test as commensurable with the hardness of the chairs; they are just radically different things, measured in radically different ways. Note, incidentally, that one could, with a slight modification, be saying something true while still making a category mistake:

The test and the chairs were hard.

If 'hard' here doesn't implicitly mean 'both hard in different senses of the term', there is a potential category mistake here. But it still could be the case that the tests were hard and the chairs were hard, and that this is all that's really intended; the category mistake is simply the confusion of thinking that these two things are the same sort of thing. The presupposition about the categories is false; but the statement as intended and understood is true. Unlike the case involving 'equally', it's easy enough to establish the truth conditions of it: you just see if the test was hard and if the chairs were hard; the category mistake is presupposed by the formulation of the meaning and not by the meaning itself. It still allows you to ask the questions that make it possible to establish truth conditions. In the 'equally' case, however, the only way it can be understood is by taking the sentence as itself committing a category mistake; we can't even formulate decently the questions that would be required to establish truth conditions.

So sentences involving category mistakes may be false, true, or meaningless; and the reason for this is the obvious fact that a category mistake is something you do, not something you say. So making a category mistake may result in a sentence that makes no sense, is clearly false, or is accidentally true, just as in other mistake of thought might.

One of the problems with the whole notion of a 'category mistake' is that it can sometimes be very difficult to determine whether one is being made. Take the above statement about Texas. Suppose it is not understood hyperliterally, but simply as the claim, "I've visited Dallas, but I've never visited (all of) Texas." This is not a category mistake. Category mistakes only rarely can be pinned on single statements, because many category mistakes are expressed in ways that admit of non-mistake interpretations. Only further inquiry can uncover it; if someone gave the above statement, we'd have to ask what they meant in order to determine whether they had made a mistake or not.

It is a question of some complexity whether figures of speech are category mistakes, or whether considering figures of speech as category mistakes is itself a category mistake, one that confuses figurative and nonfigurative speech. If, in the claim 'The dead are alive', the term 'alive' is not taken in its usual sense of not being dead, then all bets seem to be off, in any case; there might not be, for all we can say, a confusion of categories here.

In any case, the sort of thing that Atran and Norenzayan identify as their relevant sense of category (commonsense ontological domains) is not the sort of thing that could support the inference to meaninglessness even if all category mistakes resulted only in meaningless statements. Analytic and synthetic categories are both necessary for thought; we can't actually think of things that violate them, e.g., bachelors that are married or the addition of one physical object to another of the same kind, both discrete, without there being two physical objects. But commonsense ontological domains are not necessary for thought; we can think of gods and black holes and many other things that do not fit them. They are not essential to thought; they are just common-sensical. And common sense, unlike analytic necessities or synthetic preconditions of thought, is sometimes violated by reality. Thinking that common-sense categories are like analytic categories or synthetic a priori categories simply because they can all be called categories is itself a category mistake.

(Incidentally, it's worth pointing out that 'The dead are resurrected' doesn't violate any synthetic a priori categories, because whether the dead can be resurrected or not is determined a posteriori. Synthetic a priori categories have to do with things that are required for having experience of the world at all. Nothing changes about our basic experience of the world if it turns out that the dead are resurrected. Contrast this with the possibility of an empirically verifiable event that takes place outside of all space and time.)

So the justification for the terminology is just very bad, and the terminology is misleading. But, of course, this doesn't mean anything with regard to Atran's basic positions; it just means that we have to stipulate the definitions arbitrarily. That's fine; scientists do this all the time. The only danger is when people fail to realize that this is what they are really doing. In the case of Atran, we can make perfect sense of most of his claims when we grant his stipulations; the only one I can think of that would potentially be a problem is the claim that religious statements are meaningless, because he takes 'meaningless' in a technical sense to mean not having truth conditions and it seems to be based solely on the above reasoning.