Thursday, November 02, 2017

Two Roles for Belief Attribution

Belief attribution, both in philosophy and in ordinary language, normally serves two different types of role.

One role is predicting, tracking, or reporting what a person would verbally endorse. When we attribute belief to someone we are doing something like indirect quotation, speaking for them, expressing what we think they would say. This view is nicely articulated in (the simple versions of) the origin-myths of belief talk in the thought experiments of Wilfrid Sellars and Howard Wettstein, according to which belief attribution mythologically evolves out of a practice of indirect quotation or imagining interior analogues of outward speech. The other role is predicting and explaining (primarily) non-linguistic behavior -- what a person will do, given their background desires (e.g. Dennett 1987; Fodor 1987; Andrews 2012) .

We might call the first role testimonial, the second predictive-explanatory. In adult human beings, when all goes well, the two coincide. You attribute to me the belief that class starts at 2 pm. It is true both that I would say "Class starts at 2 pm" and that I would try to show up for class at 2 pm (assuming I want to attend class).

But sometimes the two roles come apart. For example, suppose that Ralph, a philosophy professor, sincerely endorses the statement "women are just as intelligent as men". He will argue passionately and convincingly for that claim, appealing to scientific evidence, and emphasizing how it fits the egalitarian and feminist worldview he generally endorses. And yet, in his day-to-day behavior Ralph tends not to assume that women are very intellectually capable. It takes substantially more evidence, for example, to convince him of the intelligence of an essay or comment by a woman than a man. When he interacts with cashiers, salespeople, mechanics, and doctors, he tends to assume less intelligence if they are women than if they are men. And so forth. (For more detailed discussion of these types of cases, see here and here.) Or consider Kennedy, who sincerely says that she believes money doesn't matter much, above a certain basic income, but whose choices and emotional reactions seem to tell a different story. When the two roles diverge, should belief attribution track the testimonial or the predictive-explanatory? Both? Neither?

Self-attributions of belief are typically testimonial. If we ask Ralph whether he believes that women and men are equally intelligent, he would presumably answer with an unqualified yes. He can cite the evidence! If he were to say that he doesn't really believe that, or that he only "kind of" believes it, or that he's ambivalent, or that only part of him believes it, he risks giving his conversational partner the wrong idea. If he went into detail about his spontaneous reactions to people, he would probably be missing the point of the question.

On the other hand, consider Ralph's wife. Ralph comes home from a long day, and he finds himself enthusiastically talking to his wife about the brilliant new first-year graduate students in his seminar -- Michael, Nestor, James, Kyle. His wife asks, what about Valery and Svitlana? [names selected by this random procedure] Ah, Ralph says, they don't seem quite as promising, somehow. His wife challenges him: Do you really believe that women and men are equally intelligent? It sure doesn't seem that way, for all your fine, egalitarian talk! Or consider what Valery and Svitlana might say, gossiping behind Ralph's back. With some justice, they agree that he doesn't really believe that women and men are equally intelligent. Or consider Ralph many years later. Maybe after a long experience with brilliant women as colleagues and intellectual heroes, he has left his implicit prejudice behind. Looking back on his earlier attitudes, his earlier evaluations and spontaneous assumptions, he can say: Back then, I didn't deep-down believe that women were just as smart as men. Now I do believe that. Not all belief attribution is testimonial.

It is a simplifying assumption in our talk of "belief" that these two roles of belief attribution -- the testimonial and the predictive-explanatory -- converge upon a single thing, what one believes. When that simplifying assumption breaks down, something has to give, and not all of our attributional practices can be preserved without modification.

[This post is adapted from Section 6 of my paper in draft, "The Pragmatic Metaphysics of Belief"]

[HT: Janet Levin.]

[image source]


Anonymous said...

Do you think that there being two roles for belief attribution is ultimately an indication that there are just two different dispositional states at work? I know that in past papers, you've gone more for the view that there are simply in-between belief states rather than multiple conflicting states (as in your Juliet case), but it seems strange that someone could so vehemently attest to having certain views while clearly acting against those views if there were simply one half-baked state underlying it all. The "what do you *really* believe?" question gets at this idea, I think. There is a sense in these cases that while some dispositional state is responsible for the person's testimony, there is, as you say, a "deep-down" state responsible for their behavior which is the "real" belief at work.

In general, I'm curious about what you think the relation is between the roles and practice of belief attribution and belief states themselves, especially whether that relation (whatever it is) permits us to take the roles of attribution as evidence for a certain metaphysics of belief.

Eric Schwitzgebel said...

Thanks for the thoughtful question, Anon 11:07!

I'm inclined to think that the folk metaphysics of belief -- while not entirely developed and self-consistent -- tends to favor the idea that there is a single deep-down state that is the cause of the outward behavior, both the utterance "P" and the other belief-that-P-ish behavior. Cases like Ralph's create a prima facie challenge for this folk metaphysics of belief. There are ways of trying to resolve that challenge and say that either he really believes that P or really fails to believe that P. But my own view is that those resolutions don't entirely succeed and the folk metaphysics is mistaken.

Michael said...

About Sellars -- in EPM you will not find the account of belief-talk that you attribute to him. What you will find is an account of thought-talk. I think this matters. No time for explanations now (although there is an interesting discussion in a paper by Mitch Parsell, "Sellars on Thoughts and Beliefs," Phenomenology and the Cognitive Sciences 10 (2011): 261-275 -- one point is that thoughts are occurrences and beliefs are dispositions to have thoughts).

Michael Kremer

Eric Schwitzgebel said...

Thanks, Michael. Interesting. Looking back at EPM, it's true that he does shift from talking about beliefs to talking about thoughts in his Rylean ancestors section. I appear to have glided over a nuance in his view. I'll check out the Parsell. Thanks for the heads up.

howard b said...

You might ascribe your stick figure of a professor's twig like dispositional split to sociological roles, to further cloud the picture. Sociologists speak of a habitus for instance.
The reason might not lie in the stars or in our genes or dendrites but in the hat we wear and with whom we hang out
Just a stray thought to shower you with

Callan S. said...

How does his assertions work? It seems like it's just virtue signaling, with no actual self monitoring going on. Does he self monitor? Does he self monitor but fail to detect how he's going against his own professed position? Indeed the outcome where he now thinks women and men are equal seems to not find this self monitoring important. Somehow he just becomes that way.

If we were to draw a diagram, it seems something like this : His deeper feeling mind (circle A) commands his speaking/thinking mind (circle B) to say he believes in equality.

But this is a one way street, just going straight out with arrows from A to B, then from B out into the world. There is no command that instead of just saying he does X, to actually self monitor as to whether he does X or not. It's just a one way street right out to the lips - I do X.

Why have we got an example with so little said about his own state of self monitoring?

I could propose a reason: The speaking/thinking mind thinks it is in charge - that it is all there is to the equation. And yet it faces the anomaly of the scenario, where the fellows thinking mind says he does X, but he does not do X.

That anomaly would be better explained that the thinker is not in control in the way it thinks it is. Indeed in the way it is told to think it is in control.

Beyond the issue of sexism, could the character also be said to have fallen into a fallacy of self control - a fallacy that actually makes them more out of control, more stampeding off into directions that aren't to do with what they profess they believe?

Eric Schwitzgebel said...

Yes, something like that seems right to me, Callan -- though probably not quite as starkly as you sketch it, in real life cases. There's not as much self-monitoring as there should be; and part of that might be due to an (implicit?) model of the mind on which one's behavior in such matters reliably reflects one's sincerely expressed opinions, without further work required.

Callan S. said...

Is that a trend you're seeing, Eric?

I was thinking of asking if there are any statistics, but I guess this is kind of an invisible thing.