I'm a superficialist about belief. On my view, to believe something is to match, to an appropriate degree and in appropriate respects, a "dispositional stereotype" composed of various behavioral, experiential, and cognitive dispositions. To believe something is just to be disposed to act and react as though you believe it -- outwardly in your planned and spontaneous behavior and inwardly in, for example, your inner speech, emotional responses, and implicit reasoning. (For more details, see here and here.) I prefer not to conceptualize belief structurally, for example in terms of the manipulation of representations stored in one or more functionally defined "boxes". It's this structural neutrality that makes my account superficial. Belief, on my view, concerns what's happening at the behavioral and experiential surface of cognition (and in patterns of transition between similarly superficial states). Facts about underlying architecture are irrelevant, except to the extent that of course there must be some underlying architectural facts or other that give rise to the surface patterns.
[Image: Cartesian vortices, from p. 55 of Le Monde]
I like to compare the architecture of belief to the architecture of personality traits. What is it to be extraverted? Plausibly, it's nothing more than to have a certain suite of dispositions -- being disposed to enjoy parties, to be talkative, to feel energized by meeting new people, to seek out social contact, etc. If you're like that, you're an extravert, and it doesn't really matter what underlying cognitive architecture implements that pattern. Maybe you have a big "E" written in some extraversion box. Maybe you've got some representation like "meeting new people is good" in your Value Box. Maybe your alpha-dongle is hooked up to your beta-flipper. It doesn't really matter. As long as you are robustly disposed to act and react, inwardly and outwardly, like an extravert, you're an extravert.
Now one disadvantage of superficialism, which has been repeatedly emphasized to me by Eric Mandelbaum (including in a published critique of my view with Jake Quilty-Dunn here) is this: Superficialism is not as explanatorily rich as a view that commits to an underlying architectural story.
Why did the extravert enjoy the party? The superficialist will have to say something quasi-circular like, "Well, she's an extravert, and extraverts tend to enjoy parties". Someone with an alpha-dongle explanation can seemingly do better: "See, it's because the party released gamma juice, flooding the alpha-dongle which then made the beta-flipper go wild!" See how much more satisfying that explanation is. More satisfying, anyway, if there really is gamma juice and an alpha-dongle that we can see and measure, lining them up with the party and the mood. An explanation that goes from the observed surface to an inner functional architecture and then back up to another observed surface will, all else being equal, be better than an explanation that just says that the two observed surfaces tend to be associated.
Back to belief: I ask Alyssa, "What is the capital of California?" and she answers "Chicago". Why? The superficialist explanation is this: To believe that Chicago is the capital of California is just to be disposed to act and react in a variety of ways, including by answering "Chicago" to questions of that sort, and Alyssa fits this pattern. (Compare again to "why did the extravert enjoy the party?) A certain type of representationalist about belief has a seemingly more satisfying answer: To believe that Chicago is the capital of California is to have the representation "Chicago is the capital of California" stored in memory. When asked the question, that representation was retrieved from memory and processed in such and such a way, generating the answer "Chicago".
The representationalist's explanation sounds (superficially?) better. But is it better? That depends, I submit, on how good the architectural story is. I don't think the architectural story is all that good. Or, less commissively, I don't think the philosophical community is currently in a position to know that it's good in the way we ought to know that it's good before hanging our concept of belief on that story.
Consider this analogy. It's the late 1500s to mid 1600s, post-Copernicus, pre-Newton, with Tycho and Galileo and Kepler and Descartes and old-school Ptolemaists all theorizing about the planets. Everyone knows, of course, that Mars will trace such-and-such a path through the sky this coming August. But why does Mars do that? What's the explanation?
Structuralists offer various explanations: It's riding on a crystalline sphere! It's hooked to the Sun on magnetic chains! It's riding in a vortex of globules!
Superficialists refrain from such structurally commissive explanations, instead simply fitting the planetary motion into a pattern: that it will appear in such-and-such a trajectory in August fits with the general pattern of what we know about its motion, its overall pattern of progression and retrogression mapped over the years. The superficialist points to the predictive equations as explanation enough for now, without positing an underlying physical mechanism.
Now if it's the case that one of the structural explanations is correct, then indeed that structural explanation is better than explanation by showing how a particular incident fits within a larger superficially observed pattern. But if we don't yet know which if any of the structural explanations is correct, superficialists have the advantage of firmer ground: They refrain from committing prematurely. The best science of the day might suggest that it's vortices, but let's wait before going all-in on that. Certainly let's wait before building our philosophical definition of "planet" in terms of vortices. The superficialist will say, superficially, "to be a planet is to be one of those things we see in the sky that tend to move like this"; the structuralist will say, "to be a planet is to be a huge sphere that rides around the Sun in a vortex of globules".
Here's where I differ from Mandelbaum and other non-superficialists. I think we're still in the pre-Newtonian days regarding the cognitive architecture of belief. Storing representations with the content P in various functional boxes in the mind, then retrieving them when relevant for use in theoretical or practical inference, coupling P with other belief-representations like "P -> Q" or with some desire-representation R to generate intention T -- that's a cartoon story which can be useful for certain purposes, but it's probably not really what the cognitive architecture of belief is like underneath for at least relatively complicated beliefs like "there's a gas station on the corner" or "my children's happiness is more important than their success in school".
Until we have the right architecture, and know we have it, it's better to go superficial -- at least insofar as we are philosophers interested in a conception of belief that's empirically robust. The representationalist's appearance of a better explanation is as misleading as the vorticist's or crystalline sphere lover's appearance of a better explanation of planetary motion.
Of course, it's good that not everyone is as restrained as the superficialist is. It's good that Descartes pushed vortices and Ptolemaic astronomers pushed crystalline spheres and Kepler hypothesized about the magnetic power of the Sun. They were all overconfident and wrong! But where would science be without them?
----------------------------------------
Acknowledgement: This post was inspired by conversations with Eric Mandelbaum and others in a series of workshops run by Jonathan Jong under a Templeton grant.
Also: Superficialism does a better job with weirdly constructed aliens (Section 6 here). And in-between belief. And implicit bias. And it better captures what we do and should care about in thinking about belief. But never mind!
12 comments:
Interesting post. Thanks!
Yet I’m having a bit of trouble seeing exactly what the disagreement is. As you admit, there must be some architectural facts that give rise to the surface patterns. And I imagine your opponent would admit that the surface patterns are important. Is the disagreement about what is *essential* to belief? While you say that S believes that p only if certain surface patterns exist in S, your opponent disagrees. And while your opponent says that S believes that p only if such and such architectural facts obtain in S, you disagree. Is that it?
It’s difficult to evaluate these claims given that the surface pattern is, as you admit, connected to the architecture. So it’s hard to evaluate the relevant counterfactuals, and thus hard to evaluate who is correct (if this is in fact the disagreement).
I take it that the disagreement isn’t merely about what the architectural facts are. That is, I take it that the disagreement runs deeper than: your opponent says the architectural facts are such and such while you remain agnostic.
My wife and I agree...
With predisposed and disposed we found posed and pose...
What fun...thanks
Mike: For belief, the practical disagreement concerns when then superficial patterns and the purported architecture give divergent results. One case is implicit bias cases, e.g., the man who sincerely says women and men are intellectually equal but who regularly treats women as less intelligent than men in his daily interactions with them. On Mandelbaum's model, for example, he has the stored representation P ("women and men are intellectually equal") and the stored representation that not-P, and each drives some of his behavior. He believes that women and men are intellectually equal and he believes that they are not intellectually equal.
On my view, it's an "in-between" case, like a case where someone is extraverted in some ways but not in others. Just like it's not quite right to say of such a person (except in special contexts) that they are extraverted or not extraverted full stop, it's not quite right to say of the implicit sexist that he believes or that he fails to believe the proposition that women and men are intellectually equal. It's a complicated case where the superficial patterns break down. (If you're curious for more about this, see my 2010 "Acting Contrary to Our Professed Beliefs" and my forthcoming "Pragmatic Metaphysics of Belief".)
Googled: Pragmatic Metaphysics of Belief...
..."To really, fully believe, you must also "walk the walk". I argue that the pragmatic approach is preferable on pragmatic grounds: It rightly directs our attention to what matters most in thinking about belief." Schwitzgebel
At some point can pragmatic mean deed and ownership of what is in front of one...
A few questions: will the future answer lie in interaction among known dimensions, or will it involve some hitherto unknown dimension? Plus, why not just say we can use a combination of behavioral, neurological, personality theorist and so forth? Behavior and consciousness are more complicated than gravity after all
If you drop the experiential requirement on superficialism, will ascribing belief to rudimentary artificial agents.
In some contexts this seems fine, like if we're trying to predict how a particular agent will behave ("It believes that doing that will cause it to win, so it will do that").
In other contexts, we'll care more about whether there are representations.
Howie: I have no objection to neuroscientific and mid-level functional explanations. I just don't think we're past the crystalline sphere phase for neuroscientific and mid-level functional explanations of belief. Early visual processing, in contrast, is a case study in the productive mix of behavioral, introspective, neurophysiological, and functional explanation.
Caleb: Right! But the phenomenal dimension is important to me -- one of the most important differences between my view and Dennett's (which ends up ascribing beliefs to chess machines).
Thanks for the reply, Eric!
A great case for distinguishing the two views would be one where a person has a stored representation that p but lacks the surface patterns that on your account constitute believing that p. You’d say that she does not believe that p. Your opponent would say that she does. But her lacking the surface patterns is reason to deny that she has a stored representation that p, thereby dissolving the disagreement. Or so it seems to me. (Perhaps it’s being assumed that if a person sincerely says that p, then she has a stored representation that p.)
In the case of personality, we long had a model of it as a global property of multiple underlying systems (including motoric) eg "...Ss having high scores on the Taylor Scale of Manifest Anxiety tend to form conditioned eyeblink responses more readily...". These go back to Pavlov at least.
In the case of geographical knowledge, say, a dispositional model seems to end up merely adding "I have a disposition to" to every statement eg "I have a disposition to say that San Francisco is south of Sacramento, that Los Angeles is south of SF, that latitude differences can be ordered", when we already know a bit about how these kind of things are represented mentally and neurologically across species. Google searching on "vagueness" and "geographical knowledge" did take me down some strange byways.
Also, when dealing with conscious states such as states of "belief", a representationalist "explanation" would really just be a bunch of correlations. It might be highly predictive of physical behavior, and it might even give a complete physical account of how specific external stimuli lead to specific beliefs which lead directly to specific behaviors.
But it wouldn't even begin to explain *why* all of that results in subjective, first-person, representational experiences of the whole process. So in that sense, a representationalist "explanation" wouldn't get at what's *really* going on in regards to the experiential aspect of the process any better than the superficialist approach. It would be a"just so" story, too.
Thanks for the continuing comments, folks!
Mike: It's unclear how much they can pull apart, but Mandelbaum argues that representationalism results in quite a low bar for believing, such that it's possible with not much of the surface patterns.
David: Sure, multiple underlying systems for personality traits, and probably also for belief. Exactly which systems, and how realistically to construe the systems -- that's another question. On geographical knowledge, there's at least a possible pretty sharp dissociation between geographical knowledge stored in linguistic format and stored in map format (though each is probably a simplification and I see no reason to think that only those two formats are possible). A non-in-between case would require alignment in dispositions, whereas if conflicting formats produced robustly conflicting dispositions, that would be an in-between case.
Ryan: Partly this depends on how abstract or functional the representations are. Where there are concrete biological instantiations of representation (e.g., in early vision), convergent behavioral, phenomenal, functional, and biological evidence supports a robust realism. Let so for the representation "I believe men and women are intellectually equal", on my view.
Post a Comment