anwst@cisunx.UUCP (Anders N. Weinstein) (02/04/89)
I agree with Harnad on one point: the interpretation of "typicality effects"
in categorization phenomena as supporting the idea that category membership
must be graded looks like a non-sequitor. I would have thought that finding
such effects for manifestly definitional categories like "triangle" or
"even number" would make this quite obvious. One understands what an even
number is if one can state and apply the definition; the typicality effects
are of course real but should be viewed as epiphenomenal from the point of
view of understanding the concept.
If one is alerted to the anti-psychologistic bent of philosophy in the Frege-
Wittgenstein tradition, one will distinguish sharply between the normative
question of which things are *correctly* classified as birds or even numbers,
and the descriptive question of how in fact our neural machinery functions to
enable us to so classify things. I agree also with Harnad that psychology
ought to keep its focus on the latter and not the former of these questions.
However, I think Harnad errs when he succumbs to the tempting view that
reliable categorization *must* be interestingly describable as application of
some (perhaps complex) rule in "featurese" (for some appropriate set of
detectable features). How does this follow? Limiting ourselves (as I think we
must) to quick and automatic observational classification, it seems to be a
safe conclusion that we have some neurally realized recognition machine. If
we suppose, as seems reasonable, that the effects of context on such tasks
are minimal, we can conclude that there must be within us some isolable
module which can take sensory input and produce a one bit yes-or-no
output for category membership.
But how does it follow that such a device must be describable as applying
some *rule*? Any physical object in the world could be treated as a
recognition device for something by interpreting some of its states as
"inputs" and some as "yes-or-no responses." But intuitively, it looks like
not every such machine is usefully described as applying a rule in this
way. In particular, this certainly doesn't seem a natural way of
describing connectionist pattern recognizers. So why couldn't it turn out
that there is just no simpler description of the "rule" for certain
category membership than: whatever a machine of a certain type recognizes?
--
Anders Weinstein ARPA: anwst@unix.cis.pittsburgh.edu
U. Pitt. Philosophy UUCP: {cadre,psuvax1}!pitt!cisunx!anwst
Pittsburgh, PA 15260 BITNET: anwst@pittvmslee@uhccux.uhcc.hawaii.edu (Greg Lee) (02/04/89)
From article <15585@cisunx.UUCP>, by anwst@cisunx.UUCP (Anders N. Weinstein): "... But intuitively, it looks like " not every such machine is usefully described as applying a rule in this " way. I share your intuition. If you don't understand how the machine works, how can you give a rule? " In particular, this certainly doesn't seem a natural way of " describing connectionist pattern recognizers. So why couldn't it turn out " that there is just no simpler description of the "rule" for certain " category membership than: whatever a machine of a certain type recognizes? It could, clearly. It's like those puzzles where you're given a few numbers in sequence -- 1, 3, 22, 23, ... -- and asked to give the next one. Except here it's inscrutable Nature talking, so there's no reason to expect to find the answer, and we really should just reply: "Ha! I'm not going to fall for that one -- you tell me." Greg, lee@uhccux.uhcc.hawaii.edu