Comments

1. #1 Chris

   April 28, 2006

   Some nice greatest hits in there. I have to admit that the moment I read the title “Is the mind like a computer? Evidence that it’s not,” I knew what was coming. I couldn’t guess exact experiment, but I knew whose experiment it would be. I have been in an extended discussion with Barsalou about this work since the ’99 BBS paper. Some of the best, or at least some of the most challenging experimental work on concpets going on right now comes out of the Barsalou lab, but their interpretations always go way beyond their data. Longer reaction times after switching modalities = perceptual simulations? Huh? It definitely implies that modality-specific priming is going on (the interesting follow up would have been to show that reaction times for same-modality trials were faster than when the second property they verified was modality neutral, like “good”), but simulations?

   I’d bet you could not only model this with something distinctly computationalist like the latest version of ACT, but I’d also bet a spreading activation account of semantic memory would predict this. That’s pretty much true of all of Barsalou’s experimental results, which shows the real problem in deciding between what sorts of representations to use: just about any result can be modeled, and perhaps even predicted, using modal or amodal representations. Barsalou makes a big deal about the fact that no one actually went about predicting these things until he came up with perceptual symbol systems theory, but that’s because perceptual symbol systems theory has led him to focus on specific types of stimuli, and not necessarily because it makes unique predictions. It’s essentially the whole Kossyln-Pylyshyn debate all over again, and it will probably result in the same conclusion.