查看文章

The orientation specificity of perceptual learning suggests a plasticity site having units tuned for orientation: V1, V2, V4. Even if learning is localized to these early areas, there still are at least two distinct possibilities as to the under--lying mechanism. The representation enhancement hypothesis explains percep--tual learning in terms of rectruitment of new units, sharpening of tuning curves, and/or any other changes that improve the signal-to-noise ratio of the represen--tations. The selective reweighting hypothesis on the other hand explains it in terms of changes in the strength of the" read-out" connections to higher taskspecific areas (Dosher & Lu, 1998). Both hypotheses are equally consistent with the observed stimulus specificity of learning.

There is neurophysiological evidence that the receptive field properties in V1 and V2 in adult monkeys do not change (or change but a little) as a result of noninvasive practice alone (Crist et al., 2001; Ghose et al., 2002; Schoups et al., 2001). Representation enhancement thus seems insufficient to account for the marked improvement in performance observed in all three studies. Psychophysical demonstr-ations of task specificity in perceptual learning also pose challenges to the representation enhancement hypothesis (Fahle, 1997; Christ et al., 2001). The tasks used in these studies, however, tend to engage disjoint as--pects of the stimulus representation and hence do not provide a conclusive test.