Talk:Categorical perception
From Wikipedia, the free encyclopedia
|
|||
| Is this a copyright violation, or does it meet deletion criteria? | ✔ pass | ||
| Is this a duplicate article? | ✔ pass | ||
| Add markup, links, layout etc. | (pending) | ||
| Rewrite where needed. | (pending) | ||
| Remove wikify tag. | (pending) | ||
| Help clear the wikification backlog! We need your help! | |||
Stevan Harnad is the expert, but as a former graduate student studying this area, I am somewhat familar with the theory, so I would like to recommend the following points. I think the aricle would be improved if it was written to include these additional points from the seminal research. It may make it slightly more complicated, but I think these points are important to capture the spirit of the original research and theory.
- The ubiquitous relationship between discrimination and identification found in the early, seminal studies: discrimination is no better than absolute identification. This is not included anywhere in this article, is it? Please correct me if I am wrong, but this distinction formed the basis for the definition of categorical perception. If a graph of the classic "inverted V" discrimination function was added, I think that could help explain the fundamental concept considerably.
- I don't understand your point. Discrimination is a relative judgments, such as same/different (for a pair of inputs). Identification is an absolute judgement, usually category-naming (for a single input). CP is defined as the discrimination function (peaks and troughs in discriminability) following the identification function (category boundaries): If its harder to discriminate differences between shades of green and shades of blue than differences between blue and green (when the size of the differences in log-frequency are equal) then you have a CP effect. The "inverted V" (peak) in the discriminability function is as at the category boundary for the blue/green identification function. Now the article n CP already says all of this: what do you want added? --Stevan Harnad
- Alternative accounts to CP; Researchers such as Dominic W. Massaro have pointed out for years that researchers have failed to consider alternative mathematical models that can account for categorical perception. The research in this area has consistently shown predicted discrimination based on identification nearly always underestimates observed discrimination. Massaro has shown that models of continuous perception predict observed discrimination better relative to categorical models when the root mean-squared deviation (RMSD) between observed and predicted values is used as a goodness-of-fit metric. Massaro has argued perception is essentially continuous - the perceptual system can discriminate within category - but the nature of the task participants have been given forces a categorical response. I think including this is important in an encyclopedia article of CP.
- CP never implied that there was no discriminability within a category. (Shades of blue do not all look identical: we can discriminate them. We just can't discriminate the as well as we can discriminate the same sized difference when it crosses the blue-green boundary. So what point of Massaro's do you think needs to be added? (I do think, though, that the CP article could be usefully updated to add the more recent findings on CP and neural imagery studies, as well as some of the animal studies on plasticity and learning. --Stevan Harnad
—The preceding unsigned comment was added by SdlV (talk • contribs) 23:24, 23 June 2006 (UTC)
