Previous: From Visual Stimuli to Color Space
Up: A Computational Model of Color Perception and Color Naming Next: Putting It All Together: From Visual Stimuli to Color Names, and Back
The second part of the mapping takes us from a color space to a
set of (basic) color names, each with a ``typicality'' or ``goodness''
measure associated. In Shepard's terminology, each color name corresponds
to a ``consequential region'' in psychological space. Categorizing a
stimulus then amounts to inferring the consequential region to which it
belongs [Shepard 1987], together with its goodness value. In this chapter
I describe the procedure I used for fitting a particular category model to
the experimental color naming data of [Berlin \& Kay 1969], present a theoretical
evaluation of the model, and outline a model for learning color names.