Pattern recognition is deemed central to cognition. It appears to follow an optimal algorithm in a wide variety of behaviors and situations. Perceiving Talking Faces proposes an invariant law of pattern recognition to describe how multiple sources of continuously perceived information, such as auditory and visual speech input, are processed to achieve perception of a category.

The book details the author's extensive series of experiments on the use of multiple cues in speech perception and other domains, and unifies the results under the fuzzy logical model of perception. It presents a blend of empirical results, theory, and mathematical model testing. The author's underlying purpose is to describe and defend a universal principle of psychological theory.