Evaluation of a dynamical model of speech perception

Area of Science:

• Cognitive Science
• Psycholinguistics
• Dynamical Systems Theory

Background:

• Speech sound categorization exhibits properties of nonlinear dynamical systems.
• Previous models utilized a two-well potential function sensitive to stimulus properties and context.

Purpose of the Study:

• To evaluate the prediction that the rate of potential slope change increases with repeated category perception.
• To investigate context-sensitive dynamical representations in speech perception.

Main Methods:

• A "say"-"stay" speech continuum was used with varying gap durations.
• Participants identified speech tokens and rated category membership goodness.
• Judged goodness served as an index for the potential's slope.

Main Results:

• Identical stimuli were perceived as better category exemplars when presented at the end of a sequence compared to the beginning.
• No such enhancement was observed for stimuli presented mid-sequence with intervening random stimuli.

Conclusions:

• Results support the hypothesis of context-sensitive dynamical representations in speech perception.
• Repeated exposure to a category strengthens its representation, influencing subsequent perception.