ASA 129th Meeting - Washington, DC - 1995 May 30 .. Jun 06
5pPP1. A dynamic, psychophysical model of adaptation in localization
experiments.
Barbara Shinn-Cunningham
Res. Lab. of Electron., MIT, 77 Massachusetts Ave., Cambridge, MA 02139
Previous psychophysical models have described how resolution depends upon
the range of physical stimuli employed in a given experiment; however, these
models generally assume that subject performance is stable over time. The
current work extends the context-coding model of Durlach and Braida [
372--383 (1969); Braida and Durlach, 483--502 (1970)] to account for
changes in subject performance that occur when feedback is used to retrain
subjects during the course of an experimental session. In the current model,
observed changes in performance are accounted for by assuming a single
exponential adaptation process. This process, which describes the adaptive
state of the subject, determines the decision criteria and context (effective
stimulus range) used in the model at a given point in time. From the dynamic
decision criteria and effective range, and knowledge of the underlying
sensitivity to localization cues, the model predicts mean response, bias, and
resolution. The model predictions are in good agreement with results from a
relatively large body of experiments, including experiments in which the number
and range of stimuli were varied.