Imperfect, Unlimited-Capacity, Parallel Search Yields Large Set-Size Effects

Many analyses of visual search assume error-free component processes. These analyses range from Sternberg's earliest serial models to Townsend's sophisticated theorems of serial-parallel equivalence. Consider a simple "yes-no" visual search task with a set-size manipulation. For the correct positive responses, an error-free, unlimited-capacity, parallel search model predicts no effect of set size on response time. In contrast, the prediction is quite different if the component processes are imperfect (even 1%-5% errors). Such imperfect processing is commonly assumed in stochastic models (such as a random walk model) but rarely have been applied to search. We have examined a stochastic model that assumes unlimited-capacity, parallel component processes which are imperfect. For this model, there are significant set-size effects for both positive and negative responses. In particular, the less discriminable the stimulus, the larger the set-size effect. In conclusion, a more realistic assumption about the error in component processing forces one to reconsider how to distinguish among serial and parallel models. Talk presented at the 1995 meeting of the Society of Mathematical Psychology, Irvine, CA. Send correspondence to: John Palmer University of Washington Psychology, Box 351525 Seattle, WA 98195-1525 E-mail: [email protected] 4 November 1995