Abstract
General recognition theory (GRT, Ashby & Townsend, 1986) can be used to formally represent the hypothesis of perceptual separability, which can be tested using response frequency data collected in a complete identification (CID) task. Systems factorial technology (SFT, Townsend & Nozawa, 1995) can address four fundamental characteristics of information processing (architecture, stopping rule, capacity, and independence in rate) using the distributional behavior of RT in a double-factorial paradigm (DFP). GRT and SFT characterize different aspects of human information processing that, to date, have been considered separately. We report on the results of one part of an empirical effort to unite these two theoretical lines using the "speedometer" stimuli, that previously have been shown (Kadlec & Hicks, 1998) to be perceptually separable. Stimuli were presented in both CID and DFP tasks, first using a payoff matrix that encouraged unbiased responding and second using a payoff matrix biased in favor of specific responses. A total of 4 observers were tested. We found that, as expected, the bias manipulation systematically moved the decision criterion either away from or toward the favored category. Perceptual separability was more likely to hold in the unbiased condition than the biased condition, regardless of the valence of the bias, as evidenced by the tests of marginal response invariance and report independence under the framework of GRT. The survival interaction contrast from SFT indicated parallel exhaustive and coactive manners were implemented in both unbiased condition and biased condition. The conclusion drawn from GRT and SFT further confirms the inference that the investigated stimulus is perceptually separable. The results indicate that combining the measures of GRT with those of SFT provides greater strength in making inferences regarding the nature of individual information and processing.
Meeting abstract presented at VSS 2017