Testing a semistochastic variant of the interactive activation model in different word recognition experiments.

A semistochastic variant of the interactive activation (IA) model of context effects in letter perception (McClelland & Rumelhart, 1981) was used to simulate response time distributions and means in different experiments investigating the effects of word frequency, neighborhood size and frequency, and orthographic priming in visual word recognition. The results provide evidence in favor of the connectivity assumption underlying the model but question the necessity of the interactivity assumption for simulating latencies in word recognition tasks. Together with those of a recent study by McClelland (1991), the present results suggest that 10 years after its appearance, the IA model's potential for testing hypotheses about the structure and dynamics of basic phenomena of human information processing in a variety of perceptual and cognitive tasks is not yet fully exploited.