Abstract: Modeling the mechanism of suppression in lexical access

GERNSBACHER, M. A., & ST. JOHN, M. F. (in press). Modeling the mechanism of suppression in lexical access. In R. Klein & P. McMullen (Eds.), Converging methods for studying reading and dyslexia. Cambridge, MA: MIT Press.

The goal of our recent research has been to computationally model the mechanism suppression during language comprehension. A computational model of any cognitive process or mechanism serves several functions. It can be used as a tool for thinking mechanistically about a cognitive mechanismÑhow it works in general and how a variety of factors affect it. A computational model can test how adequately existing theories account for behavioral data, and it can generate predictions about behavior that have not been observed or investigated. In this chapter we present the initial development of a computational model of the mechanism of suppression in lexical access. Our model, which we shall describe more fully later, is based on St. John’s Sentence Gestalt (St. John & McClelland, 1990) and Story Gestalt model (St. John, 1992). These precursors offered a viable architecture for simulating the mental representations (structures) that Gernsbacher describes in her Structure Building Framework; therefore, we used the Sentence and Story Gestalt models as a starting point. Modeling lexical access is not a novel idea. A number of computational models of lexical access already exist, but none offers a satisfying account of suppression. The goal of our work has been to develop a computationally precise account of suppression and to demonstrate its operation in lexical access. Our chapter is organized in the following way. Primarily we present a detailed look at one of the language comprehension phenomena in which the mechanism of suppression plays a vital role: the phenomenon of lexical access. We begin by presenting the behavioral data that describe the role of suppression in lexical access; we conclude by presenting our preliminary simulations.