ABSTRACT
The category of multistage activation models includes models descended from the logogen model of word recognition (e.g., Morton, 1969) and from the interactiveactivation model of letter perception (e.g., McClelland & Rumelhart, 1981; Rumelhart & McClelland, 1982). Although multistage activation models differ in many important ways, they share three characteristics: (a) the models posit the existence of multiple levels of lexical-semantic representation, such as visual features, letters, words, and semantic representations; (b) the models employ excitatory feedforward and feedback connections between successive levels; and (c) each level of representation corresponds to a stage of processing. These properties may be needed to explain some of the more complex findings on semantic priming (see, e.g., Chapters 15 and 17).
