ABSTRACT

In evolutionary terms, reading is a recent development, and the way in which different writing systems have developed indicates some of the constraints governing visual processing in reading. The deficits can be thought of in terms of a “multiple level” account of visual word recognition. For instance, some patients may be impaired at accessing whole-word descriptions, forcing reading to operate at letter or letter cluster levels. Other patients seem impaired at single letter coding, forcing reading to operate using (imprecise) whole-word descriptions. The essence of the logogen model is that word representations become abstracted from specific encounters with words, and so can be used to recognise words in different type faces, etc. The word recognition system is thus separated from an episodic memory system that holds information about specific episodes. The distributed memory account stresses the importance of both the initial episodic activation generated by stimuli and the creation of “abstract” long-term representations.