Pages 9

Given that we have just concluded that valid DD does not arise in fully distributed connectionist systems, it is not surprising that Plaut’s well-known paper entitled “Double dissociation without modularity: Evidence from connectionist neuropsychology” (Plaut, 1995) is often taken as evidence that there must be something wrong with the above discussion. His work was based on the models of deep dyslexia of Plaut & Shallice (1993), which in turn were extensions of the earlier models of Hinton & Shallice (1991). Deep dyslexia is a well-known acquired reading disorder characterized by semantic errors, such as reading “forest” as “tree”. Of particular relevance to us are two patients who provide a DD between abstract and concrete word reading. Patient CAV was able to read correctly 55% of abstract words but only 36% of concrete words (Warrington, 1981), whereas patient PW could read 67% of concrete words but only 13% of abstract words (Patterson & Marcel, 1977).