ABSTRACT
Connectionist networks have been used to support far-reaching claims about human cognitive function (Bechtel & Abrahamsen, 1991; Ellis & Humphreys, 1999; Rumelhart & McClelland, 1986; Schneider, 1987; Smolensky, 1988). In this chapter we focus on one particular application of the approach: the use of connectionist networks to simulate the patterns of performance shown by brain-damaged patients with acquired cognitive impairments (e.g. Cohen, Farah, Romero, & Servan-Schreiber, 1994; Devlin, Gonnerman, Andersen, & Seidenberg, 1998; Ellis & Humphreys, 1999; Farah, O’Reilly, & Vecera, 1993; Harley, 1998; Hinton & Shallice, 1991; Humphreys, Freeman, & Muller, 1992; Humphreys, Olson, Romani, & Riddoch, 1996; Mayall & Humphreys, 1996; McLeod, Plunkett, & Rolls, 1998; Mozer & Berhmann, 1990; Olson & Caramazza, 1994; Patterson, Seidenburg, & McClelland, 1989; Plaut & Shallice, 1993; Seidenburg, & McClelland, 1989; Tyler, Moss, DurrantPeatfield, & Levy, 2000).
