Amnesia and. the declarative/nondeclarative distinction: A recurrent network model of classification, recognition, and repetition priming

A key claim of current theoretical analyses of the memory impairments assoc iated with amnesia is that certain distinct forms of learning and memory ar e spared. Supporting this claim, B. J. Knowlton and L. R. Squire found that amnesic patients and controls were indistinguishable in their ability to l earn about and classify strings of letters generated from a finite-state gr ammar, but that the amnesics were impaired at recognizing the training stri ngs. We show, first, that this pattern of results is predicted by a single- system connectionist model of artificial grammar learning (AGL) in which am nesia is simulated by a reduced learning rate. We then show in two experime nts that a counterintuitive assumption of this model, that classification a nd recognition are functionally identical in AGL, is correct. In three furt her simulation studies, we demonstrate that the model also reproduces anoth er type of dissociation, namely between recognition memory and repetition p riming. We conclude that the performance of amnesic patients in memory task s is better understood in terms of a nonselective, rather than a selective, memory deficit.