Multiple Memory Systems: Contributions of the Human and Animal Serial Reaction Time Task

Human memory systems have been divided into two broad domains, one responsible for ‘declarative memory’ and the other for ‘non-declarative memory’. The evidence for multiple memory systems is reviewed with respect to the human SRT, a sensitive measure of non-declarative memory. A qualitative review of the human SRT literature concludes that damage to extrapyramidal brain systems disrupts SRT performance whereas limbic system neuropathology (LSN) leaves performance intact. However, a meta-analysis of the SRT literature with neuropathological patients revealed unexpectedly that patients with explicit memory disorders are impaired on the SRT task, although less severely than patients with extrapyramidal damage. Other evidence suggested that the apparent SRT impairment in humans with LSN might be due to the additional pathology (eg frontal) often evident in these patients. A brief review of the animal evidence for multiple memory systems concluded that, like humans, animals too have multiple memory systems but none of the animal tasks used to model non-declarative memory make good conceptual or behavioural contact with the corresponding human tasks. Thus a novel animal-analogue of the humanSRT task, the ‘fan-maze’, was developed. Although rats displayed a reasonable ability to perform the fan-maze SRT task it was abandoned due to technical and conceptual problems in favour of a better design. The second new SRT task used intra-cranial self-stimulation to promote prolonged, rapid and continuous responding. A control study determined that the optimal conditions for sequence learning was a single large (2820 trial) session. Intact rats that experienced a switch from the repeating to a random sequence under these conditions demonstrated a clear interference effect, the primary measure of SRT performance. A lesion study used these optimal conditions and showed that small caudate lesions impaired, whereas small hippocampal lesions facilitated, rat-SRT performance. Hence, this second task has proven to be a valid animal-analogue of the human SRT task, as rats performed it in a manner similar to that shown by humans and relied on the same neural substrate to perform the task as humans. In addition, this second task resolved the discrepancy of the LSN meta-analysis. Quantitative findings are reviewed in light of theories and studies presented earlier in the thesis. Limitations of the thesis are identified and suggestions are made as to future SRT research in animals or humans.