Several variations on the transitive inference task (e.g., if a>b, b>c, c>d, d>e, e>f, then c?e;Levy & Wu, 1996) will be simulated on a neurobiologically plausible model of the CA3 region of the hippocampus (e.g. Levy, 1996). In one variation, the task will be circularized (i.e. add f>a to the constraints above), so that there is no absolute answer to any comparison (e.g., e?a). It is hypothesized that the model will find the best local solution (e.g. em, m>n, n>o;m>c, then d?l). Model performance will be compared to existing results for humans and animals. Additional data will be provided by the animal laboratories of Dr. Howard Eichenbaum, and by human laboratories at the University of Virginia. The model will be augmented to simulate the cortical teaching function of the hippocampus. A cortical system will be added to the model, which will have a globally slower learning rate, discrete regions with few inter-regional but more intra-regional connections, and unique inputs to those regions which may or may not be correlated. Questions to be addressed will include, the capacity for permanent storage in the cortical system, improved inter-regional integration, and the time sequence for transferring information to the cortical system.
Greene, A J; Spellman, B A; Dusek, J A et al. (2001) Relational learning with and without awareness: transitive inference using nonverbal stimuli in humans. Mem Cognit 29:893-902 |
Greene, A J; Prepscius, C; Levy, W B (2000) Primacy versus recency in a quantitative model: activity is the critical distinction. Learn Mem 7:48-57 |