Using a pre-determined training schedule, we have previously shown that a hippocampus-like computational model reproduces the transverse patterning and transverse non-patterning data of Jerry Rudy's lab, two problems that require context for solution. To extend the predictions produced by the model, an extra-hippocampal decision function is added, and this function allows the model to choose among training items. Three decision functions are compared for transverse patterning, and all three decision functions reproduce the asymptotic performance of the behavioral experiments. However, two of the decision functions reproduce the experimentally observed learning curves. Finally, the enhanced model is used to predict that a novel transverse non-patterning problem is learnable. © 2006 Elsevier B.V. All rights reserved.
Wu, X., & B Levy, W. (2006). Decision functions that can support a hippocampal model. Neurocomputing, 69(10–12), 1238–1243. https://doi.org/10.1016/j.neucom.2005.12.084