| Cells throughout the hippocampal formation show spatial firing correlates as a rat navigates through space. Called place cells, they are deemed to play a crucial role in the modulation of foraging behavior, since damage to the hippocampal formation causes spatial learning deficits. Based mainly on these findings, O'Keefe & Nadel (1978) distinguish between the "locale system" for map based navigation, hypothesized to be implemented in the hippocampus, and the "taxon systems" for route navigation.; We propose here two new models, the Word Graph (WG) model, which is used to build map-based representations of the environment, and the Taxon-Affordances Model (TAM) for taxon determination of movement direction. These are composed by layers of neurons that implement Hebbian and reinforcement learning, which together allow the expression of goal-oriented behavior.; The WG model enables the animal to successfully solve tasks not only in corridor mazes, but also in open-field environments, like the Morris water maze (Morris, 1981), or the experiments proposed by Collett et al. (1986). Besides modeling explicit behaviors, the WG model also shows how path integration information may be used together with visual sensory inputs to modulate the activity of place cells In the hippocampus, as observed by Gothard et al. (1996).; The WG model creates specific nodes to represent important places in the animal's world. This process is modulated by available affordances for movement. Affordances are computed in TAM, which replicates the behavior of animals with lesions that disconnect the hippocampus from other neural systems. In particular, TAM is used to explain the results obtained with intrahippocampal lidocaine injected animals, which are unable to express place learning strategies, as observed by Packard & McGaugh (1996). It also implements the stereotypical behavior produced by fornix-lesioned rats as observed by O'Keefe (1983) and Hirsh et al. (1978).; In this way, we present a theoretical framework grounded in computer simulations aimed at explaining how map and taxon based spatial signals are generated, as well as how they may cooperate or compete in the integrated TAM-WG model to guide successful behavior. |
