< Back to Article

Model-based spatial navigation in the hippocampus-ventral striatum circuit: A computational analysis

Fig 11

Learning the latent state-space, the state-transition and state-value models.

Given (1) the last input x_t−1 and latent state s_t−1, (2) performed action a_t−1, (3) observed new input x_t, reward r_t, and inferred latent state s_t, learning consists of (5) adjusting the categorization model to make it more congruent with the state-transition model and updating the conjugate priors and of the state-transition and state-value models to accommodate the internal perception of the experienced behavioral evidence (see the text for details). Notably, the update of the state-value conjugate is a Bayesian analog of TD-learning using predicted discounted future value accumulated in (4) a forward sweep.

doi: https://doi.org/10.1371/journal.pcbi.1006316.g011