Bayesian methods for addressing long-standing problems in associative learning: The case of PREE

摘要: Most associative models typically assume that learning can be understood as a gradual change in strength captures the situation into one single parameter, or representational state. We will call this view single-state learning. However, there is ample evidence showing under many circumstances different relationships share features learned independently, and animals quickly switch between expressing another. multiple-state Theoretically, it understudied because needs data analysis approach from those usually employed. In article, we present Bayesian model of Partial Reinforcement Extinction Effect (PREE) test predictions view. This implies estimating moment responses (from acquisition to extinction performance), both at individual group levels. used analyze PREE experiment with three levels reinforcement during (100%, 75% 50%). found differences estimated states extinction, so was delayed after leaner partial schedules. The finding compatible It first time, our knowledge, are tested directly. article also aims show benefits methods bring field.