Associative Symmetry vs. Independent Associations

We develop a neural network model of paired-associate learning based upon an auto-associative learning mechanism. We show that this relatively simple neural network can replicate complex human behavioral data, but only when the correlation between forward and backward learning is highly correlated. This network based analysis is used to constrain psychological theories of association in humans. Key words: Associations, Memory, Neural networks, Symmetry. Introduction In the human memory literature, there are two competing theoretical views regarding the nature of associative formation. The independent associations hypothesis (IAH) maintains that if two symbols, A and B are encoded successively, the forward association between A and B will tend to be stronger than the backward association between B and A. Additionally, the strengths of forward and backward associations are hypothesized to be independent [8]. In contrast to this position, representatives of Gestalt psychology viewed sy..