The Role of Competition in Representational Change

Abstract

What are the principles that govern whether neural representations move apart (differentiate) or together (integrate) as a function of learning? According to supervised learning models that are trained to predict outcomes in the world, integration should occur when two stimuli predict the same outcome. Numerous findings support this, but—paradoxically—some recent fMRI studies have found that pairing different stimuli with the same associate causes differentiation, not integration. The goal of this dissertation is to provide an explanation for these paradoxical results and explore the repercussions of such an explanation. In Chapter 1, I argue that supervised learning needs to be supplemented with unsupervised learning that is driven by spreading activation in a U-shaped way, such that inactive memories are not modified, moderate activation of memories causes weakening (leading to differentiation), and higher activation causes strengthening (leading to integration). In Chapter 2, I present a series of behavioral experiments that provide a roadmap for exploring these effects. In Chapter 3, I present a neural network model that instantiates a U-shaped learning rule, and I show how it can explain a range of findings in the literature. The model can lead to integration or differentiation and makes several novel predictions—for instance, that differentiation is rapid and asymmetric. This work provides a unifying theory underlying a wide array of findings in learning and memory, and it provides a blueprint for future work in the field.