Do multidimensional representations of personality support learning?

Abstract

In this thesis, I explored how humans learn about others’ personalities. First, from a behavioral perspective, I tried to understand the strategies underlying this learning, i.e., find out what information people use and how they apply it. To do this, I used reinforcement learning models that are commonly used to explain learning across domains. I did not use these models on their own but added information structures that humans might use when learning about others. These models reveal that humans flexibly use complex knowledge structures when learning about others’ personalities. Based on these results, I next extended my focus from behavior to the brain. In recent years, developments in analysis techniques of functional brain data have allowed for queries into activity patterns rather than plain activations. Such analyses revealed multidimensional structures when learning about others’ personalities. I used these techniques to find out whether the cortex also codes for even more complex information structures that I found in the behavioral study. In general, I found evidence that the brain represents these complex knowledge structures when learning about strangers’ personalities. Finally, in the most recent work, I focused on grid-cells, which represent a specific way for representing and coding information. First established in navigational space, grid-like coding patterns have also been found for conceptual spaces in two dimensions. I investigated a personality plane based on two trait words that form the axes of this ‘trait space’ but with the used analysis methods, I found no evidence for grid-like coding for this trait space.