Content area
Abstract
Across development, learning and memory interact to support adaptive behavior. Through learning from the outcomes of their actions and remembering their past experiences, people can tailor their behavior to the demands of their environments to take actions that are increasingly likely to bring about desired outcomes. Like the behavior they support, learning and memory processes are themselves flexible — the value of different actions can be assessed and experiences can be represented in memory in myriad ways. A central challenge for development, then, is to figure out how to learn and what to remember. Fortunately, the reward statistics of the environment inform the nature of the learning computations and memory representations that can most effectively guide behavior in a given context. Thus, the central argument of this thesis is that an account of the development of adaptive behavior requires understanding changes in how learning and memory processes themselves adapt to the statistics of varied environments. The experiments presented here begin to develop this account, examining how, across development, reinforcement learning computations and memory representations demonstrate sensitivity and responsivity to the statistical structures of the environments in which they operate. The first study shows that children, adolescents, and adults flexibly adapt how they learn from valenced outcomes to make more rewarding decisions in environments in which different valence biases support efficient learning. The second study demonstrates that from childhood to early adulthood, people develop the ability to use learned environmental structure to prioritize valuable information in memory. The third study shows that across development, people use the reward statistics of the environment to adjust the extent to which they use more general versus more specific representations to guide choice and that these adaptive adjustments are reflected in subsequent memory. Together, the work detailed in this thesis highlights how the structure of the environment shapes the reinforcement learning and memory prioritization mechanisms that underlie the development of adaptive behavior.
