Theses Doctoral

Hindsight for foresight: adaptive uses of memory in value-based decision making

Nicholas, Jonathan

Effective decision making depends on using memories of past experiences to inform choices in the present. This dissertation examines several ways in which memory is used in decision making, and further aims to establish that one way we adapt to the statistics of our environment is by modifying how we use our memories to guide behavior. In chapters one and two, I focus on how incremental trial-and-error learning and episodic memories of individual events may each contribute to choice.

In chapter one, I ask how the brain may arbitrate rationally between these two systems to achieve a balance that maximizes reward. By manipulating the volatility of the environment to affect uncertainty, I show that participants rely on each system in the circumstances to which it is best suited.

In chapter two, I then ask how decisions based on these memory systems each depend on striatal dopamine. By studying patients with Parkinson’s disease both on and off their medication, I find that a lack of dopamine alters only incremental learning, and that dopamine replacement remediates this deficit with few effects on the use of episodic memory.

Finally, in chapter three, I examine a more difficult class of decisions that require individual memories to be used for planning future action. Using neuroimaging to decode memory access, I find that that the statistical structure of relationships between memories determines when they are used to support planning. Combined, these three chapters suggest that we are capable of flexibly employing multiple forms of memory, with distinct neural mechanisms, to guide a variety of choices.


This item is currently under embargo. It will be available starting 2024-06-24.

More About This Work

Academic Units
Thesis Advisors
Shohamy, Daphna
Ph.D., Columbia University
Published Here
June 28, 2023