Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article

Degree

Master of Science

Program

Neuroscience

Supervisor

Laura Batterink

2nd Supervisor

Stefan Köhler

Co-Supervisor

Abstract

It has been theorized that pattern separation and statistical learning implicate two separate circuitries within the hippocampus, one that involves the dentate gyrus subregion and one that does not, respectively. Here, we tested whether the two processes are dissociable computations that rely on separate neural pathways by examining the effect of healthy aging (Chapter 2) and dentate gyrus lesion (Chapter 3) on participants’ ability to differentiate similar trisyllabic words and extract trisyllabic words from a continuous speech stream. Neither healthy aging nor dentate gyrus damage affected the implicit expression of statistical learning, but they both impaired pattern separation and the explicit expression of statistical learning. These results suggest that the implicit expression of statistical learning and pattern separation are dissociable computations that rely on separate neural pathways whereas the explicit, high-resolution expression of statistical learning in part shares the same process and neural architecture as pattern separation.

Summary for Lay Audience

Our daily life consists of events that are similar from day to day. Yet, we are capable of distinctly remembering events from each day, due to the brain’s ability to store each event as a separate memory, a process that is known as pattern separation. At the same time, we can also extract regularities across multiple events over time through statistical learning. Both pattern separation and statistical learning have been suggested to involve the same brain region, the hippocampus. How can the hippocampus support differentiation and generalization, two very different computational processes? Some researchers proposed that these two processes are supported by two separate pathways within the hippocampus. However, this hypothesis has not been tested directly with human subjects.

Here, we examined the effect of aging on pattern separation and statistical learning to test whether the two are dissociable processes (Chapter 2). In Chapter 3, we studied the rare case of BL, a patient with a lesion to the dentate gyrus that disrupts the pathway that is thought to support pattern separation but not statistical learning. Across both studies, we used nonsense trisyllabic words (e.g., rupuni) to assess participants’ abilities to differentiate similar sounding words (pattern separation) and to extract words embedded in a continuous speech stream (e.g., babupupatubitutibu…; statistical learning).

Both older adults and BL showed impaired pattern separation. They also showed difficulties in recognizing words embedded in the speech stream with high precision, even though they demonstrated successful learning of the words on statistical learning tasks that did not require high-precision memory. These results suggest that statistical learning is not a single process but rather consists of separable components. Extracting the hidden triplet structure and acquiring vague, non-verbalizable knowledge of the words is a process that is independent of pattern separation, as hypothesized by previous research works. In contrast, the formation of clear, verbalizable knowledge of the words likely shares the same process and neural correlates as pattern separation. These findings contribute to our understanding of the neural mechanisms involved in the differentiation and generalization of memories, as well as the impact of aging on these two processes.

Download

Share

COinS