Episodic memory retrieval is believed to require recovering a slowly changing spatiotemporal context. Two findings supporting this view are the recency effect (recent events are more likely to be remembered) and the contiguity effect (after recollecting an event, things that occurred near the event are more likely to be remembered). Critically, contiguity and recency effects are similar across time scales (i.e., they are scale-invariant). As a result, a scale-invariant model of context was proposed which makes several empirical predictions about the brain and behavior. Neurally, this model predicts neurons that respond shortly after an event and exponentially return to baseline at various rates. Experiment~1 analyzed single-unit recordings from the entorhinal cortex of 2~male rhesus macaques as they freely viewed images. In support of this hypothesis, I report units that fired in response to image onset and then relaxed their firing with a spectrum of rates.
Behaviorally, this model suggests that the time required to initiate memory retrieval can increase logarithmically with recency. Experiment~2 analyzed results from 6~continuous recognition experiments, where 202~human adults judged on each trial if a stimulus is new or old without separate study and test phases. Response time distributions demonstrated that the time required to initiate memory retrieval increased as a logarithmic function of that memory's recency out to at least 8~minutes, supporting this behavioral hypothesis.
Finally, although there is a spectrum of decay rates, there must be a fastest decay rate—items presented at a rate faster than this fastest decay rate should disrupt the contiguity effect. Experiment~3 analyzed behavioral results from 330~human adults performing an immediate free recall task in which lists were presented at 2, 4, or 8~Hz. In immediate free recall, participants studied a list of items and began recall as soon as the list ended. Critically, for lists presented at 8~Hz, the contiguity effect was eliminated, consistent with the hypothesis that the contiguity effect does not occur at very short time scales. Together these results suggest that temporal context is scale-invariant and offer additional constraints to the form of this representation.
| Identifer | oai:union.ndltd.org:bu.edu/oai:open.bu.edu:2144/48503 |
| Date | 26 March 2024 |
| Creators | Bright, Ian M. |
| Contributors | Howard, Marc W. |
| Source Sets | Boston University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis/Dissertation |
| Rights | Attribution 4.0 International, http://creativecommons.org/licenses/by/4.0/ |
Page generated in 0.0205 seconds