Event boundaries modulate neural representations of temporal context
Lynn Lohnas1, M. Karl Healey2, Lila Davachi1; 1New York University, 2Michigan State University
Although we experience everyday life as a continuous stream of information, we generally perceive and remember this information as discretized events. It has been hypothesized that as experiences are segmented, they are separated by event boundaries (Zacks et al., 2001). Interestingly, recent work has shown that segmenting experiences also influences how those experiences are remembered. Critically, stimuli presented in different events are more weakly associated than stimuli presented in the same event (DuBrow & Davachi, 2013, 2014, 2016; Ezzyat & Davachi, 2011, 2014). Here we examined how boundaries may weaken associations by modulating context representations. We examined a slowly changing neural representation used to query temporal context (Manning et al., 2011) and the implications for behavior. We recorded scalp electroencephalography while participants (N=147) performed a free recall task, with events operationalized as 2-6 words studied with the same encoding task. Each studied or recalled word was associated with a temporal context vector, defined as an autocorrelated vector of power values across electrodes and frequencies. Our results were motivated by predictions of the Context Maintenance and Retrieval model (CMR; Polyn et al., 2009), which assumes that boundaries disrupt temporal representations. Thus, CMR predicts that, controlling for objective time, the similarity in temporal context should be greater between two words from the same event than two words presented in different events. We found the neural data were consistent with CMR’s prediction. Further, across participants, behavioral performance correlated with the extent of the decrease in neural similarity for items within versus across events.
Topic Area: LONG-TERM MEMORY: Episodic