Schedule of Events | Symposia

De-Confounding Associations between Neural Activity and Memory Performance

Poster Session C - Sunday, March 30, 2025, 5:00 – 7:00 pm EDT, Back Bay Ballroom/Republic Ballroom

Riley DeHaan¹ (rdehaan@sas.upenn.edu), David Halpern¹; ¹University of Pennsylvania

Decades of work have demonstrated associations between neural activity at the time of memory encoding and subsequent memory performance. However, the vast majority of this literature estimates the subsequent memory effect without simultaneously accounting for stimulus properties such as study position and item identity, which are known to affect both memory and neural activity. It therefore remains unclear whether these effects from observational data reflect causes of successful memory encoding or task-related confounds. Here we analyze intracranial EEG recorded from 375 subjects during a delayed free recall task and from 72 subjects during a paired associates learning task. We compare estimates of the subsequent memory effect, computed with and without adjustments for confounding variables of serial position and item identity. We find that accounting for these confounds significantly attenuates the magnitude of the subsequent memory effect. Furthermore, stimulus features are twice as effective at predicting memory outcomes than neural data alone. After removing stimulus-related effects, neural activity remains a significant behavioral predictor but only explains an additional 3.9% of the variance in recall performance. We further analyze differences in subsequent memory effects across anatomical regions, frequency bands, and time periods relative to item presentation after accounting for task variables. These results show qualitative differences in the regions associated with successful memory as opposed to item identity and serial position. Confounding is large in these memory tasks and should be expected to be large across much of neuroscience. These results suggest careful de-confounding should be a central step in systems neuroscience.

Topic Area: LONG-TERM MEMORY: Episodic