Schedule of Events | Symposia

Neural Decoding of Anticipation

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

Joseph Kahana¹ (kahana@psych.upenn.edu), Michael Kahana¹; ¹University of Pennsylvania

You know it's coming: the start of a race, the beginning of a quiz, the onset of a memory task, but how does the brain time the anticipation period? Neuronal recordings reveal time cells that fire at specific delays and ramping cells that increase or decrease their firing across an interval. Here, we asked whether spectral EEG components similarly represent temporal information. We analyzed intracranial recordings as 396 patients watching ten-second countdown videos preceding the onset of a memory task. We decomposed these signals into frequencies from 3-280 Hz. We first asked how theta [3 Hz], alpha [12 Hz] and gamma [110 Hz] power varied across the countdown period across frontal (N = 337), temporal (N = 352), and hippocampal (N = 173) electrodes. Theta increased throughout the interval, gamma decreased, and alpha exhibited spikes corresponding to the ten countdown ticks. Next, we used penalized regression to predict time as a function of spectral power, evaluating these models in hold-out sessions in 50 subjects who contributed multisession data. Both lasso and ridge regression reliably predicted time within the interval, with correlations of 0.12 (p < 0.001) and 0.11 (p < 0.001), respectively. These results, however, could have arisen from the transient neural response to countdown onset. To rule this out, we repeated our analysis, excluding the first two seconds of the interval, and found similarly reliable correlations 0.11 (p < 0.001) and 0.10 (p < 0.001), respectively. These results demonstrate that spectral EEG components reliably decode temporal information during anticipation.

Topic Area: LONG-TERM MEMORY: Episodic