High-frequency broadband activity demonstrates slow theta phase preference for sequential order in working memory maintenance

Poster Session B - Sunday, April 14, 2024, 8:00 – 10:00 am EDT, Sheraton Hall ABC

Samantha Gray¹ (samantha.gray@northwestern.edu), Adam Dede¹, Jack Lin², Ignacio Saez^2,3, Fady Girgis^2,4, Edward Chang⁵, Kurtis Auguste^5,6, Ammar Shaikhouni^7,9, Robert Knight⁸, Elizabeth Johnson¹; ¹Northwestern University, ²University of California, Davis, ³Icahn School of Medicine at Mount Sinai, ⁴University of Calgary, ⁵University of California, San Fransisco, ⁶UCSF Benoiff Children's Hospital, ⁷Ohio State University, ⁸University of California, Berkeley, ⁹Nationwide Children's Hospital

Neurons representing place and time in the hippocampus (HPC) encode information through cell specificity and theta phase preference. Dominant models of working memory (WM) propose a theoretical framework for theta-phase-coding of slots for stimuli representations. Previous findings support these models by demonstrating slowing theta with increased WM load, allowing more neuronal spikes representing information to fit within one theta cycle. Additionally, the orbitofrontal cortex (OFC) has been causally implicated in representing temporal order in humans. Here, we use macro- and micro-wire stereotactic EEG to uncover mechanisms of maintaining temporal order in humans. We analyze 20 WM delayed match-to-sample datasets (7F; aged 30.65 ± 11.15) wherein patients memorized sequences of three shapes (sample), followed by a 2-s delay and presentation of three more shapes (match). Slow theta oscillations slowed from the sample to match in HPC (p=0.004) and OFC (p=0.01), and we found evidence for HPC-OFC synchrony during maintenance of the sample. We observed phase-amplitude coupling between slow theta and high-frequency broadband activity – a measure of multiunit activity – in HPC (p=0.04) with a trimodal distribution of phase preference, suggesting that sequential order is slow theta phase-coded. We then isolated 152 HPC neurons from a subset of 10 patients (88 micro-wires) and estimated stimulus preference based on maximum firing rate. We identified 11.0% of neurons with stimulus 1 preference, 8.5% with stimulus 2 preference, and 9.2% with stimulus 3 preference. Further analyses will formally test phase-coding of HPC neurons by extracting the instantaneous theta phase at spike times during maintenance.

Topic Area: EXECUTIVE PROCESSES: Working memory