EEG biomarkers of free recall

Abstract

Brain activity in the moments leading up to spontaneous verbal recall provide a window into the cognitive processes underlying memory retrieval. But these same recordings also subsume neural signals unrelated to mnemonic retrieval, such as response-related motor activity. Here we examined spectral EEG biomarkers of memory retrieval under an extreme manipulation of mnemonic demands: subjects either recalled items after a few seconds or after several days. This manipulation helped to isolate EEG components specifically related to long-term memory retrieval. In the moments immediately preceding recall we observed increased theta (4-8 Hz) power (+T), decreased alpha (8-20 Hz) power (-A), and increased gamma (40-128 Hz) power (+G), with this spectral pattern (+T-A + G) distinguishing the long-delay and immediate recall conditions. As subjects vocalized the same set of studied words in both conditions, we interpret the spectral +T-A + G as a biomarker of episodic memory retrieval.

Keywords: Alpha; Gamma; Reinstatement; Retrieval; Theta.

Fig. 1.. Experimental Paradigm and Behavioral Data.

A. During Sessions 1–5, subjects performed an immediate recall task for each of 576 words, silently reading a word which they verbally recalled after waiting approximately 1 s. On average, subjects responded 1.53 s after word onset. Each of sessions 6–10 began with subjects attempting to freely recall the 576 words that they had seen on each of the preceding sessions along with any other words that come to mind during a 10 min retrieval interval. Subjects then performed the same immediate recall task as on earlier sessions. B. Subjects exhibited very high levels of immediate recall across all sessions, with performance dropping modestly across blocks and recovering following breaks. C. Subjects exhibited modest levels of delayed recall on the first (surprise) recall test given on Session 6, but performance rose sharply across subsequent sessions, hitting an average of 103 correct recalls by the final session. D. Intrusions similarly rose across sessions, but much less quickly than successful recalls. E. Inter-response times (IRTs) increased with output position; on sessions where subjects recalled a larger proportion of items, IRTs were generally faster throughout the retrieval period but rose sharply during the last few correct recalls.

Fig. 2.

Statistical maps illustrating relative increases (red) and decreases (blue) in spectral power across key memory contrasts for eight regions of interest. A. Delayed Recall vs. Deliberation. B. Delayed Recall vs. Immediate Recall. C. Deliberation vs. Immediate Recall. Panels A and B use matched time periods from the first five sessions of Immediate Recall and the five sessions of Delayed Recall following those first five sessions. Black-bordered regions indicated significant FDR-corrected t-tests on within subject difference scores (p < 0.05). Color bar corresponds to t-stat differences in panels A, B, and C. Electrodes locations corresponding to each region of interest appear on a schematic view of an electrode net (LAI,RAI: left/right anterior inferior; LSA,RAS: left/right anterior superior; LPI,RPI: left/right posterior inferior; LPS,RPS: left/right posterior superior) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).

Fig. 3.. Time course of high-frequency activity leading up to correct recalls.

Delayed recall, immediate recall, and delayed recall baseline log high-frequency activity (98–110 Hz power) at posterior electrodes in the 1.0 s leading up to vocalization of recalled items, averaged for all correctly recalled items and across subjects. Results are shown separately for phases one and two. Error bands reflect 95% confidence computed by the method of Loftus and Masson (1994).