Persistent Single-Neuron Activity during Working Memory in the Human Medial Temporal Lobe

Abstract

Working memory is an essential component of human cognition. Persistent activity related to working memory has been reported in many brain areas, including the inferior temporal and prefrontal cortex [1-8]. The medial temporal lobe (MTL) contains "concept cells" that respond invariantly to specific individuals or places whether presented as images, text, or speech [9, 10]. It is unknown, however, whether the MTL also participates in working memory processes. We thus sought to determine whether human MTL neurons respond to images held in working memory. We recorded from patients with chronically intractable epilepsy as they performed a task that required them to remember three or four sequentially presented pictures across a brief delay. 48% of visually selective neurons continued to carry image-specific information after image offset, but most ceased to encode previously presented images after a subsequent presentation of a different image. However, 8% of visually selective neurons encoded previously presented images during a final maintenance period, despite presentation of further images in the intervening interval. Population activity of stimulus-selective neurons predicted behavioral outcome in terms of correct and incorrect responses. These findings indicate that the MTL is part of a brain-wide network for working memory.

Keywords: Sternberg Task; human; medial temporal lobe; microwires; persistent activity; single units; working memory.

Figure 1. Experimental Design and Example Response

(A) Behavioral task. In each trial, the subject saw a stream of four or three images, chosen from a pool of eight or nine, respectively, for 200 ms each. In three-quarters of trials, a blank screen was presented after each image, such that the inter-stimulus interval was either 200, 500, or 800 ms. In the remaining one-quarter of trials, there was no intervening blank screen. Following image presentations, subjects saw a mask followed by a fixation cross, which was presented for a minimum of 2.4 s. The fixation cross then disappeared, and subjects saw two probe pictures simultaneously, one of which had been presented in the preceding stream of images. After these probes disappeared, the subject pressed a key to signal the previously presented image. (B) Spike rasters and peri-stimulus time histograms for a stimulus-selective single unit recorded from the right entorhinal cortex at sample presentation, including presentations at all inter-stimulus intervals. Images at the top indicate presented stimuli.

Figure 2. Activity after Stimulus Offset

(A) Spike rasters and peri-stimulus time histograms for a single unit recorded from the right amygdala that showed increases in the duration of stimulus responses with increasing inter-stimulus interval (ISI). At ISI = 0: latency = 248 ms, duration = 188 ms; at ISI = 200 ms: latency = 230 ms, duration = 394 ms; at ISI = 500 ms: latency = 236 ms, duration = 651 ms; at ISI = 800 ms: latency = 230 ms, duration = 936 ms. (B) Duration of image responses of 107 units with significant stimulus information computed separately for each inter-stimulus interval. Points are jittered to reveal their distribution. Error bars are ±SEM. p values indicate significance according to unequal variance t test. (C) Debiased proportion of variance explained by image identity (partial ω²) averaged over the same 107 units shown in (A). Information was computed over all spikes in a 200 ms window centered at the corresponding time on the x axis. Colored dotted lines indicate the onset of the next picture in the corresponding ISI conditions. Bars at the top indicate significant image information at the given time point in the corresponding ISI condition (p < 0.05 corrected for all time points by permutation test of mean ω²). Error bars are ±SEM. See also Figure S1 and Tables S1 and S2.

Figure 3. Maintenance Period Activity

Spike rasters and peri-stimulus time histograms for a single unit recorded from the right amygdala that encoded a previously presented image during the maintenance period. Images at the top indicate presented stimuli, with photos of lab personnel and photos provided by the patient replaced with blue placeholders. A red line separates trials where the preferred stimulus (the spider) was not presented (top) from trials where it was (bottom). Black histograms indicate average firing rate over all trials; red histograms indicate average firing rate after removing trials that included the preferred stimulus. Responses following image presentation (A) and responses during the maintenance period (B) are shown. See also Figures S2 and S3 and Table S3.