Transient storage of a tactile memory trace in primary somatosensory cortex

Abstract

Working memory is known to involve prefrontal cortex and posterior regions of association cortex (e.g., the inferior temporal lobes). Here, we investigate the potential role of primary somatosensory cortex (SI) in a working memory task with tactile stimuli. Subjects were required to compare the frequencies of two vibrations separated by a retention interval of 1500 msec. Their performance was significantly disrupted when we delivered a pulse of transcranial magnetic stimulation (TMS) to the contralateral SI early (300 or 600 msec) in the retention interval. TMS did not affect tactile working memory if delivered to contralateral SI late in the retention interval (at 900 or 1200 msec), nor did TMS affect performance if delivered to the ipsilateral SI at any time point. Primary sensory cortex thus seems to act not only as a center for on-line sensory processing but also as a transient storage site for information that contributes to working memory.

Fig. 1.

Summary of the procedure for experiments using TMS. Subjects felt two 1000-msec-long vibrations, separated by a 1500 msec retention interval during which they received a single pulse of TMS. TMS was delivered either 300, 600, 900, or 1200 msec after the end of the first vibration (1200, 900, 600, or 300 msec before the start of the second vibration). TMS was applied to the left or right SI, and the vibrations were presented to the left or right index finger. Thus, on half the trials TMS was applied to the SI contralateral to the vibrations, and on the remaining trials, TMS was applied ipsilateral to the vibrations.

Fig. 2.

Results of experiment 1, in which subjects compared two vibrations separated by a retention interval of 300, 600, 900, or 1200 msec. At all intervals, performance was above chance for vibrations presented both on the same side and on opposite sides. However, at the shorter intervals, the subjects were significantly more accurate when the two vibrations were presented on the same side than on different sides (p < 0.05). There was no such laterality effect at the longer intervals. Error bars indicate SEM.

Fig. 3.

Effects of TMS on vibration discrimination in experiment 2. The plot shows the mean difference in accuracy between trials in which TMS was applied to the contralateral SI and trials in which TMS was applied to the ipsilateral SI. This difference score is significantly below zero when TMS was delivered 300 or 600 msec into the 1500 msec retention interval, but not when TMS was delivered 900 or 1200 msec into the interval. Therefore, TMS disrupted performance when applied to the contralateral SI in the first half of the retention interval. Error bars indicate SEM.

Fig. 4.

Diagrams showing possible neuronal mechanisms involved in working memory for vibrotactile stimuli. During delivery of the first vibration to a fingertip (phase i), the frequency of the vibration is encoded by the firing rate or firing pattern of populations of neurons in primary and secondary somatosensory cortex (SI and SII). SI neurons fire in phase with the indentation cycle of the vibration, whereas the firing rate of neurons in SII is a monotonic (increasing or decreasing) function of the vibration frequency. Both patterns are depicted here by peristimulus time histograms of neuronal activity, as reported by Salinas et al. (2000). Across the retention interval, subjects must remember the frequency of the first vibration to compare it with the second vibration. We propose that this memory trace is supported initially by ongoing neuronal activity in both SI and SII (phase ii), but by 900 msec into the interval (phase iii), the memory is no longer held in SI. Populations of neurons in distinct areas of the premotor and prefrontal cortex (PFC) also contribute to sustaining the memory trace, especially toward the latter part of the retention interval (Romo et al., 1999; Hernández et al., 2002).