Working Memory 2.0

Abstract

Working memory is the fundamental function by which we break free from reflexive input-output reactions to gain control over our own thoughts. It has two types of mechanisms: online maintenance of information and its volitional or executive control. Classic models proposed persistent spiking for maintenance but have not explicitly addressed executive control. We review recent theoretical and empirical studies that suggest updates and additions to the classic model. Synaptic weight changes between sparse bursts of spiking strengthen working memory maintenance. Executive control acts via interplay between network oscillations in gamma (30-100 Hz) in superficial cortical layers (layers 2 and 3) and alpha and beta (10-30 Hz) in deep cortical layers (layers 5 and 6). Deep-layer alpha and beta are associated with top-down information and inhibition. It regulates the flow of bottom-up sensory information associated with superficial layer gamma. We propose that interactions between different rhythms in distinct cortical layers underlie working memory maintenance and its volitional control.

Keywords: bottom-up; cognition; cortex; oscillations; prefrontal cortex; synchrony; top-down; working memory.

Figure 1 –. Gamma and beta bursts underlie working memory.

A) A single trial example of LFP power in time and frequency. Two stimuli were presented (S1 and S2) and later tested following a delay. Narrow bursts of power in the beta and gamma bands are evident both during cue processing and delay. LFP data from sites that contained spikes which carried information about the presented cue (B) vs. those that did not (D) are shown. Only sites containing informative spiking (D, population average) showed modulation of beta and gamma. This effect remained after controlling for differences in spike rate between informative (B) vs. non-informative (C) sites. C) On gamma-modulated sites, the beta and gamma burst rates are mirror images of each other. Gamma bursting increases during stimulus presentation and towards the end of the delay, and beta does the opposite. D) On sites without informative spiking, only beta is task modulated and less so. Modified from Lundqvist et al (2016)

Figure 2 –. Laminar organization of gamma/beta rhythms and delay activity.

A) Gamma and alpha/beta power are segregated into distinct layers. Gamma power peaks 400 um above layer 4 whereas alpha/beta power peaks at 600 um below layer 4. Gamma bursts in superficial, but not deep layers, carry significant information about the cued item during the WM delay period (quantified by the Percent Explained Variance, or PEV, statistic). Beta bursts do not carry significant information during the delay (not shown). Dotted lines are +/− 1SEM across sessions (N=60). B) Spiking activity, quantified by multiunit change from baseline (arbitrary units) during the delay period is strongest in superficial layers. The pattern of laminar pattern of delay activity correlates strongly with gamma, and is strongly anti-correlated with alpha/beta. C) Correlation map between gamma and beta power across layers. Deep layer beta power is anti-correlated with superficial layer gamma power during the WM delay. From Bastos et al (2018).

Figure 3 –. A model of WM.

Denoted by two rectangular, dashed boxes, two cortical compartments, superficial and deep, are made up of densely interconnected excitatory pyramidal (black) and inhibitory (red) interneurons. Inhibitory connections are line segments with a red, rounded end, and excitatory connections are line segments with a black, arrow end. Two separate PING networks in superficial vs. deep layers are responsible for generating gamma in superficial layers and beta in deep layers (sustained by connections to thalamus and basal ganglia, not shown). The looping arrow returning on itself in the superficial layers represents the recurrent connectivity found within layer 3 pyramidal cell networks in prefrontal cortex. The sinusoidal red-line in deep layers reflects beta oscillations and their driving influence on superficial beta oscillations. Beta oscillations are phase-amplitude coupled with gamma oscillations (blue squiggly lines), and these gamma oscillations organize delay-period spiking representing WM content (straight black marks). Spiking activity inside gamma bursts is more informative than outside. Over time, moving from left to right in the figure, the deep beta reduces in power and releases inhibition onto the superficial layers. This results in enhanced superficial gamma and spiking, i.e., enhanced maintenance of WM, as is seen when transitioning between baseline to WM task performance. The reversed process (enhancement of deep layer beta, enhanced suppression of superficial layer gamma/spiking) which would “clear out” the contents of WM, as seen at the end of the trial, or when WM contents are no longer needed. From Bastos et al (2018)