We report electroencephalography (EEG) results from a non-patient pilot study conducted whilst developing a neuromodulation approach for improving visual spatial working memory (vSWM) in people with schizophrenia. Working memory impairments are common in people with schizophrenia yet respond poorly to current drug treatments. Transcranial magnetic stimulation (TMS), a minimally-invasive, well-tolerated, brain stimulation technique that is performed whilst a person is awake and alert, may improve working memory performance. However, results have been inconsistent, possibly because TMS was delivered during the heterogenous “resting-state”. We delivered TMS to left dorsolateral prefrontal cortex time-locked to specific events in a vSWM task, aiming to modulate functional networks involved in encoding spatial data into working memory.

Each trial in the vSWM task started with a 2-second-long sample display containing either three or four coloured circles positioned at random locations. This was followed by a 2-second delay period. At the end of the delay period, a visual cue appeared, indicating the target colour. Participants moved a crosshair to the screen location where the target had appeared. We recorded 64-channel EEG throughout. In Experiment 1, twelve participants completed three- and four-item task versions. In Experiment 2, eighteen participants completed the four-item task in three separate blocks within a single session. Between blocks, they completed a short task version alongside TMS. TMS (intermittent theta burst stimulation, 600 pulses, 3.3 minutes) was delivered over the F3 electrode position. Each stimulation on-phase was synchronised to coincide with the onset of sample display. In a random order, one TMS block was active, and one was sham (90° coil rotation).

In Experiment 1, EEG showed decreases (“desynchronisation”) in beta (13–30 Hz) power during sample display and increases (“synchronisation”) during the delay period. Both effects were greater in the four-item condition, and in posterior electrodes. In Experiment 2, posterior beta desynchronisation during sample display was greater following either active or sham stimulation. However, synchronisation during the delay period reduced following sham and increased only following active stimulation. Likewise, performance declined following sham but remained stable or improved following active stimulation.

We examined the effects of TMS on electrophysiological signals evoked during a spatial working memory task. We found that beta-band oscillatory activity, thought to safeguard stored information during memory delays, was increased by memory load and maintained or restored in blocks following active TMS. These effects were greatest over parietal/occipital areas. It is suggested that this beta activity serves to protect memory traces from distractors (in the current case, internal distractors). Notably, if TMS enhances delay activity within areas of the brain involved in stimulus representation that are distal from the stimulation site, then its effects are best understood as network level modulations of brain activity.