Human oscillatory brain activity dissociates sub-processes of visual working memory

Keeping a small number of visual items in memory for a few seconds comprises a variety of cognitive functions. Typically, target items have to be selected from a rich visual scene; targets have to be encoded efficiently while distracter items need to be suppressed; relevant items have to be maintained and distracting items are required not to be stored; a memory trace needs to be protected against incoming new information during retention and should be matched or compared to visual input during retrieval; etc. The usage of event-related brain oscillatory dynamics bears the possibility to investigate such sub-functions of visual working memory processes that are very hard to dissociate from each other with, for instance, event-related potentials. Here, results from electroencephalography (EEG) and transcranial magnetic stimulation (TMS) studies in healthy human subjects are reported. They suggest that dynamic amplitude fluctuations at theta frequency range (4–8 Hz) within a fronto-posterior cortical network are related to specific encoding of target items during encoding of visuospatial locations. A strong target-related theta response predicts high visual working memory capacity, whereas a strong response of this theta network to encoding of distracting items predicts low memory capacity. During maintenance of visual items, on the other hand, sustained cross-frequency phase coupling between theta and gamma (50–70 Hz) is obtained at brain areas storing target items. This theta–gamma phase coupling shows a load-dependent increase of up to four items to be retained. When memory load is further increased, theta and gamma phases are decoupled again. The magnitude of memory load-dependent theta–gamma phase coupling predicts individual working memory capacity. Efficient suppression of distracter items during a retention period, however, is reflected by increased EEG alpha (8–12 Hz) amplitude at brain areas processing distracting visual information. Memory capacity can also be predicted by alpha amplitude, in this case, based on the efficient inhibition of distracter maintenance. Moreover, by entraining alpha activity using repetitive TMS, it can be shown that memory capacity is increased as the neural mechanism of distracter suppression is reinforced by the stimulation