Oscillatory signatures of unimodal, bimodal, and cross-modal sensory working memory

Neural oscillatory activity is an essential brain mechanism that enables and subserves a vast range of cognitive functions. Studying them non-invasively through electroencephalography (EEG) has proven to be an effective method of discovering associations between oscillations in different frequency bands and various cognitive functions. Studying the oscillatory dynamics of human working memory (WM) – a core component of human higher cognitive functioning – has been particularly fruitful, leading to insights about the mental processes, frequency bands, and brain areas involved. In addition to frequency band specificity, the application of source reconstruction methods has led to further insights by revealing specific brain areas associated with WM related processing. In the present study, we focused on the oscillatory power dynamics during sensory working memory (SWM) in auditory and tactile modalities in the alpha band. In a delayed comparison two alternative forced choice task participants received two seque ntial stimuli and had to respond whether the intensity of the second stimulus was stronger than that of the first stimulus. In three related EEG experiments we examined SWM processing under unimodal (stimulation in one modality), bimodal (stimulation in both modalities simultaneously), and cross -modal (sequential stimulation of the modalities) conditions. An additional non -WM control condition allowed us to explore not only the differences between auditory and tactile WM, but also the effects of the WM task itself on the delay period oscillatory activity within each sensory modality. Our results showed that, while the bimodal stimulation condition led to behavioral enhancement, an increased stimulus difference was necessary to maintain the same level of performance also in the cross-modal conditions. Localizing the oscillatory activity in the alpha band (8 – 12Hz) revealed a clear disinhibitory effect over the somatosensory cortex during the early and the late delay period, while the mid-delay did not show any differences in SWM between the two modalities. A similar, albeit weaker , effect was observed over the auditory cortices. A right parietal reduction of alpha power emerged during the late delay when a tactile stimulus had to be compared cross-modally. This suggests the involvement of parietal somatosensory association cortex in the cross-modal transformation of the tactile stimulus. Lastly, the differences between cortical source distributions when contrasting unimodal and cross-modal conditions demonstrated that late delay effects do not reflect only anticipatory effects due to the upcoming modality, but also reflect the influence of the stimulus modality kept in WM. Contrasting the bimodal condition with the unimodal ones revealed a parametric beta band effect in a right parietal area during the early delay only in the bimodal condition, which suggests that beta oscillations might play a role in multimodal integration under SWM conditions. A second effect during the early delay period was observed in the theta (4 – 7Hz) band. An early effect appeared when contrasting conditions in which the first stimuli were identical while the second stimuli differed across the conditions. This result suggests that the early delay period is already shaped by the anticipated comparison context. The clearest differences in the contrast between WM and non-WM task were observed in theta and gamma bands. Source localizing the condition differences suggested the involvement of hippocampal and fronto-central areas in carrying out the WM task. Furthermore, sensory cortices of the respective modality conditions showed the highest levels of connectivity with the rest of the brain during the late delay, further highlighting the involvement of gamma band oscillations in SWM related processing. Overall, this study demonstrates that the results obtained when studying SWM related processing strongly depend on the sensory modality examined and the type of WM task employed. Any observations with regard to SWM related oscillatory power dynamics should be explored in multiple contexts before drawing any generalized conclusions