The Role of Theta- / Gamma-Coupling in Working Memory Dysfunction in Schizophrenia

Abstract

Background: Prominent working memory (WM) deficits have been repeatedly observed in people with schizophrenia (PSZ) and their unaffected relatives (REL), including in the spatial domain. Given the apparent association between spatial WM dysfunction and genetic liability for schizophrenia, spatial WM deficits have been proposed as a potential endophenotype for the disorder. Deficits in the neural correlates of WM performance have likewise been observed in PSZ and REL. Lisman and Idiart (1995) offered a model delineating a mechanism for the representation of multiple stimuli in WM through systematic interactions between activity in the theta- and gamma- frequency ranges, and much experimental evidence in support of this model has been obtained since. Activity in these frequency ranges has also been proposed as a potential underlying factor in WM dysfunction in PSZ and REL, especially in light of documented deficits in the theta- and gamma- bands independently. Methods: Theta- and gamma-band oscillatory activity recorded during a spatial WM task was examined through time-frequency analyses in PSZ, REL, and CTRL. Indices of power, phase-synchrony, cross-frequency coupling and their relationships to task performance were explored. Results: PSZ demonstrated abnormalities in measures of both theta- and gamma-band power as compared to CTRL and REL, whereas deviance in power measures were limited to gamma-activity in REL. Both PSZ and REL showed reduced phase-synchrony across examined frequencies for electrodes where synchrony was observed in CTRL. Theta-gamma coupling increased significantly in response to WM stimuli, though minimal differences were observed across diagnostic groupings. Behavioral performance was generally predicted by measures of low frequency power and high-frequency phase synchrony, though the predictive ability of focused measures of gamma-band power, synchrony and phase-amplitude coupling was increased for PSZ as compared to CTRL and REL. Discussion: Disturbances in various measures of theta- and gamma-band oscillatory activity was observed in PSZ and to a lesser extent in REL. PSZ showed unique predictive relationships between certain neural indices (including limited indices of cross-frequency coupling) and behavioral performance, even in measures where no group differences were observed, suggesting PSZ are more impaired by normal variability in neural processes related to these measures than CTRL and REL. These findings, in light of preserved WM performance in REL, may further support the presence of a compensatory mechanism in REL that insulates them from deficits in performance. Cross-frequency coupling appears to have some predictive utility regarding WM ability, though further work in determining particular frequency pairs of relevance is needed.