Effect of Transcranial Magnetic Stimulation on Memory

Psychiatrists have recognized cognitive dysfunction as the principal impediment to recovery and psychosocial rehabilitation of patients with schizophrenia since the first edition of Emil Kraepelin's 1893 masterwork, Dementia Praecox. While currently available pharmacotherapy may improve psychotic symptoms, there is still no largely effective treatment for impairments of critical executive functions - attention and working memory - that are widely considered the functional nexus of this extremely costly illness. Thus, it appears heuristically reasonable to consider putative modes of treatment that may specifically target these deficits. For example, abundant evidence supports the assumption that pre-task alpha band power, measured as mean peak frequency, is inversely associated with reaction time (RT) - an index of encoding efficiency - in a working memory task. Additionally, recent evidence suggests pre-task alpha peak frequency predicts outcome on tests of working memory. Evidence also suggests patient with schizophrenia have low pre-task alpha peak frequency, and slower RT on working memory paradigms, such as the n-back and Sternberg compared with healthy controls; taken together, the weight of evidence lends plausibility to the assumption that an increase in patients' pre-task alpha band power might be associated with a faster reaction time. Notably in this regard, recent evidence suggests transcranial magnetic stimulation (TMS) may increase pre-task alpha, and is associated with a faster RT in healthy subjects. There are as yet no reports of the effect of TMS on alpha band peak frequency, or reaction time, in patients with schizophrenia. A critical obstacle to progress in the clinical neuroscience of TMS effects in human subjects until now has been the absence of a means to determine the optimal subject-specific brain region to target with TMS. Cortical targets for stimulation are customarily selected on the basis of group mean data, with the tacit assumption that candidates for TMS all utilize the same distributed neural circuit to perform a particular cognitive task. This untested assumption is a conceivable source of the confusing discrepancy of TMS clinical trial outcomes, for example, in the depression treatment literature. In this regard, a group from our laboratory recently reported the use of structural equation modeling (SEM) based path analysis to construct group and subject-specific models of the 2-back working memory task in healthy subjects studied with PET. The results of this investigation suggest there may be a significant association between cognitive strategy, circuit path, and memory performance. We found higher scores were associated with activation of a left hemisphere distributed circuit, and a verbal cognitive strategy, while lower scoring subjects utilized a right hemisphere circuit, and a visuo-spatial strategy. Notably, these findings are congruent with those of previous studies that used different methods of analysis and a similar paradigm. Taken together, these data suggest that a trial of TMS aimed at improving working memory reaction time by elevating pre-task alpha power should first identify the predominant nodes of a subject-specific working memory sub-network as putative sites of stimulation. There are no reports of such an approach to TMS target selection in either healthy controls, or patients with schizophrenia. In this proof of concept with healthy subjects protocol, our principle assumption is alpha frequency TMS directed to a subject-specific predominant node (right or left DLPFC) of a distributed neural network reduces reaction time on the n-back and Sternberg working memory paradigms.