Philipp Deutsch^1,2 (), Cora Fischer^1,2, Jochen Kaiser^1,2, Christoph Bledowski^1,2, Benjamin Peters^3,4; ¹Institute of Medical Psychology, Goethe University Frankfurt, ²Cooperative Brain Imaging Center, Goethe University Frankfurt, ³Institute for Language, Cognition and Computation, University of Edinburgh, ⁴Centre for Cognitive NeuroImaging, University of Glasgow

A hallmark of working memory is its capacity limitation: memory performance decreases with more simultaneously memorized items. The neuronal origins of this limitation remain unclear. While capacity limitations are typically assessed by asking participants to report a single item, we applied a whole-report paradigm requiring participants to report all items. Specifically, participants memorized one, two or four orientations of Gabor gratings. After a delay, they sequentially reported each item at the locations indicated by external cues. For each item, participants provided two responses: 1) a confidence rating about memory quality (scale from 0=forgotten to 3=best possible remembered) and 2) a reproduction of the memorized orientation on a continuous scale. This allowed us to assess the number of items rated as forgotten and to estimate the memory precision for each item. We developed a composite measure of memory precision by integrating the confidence rating and orientation report, while accounting for the systematic impact of report order on memory. We recorded brain activity using magnetoencephalography while participants performed the task. Using beamforming, source level estimates of brain activity were obtained. We then applied multivariate pattern analysis to the activity across the human cortical surface. Preliminary results suggest that the activity patterns in posterior cortical regions predicted the aggregated precision of the reproduced orientations. This prediction was most pronounced during memory encoding but remained present until report. Moreover, activity patterns from posterior regions during similar task phases predicted whether a subject rated items as forgotten. Together, these findings suggest that working memory capacity is already limited by the initial processing of the stimuli. Future analyses will examine which factors contribute to the prediction signals including fluctuations of attention between trials or spatial positions and stimulus characteristics.