Anatomy of early visual cortex predicts visual working memory capacity

Abstract

Visual working memory (VWM) provides a vital link between sensory input and higher-level cognitive processing and is subject to large interindividual variation. Previous research has suggested that limitations to VWM capacity are set by higher-order areas such as prefrontal, posterior parietal, inferotemporal and lateral occipital regions and their interactions. However, the role of early visual cortex has recently caught the interest of researchers, shedding light on the question how individuals are able to remember more fine-tuned information about a stimulus, such as a grating’s orientation. Using a combination of psychophysical testing, functional and structural magnetic resonance imaging methods (MRI), we show that anatomical properties of early visual cortex are also strongly predictive of an individual’s visual working memory capacity. In a two-alternative-forced-choice task, we asked our participants to remember the orientations of Gabor gratings that surrounded a fixation point in a circular fashion. Using the standard retinotopic mapping procedure to functionally define the boundaries of early visual cortices in each participant individually, we found that participants’ performance in the task was strongly positively correlated with the surface size and cortical thickness of primary visual cortex (V1) and with cortical thickness of secondary visual cortex (V2). That is, participants with a larger V1 or thicker V1 and V2 tended to have a higher VWM capacity. In contrast, we found no such relationship when using a numerical working memory version of the task, thereby confirming the specificity of the observed relationships. Whole brain cortical thickness analyses additionally supported the importance of low-level early visual cortex as a decisive stage for setting individual capacity differences. Our findings insinuate that early sensory areas might play a larger role in shaping individual differences in higher cognitive functions than hitherto thought.

Meeting abstract presented at VSS 2013