Abstract
Identification of visually presented numbers and letters is an essential task in literate societies, yet the neural mechanisms of this behavior are still unclear. Recent studies have argued that specific regions of ventral visual cortex are selective for numbers (Grotheer et al., 2016) and letters (Thesen et al., 2012). Here, we completed an ultra-high resolution (7T) fMRI study to investigate whether these regions encode information about individual symbols. First, we determined whether separate number and letter selective regions could be found in ventral visual cortex. Next, we used multivariate pattern analyses (MVPA) to explore which regions of ventral visual cortex encode numbers and letters through different patterns of response to individual symbols. Two experiments were interleaved during a single scanning session: 1) a block design localizer using a one-back task, 2) an event-related experiment with presentations of individual letters and numbers. The localizer included blocks of numbers, letters, false numbers, and false letters. The event-related runs included five letters (c, d, e, f, g) and five numbers (3, 4, 5, 6 ,7) presented in four fonts. Candidate number and letter selective regions were defined as in previous research by contrasting numbers against false numbers and letters against false letters. These regions showed high overlap. Moreover, direct contrasts of numbers vs letters failed to reveal regions of interest. However, the localizer revealed bilateral regions responsive to both numbers and letters, and MVPA was completed within these regions using data from the event-related runs. These symbol-responsive regions showed different patterns of response for individual numbers and for individual letters. Thus, our results argue against separate regions of visual cortex showing univariate differences in response to numbers and letters. However, bilateral regions of ventral visual cortex responsive to both letters and numbers contain information about individual symbols in multivariate differences in neural response.
Meeting abstract presented at VSS 2017