Purchase this article with an account.
Daniel Janini, Brett Bankson, Chris Baker; Representational dynamics of number processing in symbolic and non-symbolic formats. Journal of Vision 2018;18(10):827. doi: 10.1167/18.10.827.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The human brain can rapidly form number representations from a variety of visual formats including digits, number words, and dot displays. While an extensive literature has investigated where these symbolic and non-symbolic number representations are formed in the brain, less is known about the temporal aspects of this process. Here, we explored the emergence of number representations by applying multivariate pattern analyses to MEG data. Participants (n = 22) encoded the magnitude of visually presented numbers across three different formats (digits, words, and dot displays). First, we investigated how quickly the brain forms individual number representations from visual input, and if this differs between formats of presentation. We used a support vector machine (SVM) to classify number within each format at each time point. The SVM results revealed peak decoding at 110-120ms following stimulus presentation within each of the three formats. Next, we assessed whether number magnitude and visual shape models could explain the variance in neural response to different stimuli. Representational similarity analyses (RSA) were used to compare number magnitude and visual shape models to the neural responses. RSA yielded the highest correlations to the number magnitude model at around the same time point as peak decoding accuracy - 120ms for digits, 135ms for words, and 120ms for dot displays. With regard to the shape model, the highest correlations were found at different time points depending on the format - 170ms for digits, 70ms for words, and 120ms for dot displays. Thus, individual number representations are formed quickly by the brain for both symbolic and non-symbolic formats of presentation. These representations emerge concurrently with magnitude information, while shape information appears to drive the MEG signal in a more format-specific manner.
Meeting abstract presented at VSS 2018
This PDF is available to Subscribers Only