Fig. 1: The above-chance classification of activity patterns for nouns and verbs in area V5/MT in congenitally blind individuals.

Results of support vector machine classification of activity patterns for noun blocks and verb blocks in the visual areas in congenitally blind (n = 20) and sighted (n = 20) participants. LO the lateral occipital area, FG the fusiform gyrus. Statistical testing against classification chance level was performed separately in each participant group using the permutation procedure, in which the actual results were compared with the null distribution of 1000 classification values obtained with the labels of noun blocks and verb blocks randomly reassigned. The results were corrected for multiple comparisons across the visual areas, within each group (Bonferroni correction for 8 tests). The statistically significant results were found only in area V5/MT in blind individuals (mean classification accuracy = 54.3%, **p = 0.008). The between-group test was performed only in area V5/MT, in which significant results were observed in the blind group, using two-tailed, two-sample t-test. Since only one between-group comparison was performed, the correction for multiple comparisons was not necessary. This analysis confirmed that the classification of activity patterns for nouns and verbs in area V5/MT was more accurate in the blind group than in the sighted group (mean difference = 5%, 95% CI [1.6%, 8.5%], t(38) = 2.94, Cohen’s d = 0.93, **p = 0.006). Error bars represent the standard error of the mean calculated across the results for individual participants in each group. The a priori chance classification level (accuracy = 50%) is marked with the black line. Source data are provided as a Source Data file.