Extended Data Fig. 3: RSA replication in a motion matched subsample and longitudinal effects.
From: Infants have rich visual categories in ventrotemporal cortex at 2 months of age
(a,b) RSA results in a subsample of participants with matching motion distributions to the adult cohort, excluding the NICU group. Correlations are normalised by the motion-corrected noise ceiling per ROI/age-group. Boxes bounds are the 25th and 75th quartiles of the bootstrapped 95% confidence interval across subject pairs (2-months (n = 101): 14108 unique subject/run pairs, 9-months (n = 44): 1995 pairs, adults (n = 17): 136 pairs). The middle line is the median, whiskers extend to 1.5 x IQR, individual points are outliers (c) The average within subject correlation across 2 and 9 months (n = 38) is depicted in the red dashed line. This was compared to a null distribution, estimated using 1000 permutations shuffling across subjects and ensuring the same subject was not being correlated with themselves in the permuted test. We find no effect of longitudinal similarity in the representations of EVC (p = 0.656), VVC (p = 0.378) and LO (p = 0.507).
