Fig. 6.

Traditional representational similarity analysis using sample correlations from a real data set. (A) Correlation-matrix for primary somatosensory cortex (S1) for two conditions (sense and move) and four levels (fingers 1, 2, 3 and 5). Bright squares indicate high correlations, dark squares zero or slight negative correlations. (B) In S1, the average correlation between movement and sensory patterns for the same finger (1) was elevated compared to those between different fingers (2). This was also the case for M1, but not for lobule V of the cerebellum. (C) The correlations between patterns for different fingers within the same condition were higher and more pronounced in the movement (4) than in the sensory (3) condition, suggesting different strength of the common activation patterns. (D) These correlations were much higher for patterns estimated within the same run than for different runs, indicating a strong covariance in the estimation errors. Error bars indicate between-subject (N = 7) standard error.