Learning Δs per subject and per model calculated separately for observed successes vs. errors.

Each learning Δ represents the gain or loss observed in the number of errors committed over 10 hands-on trials for pairs preceded by observation of a model vs. pairs learned purely individually (individual score – social score/individual score *100). Positive Δs indicate that individual learning after observation of a model was better (i.e. accompanied by less errors) than purely individual learning, whereas negative Δs correspond to a loss of performance after observation, i.e. more errors.