Bulf, H. 1, 2 , Bettoni, R. 1, 2 , Westphal-Fitch, G. 3 , Fitch, T. 3 , Cecchetto, C. . 1, 2 & Macchi Cassia, V. . 1, 2
1 University of Milano-Bicocca
2 Milan Center for Neuroscience (Neuro-MI)
3 University of Vienna
Rule learning (RL) refers to the ability to detect high-order repetition-based rules (ABB, AAB, ABA) from a sequence of elements and to generalize them to new items. While infants can learn high-order rules from both auditory and visual sequences, it is not clear whether RL operates on perceptual features -by extracting the mere repetition of the same element- or on a more abstract level. Here we investigated 7-month-olds? ability to learn ABB (adjacent repetition) or ABA (nonadjacent repetition) rules embedded in triplets of visual shapes in which adjacent and nonadjacent repetitions were provided by two (perceptually) different elements that belonged to the same category. Visual categories were closed (e.g. diamonds) and open (e.g. Xs) geometrical shapes, each of which was instantiated by 6 different exemplars. Infants (n=18) were habituated to ABB or ABA rules, and subsequently tested with both ABB and ABA triplets composed of shapes that differed from those shown during habituation. At test, infants looked longer to the novel sequences when habituated to the ABB rule, but not when habituated to the ABA rule. These findings show that infants are able to extract and generalize repetition-based rules at an abstract level only in the presence of adjacent repetitions.