Rule learning in an animal model (Gallus gallus).

Rule learning involves the ability to detect and generalize rule-like identity patterns, such as those differentiating the sequence AAB from ABA. This faculty, possibly involved in language acquisition, has been studied in human infants who, after a familiarization phase with sequential auditory or visual stimuli, generalize to new sequences with the same structure. Similar forms of sequence learning have been demonstrated in non-human animals, such as cotton-top tamarins (tested with stimuli and procedures resembling those used in infants). We investigated this ability in visually naïve chicks, which were exposed to a video sequence of arbitrary visual elements (coloured geometric shapes). In Exp. 1, the imprinting phase was followed by a recognition test between the imprinting sequence (e.g., ABA) and a structurally different unfamiliar one (e.g., AAB). Chicks revealed the ability to discriminate the two sequences, being quicker when approaching the unfamiliar one (expressing a preference for a stimulus representing a slight novelty with respect to the imprinting object). In Exp. 2, after the recognition test, we introduced a generalization phase during which subjects had to choose between two new sequences, composed of perceptually novel elements differing in shape and colour. Only one of these was structurally identical to the imprinting stimulus. Chicks successfully generalized their initial learning, preferentially approaching the novel sequence whose structure resembled that of the imprinting stimulus. This reveals that rule learning, a mechanism involved in sophisticate cognitive operations like language acquisition, might be present in our animal model.