[The capacity for transitive inference in birds: the solving of the Gillan test by Corvidae and pigeons].

Transitive inference in birds was investigated using D. J. Gillan (1981) experimental procedure. In the initial experiment 11 pigeons and 14 crows were trained to differentiate colour stimuli in pairs of adjacent ones from five-stimulus succession A, B, C, D, and E. The birds learned that stimulus E was associated with more food items than D, D, in turn, signalled less food items than C, and so on. Then the birds were tested by newly formed pairs of stimuli from the same succession (BD, CE, and BE). There were two series of testing. Both crows and pigeons solved the transitive inference test successfully (choosing D and E) in the series with small numbers of food items. In the series with larger numbers of food items the pigeons shifted to random performance while in crows the percentage of correct choices decreased. The proportion of adequate test solutions grew with the absolute difference between the numbers of food items associated with stimuli to compare. The capacity for solving the test in our experiments can be considered as the result of immediate comparison of the absolute numbers of food items associated with each stimulus. Thus, the data cannot be regarded as the final proof that these species are capable for transitive inference.