Hooded Crows Solve a Transitive Inference Problem Cognitively
We investigated the ability of hooded crows to form transitive inferences. Subjects were trained to discriminate a series of overlapping pairs of stimuli: A+ B−, B+ C−, C+ D−, D+ E, where the letters stood for colour stimuli and plus and minus for rewarded or non-rewarded
choices. The stimuli were cards of different colours with a circle of the same colour on the reverse side and diameters decreased from A to E. To preclude an influence of the reinforcement history on choices with the test pair BD, an overcompensation phase was instituted after training. It
consisted of the presentation of all training pairs with frequencies selected so that the reward to non-reward ratios for stimulus D would be between 1.5 and 2.0 times greater than for B. If, during the BD test, the bird chose the stimuli according to these ratios they should prefer D. If
they chose according to diameter relation they should prefer B. During these tests, the crows strongly preferred B over D (83.1%). In a second experiment, subjects were trained with the same procedure except that the diameters of the circles were all the same. During this test, the performance
of two crows was not significantly different from chance level (53.1%), and the other two crows preferred D (80.0%). We conclude that crows can solve transitivity tests using cognitive mechanisms if they are offered additional information (in this case circle diameter) which, presumably, allows
them to represent the relevant stimuli in an ordered series.