Transitive responding in pigeons: influences of stimulus frequency and reinforcement history

Fersen et al. (1991) (J. Exp. Psychol.: Anim. Behav. Process., 17: 334-341) trained pigeons to discriminate four overlapping pairs of stimuli A + B -, B + C -, C + D - and D + E - (+ rewarded, - penalized). When subsequently tested with a pair BD the pigeons showed a strong preference for stimulus B. A special value transfer theory was offered as an explanation for this transitive responding. A simpler reinforcement ratio account based on certain inequalities factually affecting the accuracy performance on training pairs is proposed. To explore its implications an experiment employing a novel grit-grain conditioning method was carried out. The presentation frequencies of the training pairs were biased so that the choice accuracies obtained at the end of training were approximately equal for all pairs. Testing with pair BD still yielded high preference for B, documenting the robustness of the transitive responding phenomenon. When suitably adjusted to the training design the reinforcement ratio account was still viable. The transitive responding with the BD and other test pairs could also be simulated with a simple reinforcement based conditioning model. Some of the subjects were then retrained with modified presentation frequencies so that the subjects ended up with an even overall exposure to all training pairs. Test pairs continued to yield strong transitive responding. It is concluded that this behavioral effect is a robust phenomenon which is largely unaffected by training design modifications.