Suboptimal Choice Behaviour in Hens

Abstract

Choice performances of six Brown Shaver hens were studied across three series of conditions, using a partial replication of Stagner and Zentall’s (2010) procedure. Hens chose between two alternatives during choice-trials. Choice of the low-probability alternative provided one of two stimuli that reliably predicted 100% reinforcement on 20% of the trials, and 0% reinforcement on the remaining 80% of trials. Choice of the non-discriminative alternative provided one of two unreliable stimuli that non-differentially signalled 50% reinforcement, regardless of the stimulus presented. In Conditions 1-4 and Condition 6, the stimuli and reinforcement contingencies associated with the side keys were repeatedly reversed across conditions. In Condition 5, stimulus and reinforcement contingencies were each equated at 50%. Results replicated Stagner and Zentall’s (2010) findings demonstrating that choice for the low-probability alternative for most hens remained almost exclusive across reversals, except Condition 5 where previous choice performances for all subjects, was disrupted. In Conditions 7-11, the reliability of the low-probability stimuli was gradually decreased across conditions. Results showed that the decreasing reliability of the low-probability stimuli did not deter many hens from choosing suboptimally. Conditions 12 and 13 reversed Conditions 8 and 11, respectively. Results showed that choice performances for most hens in Condition 13 varied considerably from Condition 11. The main finding from this study is that hens do not exclusively choose signals that reliably signal a reinforcing outcome. This is because choice for the low-probability alternative for many hens was not affected when the reliability of those signals decreased. This suggests that the signals themselves may be more reinforcing than their subsequent outcome which may have implications for human gambling behaviour.