Abstract:
An historical emphasis on the behaviour of individual organisms has ensured that studies of group behaviour are rare in the experimental analysis of behaviour. Moreover, due to the impracticability of automating the collection of data from individual group members, the few existing studies of group behaviour have a common shortcoming: An inability to measure the contribution of individuals to the collective performance of the group in toto. An unique approach to this problem is taken in the present experiments: "Group members" are housed in separate experimental chambers, but are required to compete with one another for access to concurrent food sources.
In Experiment l, five pigeons competed for their entire food intake on a single, continuously available concurrent variable-interval variable-interval schedule. The subjects were ranked according to an arbitrary dominance hierarchy which determined the priority of a subject's access to the concurrent alternatives. Each subject's responding constrained the alternatives available to those lower on the hierarchy. The availability of a schedule to a given subject was signalled by whether the appropriate key was illuminated in that subject's chamber. Across six experimental parts, the level of between-subject competition was successfully manipulated via the overall reinforcer rate arranged by the concurrent schedule. Within each part, the arranged reinforcer distribution was varied over a wide range.
These manipulations produced orderly changes in both individual and group performances. Dominance rank exerted a progressively greater influence over individual performance with increasing competition. The generalised matching law accurately described the changes in behaviour allocation within experimental parts, with the majority of parameter estimates being consistent with values representative of the concurrent-schedule literature. However, the data did not provide good support for the model of competitive concurrent-schedule performance proposed by Gray and Davison (in press). Essentially, the model failed to account for the effects of the "communal" nature of the concurrent schedule.
The effects of priority access and the "communal" versus "personal" nature of the concurrent schedule on the performance of six pigeon dyads were examined in Experiment 2. Consistent with Experiment 1, dyad members received simultaneous experimental sessions in separate chambers. However, unlike Experiment 1, sessions were limited to a maximum of 40 minutes and each subject's consumption of food was independent of its sessional performance. These modifications afforded a more straightforward test of the Gray and Davison (in press) model.
Priority access and the nature of the concurrent schedule produced distinct effects on the performance of the dominant and subordinate members of each dyad. Once again, the generalised matching law accurately described the relationship between behaviour allocation and the obtained distribution of reinforcers. Consistent with expectations, when dominant and subordinate dyad members earned reinforcers on "personal" concurrent schedules, the Gray and Davison (in press) model accurately predicted individual and dyad performances; yet the model again failed to account for the performance of individuals competing for "communal" reinforcement. However, a series of modifications to the model improved its predictive accuracy considerably, at least for t1g simpler two-forager situation.
The orderly changes in individual and group performances demonstrated in these experiments establish the merit of an individual-organism approach to the study of group behaviour.