The Temporal Pattern of Behaviour in Isolated Male Zebra Finches : Transition Analysis

A transition analysis has been carried out on the behaviour of single male zebra finches, isolated as far as possible from all varying environmental stimuli. As wide a variety of behaviour patterns as feasible was included and grouping of these was kept to a minimum. The dangers of grouping acts on functional rather than causal grounds are stressed. The method of analysis differs from most of those used previously in the following ways: 1. Data from different individuals are not massed for statistical treatment. This avoids a source of bias which makes it difficult to reach conclusions about the population as a whole. 2. Transitions between one act and itself are not considered. It is argued that models including such transitions have a strong arbitrary component and are predisposed to demonstrate merely that certain acts occur in bouts. A method is employed for calculating the expected frequencies of transitions which does not appear to have been used before in this field. It is based on a random model which excludes from consideration homogeneous transitions and any others which cannot occur because of environmental constraints. The analysis suggests that zebra finches show cycles of behaviour having active and inactive phases, with ingestion occurring mainly in the former and grooming in the latter. Song appears to be transitional between the two and to occur during periods when other tendencies are low, while stereotypes often follow such periods. Certain pairs of behaviours which occur frequently in sequence are discussed. In some cases one probably generates the peripheral state appropriate to the other and in others the pairing appears to have central causation. Triplet analysis on the massed data from all birds suggests that for most behaviours triplets of the form A-B-A are commoner than would be expected from the frequencies of the pairs A-B and B-A. It is suggested that factor analysis is not a very useful method for dealing with transition data. Detailed closer study of particular interactions, and of the temporal pattern of individual behaviours, are likely to prove more fruitful.