The Interpretation of Individual Differences in Rhesus Monkey Infants

Abstract

A varimax rotated solution of a Principal Components Analysis (PCA) was used to discover how in the rhesus monkey behavioural measures of interactions between mothers and infants and between infants and other members of the group clustered, in the sense that they were more strongly correlated with each other than with other measures. Eight week-old infants, living with their mothers in captive social groups, were the subjects, and the males and females were studied separately. A form of cluster analysis, the construction of Maximum Spanning Trees (MST), was used to display clusters of behaviours, corresponding to the PCA groupings, and to show how separate groups of measures could nevertheless be connected to each other. In both sexes, a group of measures called Mother protects proved to be separate from a group called Infant social, and these groups were both separate from various measures reflecting maternal rejecting behaviour. Also separate was a single measure of the infants' tendency to show distress by squeaking, geckering and screaming, and a measure of the time the infants spent alone while out of contact with their mothers. The separation of these groups of measures means that, in working with the populations of male and female infants as wholes, it is not possible to predict an individual's score on a measure from one of the above-mentioned groups from the score of a measure from another group. The sexes differed in that the mother's act of moving away after her infant approached, possibly a rejecting or leading tactic, was associated with the other kinds of rejecting behaviour in female infants only. In males, mothers who groomed them for shorter times tended to restrain them more often, while in female infants maternal grooming was associated with other monkeys initiating contact with the infants. In females, but not males, the frequency with which others made contact was not associated with the frequency with which the infants played with others. We showed how measures can be interpreted in terms of the groupings of their component measures. Two derived measures, % Rejections and the Proximity Index, could (like a monkey's weight) be correlated with a variety of measures, themselves uncorrelated. We discussed the consequences of this, together with certain other complications that can arise when measures are used. Where we failed to find correlations between measures or groups of measures, we thought it possible that correlations might nevertheless be found in sub-populations of the infants. For example, only those infants whose mothers rejected them at above a certain frequency might show a correlation between the frequency of their distress and the frequency with which others make contact with them.