Psychosomatic Medicine 23:323-343 (1961)
© 1961 American Psychosomatic Society

Cluster Analyses of Skin Resistance and Heart Rate at Rest and Under Stress

¹ University of California, Berkeley, Calif.

This study was undertaken as part of a research program on personality and psychological stress (for methodological as well as theoretical reasons) to obtain systematic information on the structure of autonomic reactivity under neutral and stressor conditions. Thirty-five male and 35 female undergraduate students participated in an experimental procedure consisting of a personality assessment session, a control session (neutral film on corn farming), and an experimental session (stressor film on primitive subincision rites). During the control and experimental sessions continuous recordings of heart-rate and skin resistance were taken, and just prior to the S's exposure to the films, a base-line rate (no obvious stimulus) was taken on each autonomic channel. This provides four experimental conditions under which to compare autonomic response: control base, control film, and stress base, stress film.

The autonomic nervous system records were then scored by means of the most commonly used techniques as indicated in the current literature. For skin resistance these scoring procedures may be classified into five main categories: level, variability, lability, response, and recovery. The heart-rate records were analyzed in essentially the same ways except that lability could not be scored. The various methods of scoring were followed across the four experimental conditions, and this provided us with correlation matrices of 63 variables for skin resistance and 53 variables for heart-rate. These matrices were then cluster analyzed, and 10 oblique clusters for skin resistance and 11 for heart-rate were located on orthogonal dimensions.

The main conclusions drawn from the cluster analyses are as follows:

The claim that certain types of autonomic measurement are independent of each other has been substantiated in our data. Such categories of measurement as level and variability are distinguishable from each other, and since they sometimes define separate clusters, it is clear that they have different psychophysiological meanings.

A second finding of significance, and representing a qualification of the above statement, is that the structure of autonomic measurement varies with the specific channel (e.g., heart-rate or skin resistance) being studied and the experimental conditions under which the measurements are made. An example of these complex interactions occurs in the case of heart-rate measures of level and variability. These measures are independent (define separate clusters) under resting conditions, but under stressor conditions they both load on the same cluster indicating that under strong arousal conditions level and variability appear to have essentially the same psychological meaning.

Another important example of the effects of experimental conditions on the organization of autonomic nervous system activity can be seen with respect to skin resistance measures of variability. Variability appears as a separate cluster under each of the experimental conditions, control base-line, control film, stressor base-line, and stressor film. It is highly sensitive to conditions and means different things under the different experimental conditions. As another instance, lability emerges as a separate dimension only under the stressor film condition, while under control conditions it is indistinguishable from variability and response (biggest GSR).

Thus, although it is correct to say, as in (1) above, that such categories of measurement as level and variability are frequently independent of each other, this depends not only upon the particular autonomic nervous system channel involved, but also upon the nature of the conditions under which the measurement is made. What emerges as an independently defined dimension under one condition may disappear or be fused with other dimensions under another.

Some measurement classes of autonomic nervous system activity, which are often employed in research in personality dynamics, are never observed to load up to the criterion of acceptability (as used in this research) on any of the cluster dimensions in the analyses. Such measurements as level and variability appear to be particularly meaningful since they often either define independent dimensions themselves or load heavily on other cluster dimensions. Other classes such as recovery or response appear less useful even though concepts which define them are much better known and referred to more often in the literature. This could well be explained on the basis of the special conditions of the present research as compared with others. However, the work of Lacey⁴ and Lacey and Lacey, to which we have often alluded, reflects increasing interest in such dimensions as variability, level, and lability as contrasted with the more traditional ones of response and recovery.

Our findings support the independence and sensitivity to conditions of level and variability but do not support the usefulness of the dimensions of response and recovery. At the same time, the observation that certain classes of autonomic activity emerge only under certain experimental conditions cautions against overzealous generalization about what types of measurements are meaningful.

From a methodological standpoint, the research has greatly simplified our own future analysis of autonomic activity because it has revealed a great many redundancies among the many possible specific techniques within classes of measurement. For example, many measures of level of skin resistance activity are possible, and load very highly together within the same cluster dimension. Judicious selection of scoring systems can be tailored to fit given experimental conditions, and there can be considerable confidence that many of the specific techniques are interchangeable. The present research provides a small catalogue of these interchangeable techniques, along with examples of some which appear to be relatively independent.

Two main conclusions (outside the scope of the cluster analyses) are also provided from the analysis of the effects of the stressor film conditions on autonomic activity, and from the correlations between the skin resistance and heart-rate channels. These are:

The large and consistent effects on the important autonomic indices such as level and variability that resulted from the presentation of the subincision film fully justify our reference to that experimental condition as a stress producer.

There is little relation between heart-rate and skin resistance activity, with only isolated exceptions which are invalid because of spuriousness or because some apparently significant correlations will occur merely by chance in cluster analyses. The lack of correlation across the two autonomic channels is consistent with the findings from other laboartories.

We believe that further systematic research of this sort will clarify the structure of the autonomic nervous system, and what is more important to the personologist, highlight its role in stress arousal.