Response Specificity in Psychophysiology

Abstract

In the field of response specificity, we are faced with a great diversity of specificity measures and definitions which make it difficult to compare the results presented in the literature. The objective of this paper is to assess different specificity calculations (individual, situational, and motivational response specificity) and to elucidate the underlying meaning of those measures which are based on correlations and on specific sources of variance. It will be shown that an approach on the basis of variance components is suitable to elucidate different specificity measures. The main advantage of this approach is that it explicitly describes the frame of reference of the specificity measures and that it clarifies the relationships between specificity, consistency, and covariation. An empirical investigation is presented which demonstrates the application of the approach and which gives an idea of the magnitude of the differences between the specificity measures. Special features (averaging procedures, fixed/random factors) are discussed and recommendations for further research are derived.

An affective computing approach to physiological emotion specificity: Toward subject‐independent and stimulus‐independent classification of film‐induced emotions

“…Third, whereas most of the emotion classification studies did not test their classifiers with a different data set than it was trained with, cross‐validation is an essential step for evaluating the performance of a given classifier in case of unknown data. Moreover, as individual and situational specificity has long been acknowledged in psychophysiological research (Hinz, Seibt, Hueber, & Schreinicke, 2000; Lacey & Lacey, 1958; Marwitz & Stemmler, 1998), it is important that also emotion classification studies acknowledge the need for a differential treatment of subject and situation dependence in trained classifiers. Supporting the importance of considering the influence of such varying factors, generalizability theory (Webb & Shavelson, 2005) suggests that facets, across which generalization is sought, such as subject and stimulus characteristics should be taken into account.…”