Fingerprints or populations?

Abstract

For as long as scholars have been writing about the human mind they have been speculating on the nature of affect and emotion. Currently, scientists agree that affect and emotions are tools that evolved for dealing with the challenges of life. Scientists also agree that both affect and emotions involve physiologically-mediated changes in heart rate, respiration, activity of the sweat glands, etc., although the nature of these changes have been debated for over a century. Classical theories hypothesize that certain emotion categories have biological essences resulting in physiological "fingerprints" that are specific to one category and distinct from other categories. An alternative hypothesis, what we call the "populations" hypothesis, suggests that there are no fingerprints for emotion categories. Rather, categories name a population of variable instances that are context sensitive. We present data from two meta-analyses of over 350 published studies of physiological reactivity during instances of emotion and affect in which we directly test whether there are specific biobehavioral patterns that form a distinct "fingerprint" for affect and emotion categories. We utilized traditional univariate meta-analysis alongside multivariate pattern classification to search for fingerprints. In our emotion meta-analysis, both traditional univariate meta-analytic techniques and multivariate pattern classification failed to reveal evidence of distinct autonomic fingerprints for emotion categories. Instead, we found tremendous variation within and across emotion categories. We also found that experimental context explained a significant portion of the variability in many comparisons. We found significant within category variability in our affect meta-analysis as well. However, in our multivariate pattern classification analysis, we were able to classify affect categories slightly above chance, but found no evidence of specific patterns in our univariate results. Overall, our findings did not suggest robust fingerprints for either affect or emotion. Rather, pervasive variation across and within categories suggested that both affect and emotion may be more accurately described as abstract categories that describes a population of variable instances.