What are emotions and how many are there?

Human emotion has been a focus of scientific research across a wide variety of scientific disciplines and, in spite of careful research dating back as far as Charles Darwin, no consensus has emerged as to what emotions are, how many there are, and whether they are sharply bounded biological processes or culturally defined processes with diffuse boundaries, changing in response to contextual factors. Recently, Cowen and Keltner published research applying what they call a “semantic space approach” to the study of emotion (Cowen in Trends Cogn Sci 22:274–276, 2018; in Proceedings of the National Academy of Sciences of the United States of America, 2017; in Am Psychol 2019). Their method finds that respondents recognize 28 emotional categories arrayed in a 27-dimensional space. These categories are not sharply bounded, but tail off into one another. Some researchers (Barrett in Trends Cogn Sci 22(2):97–99, 2018) have called their methodology into question, and recommended alternative procedures to check these results. In this article we apply an alternative, widely used methodology for generating semantic spaces to Cowen and Keltner’s emotional categories. Results of this alternative (“Galileo”) analysis conducted in the United States and Singapore support the finding that the space in which these 28 emotional categories lie is a high dimensional space, and that they are indeed not sharply bounded. Additionally however, the Galileo method finds that the underlying space is non-Euclidean and it provides information about the actual sizes of the dimensions. Results also show that some of Cowen and Keltner’s emotional categories (such as frustration) may not be best described as emotions themselves, but rather as situations in which other emotions (e.g., anger, sadness, etc.) may be generated.