ABSTRACT
The overreliance on verbal models and theories in psychology has been criticized for hindering the development of reliable research programs (Harris, 1976; Yarkoni, 2020). We demonstrate how the conceptual space framework can be used to formalize verbal theories and improve their precision and testability. In the framework, scientific concepts are represented by means of geometric objects. As a case study, we present a formalization of an existing three-dimensional theory of emotion which was developed with a spatial metaphor in mind. Wundt posits that the range of human emotion can be represented along three axes of basic emotions (pleasure/displeasure, excitement/inhibition, tension/relaxation), just as color vision can be represented using three basic colors. We use dimensions to represent basic emotions, points to represent emotional states, and regions to represent broader emotional concepts. We then compare our formalization to an existing structuralist formalization of Wundt’s theory. Further, we discuss the empirical predictions that our formalization generates, such as comparisons of similarity and intensity. We conclude by demonstrating how the tools developed in the conceptual space framework can be used to formulate a theory of emotion based on empirical observation.
Acknowledgements
Frank Zenker is the author of the idea of using conceptual spaces to formalize psychological theories. We would also want to thank Müge Kuyumcuoğlu Tütüncüoğlu, Aran Arslan, Bican Polat, Seçil Aracı, Zeynep Burçe Gümüşlü, Rainer Reisenzein, Disa Sauter, Dilectiss Liu, Rafał Urbaniak, Bartosz Wcisło, and Frank Zenker for their useful comments.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Author contribution
Michał Sikorski’s contribution amounts to 70% of all work related to the manuscript and Ohan Hominis contributed the remaining 30% of the work.
Notes
1. We do not claim that the formalization by itself solves the replication crisis. Of course, improvements to statistical designs and to the reward system are necessary to successfully ameliorate the crisis (see Eronen & Bringmann, Citation2021, p. 785). However, in light of the problems caused by the verbal nature of psychological theories, it seems that formalization has to be a part of the solution.
2. These terms are variously translated, here we use translations from Reisenzein (Citation1992).
3. The terms are sometimes used differently; Damasio (Citation2004) and Hansen (Citation2005) argue that emotions are fundamental, and the experience of the emotion is to be considered the feeling. To maintain coherence in the case study, we adopt the terminology of Wundt and Reisenzein.
4. It would be a disservice to Wundt to not mention the distinctions seeing as he took pains to draw them. Yet, it would similarly be a disservice to our readers to provide a formalization laden with outdated and needlessly confusing terminology.
5. According to Wundt, a state in which no emotions are felt is unlikely. Our formalization is not committed to the realizability of such a state. Yet, it seems to be a natural interpretation of the point (0,0,0). Alternatively, point (0,0,0) can be interpreted as a neutral emotional state.
6. Berrios et al. (Citation2015) conduct a meta-analysis of mixed emotion. The results suggest that “mixed emotions are a robust, measurable and non-artifactual experience”.
7. Measurements play a central role in quantitative theories of emotions. At the same time, such theories need to relate their explanations to measurable variables. We do not argue here that theories need to be reduced to formal models. Rather, we argue that the increase in formalization, particularly of the quantitative components, would help the appraisal of theories. As we have seen in the introduction, there have been tendencies for papers with opposing viewpoints to speak past one another instead of explicitly addressing one another’s concerns. Formalization offers an avenue toward more explicitly defined realms of debate.
8. The main aim of Douven (Citation2016) is to test the theory of graded membership developed in Kamp and Partee (Citation1995). Because of that, he does not focus too much on the details of the conceptual space he developed. For example, he does not interpret all of the generated dimensions.
9. Arguments supporting the convexity hypothesis strongly suggest that non-convex concepts should not be used in inductive reasoning. Consequently, it seems clear that such concepts should not be employed in formulating empirical hypotheses. However, it is not clear if these concepts can play an auxiliary role in the process of theory construction, for example, during the derivation of consequences of the theory in question. Perhaps a good way to study this issue and provide additional support for the convexity hypothesis is to conduct historical studies to determine if non-convex concepts were present in successful scientific hypotheses and theories, and if so, what role they served.
10. Theories that aim to identify a set of universal, innate emotions are able to provide a plausible starting list of empirically testable affective concepts (Ortony, Citation2021).
11. The validity of a newly constructed conceptual space can be tested. First, one can test whether the generated dimensions correspond to quantitative and measurable attributes (see e.g., Trendler, Citation2009). Then, they can test whether the construct validity of the space and measurements correlated to the posited dimensions (see e.g., Embretson, Citation2016). Plausibly such tests and the adjustments based on their results can be used to further develop the space into a robust framework for the classification of emotions.