Matching Skin Conductance Data to a Cognitive Model of Reappraisal

Matching Skin Conductance Data to a Cognitive Model of Reappraisal Tibor Bosse 1 , Jessica Brenninckmeyer 2 , Raffael Kalisch 2 , Christian Paret 2 , and Matthijs Pontier 1 VU University Amsterdam, Department of Artificial Intelligence, de Boelelaan 1081, Amsterdam, 1081HV, the Netherlands tbosse@few.vu.nl, mpr210@few.vu.nl University Medical Center Hamburg-Eppendorf (UKE), Institute for Systems Neuroscience, Martinistr. 52, 20249 Hamburg, Germany Jessica.Brenninkmeyer@stud.uni-hamburg.de, rkalisch@uke.uni-hamburg.de, christian_paret@hotmail.de Abstract In the present paper we show that an existing mathematical model of emotion regulation can, if reduced to its reappraisal- specific components, fit skin conductance data obtained from an empirical study of reappraisal. By applying parameter tuning techniques, optimal fits of the model have been found against the (averaged) patterns of the skin conductance data. The errors that were found turned out to be relatively low. Moreover, they have been compared with the errors produced by a baseline variant of the model where the adaptive cycle has been removed, and were found substantially lower. Keywords: emotion regulation, reappraisal, mathematical modeling, adaptation, skin conductance data. Introduction Emotion regulation refers to ‘all of the conscious and nonconscious strategies we use to increase, maintain, or decrease one or more components of an emotional response’ (Gross, 2001). This ability to regulate our own emotional states provides us with behavioral flexibility and is related to well-being and mental health (e.g., Gross, 1998, 2001; Ochsner and Gross, 2005; Thompson, 1994). Recently, a number of authors have developed computational models of the processes related to emotion regulation and coping (e.g., Bach, 2008; Bosse et al., 2010; Gratch and Marsella, 2004; Marsella and Gratch, 2003; Reisenzein, 2009; Silverman, 2004). Computational models of emotion regulation may be useful for various reasons (see (Wehrle, 1998) for an overview). From a Cognitive Science perspective, they may provide more insights into the nature of affective disease and the working mechanisms of therapy. From an Artificial Intelligence perspective, they may be used to develop virtual agents with more human-like affective behavior. In previous work (Bosse et al., 2010), we presented CoMERG, a Cognitive Model for Emotion Regulation based on Gross. Inspired by the theory put forward in (Gross, 2001), this model distinguishes five different strategies that humans typically use to affect their level of emotional response (for a given type of emotion) at different points in the process of emotion generation: situation selection, situation modification, attentional deployment, cognitive change, and response modulation. The different strategies and their effects are represented in the model via a set of difference equations. An important asset of CoMERG is that the model is adaptive (see Bosse et al., 2007b). That is, based on the perceived success of an emotion regulation strategy that is performed, a person may adjust the degree of sensitivity of the process on the fly (e.g., in case a certain strategy does not decrease an undesired emotion sufficiently fast, the person may put more effort in the regulation). However, although a preliminary evaluation indicated that CoMERG produced plausible patterns (Bosse et al., 2010), to date the output of the model has never been compared with empirical data. In order to assess to what extent CoMERG is able to reproduce empirical data, we here fit the model to skin conductance data that resulted from two empirical studies of reappraisal (unpublished material). Reappraisal, a variant of the cognitive change strategy aimed specifically at down- regulating emotion, is one of the most widely studied emotion regulation strategies. Gross (2001) defines reappraisal as a process where ‘the individual reappraises or cognitively re-evaluates a potentially emotion-eliciting situation in terms that decrease its emotional impact’. For example, losing a tennis match is usually appraised as negative and would induce anger or sadness. To reduce these negative reactions, one could reappraise the situation by blaming the weather circumstances instead of the own capacities or by considering sportive success as irrelevant. In (Kalisch, 2009), a novel (informal) model for reappraisal is presented, based on recent insights from imaging neuroscience. This model, called the implementation-maintenance model of reappraisal (IMMO), is characterized by its focus on the necessity of a mental reappraisal effort that needs to be maintained over the course of the emotional episode and is continuously adapted. Adaptation is realized through a loop of iterative evaluation and readjustment of the regulation process. IMMO thus shares a critical adaptation component with CoMERG. To be able to better fit the results of CoMERG to the skin conductance data, the general model needs to be tailored specifically to reappraisal. Thus, the current paper has two main goals, namely 1) to refine the generic computational emotion regulation model CoMERG to the reappraisal