Affects affect affects: A Markov Chain

In psychology, the term affect has been increasingly used to indicate an overarching state including a wide range of phenomena, including the experience of feelings, moods, and/or emotions (Schiller et al., 2022). Affective states, in particular, refer to individuals' current emotional state or mood toward allostatic goals. The first attempt to outline how the affective space was organized, using standard coordinates, can be traced back to James A. Russell (Russell, 1980; Russell and Barrett, 1999; Posner et al., 2005). Russell's circumplex model of affect was increasingly interpreted as being able to plot each emotion precisely on a two-dimensional plane, with arousal on one axis and valence on the other (Posner et al., 2005; Britton et al., 2006; Jefferies et al., 2008). Arousal refers to the intensity of our emotional experience whereas emotional valence refers to whether our emotional experience is positive or negative in nature. This approach makes it possible to outline the emotional “identikit” of each affective state and emotion in a punctual way. That is, sadness has been progressively conceived as a negative valence emotion with a low level of arousal. Conversely, joy has been deemed as a positive valence emotion with a medium to high level of arousal intensity. Consequently, in the last 20 years, a wide array of stimuli has been proposed to elicit and study affective states in this punctual way, ranging from pictures and videos to sounds, narratives, real situations, and virtual reality and music. While treating affects as independent states created easier ways to elicit, measure, and operationalize them, this assumption may have held some drawbacks in terms of validity when dealing with the continuous stream of affect that characterizes real life. In real life, an affect can affect the current affective state and/or the next one. Moreover, an increasing number of studies have developed mathematical indexes to compute the mixed nature of specific affective states (Picard et al., 2001; Calvo and D'Mello, 2010; Cipresso et al., 2015, 2017; Hamaker et al., 2015; Cipresso and Immekus, 2017; Poria et al., 2017; Waugh and Kuppens, 2021), thereby suggesting that affect features several nuances and also embeds paradoxes that may be not fully explainable with a linear and punctual model of affect. To put it differently, the effect of one affective state carries over to the next state and cannot be treated as an isolated occurrence because, in real life, each state is connected to the previous (and/or to the future) one. Thus, the affect of one state will have an effect on the following one. For instance, if a student feels stressed at the start of an exam, the stress will likely carry over to the following task, making it difficult for the individual to focus and answer questions correctly. Therefore, it becomes far more important to recognize the connection between affective states and the influence that one state may have on subsequent ones. However, most (if not all) current psychological approaches in the study of affect have not drawn from mathematical and statistical paradigms, thereby allowing for this operationalization of affect (Nummenmaa and Niemi, 2004; Mauri et al., 2010).