Table 2. Affective Concerns and Features.

This table describes two sets of algorithms among affective phenomena—affective concerns and features—providing examples and their theoretical rationale. Each phenomenon’s computational problem is also specified, as well as algorithmic solutions existing in the field already.

Affective Phenomenon	Examples	Theoretical Rationale	Computational Problem	Existing Algorithmic Solutions
AFFECTIVE CONCERNS inference of objects as actionable and thus relevant, which is reflected in the felt implication of those objects	physiological, operational, global	Intentionality: feelings are about something (Frege, 1948; Russell, 1905; Dennett, 1989; Bedford, 1956; Brentano, 2012; de Sousa, 1990; Kenny, 2003; Leighton, 1985; Pitcher, 1965; Teroni, 2007; Deonna and Teroni, 2012; Clore and Huntsinger, 2009) Embodiment and Enactivism: all cognition is encoded as action (Varela et al., 2017; Colombetti and Thompson, 2008; Shargel and Prinz, 2017; Slaby and Wüschner, 2014; Hufendiek, 2015; Hutto, 2012) Affordances: the environment provides actionable meaning for the organism (Gibson, 1977) Motivation Theories: feeling states as motivational (Duffy, 1957; Frijda, 1986, 2017; Hommel et al., 2017)	how to infer relevance of objects	Bayesian: using observable (interoceptive or exteroceptive) sensory data to infer and update the non-observable meaning (hidden or latent conditional probabilistic states) of those sensory data (Barrett, 2017; Dayan et al., 1995; Doya, 2007; Friston et al., 2016; Knill and Pouget, 2004; Lee and Mumford, 2003; Neal, 2012; Seth, 2013; Seth and Friston, 2016; Smith et al., 2019; Wolpert et al., 1995; Palacios et al., 2020)
Immediate to Distal gradient of relevance according to the distance from metabolic impact that the actions demanded would have (i.e., complexity, timescale, abstractness)	physiological, operational	Evolutionary and Biological Theories: courses of action can be more immediate or more distal (Jonas, 2001)	how to organize needs with varying extents of actionability	Bayesian: hierarchical inference with varying levels of complexity, timescale, or abstractness; immediate concerns are hidden states inferred at a lowest level and distal concerns are those at the highest (Pezzulo et al., 2015; Pezzulo and Levin, 2016; Pezzulo et al., 2022)
Physiological concerns require the most immediate or concrete actions	nourishment (hunger), hydration (thirst), internal integrity (nauseous); sensation	Interoception and Homeostasis: internal state as indicative of homeostatic status (Craig, 2002, 2013; Pace-Schott et al., 2019, this issue)	how to address immediate needs	Reinforcement Learning: reflexive decision-making in reinforcement learning (i.e., model-free; van Swieten et al., 2021); pain as aversive prediction errors (Roy et al., 2014) Bayesian: fatigue as metacognitive inference (Stephan et al., 2016)
Operational concerns range from proximal to distal, wherein an organism has a feeling toward an object that, if acted upon, has proximal to eventual metabolic impact	safety (joy, happiness, exhilaration), danger (fear, worry, dread), obstruction (frustration, annoyance, anger), loss (disappointment, sadness, grief), epistemic (curiosity, intrigue, fascination), cooperation (care, love, belonging, trust, empathy), moral (pride, admiration, shame, moral disgust), aesthetic (awe, appreciation, beauty); emotion	Evaluative or Action-Oriented Theories: feelings or emotions are evaluation of readiness for action (Dewey, 1895; Frijda, 1986, 2017; King, 2009; Deonna et al., 2015; Deonna and Teroni, 2012, 2015; Scarantino, 2014, 2015; Simon, 1967; Tooby and Cosmides, 1990, 2008; Clore and Palmer, 2009; Seth, 2013; Oatley and Johnson-Laird, 2014; Bach and Dayan, 2017; Atzil and Gendron, 2017; Hommel et al., 2017; Hommel, 2019; Suri and Gross, 2022; Quadt et al., 2022; Del Giudice, 2021) Basic and Discrete Appraisal Theories: differentiating emotion types by respective evaluations (Kragel and LaBar, 2016; Adolphs, 2017; Ekman and Cordaro, 2011; Lazarus, 2001; Roseman and Smith, 2001; Izard, 2007; Levenson, 2011; Panksepp and Watt, 2011; Scherer et al., 2010) Dimensional Appraisal Theory: infinite combinations of concerns differentiate an infinite typology of emotions (Grandjean and Scherer, 2008; Scherer, 2009; Scherer and Moors, 2019; Lerner and Keltner, 2000, 2001) Constitutive Appraisal Theory: emotions are these evaluated concerns (Clore and Ortony, 2000, 2013; Ortony et al., 1988) Constructionist Theories: emotions are constructed from lower-level ingredients (Damasio, 2003; Russell, 2003; Barrett, 2017; LeDoux, 2012)	how to address proximal to distal needs	Affective Computing, Reinforcement Learning, and Bayesian: models for differentiating between emotions (Gratch and Marsella, 2004; Scherer et al., 2010; Broekens et al., 2013; Lee et al., 2021; Marsella and Gratch, 2009; Marsella et al., 2010; Poria et al., 2017; Bach and Dayan, 2017; Sennesh et al., 2022; Smith et al., 2019).
Global summative adaptive states	trajectory, optimization	Constructionist Theory: summarizing overall organism state (LeDoux, 2012)	how to characterize own adaptive performance across time	See below.
Trajectory the direction of adaptation with regard to comfort zone	positive, negative; mood	Philosophical Theories: increased likelihood of positive/negative occurrences (Price, 2006; Railton, 2017)	how to characterize local direction of environment	Reinforcement Learning: momentum or trajectory of reward and punishment prediction errors (Eldar et al., 2016) Bayesian: mood as hyperpriors (Clark et al., 2018)
Optimization optimal match between the organism and the environment	low, high; wellbeing	Life Satisfaction: global evaluation of one’s life (Schwarz and Strack, 1999; Strack et al., 1991; Diener and Ryan, 2009; Diener, 2009; Diener et al., 2009; LeDoux, 2012; Krueger and Schkade, 2008; Maddux, 2017; Oishi et al., 2020)	how to recognize best adaptive performance across an extensive period of time	Bayesian: maximizing momentary valence—using momentary judgments of adaptiveness to evaluate global optimality in adaptiveness (Smith et al., 2022; Miller et al., 2022) Information Theory: flow as mutual information (Melnikoff et al., 2022)
AFFECTIVE FEATURES momentary information on the adaptive process	valence, arousal	Allostasis: measures of predicted homeostatic need (Cannon, 1929; Cooper, 2008; Sterling and Eyer, 1988; Carpenter, 2004; McEwen and Wingfield, 2003; Sterling, 2012, 2020; Schulkin and Sterling, 2019; Sennesh et al., 2022)	how to characterize momentary status of adaptive process	See below.
Valence metric of evaluation of goodness or badness	positive, negative	Core Affect: valence as ubiquitous across all affective experience (Russell et al., 1989; Russell, 2003; Posner et al., 2005; Kuppens et al., 2013) Philosophical Theory: minimal metacognition in self-assessment of own adaptiveness (Van de Cruys, 2017)	how to characterize momentary suitability for organism’s adaptivity	Reinforcement Learning Implementation: predicted rewards and punishments as ‘happiness’ (Rutledge et al., 2014) Bayesian Implementation: organism’s predictive evaluation of its adaptiveness or preparedness for its environment (Hesp et al., 2021)
Arousal metric of activation of various systems	low, high	Core Affect: arousal as ubiquitous across all affective experience (Russell, 2003; Russell et al., 1989; Posner et al., 2005; Kuppens et al., 2013) Wakeful, Sexual, Autonomic, Physical, and Affective Arousal Theories: activation can occur within different systems at different levels (Duffy, 1957; Zuckerman, 1971; Pribram and McGuinness, 1975; Thayer, 1978; Robbins and Everitt, 1995; Robbins, 1997; Cahill and McGaugh, 1998; Jones, 2003; Eysenck, 2012; Satpute et al., 2019; Satpute et al., 2019; Neiss, 1988; Griffiths, 2013)	how to characterize momentary activation of system	Vigor: effort as an outcome of arousal (Niv et al., 2007)