Extending the Appraisal Tendency Framework to Improve Health and Healthcare

The growth of a robust body of research examining emotions and decision-making (Lerner, Li, Valdesolo, & Kassim, 2015) and an unprecedented societal focus on behavioral prevention of disease suggests that now is the time to leverage emotion science to improve health and health care. Extending the appraisal tendency framework (Lerner & Keltner, 2000), we predict how emotions may interact with situational factors to improve or degrade health-related decisions. We also discuss how policymakers can leverage emotional influences on judgment and decisionmaking to improve health decisions and healthcare. Our review examines four categories of judgments and thought processes of clear relevance to health decisions: risk perception, valuation and reward-seeking, interpersonal attribution, and depth of information processing. By building on prior research and theory, we illustrate ways in which a better understanding of emotion can improve judgments and choices regarding health. EMOTIONS AND HEALTH DECISIONS 3 Emotions and Health Decision-Making: Extending the Appraisal Tendency Framework to Improve Health and Healthcare In an era of unprecedented focus on health policy and behavioral prevention of disease (e.g., Barry & Edgman-Levitan, 2012), understanding the relevance of behavioral science to health is critical. The decisions that people make about their health and the health of others significantly affect the quality, trajectory, and length of human life. Given that many causes of mortality and reduced quality of life, such as heart disease, diabetes, and cancer, can be prevented with behavioral modifications (Fisher et al., 2002; Ford, Zhao, Tsai, & Li, 2011; Khaw et al., 2008; Pearson et al., 2002; Stefanek et al., 2009), recent findings from the field of behavioral science specifying how emotion alters judgments and decision making (Lerner et al., 2015) may offer a key to better outcomes. Specifically, contrary to the popular view that emotions generally contaminate rational decision-making, converging evidence indicates that they actually can improve decisions (e.g., Damasio, 1994; Lerner & Keltner, 2000; 2001; Loewenstein & Lerner, 2003). The field of behavioral economics has begun to make important connections between behavioral science and systems-level interventions. Behavioral economic principles are beginning to be incorporated into health policies and interventions at a variety of levels (see Thaler & Sunstein, 2008). These policies have achieved varied success (see Marteau, Ogilvie, Suhrcke, & Kelly, 2011), but are currently limited to leveraging basic research on economic decision-making, and have not yet capitalized on research demonstrating that emotion influences decisions. In the health domain, research suggests that global affective states – feeling good or bad – contribute to unhealthy behaviors such as smoking (e.g., Addicott, Gray, & Todd, 2009; Perkins et al., 2008), alcohol consumption (e.g., Kelly, Masterman, & Young, 2011; Ostafin & Brooks, 2011), and overeating (e.g., Loxton, Dawe, & Cahill, 2011). Emotions also contribute to healthrelated risk perceptions (Peters, Lipkus, & Diefenbach, 2006) and health decisions made in response to numeric information (Peters et al., 2009). However, less systematic attention has been paid to discrete emotions such as anger, fear, sadness, or disgust (and positive discrete emotions such as gratitude or pride), despite evidence in other domains demonstrating that emotions of the same valence (e.g., anger and fear) can yield dramatically different decisions and behaviors (e.g., Lerner & Keltner, 2000; 2001). Integral emotions – those that are normatively relevant to a decision because they are elicited by a component of the decision or would be influenced by an outcome of the decision – can predict health decisions (see DeSteno, Gross, & Kubzansky, 2013). For example, worry about a health threat may trigger preventive behavior (e.g., Hay, McCaul, & Magnan, 2006). From a functional perspective that views the purpose of emotions as means to motivate fulfillment of goals (see Keltner & Gross, 1999), integral emotions may produce adaptive decisions because they highlight threats, motivate mitigating actions, or signal that a goal has 1 See also http://www.pcori.org/about-us/landing/ 2 We define emotion as a relatively brief affective reaction to a specific person, situation, or sensory stimuli (see Keltner & Lerner, 2010). Unlike moods, which tend to be viewed as less intense positive or negative affective states that are sustained over some period of time, we use the term emotion to refer to discrete categories of feeling state that differ not only in terms of valence but also on a variety of other cognitive appraisal dimensions (e.g., Smith & Ellsworth, 1985). Our use of emotion is closely related to the concept of an “emotion schema” (Izard, 2007). 3 Research has also examined how health behaviors might influence emotions, such as with exercise and positive affective outcomes (e.g., Hall, Ekkekakis, & Petruzzelo, 2002). EMOTIONS AND HEALTH DECISIONS 4 been achieved. However, these are not the only affective influences on judgment and decisionmaking; consumer and decision scientists have also focused on incidental emotions – those elicited by a person, situation, or stimuli not normatively relevant to the decision – can also influence unrelated decisions (Loewenstein & Lerner, 2003; Loewenstein, Weber, Hess, & Welch, 2001). For example, sadness elicited by a prior event has been found to influence eating behavior (Garg & Lerner, in press). The influence of incidental emotions can linger, even when decision makers face substantial incentives to avoid bias (e.g., Lerner, Small, & Loewenstein, 2004) and after the emotion experience itself has ceased (Andrade & Ariely, 2009; Schwarz & Clore, 1983). Sometimes the influence of incidental emotions may be overwhelmed by integral emotions– for example, a patient’s amusement over a film may dissipate when she receives a disease diagnosis. However, many complex emotions may contribute to affective experiences at any given time (see Wilson & Gilbert, 2003, for a discussion), and incidental emotions may be equal contributors even when integral emotion is powerful, particularly when the incidental emotion is felt intensely or is the result of a very personally relevant event. Thus, the same cancer patient could feel fear at a diagnosis, but not forget – or stop feeling – the anger she feels over a previous argument with her spouse. Objectively, this incidental anger should not factor into a treatment decision, as it is not normatively relevant to the decision (e.g., Han, Keltner, & Lerner, 2007). However, that anger is still meaningful and salient, and may carry over to influence her subsequent cancer treatment decisions. Thus, it is imperative to consider the influence of both incidental and integral emotions on health judgment and decision-making. Health-related interventions could capitalize on this basic knowledge of the role of emotion in decision-making. Currently, behavioral economics interventions for population-level health behaviors and decisions tend to take a one-size-fits-all approach. For example, some countries have successfully leveraged basic knowledge about human decision-making and defaults to improve organ donation rates by creating conventions where choices to donate are “opt-out” rather than “opt-in” (Johnson & Goldstein, 2003; van Dalen & Henkens, 2014). This suggests that leveraging defaults may be a promising direction for other behavioral economics health interventions. However, emerging evidence cautions against treating defaults as a panacea: in one instance, an opt-out colorectal cancer screening intervention actually decreased screening rates (Narula, Ramprasad, Ruggs, & Hebl, 2013). Contextual factors such as emotion may explain the varying success or failure of defaults and other behavioral economics interventions. As it turns out, individuals lean more toward the default choice when a decision is emotionally laden (Luce, 1998), as is presumably the case with organ-donation decisions. Moreover, emotions such as anger may reduce (or reverse) reliance on defaults (Garg, Inman, & Mittal, 2005), suggesting the possibility that implementing a default that angers individuals – such as imposing a default option on a behavior that people are extremely resistant to – may backfire. Other examples of behavioral economics interventions that have failed or induced unhealthy behaviors (e.g., Cherney, 2011; Wansink & Chandon, 2006) underscore the importance of understanding contextual factors such as emotion that could predispose success or failure. Connecting Emotion Research with Health Decisions The Appraisal Tendency Framework (ATF; Han et al., 2007; Lerner & Keltner, 2000; 2001; Lerner & Tiedens, 2006) provides a useful framework for clarifying and predicting how specific, discrete emotions systematically improve or degrade health-related decisions and interventions. The ATF can identify (1) individual differences in the tendency to respond to EMOTIONS AND HEALTH DECISIONS 5 situations with certain discrete emotions (Ambady & Gray, 2002; Lerner & Keltner, 2001) and (2) certain health situations that routinely evoke a particular discrete emotion (such as cancer and fear; e.g., Holland, 2003). We note that the influence of emotion on particular decision-making tendencies depends on the properties of a decision (Lerner & Tiedens, 2006). For example, the fact that anger increases risk taking may lead to increased benefits when the option associated with the most likely benefit is also uncertain, ambiguous, or risky (Ferrer, Maclay, Rim, Litvak, & Lerner, in preparation), as is the case with some treatments for cancer or other diseases. As such, emotions can facilitate or hinder decision-making – or augment or degrade intervention efforts –depending on the circumstances (Reyna, Nelson, Han, & Pignone, in press). Thus, rather than predicting that a specific emotion is always beneficial or deleterious, the ATF may pinpoint how specific emotions interact with certain types of health decisions, thereby shedding light on decisions that would benefit or be hindered by particular emotions. Our review focuses largely on research on incidental emotions, because such research involves highly controlled paradigms and experimental inductions, allowing us to draw causal conclusions about the general influence of discrete emotions on judgment and decision-making patterns. Notably, some studies have targeted discrete integral emotions (e.g., fear or worry) in a health context, but typically have not isolated the influence of such emotions on subsequent judgment and decision-making. Rather, these inductions often occur in the context of health behavior change interventions that are designed to intervene on many other constructs and processes (e.g., Portnoy, Ferrer, Bergman, & Klein, 2014; Witte & Allen, 2000). When inductions take this inclusive approach, it is not possible to infer mechanism (Suls, Luger, & Martin, 2010). Thus, such studies cannot fully identify systematic ways that particular emotions can influence patterns of health-related decision-making. For this reason, these studies are beyond the scope of this review. Because theory (Han et al., 2007; Keltner & Gross, 1999; Lerner, Han, & Keltner, 2007) and research (Isen & Erez, 1007; Lerner, Gonzalez, Small, & Fischhoff, 2003) suggest that that the pattern of judgment and decision-making arising from an emotion will be similar regardless of whether it is integral or incidental, studies of incidental emotion allow us to infer patterns of the general influence of discrete emotions, both incidental and integral, on health-related decision-making. In this chapter, we consider the effects of emotion on four general categories of judgments and thought processes relevant to health decisions: risk perception, valuation and reward-seeking, interpersonal attribution, and depth of information processing. We discuss ways in which emotions may improve or degrade health decisions through their influence on these judgments and thought processes in two health decision domains: Choices about health promotion and prevention behaviors (e.g., choices about food, tobacco, physical activity) and medical decisions (e.g., decisions about preventive care and treatment). We then discuss broad policy implications of these areas. We define decision-making broadly, extending beyond single-event decisions (e.g., cancer screening) to include decisions and choices that contribute to behavioral patterns or maintenance (e.g., decisions to quit smoking or food choices as contributors to a pattern of adhering to smoking cessation or weight loss programs), given that similar underlying psychological, affective, and decisional processes contribute to a diverse array of decisions (e.g., Reyna, 2008). Although maintenance choices are made over time and can require frequent 4 Although health researchers have advocated for small –scale experiments that isolate and control constructs in isolation (Suls et al., 2010), in practice this has not occurred with emotion inductions. EMOTIONS AND HEALTH DECISIONS 6 decision-making (Rothman et al., 2004), behavioral patterns or maintenance initiated by a single decision (e.g., whether to enter a smoking cessation program) can be influenced by emotion. Moreover, frequently experienced emotions (e.g., those repeatedly triggered by a volatile relationship, a frustrating job, or a satisfying friendship) can systematically influence repeated decisions that contribute to patterns of behavioral maintenance. The Appraisal-Tendency Framework The Appraisal-Tendency Framework provides a useful theoretical foundation for understanding how emotions influence health-related decisions. The ATF assumes that specific emotions give rise to corresponding cognitive and motivational processes that are related to the target of the emotion (i.e., the situation, person, or other stimulus that elicited the emotion), which account for the effects of each emotion upon judgment and decision making. In contrast to theories that predict how broad mood states (positive or negative) may influence judgment and decision making (e.g., Bower, 1991; Forgas, 2003; Isen, 1993), the ATF offers specific predictions for how discrete emotions will influence judgment and decision making (See Tables 1 and 2). Emotion theorists have argued that a range of cognitive appraisal dimensions, or categorical dimensions characterizing cognitive tendencies associated with emotion, usefully differentiate emotional experience. In one empirical examination of appraisal dimensions, Smith and Ellsworth (1985) identified six dimensions that categorize patterns of thinking associated with different emotions: pleasantness/valence (whether the emotion is pleasant); certainty (whether the emotion was elicited by a predictable stimulus); personal control (whether emotion was elicited by something under one’s personal control); other or situational responsibility (whether the emotion was elicited by a stimulus controlled by another person or a situation); attentional activity (whether the emotion was elicited by a stimulus that demands attention); and anticipated effort (the amount of effort an individual anticipates will be necessary to deal with the emotion or its elicitor). According to the ATF, patterns of cognitive appraisals along these dimensions provide a basis for comparing and contrasting discrete emotions. For example, certainty and control are the central dimensions that separate anger from fear. Anger is associated with appraisals of certainty about an event and individual control for negative events. Fear, by contrast, is associated with appraisals of uncertainty about what happened and situational control for negative events. Despite its positive valence, happiness, like anger, is associated with an elevated sense of certainty and individual control (Averill, 1983; Smith & Ellsworth, 1985; Weiner, 1986). Therefore, happiness, at least in one respect, resembles anger more so than fear. Each emotion is also accompanied by a core appraisal theme (Lazarus, 1991), which is a mental schema associated with the emotion that summarize the specific harms or benefits associated with the target or elicitor of the emotion. Emotion-specific core appraisal themes affect the likelihood of specific courses of action (Lazarus, 1991; Frijda, 1986; Roseman, Wiest, & Swartz, 1994; Scherer, 1999, 2001). For example, sadness is accompanied by a core appraisal theme or mental schema of loss; anger involves a core appraisal theme of being slighted or demeaned (Lazarus, 1991). The ATF proposes that these appraisal themes systematically trigger a predisposition toward specific action tendencies, behavioral patterns aimed at overcoming obstacles or meeting goals made salient by the emotion and its core appraisal theme (Frijda, 1986). These action tendencies are triggered when the appraisal dimensions associated with an emotion are also relevant to a EMOTIONS AND HEALTH DECISIONS 7 particular judgment or decision. For example, fear is associated with high uncertainty and reflects core appraisal themes of being threatened; thus, it is relevant to judgments about risk and triggers risk-avoidant behavior (see also Rivers, Reyna, & Mills, 2008). Sadness, by contrast, is characterized by appraisals of experiencing irrevocable loss (Lazarus, 1991) and thus accompanies the action tendency to change one’s circumstances, perhaps by seeking rewards (Lerner et al, 2004). In sum, the ATF predicts that each emotion has motivational properties that fuel carryover to subsequent judgments and decisions. The form of that carryover is termed appraisal tendencies – where the appraisal dimension and appraisal theme are together activated by the properties of a situation to shape behavioral action tendencies that predispose certain judgments, decisions, and actions. Although tailored to help the individual respond to the event that evoked an emotion, appraisal tendencies persist beyond the eliciting situation and affect both the content and depth of thought. Broadly speaking, appraisal-tendency influences on judgment and decision making can be divided into two categories: content effects and depth-of-processing effects.