1   Beyond Willpower: Strategic Solutions for Reducing Self-Defeating Behavior Angela Duckworth, University of Pennsylvania David Laibson, Harvard University Katherine L. Milkman, The Wharton School of the University of Pennsylvania 2   Beyond Willpower: Strategic Solutions for Reducing Self-Defeating Behavior Self-defeating behavior is a singularly timeless—and timely—topic in both psychology and economics. Self-control failures contribute to an enormous number of pressing policy challenges, from improving educational achievement (Duckworth & Seligman, 2017) and increasing retirement savings (Benartzi & Thaler, 2013) to tackling the obesity epidemic (Van Epps et al., 2016). Scholarly attention to self-control has grown dramatically over the last two decades, as shown in Figure 1, which depicts the percentage of articles in Psychological Science about self-control from 1995 to 2016. But the literature on this topic stretches back thousands of years (Proverbs 25:28; Plato, 370 BCE; James, 1899; Freud, 1916; Smith, 1759; Schelling, 1978; Thaler & Shefrin, 1981). [INSERT FIGURE 1] Why has the topic of self-control continued to play an important role in the analysis of human psychology? In part, because of the stubbornness of self-defeating behavior: often, unwise personal choices persist even after we recognize the error in our ways. From forgoing dessert to exercising regularly to saving for retirement, many of us feel as if we’re in a perennial battle with ourselves. What’s more, most people counter-factually predict that they will overcome this battle in the near future (e.g., Augenblick & Rabin, 2017), even as they recognize that other people’s self-control problems will persist (Fedyk 2017; Pronin, Olivola, & Kennedy, 2007). Finally, temptations—rewards that have short-term gratification but distract or impede us from long-term goals—are harder to avoid than ever, thanks to convenience stores, social media, designer drugs (e.g., “street” opioids), and other vices that did not exist for our forebearers. What precedes self-defeating behavior is a self-control dilemma; an internal goal conflict in which an individual recognizes one option (especially when viewed at a distance) as more 3   valued than another more immediately compelling option. When an individual adjudicates this goal conflict by pursuing the more enduringly valued option, he or she has successfully exercised self-control. On the other hand, when an individual pursues the more tempting option, he or she has failed to exercise self-control and instead engages in self-defeating behavior. The most straightforward approach to resolving a self-control dilemma is to suppress the impulse to gratify immediate urges and/or to elevate the more enduringly valued goal. The vernacular term “willpower” describes the capacity to directly override one action tendency and privilege another. The ability to intentionally modulate such responses is more advanced in our species than in any other. Accordingly, this feature of our behavioral repertoire relies on the most recently evolved areas of the human brain (Cohen, 2005). Exhortations to use willpower thus constitute the most obvious—albeit least sophisticated—public policy approach to reducing self-defeating behavior. However, recent advances in the science of self-control have illuminated more efficient policy solutions. In particular, we now know a great deal about the situational and cognitive precursors to the urge for instant gratification. Well before willpower is required, for example, it is possible to strategically avoid temptations altogether or to frame choices in ways that make temptations less alluring (Duckworth, Gendler & Gross, 2016). We have three objectives in this article. First, we provide a theoretical taxonomy for understanding diverse strategic approaches to reducing self-defeating behavior. We highlight one dimension that distinguishes situational vs. cognitive strategies, acknowledging that most interventions are a hybrid of the two. We also highlight a second dimension that distinguishes strategies deployed by the individual (i.e., self-initiated self-control strategies) and those deployed by third-parties (e.g., nudges initiated by policymakers, employers, etc.). 4   Second, we aim to synthesize policy-relevant empirical research in both psychology and economics with the goal of identifying varied approaches to decreasing self-defeating behavior. Finally, we conclude with a discussion of the relative advantages and disadvantages of each approach. We argue that strategic interventions designed to reduce self-defeating behaviors are more desirable than simply exhorting individuals to rely on sheer willpower. Strategic Interventions to Reduce Self-Defeating Behavior From early childhood, all individuals have some capacity to directly suppress one urge while directly augmenting a goal-congruent rival (Duckworth & Steinberg, 2015). Unfortunately, willpower—directly modulating responses in accordance with enduringly valued goals —can feel exhausting (Hagger, Wood, Stiff, Chatzisarantis, 2010; Kurzban, Duckworth, Kable, & Myers, 2013), and failures to do so are common. Likewise, public policies that essentially exhort people to resist immediate gratification tend to be ineffective. Consider, for example, the “just say no” campaign, launched from First Lady Nancy Reagan’s three-word response to a schoolgirl who asked what she should do if someone offered her drugs. The subsequent Drug Abuse Resistance Education (DARE) program implemented by a majority of U.S. school districts in the 1980s has been shown in meta-analyses to have no measurable impact on youth alcohol, drug, and tobacco use (West & O’Neal, 2004). We propose a classification of four types of interventions that do not rely primarily on willpower and instead strategically help people reduce self-defeating behavior. Our classification distinguishes between approaches that modify a person’s situation versus her cognitions, depending on whether the target of the intervention is the objective, physical situation or, in contrast, the subjective mental representation of the environment. We further differentiate between strategies that are self-deployed versus other-deployed. In the former, individuals take 5   deliberate action to improve their decisions; in the latter, individuals may be oblivious to the actions that other parties take on their behalf. A similar distinction—between boosts and nudges—has been made by Hertwig and Grune-Yanoff (2017): Whereas self-deployed strategies are executed by individuals with intention, other-deployed strategies do not require the explicit cooperation of individuals and, indeed, are often executed without individual’s conscious awareness or consent. Figure 2 highlights four illustrative examples that can be categorized within this framework. Note that most examples don’t fit easily into one of these four boxes, but are instead better understood as hybrid cases.1 There is arguably a continuum of self- versus other- deployment. And, arguably, all situational changes influence decision making via cognitive mechanisms (Duckworth, Gendler, & Gross, 2016). [INSERT FIGURE 2] Let’s begin with self-deployed cognitive approaches, which are often initiated spontaneously and sometimes gently suggested by others. These strategies shift the way an individual subjectively interprets a given situation in order to make long-term choices more appealing and actionable, and short-term temptations less so. Such self-deployed cognitive interventions do not target the objective situation itself, but rather the way a decision maker subjectively experiences or interprets her situation. Examples include goal setting and planning, actively distracting oneself from temptations, and thinking about long-term goals in ways that enhance their salient benefits. 1The distinction between self-deployed and other-deployed interventions reflects how they are typically implemented: by the individual herself or by other parties on her behalf. In many cases, a strategy categorized as self-deployed could be deployed by others, and vice versa. 6   Moving to other-deployed cognitive interventions, we define these as strategies an outsider, such as a policymaker or employer, could deploy to shift the way an individual interprets her situation. Examples include establishing a social norm of good behavior (e.g., the norm that alcohol is not ordered at a business lunch) and prompting people to make decisions requiring self-control at times when their willingness to exert willpower is known to be high (e.g., when they are well-rested and nutritionally satiated). In another approach to favorably resolving a self-control dilemma (second row of Figure 2), a decision maker voluntarily changes her environment to reduce the likelihood of self- defeating behavior by creating incentives, obstructions, and affordances that favor long-term goals over short-term temptations. Examples of such interventions include people purchasing small (rather than large) packages of junk food at a premium to prevent their own future overindulgence, installing apps on their computers to limit the time they spend on social media, and deleting a video game from an iPad because they find that they can’t control their own usage. We will refer to this class of approach as self-deployed, situational interventions. Finally, we move across the bottom row of Figure 2 to other-deployed situational interventions. In this category, a third party, such as a policymaker, employer, or family member, takes actions that motivate less self-defeating behavior from the individual. Such interventions create situations whose incentives and affordances reduce the burden on individual willpower. Examples include imposing taxes on cigarettes, outlawing heroin, and auto-enrolling employees into savings plans. Some of the interventions in these four categories have been tested in the field, while others have been tested only in the laboratory. This distinction is important, as interventions that are effective in highly controlled, sometimes contrived laboratory settings may not be effective 7   in uncontrolled, policy-relevant environments. In this article, as we synthesize past research on interventions that decrease self-defeating behavior, we primarily describe interventions that have been studied in field settings, but as appropriate, we also describe promising interventions that have primarily been tested in the laboratory. Empirical Research on Interventions that Reduce Self-Defeating Behaviors In this section, we review research about different interventions that have proven effective for reducing self-defeating behaviors. We relate these interventions back to Figure 2 but note that many of them fall into multiple categories. Cognitive interventions. Cognitive interventions that are typically self-deployed. As described previously, self-deployed cognitive interventions allow an individual to change the way he or she subjectively interprets a given situation in order to make long-term choices more appealing and actionable, and short-term temptations less so. We begin with the purest examples of self-deployed cognitive interventions, then move to increasingly hybrid approaches that can be self- or other-deployed. Goal setting and planning. Setting specific (versus general) and feasible (versus impossible) goals helps people make more self-controlled choices (Locke & Latham, 2002). Mental contrasting is one technique that facilitates this process. In sequence, an individual identifies a goal and then contrasts the positive outcome of attaining that goal with the obstacle that currently stands in the way (Oettingen, 2012). Particular emphasis is given to obstacles within the individual’s control; however, one possible outcome of mental contrasting is that the individual recognizes that external obstacles make the goal infeasible—in which case, the individual appropriately discards the goal. In one study, elementary school students taught 8   mental contrasting outperformed classmates randomly assigned to a placebo control condition on a quiz a week later (Gollwitzer, Oettingen, Kirby, Duckworth, & Mayer, 2011). Other studies have shown the benefits of this technique for increasing physical activity among overweight, middle-aged men (Sheeran, Harris, Vaughan, Oettingen, & Gollwitzer, 2013) and improving healthy eating and physical activity among college students (Johannessen, Oettingen, & Mayer, 2012). Goal commitment is necessary but not sufficient for goal-directed action. After setting a goal that cannot be immediately accomplished, follow-through is facilitated by making specific plans regarding when, where, and how a person will take action (i.e., implementation intentions, Leventhal, Singer, & Jones 1965; Gollwitzer, 1999). Advance planning reduces procrastination as well as forgetfulness, and also supports the enactment of difficult actions. Thinking through the when, where, and how of an intention creates a commitment that is memorable, can be automatically enacted, and is psychologically costly to break (Gollwitzer & Sheeran, 2006; Rogers, Milkman, John, & Norton, 2015). Prompting people to make plans has been shown to increase self-controlled decisions, such as the decision to exercise (Arbour & Martin Ginis, 2009; Milne, Orbell, & Sheeran, 2002), to meet deadlines (Bagozzi, Dholakia, & Basuroy, 2003), to get a flu shot (Milkman, Beshears, Choi, Laibson, & Madrian, 2011), to get a colonoscopy (Milkman, Beshears, Choi, Laibson, & Madrian, 2013). A meta-analysis of close to 100 studies found a medium-sized effect of making plans on goal attainment across age groups, life domains, and types of obstacles (Gollwitzer & Sheeran, 2006). Recently, the goal-setting technique of mental contrasting has been paired with the planning technique of implementation intentions and shown to support self-control better than either component alone (Adriaanse et al., 2010; Kirk, Oettingen, & Gollwitzer, 2013). A series of 9   field studies has demonstrated the particularly powerful benefits of this sequence. For example, fifth graders taught to set goals and make plans earned higher report card grades, came to school on time more often, and were rated by their teachers as superior in classroom conduct (Duckworth, Kirby, Gollwitzer, & Oettingen, 2013). Notably, these benefits were measured over one marking period but diminished to non-significance the following marking period, suggesting the need for additional support and reinforcement. This technique has also been shown to increase physical activity for several months post-intervention (Stadler, Oettingen, & Gollwitzer, 2009; Christiansen, Oettingen, Dahme, & Klinger, 2010). Self-monitoring. Most self-controlled behaviors must be enacted consistently over time to yield significant benefits (Rachlin, 2004). For example, resisting dessert, studying, going for a run, or saving a few dollars for retirement all pay dividends for long-term well-being only if repeated again and again. This presents a challenge, as attention to goals typically lapses over extended periods of time. New Year’s resolutions, for example, are at the top of one’s mind in early January but quickly lose their urgency. Moreover, we may not be fully aware of our snacking, web surfing, couch sitting, impulse shopping, and other bad habits that undermine long-term goals. Self-monitoring is the intentional and consistent observation of one’s own behavior (Kanfer, 1970). The benefits of self-monitoring are especially well-documented in the domain of weight loss. For example, one early study found that dieters who consistently monitored their food intake lost more weight than those who did not (Baker & Kirshcnbaum, 1993). A more recent systematic review confirmed a consistent relationship between self-monitoring and weight loss, though this review noted that correlational studies are much more common than randomized clinical trials and that studies using objective outcome measures are needed (Burke Wang, & 10   Sevick, 2011). With that caveat in mind, we note that self-monitoring has been shown to help alcoholics drink less (Hester, 1995) and to help students improve academically (Schmitz & Perels, 2011; Zimmerman & Paulson, 1995). Psychological distancing. Several decades of laboratory research have demonstrated the capacity of children and adults to evaluate their situation in more versus less psychologically distanced terms (Kross & Ayduk, 2017; Trope & Liberman, 2010). Temptations are most potent when there is no psychological distance separating us from their alluring features. In other words, we are most likely to succumb to indulgences we later regret when they are available in the present moment. There are four dimensions of distance: space, time, social, and hypotheticality. In other words, the more a temptation is not here, not now, not for me, or not real, the less tempting it is. A large body of laboratory studies has found that psychological distance tends to facilitate resistance to temptation and, instead, to promote the pursuit of more valued goals, whose benefits are typically psychologically distant (Fujita & Carnevale, 2012). For instance, when preschoolers are encouraged to delay consumption of a tempting marshmallow in order to earn a second treat, children encouraged to think of the marshmallows as “fluffy white clouds” are able to wait more than twice as long as children encouraged to think of how “sticky and sweet” they are (Mischel & Rodriguez, 1993). Likewise, both children and adults who process emotionally upsetting events in the third person (using their name or a third-person pronoun) rather than the first person (“I”) demonstrate superior emotion regulation (Kross et al., 2014, 2017; White, Kross, & Duckworth, 2017; Streamer et al., 2016; Nook, Schleider, & Somerville, 2016). Psychological distancing has also been posited to be one of the active ingredients