THE ACTION-HESITANCY MODEL 1 “He Who Hesitates Is Lost, But Only for the Moment” An Action-Hesitancy Model of Decision Making Zeljka Buturovic1 1 Phone: 917-514-4891 Email: zeljka@paradox.columbia.edu THE ACTION-HESITANCY MODEL 2 Abstract Here I propose an approach to understanding decision making, an approach I dub the actionhesitancy model. People strive to make obvious decisions and act without hesitation. When obviousness is lacking, they hesitate. They seek new information and insight, to reappraise their situation and to come to a course of action of action that seems obvious. Much of current decision-making theory sees the essence of decision making in weighing costs against benefits, with a high degree of precision. In contrast, the action-hesitancy model emphasizes human resourcefulness in sifting and gathering knowledge, separating relevant from irrelevant, and engaging in self-correction. THE ACTION-HESITANCY MODEL 3 Decision-making is primarily a psychological, not an economic phenomenon. For one, humans make decisions about many things, and decisions on how they will spend their money are just a fraction of them. Though these other, apparently non-economic decisions, such as a choice or a career or a spouse, often do have economic consequences that make them interesting to economists, they are rarely seen as economic decisions from the point of view of the person making them or other observers. In addition, the primary interest of psychologists has traditionally been behavior an experience of an individual (Wegner & Gilbert, 2000), perhaps a small group, and, only exceptionally, behavior of groups of tens of millions of individuals that are often of interest to economists. Yet, a model of human decision-making developed for aggregate purposes, such as those of economics, can be misleading when it comes to understanding individual behavior. For example, a conception of atoms as solid balls is very useful for the aggregate purpose of statistical mechanics but completely misguided when it comes to the study of atoms themselves. Similarly, a crude, mathematical approximation of a decision-maker that is very convenient for aggregation might turn out to be profoundly mistaken about the very nature of decision processes at the individual level. For economists, the choice under uncertainty is modeled by the expected utility theory (Mas-Collel et al, 1995). Briefly, expected utility theory describes a person making a decision as having a preference relation over a set of lotteries, where each lottery is a set of probabilities over possible outcomes. These lotteries are different in a sense that each of them “favors” some outcomes at the expense of others. If the preference relation over this set of lotteries satisfies certain assumptions, then these preferences can be represented by a utility function of the expected utility form. This means that each of the outcomes (for various alternative courses of THE ACTION-HESITANCY MODEL 4 action) can be assigned a number (utility) such that numerical ordering of the expected utility of each alternative (lottery) perfectly mimics the person’s preferences over lotteries. Thus, the most preferred alternative (lottery) is the one with the highest expected utility. Conceptually, utilities and probabilities are supposed to be derived from the observed behavior – that is, one’s preferences are assumed to be revealed through his actual choices. However, it has been a common practice, especially in psychology, to treat those utilities and probabilities as introspectively available, and argue that a rational agent should follow a course of action with the highest expected utility. Almost everything that has been done in psychological research on decision-making has its roots in this theory. The classic anthology “Choices, Values and Frames” (Kahneman & Tversky, 2000), for example, has 42 papers that are largely critical of the view that expected utility theory adequately describes human decision-makers. Yet, none of these papers discusses choices or values conceived in any other sense but that of choices with clear financial implications. In addition, all papers are couched within basic conceptual framework of expected utility theory and its conceptual heirs, such as Prospect Theory (Tversky & Kahneman, 2000) and novel versions of regret theory (Humphrey, 2004). The lay person interested to learn basics of human decision-making psychology is likely to be quite surprised by what sort of problems decision-making experts actually study. This research is often perceived as data-driven, but that impression is - at best - justified in comparison with the sort of research that the same theory instigated within mainstream economics. But when this research is compared to what has been done in mainstream psychology, we see that it is, to the contrary, very much theory-driven. The massive empirical evidence that has accumulated as a result of this research program was to a large extent directed THE ACTION-HESITANCY MODEL 5 at falsifying a very specific theory, namely, the expected utility theory, and not at collecting robust data on actual decision-making, on which a psychologically plausible theory will then be built. More significantly, the theory in question was embraced by economists, not by psychologists, and, representation theorem which made its conceptual foundations was discovered by a mathematician (Ramsey, 1926). At the time when the decision-theory within psychology was developed, many psychologists were behaviorists. Whatever weaknesses of his research program, Skinner cannot be accused of assuming that humans are perfectly rational agents (Schultz, 2006) – quite the opposite. And while expected utility theory is still defended by economists though by saying that Kahneman at al. misunderstand it (Gul & Pesendorfer, 2005), it is quite difficult to find a single psychologist who ever seriously defended it and there are probably not many who really know what it is or whether its exposition by other psychologists is adequate. Thus, criticisms of expected utility theory were, to a large extent, conducted within scientific community that was ill-prepared to assess any successes that that theory had in addressing problems relevant to economists and consequently offer it proper defense. This is not to say that concerns about the conceptual underpinnings of decision-making models in psychology are new – just that previous ones don’t go far enough. The appropriateness of couching the study of decision-making within the consequentialist framework of the expected utility theory has been questioned for a number of years (Loewenstein et al, 2001). Starting with the early arguments about limitations of working memory and attention and pervasive use of heuristics and more recent arguments about the role of feelings and automatic processes in decision making (e.g. Wilson et al, 1993; Wilson & Schooler, 1991; Damasio, 1994), empirical adequacy of consequentialist frameworks has been thoroughly challenged, and, as a result, semi THE ACTION-HESITANCY MODEL 6 non-consequentialist models have started to emerge. For example, Loewenstein et al. have proposed a risk-as-feelings hypothesis – a non-consequentialist model in which behavior partly results from goal-oriented considerations but partly from feelings unmediated by cognitive considerations. Kahneman (2003) proposes a two-system view of human cognition, in which fast, parallel, automatic processes are supplemented by slow, serial, effortful processes. This general framework then gets supplemented by a clearly consequentialist model of prospect theory to yield explanation for some of the decision-making phenomena. Higgins’s regulatory focus theory (1997) is an attempt to step back from consequentialist thinking by assuming that, apart from goals themselves, the way of approaching goals (promotion vs. prevention) and its fit with the kinds of goals pursued needs to be taken into consideration for a number of phenomena to be explained. What is common to these attempts is that they have retained the basic consequentialist framework imported from economics and then tried to attach to it a non-consequentialist supplement. This did not and cannot come without a cost. For example, the cost of adding a direct route from feelings to outcome of a decision in the risk-and-feelings hypothesis (henceforth RAF) is abandonment of the notion of a decision altogether. That notion is now replaced by ‘behavior’2. In some sense, this had to happen. When, in what began as a decisionmaking model, the route opens for situations in which no consideration of options is available, it seems that the fragile notion of a decision has to be discarded since its essence is consideration of options. However, since the consequentialist kind of thinking and resulting models emerged precisely because of the way situations normally described as ‘decisions’ are conceived, once the concept of decision is removed from the model, the appeal of all other major decision-making The replacement is intentional and significant. In the words of authors: “…the term decision… is deliberately replaced with behavior… [which] reflects the observation that many types of emotion-driven risk-related behaviors…. Do not seem to reflect decisions in the sense that the term is usually used. 2 THE ACTION-HESITANCY MODEL 7 concepts (outcomes, probabilities) for explaining human behavior is no longer obvious. Put simply, once the model starts explaining all behavior rather than just what is meant to be decision-making it is no longer clear that personal probabilities (as opposed to say, expectancies, scripts, motives, skills etc) should be an interesting concept to work with. In addition, the novelty of questioning sufficiency of consequentialism which gives RAF competitive advantage compared to other models of decision-making (had RAF remained a decision-making model) loses its importance. For, while consequentialism is pervasive among decision-making models, it is not necessarily pervasive among models of human behavior in general. For example, connectionist models of human behavior such as CAPS (Mischel, 2004) go beyond consequentialist frameworks. The case of the two system (henceforth TS) model (Kahneman, 2003) illustrates on its own problems of integrating non-consequentialist considerations into a consequentialist model. Prospect Theory is a clearly consequentialist model – in terms of conceptual understanding of decision-making it is not very different from the expected utility theory. Because this prevents it from adequately addressing non-consequentialist aspects of decision-making, it gets supplemented by another model – a model about two processes, the TS model. That model, however, is not a model of decision-making. It is a model of human mind, regardless of whether it is making decisions or not. Both models are then used as explanations for decision-making phenomena but their mutual relationship is not worked out theoretically and it is quite unclear how it could be. This problem is not accidental or limited to the TS model. It reflects the underlying difficulty that that all these attempts have which comes from the fact that, despite original additions, they have retained the basic consequentialist framework. The idea that making a THE ACTION-HESITANCY MODEL 8 decision is, at the core, evaluation of outcomes of alternative courses of action appears to be so central to the very notion of a decision and so powerful in explaining broad patterns of human behavior that it is understandably difficult for any model to part from it. But its very power is, I believe, what makes supplementing it so difficult. Consequentialist models essentially do not need theoretical supplements. They are largely silent about the psychological processes involved in decision-making because they do not need them to explain broad patterns of human behavior – in fact, they were never meant to describe those processes. The point was, rather, to separate pattern of choices (that satisfies the assumptions of the expected utility model) into two independent factors: subjective probabilities and perceived utilities. Fitting additional cognitive processes into this basic model makes the model look either unnecessarily complex or as a model of something other than decision-making. Yet, looking from a broader psychological perspective, consequentialism must be at least partly inadequate as a model of decision-making. One area in which that was never really in doubt is the area of moral judgment and decision-making. Moral decisions are clearly decisions in a sense that they involve considerations of options, but, as a rule, they involve more tension between utility of the outcome and permissiveness of the action itself. Still, any account of decision-making should be able to explain how they occur. Expected utility model is quite capable of explaining them, as notions of utility and preference are able of addressing all kinds of decision-making situations, including those that are perceived as dealing with morals. This is reflected in the fact that one of the most prominent traditions in moral philosophy is utilitarianism, a less rigorous exposition of the same idea. However, perhaps because of its prima facie psychological implausibility, the study of moral psychology did not start with utilitarianism at its foundation (rather, it began with another species of rationalism – deontology). THE ACTION-HESITANCY MODEL 9 Bypassing utilitarianism turned out to be a good choice for moral psychology. There exists an ample demonstration of the existence of taboo-tradeoffs (Turiel, 1983; Tetlock, 2003), that is, decisions where people refuse to engage in cost benefit analysis (for example, how much would they ask to sell their right to vote). In minds of people cost-benefit analyses which might be interpreted as eminently rational from an economic perspective, are deemed immoral and impermissible. Furthermore, studies have shown that moral behavior is greatly influenced by contextual information. The study of “trolley problems” - a family of moral dilemmas in which the agents are forced to inflict harm in order to obtain a greater good (for example, direct a train towards killing one innocent person in order to save five innocent people tied to train-tracks) showed that subtle changes in context with no clear cost-benefit implications (such as pushing a person to the tracks as opposed to merely flipping a switch) have great consequences to judgment of permissibility of actions (Greene, 2007). Finally, a number of studies have shown that justifications of judgments, which often have a form of cost-benefit analyses, often follow rather than precede judgments themselves – a fact demonstrated both in moral judgments (e.g. Haidt, 2001) as well as decision-making proper (Nisbett & Wilson, 1977). As a result, attempts have been made (most significantly by Haidt, 2001) to provide models of moral judgment that would soften the overly rationalistic, deontological view of moral decision-making. Haidt’s (2001) model, in particular, postulates that moral judgments mainly result from moral intuitions and, to some extent, posthoc reasoning, both of which can influence and be influenced by intuition and reasoning in a surrounding social environment. However, this particular model makes no attempt to relate itself to judgments of non-moral issues, perhaps because consequentialism is perceived to be indispensable in modeling the latter. Thus, inadequacy of consequentialism in explaining moral choices and THE ACTION-HESITANCY MODEL 10 judgments has cemented a separation of theoretical accounts of decision-making that is explicitly related to morals from the “mainstream” decision-making. Empirical discoveries within both traditional decision-making research and research in moral psychology have increasingly put pressure on the conceptual foundations of the expected utility theory, with many attempts to graft these discoveries on this theory. In addition, psychological research unrelated to decision-making has, over many years, uncovered robust facts about human mind and behavior. Among these facts are the following: Subjective construal The idea that human mind interprets the environment is perhaps the least controversial thesis in all of psychology (Griffin & Ross, 1991). From the infinite multitude of physical stimuli, human mind extracts a tiny fraction and organizes it into a meaningful whole. This idea is fundamental to the Gestalt approach to psychology (Koffka, 1935). In a famous study, Asch’s (1946) showed how a set of terms changes meaning when additional term (such as warm vs. cold) is added to it. It is further supported by the long list of effects produced by mere labeling on behaviors such as opinions in politics (Gilovich, 1981), and amount of money offered in an ultimatum game (Camerer, 2003). The subjective construal by definition depends on the cognitive-affective state of the agent, which might or might not be phenomenologically accessible. This has been demonstrated by priming effects (e.g. Nisbett and Ross, 1980; see Higgins & Bargh, 1987; Srull & Wyer, 1986 for reviews). From this follows that, whenever an agent tries to make a decision, he see himself as being in a certain kind of situation. One’s history, mood (Bodenhausen, Gabriel, & Lineberger, 2000), knowledge or perceived social role are thus all likely to be factors in his decisions. There is no a priori reason to assume that this construal always involves subjective probabilities and THE ACTION-HESITANCY MODEL 11 utilities as psychological upgrades of the economic model imply. In principle it might be argued that all types of subjective construal can be reinterpreted in terms of personal probabilities and utilities – but if that is so, then an empirical argument must be made that psychological findings well outside of decision-making literature might be reasonably re-interpreted within it (Goldstein & Weber, 1997). It cannot be merely assumed that a forced choice in a gamble with explicitly stated probabilities can be extended to all other, or, for that matter, any other, situation. Another possibility is to restrict the study of decision-making to whatever situations are obviously construed in terms of probabilities and utilities of different outcomes or alternatively, impose such appraisal of the decision-situation on the subject whose decision-making is being studied. This in fact has largely been the case, but it is not clear whether there is any reason, apart from historical development of the field, why should psychologists limit themselves in this way. Inconsistency between attitudes and behaviors To talk about what is right is one thing, to notice that such a talk bears on a situation at hand a whole other. This has been pointed out at least since Mill (1869) and Dewey (1924). “The power of the situation” is a set of empirical observations that speaks to the fundamental role of environment in determining the actual course of actions taken. Every theory of decision-making has to address the question of why making a decision about a behavior in a hypothetical or verbally described situation, such as asking subjects what they would do or what ought to be done in an experimental set-up that tests for conformity (e.g. Asch, 1955) and making the actual decision as reflected in the course of action actually taken, can be so very different. The focus of decision-theories on judgment as opposed to actual behavior possibly reflects a general conviction within cognitive psychology that psychology should study competencies (such as judgments of grammaticality) as opposed to performances (that is, actual spoken or sign THE ACTION-HESITANCY MODEL 12 language). In fact, early on Chomsky (1975) explicitly claimed that study of performances, that is, actual behavior, is intractable. While this assumption allows one to separate signal and noise, it also has the inconvenient implication that behavior of actual subjects in the Milgram’s experiment is irrelevant for the science of psychology. What matters is their competency to judge such behavior as inadequate when presented with the case in abstraction. It is not at all obvious that such focus in study of decision-making is justified (Agnew, Carlston, Graziano & Kelly, 2009). Decision-making is eclectic Decision-making is not a single cognitive process. In fact, almost all mental processes can be utilized in some decision-making situations. But this fact has been unintentionally obscured in research driven by a very specific theory, focused on a single way of making decisions. As we have seen above, the expected utility framework assumes that decision process begins with a choice within a fixed set of options which is, from that point, effectively sealed from any new information regarding either options or their likelihoods of occurring. In experimental situation these options are most often provided by the experimenter. However, in a natural environment, figuring out what the options are is one of the most important parts of decision-making. In addition, the whole universe of options can be radically transformed by introduction of outside information. If one is trying to choose a fabric for his sofa, for example, and is offered various options with available prices etc, his understanding of his decision problem will transform radically if he is to learn over the cell-phone that his house is on fire. Decision-making theory must leave the agent the option to radically change his mind and discover new options at any point in the decision-making process. Following economic tradition, THE ACTION-HESITANCY MODEL 13 theories of decision making have instead focused on a state of the agent faced with fixed options with no access to either outside information or even his own memory. The action-hesitancy model This paper proposes a new, non-consequentialist decision-making theory that can satisfy these major properties of decision-making as well as some others. The first requirement in approaching decisions in a psychologically fresh way is to address the issue of why decisions happen at all – that is, why there exists hesitancy. The state of hesitancy is a failure of decision-making processes that normally produce action. We need to make a decision when the course of action - because of an inner conflict, “objective” uncertainty, cognitive ambiguity, conflicting information or some other reason – is not obvious to us. Decision-making, then, should not be construed as aiming at any particular goal (such as hedonic maximization) but as process that returns us to the state where the course of action is obvious. The intuitive appeal of hedonic maximization stems from the fact that actions with clearly hedonically superior outcomes often are obvious. Consequently, evaluation of possible outcomes is just one of many strategies used to resolve hesitancy, not a foundation or a model for all others. If this distinction between obviousness and hesitancy seems a little nebulous that is because it inherited its nebulousness from a distinction between cases in which decisions are made and cases in which no decision-making is involved that is embedded in most of the current models of decision-making (see Loewenestein et al, 2001 for the exception already discussed). Thus, obvious decisions are usually not considered to be decisions at all, so what are usually thought of decisions are cases of hesitancy. But rather than pinning the distinction on a property THE ACTION-HESITANCY MODEL 14 such as intentionality, as some models do (Kahneman, 2003) the model I am proposing pins its distinction to a less metaphysically burdened notion of mere hesitancy. However, the change goes beyond merely renaming the old distinction. The reason why previous “no-decisions” are now made an integral part of decision-making model is because they serve as a model for all decisions. In other words, each agent is trying, when hesitant, to bring himself into the state of action, that is, the state in which decisions become obvious again (Figure 1). The appreciation of the connection between certainty of attitudes and behavioral action has a long history and a strong empirical record (e.g. Abelson, 1988; Barden & Petty, 2008). In addition, it has been shown (Buturovic-Bradaric, 2010) that experimentally manipulated feeling of obviousness influences one’s willingness to act on it, which corroborates the importance of certainty in decision-making. The way the agent leaves the hesitancy stage is by acquiring new beliefs which in turn change his perception of his situation in a way that makes some course of action obvious. Thus, in order to understand what happens when the agent is hesitant we have in parallel to address the case when he is not hesitant, that is, when decisions he makes are obvious to him. In order to understand what kind of beliefs the agent is trying to acquire we need to have on mind the ideal decision that this process aspires to, which is an obvious decision, as shown in Figure 1. In other words, though considerations of happiness and utility play an important role in decision-making, when making a decision the agent is not trying to be “happy” - what he is trying to accomplish, using, deliberately and/or unconsciously, a great number of psychological processes, is to be certain. Analyses of goals and desires, costs and benefits, probabilities and utilities are, on this account, just one important tool that agents use in trying to make the choice obvious to themselves. THE ACTION-HESITANCY MODEL 15 Obviousness relations Obviousness relations relate composites, which are psychological “summaries” of environmental situations (appraisals, gists), with courses of action. They have an if-then form and they are categorical and conditional. A condition needs to be met for any particular obviousness relation to be activated. But once the condition is met, the course of action is determined categorically. As far as I can tell, the nature of this categorical activation is best described as obviousness. The relation of obviousness between situations and actions can be a result of a variety of psychological processes. What is common to all of them is that, when in such situation, the agent acts without seriously considering alternative courses of action. Thus, it is important to appreciate that the relation of obviousness is not at all the same as one in which, say, “hot” system of CAPS network is involved (e.g., Mischel, 2004, Ayduk et al, 2002). It is true that when the agent is in some states, such as when he is very angry, some courses of action (e.g. slamming the door) might become obvious to him. But the relation of obviousness also exists in cases that are considered as purely cognitive as any, such as a relation between a mathematical problem and its solution. There are usually no intense feelings involved in recognizing that, if one is not sure what x is such that x+3567 = 3578, then, obviously, one should subtract 3567 from 3578, and, obviously, x=11. This process is highly deliberate but, at the same time, it is not decision-making, because it does not involve serious consideration of different options. Nor is the relation of obviousness at all the same as “intuition” or “gut feelings”. Intuiting about a problem or a decision rarely leads to an obvious conclusion (e.g. “my intuition tells me that I should study law but…”), and some obvious solutions are not at all intuitive. That the Earth rotates, to take one example, is entirely obvious yet completely unintuitive. In fact, it is THE ACTION-HESITANCY MODEL 16 often precisely results that are – after some substantive learning through hesitancy cycles - both obvious and non-intuitive, that command our greatest admiration. This is a very important point, as there exists a great number of “dual-process” psychological theories that make distinction between intuitive and thoughtful, affective and cognitive etc, the centerpiece of their theory of judgment (Sloman, 1996). This is not at all the case with the action-hesitancy (a-h) model. This model can very well acknowledge that there exists an important and even principled difference between these processes for the simple reason that it does not specify the single type of a process that brings about relations of obviousness. The distinction between action and hesitancy is conceptually orthogonal to the distinctions between emotions, intuition and reflective cognitive processes. It is basically a distinction between effective (those that deliver actions) and non-effective processes (those that keep the agent in the state of hesitancy) in a particular context, regardless of the phenomenological nature of those processes on the cognitive-affective-intuitive dimension. Unlike some other models (e.g. Reyna, 2004) the action-hesitancy model does not put emotions, gut-feelings or intuitions on a pedestal. But it does allow that, in many cases, emotions, gut-feelings and intuitions play a crucial role in the pursuit of action – specifically, they play it in those cases in which they leave no room for hesitation. If they do leave room for hesitation they have, in essence, failed in that particular situation. The same holds for reasoning: if reasoning does not deliver certainty then it has failed in its task. Logic is, on this view, of such a significant value because it is so efficient in delivering certainty – but only in some contexts. However, it is the certainty of action, that is, “whatever works”, that is of paramount importance – and not any particular process across the board. THE ACTION-HESITANCY MODEL 17 The relations of obviousness also cut across a distinction between processes that are innate (meaning, have a similar form in a variety of environments) and learned, as well as those that are universal for almost all humans and those that are specific for one person. Some relations of obviousness are, indeed universal, and some of those almost certainly have a strong genetic underpinning (for example, it is obvious to many mothers that when their baby cries they should try to sooth it). But some of them are obviously learned and seen as obvious by only small numbers of people. For example, only a very small proportion of individuals in the history of our species have found it obvious that exp(i*π)=-1. But for those that have, this is as obvious as anything ever is. Thus, the relations of obviousness between situations and actions subsume variety of very distinct relations such as instincts, associations, habits and can initiate behavioral scripts, motor plans, and algorithms for solving classes of problems. These relations can arise in many ways – through development of innate capacities, one-trial learning, use-it-or-lose-it principle, classical and operant conditioning. It is very important to note that actions performed influence relations of obviousness (it looked obvious to Josh that he should drink milk in a carton, but, having found it to be sour, it is now obvious that he shouldn’t), cognitive-emotional state of the agent and environment. The a-h model puts a great weight on human ability to learn from their mistakes – something that is almost entirely absent from decision-making models in the current tradition which, due to their origin in a static consequentialist model, tend to lack time dimension. Composites A composite is a psychological summary i.e. a subjective interpretation of an environmental situation. The main function of composites is to extract and interpret relevant features of a context, which allows agents to classify physically different situations under the THE ACTION-HESITANCY MODEL 18 same composite and apply appropriate relation of obviousness, if there are any. The number of relevant features might be very small, allowing for quite different contexts to be classified under the same compositional type. Crucially, the composites can change. They can change because the environment has changed, because the agent’s state has changed (he turned on the music and now it seems obvious that he should call his girlfriend), because the agent has endorsed a new belief or because the agent is convinced (by analogical reasoning, for example) to see his situation in a different way. In either of the latter two cases, the situation might seem radically different to the agent, although the environment hasn’t changed. The notion of belief figuring here should be treated loosely, as possibly having a motivational component – what it means is that the agent was exposed to some new information, such as advice, he believed it or in some other psychological way endorsed or accepted it. This event changed the composite, by adding a new feature, highlighting hitherto ignored feature of the environment or by changing the set of relevant features altogether. Hesitancy When the agent is faced with a new environmental situation he tries to match the resulting composite to one of composites for which an obvious course of action exists. That is, he tries to match it to the “if” in the one of the existing “if-then” obviousness relations. As explained above, if the obvious course of action is available, the agent pursues it. Hesitancy happens when there is no obvious course of action available for the current composite, because the agent is torn among multiple possible interpretations or, alternatively, conceives of situation too broadly and “does not know what to think”. THE ACTION-HESITANCY MODEL 19 One thing that the agents can do in this situation is to search for a subset of composites which do have obvious courses of action associated with them and that are, at the same time, cognitively close to his present composite. The courses of action associated with this subset of composites are what are normally considered to be ‘options’ – something that the agents normally need to construct as opposed to be given by the experimenter. Starting with the closest option, our agent might try to understand what sort of cognitive state he needs to be in – for example, what additional beliefs he needs to acquire - in order for things to be obvious to him. He then goes on to search for information that will bring him to this state. Another possibility is that the agent reinterprets his current situation. One way to do this would be by reasoning from analogies. Analogies can help discovery entirely new and obvious options, or help make non-obvious possibilities obvious. If one is considering, for example, whether to be a vegetarian and is hesitant about that, he can be convinced to become one by the analogy animals=children. Since he would obviously not eat children, and animals are, by assumption, identical to children, then he should not be eating animals. So, by reasoning by analogies, the agent corrects his view of his circumstances. However, this correction is always tied to a certain environment and state, so it does not follow that agent will equate animals and children in all circumstances. There exists a tradition within decision-theory (Keynes, 1921; Selten, 1978; Cross, 1983) of modeling all decisions “under uncertainty” as cases of analogical reasoning. According to a recent version of this approach (Gilboa & Schmeidler, 2001), when a person is faced with a decision problem, it calculates the value of similarity function between that problem and cases available in memory. For each course of action, the agent then calculates the utility of that course of action as a linear combination of similarity function between the new problem and memory THE ACTION-HESITANCY MODEL 20 cases and utility of outcomes associated with these memory cases. There is an important similarity of this approach with my basic framework – namely, the idea that the person tries to match the current composite to some set of composites available, and that she often does so by using analogical reasoning. There are several fundamental differences. For example, in the a-h model, the agent does not associate utilities with composites. It is obviousness relations that implicitly represent hedonic consequences of decisions, but only at a crude level – options that would be considered hedonically different if measured on a continuum can nevertheless be approximated by the same psychological composite. Also, the nature of composites that the agent uses is such that composites are not necessarily based on actual experience or statistical learning of categories but can be hypothetical in nature. The most important difference however is the existence of the feedback loop, that is, active attempts on the side of the agent to reshape his composite. This does not exist in Keynes’ model or does that are derived from it. This is a theoretical consequence of the fact that Keynes and those who followed on his steps did not make a distinction between successful (obvious) and unsuccessful matching. As a consequence, on the ah account, the matching of the current problem (composite) and available problems does not have to be based on analogies – whether it is will depend on the information that the agent acquired and the problem at hand. The creation of composites depends on the obviousness relations that are available to the agent. A scientist devoted to a theory will tend to see many different situations as belonging to the same composite. Once one learns about mass and acceleration, he sees great variety of different situations as belonging to “accelerating mass” composite. Wildly different physical situations (rotating Earth and moving truck) can be, in a given context, perceived as being of the THE ACTION-HESITANCY MODEL 21 same type i.e. reducing to the same composite. The composite depends on the situation and the state of the agent. It is a psychological fact of priming, predated many centuries by the idea of “laws of association”, that the psychological state of the agent will be influenced by his previous states – either those that he has been recently in or those that are chronically accessible to him. The composites that are more often accessed will be the first options to be considered in most cases, and often new options will be recognized within old composites, resulting in hierarchical growth of composites. One of the most prominent forms of belief-acquisition behavior is a consultation with other agents. But it is not the only one. Other ways of belief-acquisition is memory-retrieval and computation – figuring out what one thinks about an issue given his commitments to other statements. For example, one can believe in the axioms of classical geometry but not hold a belief in any particular theorem that follows from them, say about the sum of angles in a triangle. When faced in with a new problem, that asks for calculation of the third angle of a triangle with two of them known, he is incapable of providing an answer that seems obvious. Thus, he embarks on acquiring new beliefs – most likely by reading a book on geometry, asking a teacher or deriving a theorem on his own. Once the theorem is derived, his initial problem gets reduced to a class of problems in which a simple subtraction is asked for, and the answer becomes obvious. But while computation is a prominent form of belief-acquisition behavior it should not be assumed that all such behaviors are either effortful or reflective. Observational learning, for example, also counts as a form of belief-acquisition behavior. If one is hungry yet hesitant to open a snack during the class-time and then observes somebody else snacking, this might make it obvious to him that he is permitted to snack. Thus, like there is no single psychological process (reflection or intuition, THE ACTION-HESITANCY MODEL 22 emotion or habit) that drives action, there is no single psychological process that drives belief acquisition. This might all sound trivial, and in a sense the observation that people seek new information when hesitant is entirely trivial. But, in another sense, it is precisely the trivial nature that makes it important. The reason why it is trivial is because it is very robust. It is ubiquitous and widely recognized as such. It is not at all clear that subtle, narrow and unexpected gambling phenomena serve as a more proper foundation for a decision-making theory than phenomena so robust to be considered trivial. In any case, the account cannot be judged on the triviality of its founding observations. It can only be judged regarding how it handles empirical data such as: The role of authority in decision making Milgram’s experiments are one of the most famous of all psychology. They deliver a non-intuitive answer to an important question: what will a person decide when faced with a request to deliver an extremely painful and dangerous electric shock to a fellow participant who failed in performing an inane task and who alternatively begs and vehemently protests the treatment, requesting to be released from the “study”? Famously, about 65% of people are ready to give 450 V electric shocks because experimenter told them to (Milgram, 1969; Doris, 2002) – a result that has recently been corroborated (Burger, 2009). More than 90% are ready to perform tangential collaborating tasks when another participant is giving such shocks. In a variant of Milgram’s experiment, 77% gave a (this time real) painful shock to a hapless “cute, fluffy” puppy (Sheridan & King, 1972). Hume’s folk-psychological assertion that “Reason is, and ought only to be the slave of the passions, and can never pretend to any other office than to serve and obey them” (Hume, 1739) is plainly false. Whatever reason is the slave to here, it is not passion. People giving THE ACTION-HESITANCY MODEL 23 shocks were extremely upset while doing so. Thus, any decision-making theory (such as RAF) that puts faith into feelings to compensate for the difficulties of consequentialist foundation will face problems trying to account for this pattern. The problem is not just that feelings work in the direction that is the opposite of observed behavior; it is previously observed gap between a decision made hypothetically and a decision carried out in actual circumstances. This is reflected in the fact that those asked to predict the results of Milgram experiment grossly underestimated the proportion of obedient subjects, presumably using a simulation of their own assumed behavior as a measuring stick. But consequentialist models give little reason to assume that utility and probability of various outcomes associated with either refusal to participate or follow-through should be different between the case in which a person participates in the experiment and in which he is merely told about the experimental set-up. The power of the notion of ‘salience’ to account for the stark contrast in judgments between two contexts is debatable. It is true that in the real experiment the authority is very salient compared to its hypothetical version – but so is the suffering of the victim. Finally, the experiment has shown that participating subjects regularly ask the experimenter what they should do – a behavior not easily accounted for by the RAF. The heterogeneous mixture of Kahneman’s Prospect Theory and two-system theory also faces some difficulties with Milgram’s experiments. To account for the behavior of experimental participants we need to assume that with each increase in voltage the reference point changes in such a way to make expected value of following the instructions and possibly killing a person greater than the value of not following the instructions and saving the person. This seems implausible because, as already mentioned, few of those assessing those values outside of the experimental situation see this relation. Perhaps the most generous application of Prospect THE ACTION-HESITANCY MODEL 24 Theory would have hypothetical experiment treated as one decision (whether to follow-through till the end – that is, 450V), where reference point is the begging of the experiment, and actual experiment treated as a series of decisions for which reference point changes with each decision (i.e. each voltage increase). However, it still remains unclear that the expected gains and losses work out to lead the subject to keep giving shocks, not least because prima facie losses seem to be much bigger than gains, which is reflected through enormous emotional distress. Additional problems are the marked hesitancy of the subjects reflected in their asking the experimenter what to do, as well as sensitivity of their behavior to variety of factors in many versions of the experiment (when subjects are asked to administer shocks physically, when instructions are delivered to them over the phone, when they are given the instructions by other “participants” as opposed to authority, etc) – neither of which is easily accounted for by consequentialist accounts. This is a result of the fact that behavior of the subjects in the experiment is, in a sense, neither intuitive nor reflective nor emotional. The participants are very hesitant and repeatedly seek help from the experimenter; their emotions tell them to stop and it is unlikely that it is higher reflective processes that tell them to continue. Haidt’s (2001) model also has difficulty with the fact that when people are presented with the case of experimental scenario outside experimental environment, they, according to this model, intuitively judge that it is wrong to deliver shocks to the victim. Yet, when asked to follow through one’s own judgment – or rather, to make it when it is most pertinent, that turns out to be quite difficult. It is rather implausible that moral intuitions of the subjects in the experiment change, as Haidt’s model would suggest – it is more likely that they didn’t see it as the same situation as the one when they are posed “a moral problem” to reflect on. There is nothing in Haidt’s model that suggests this should happen. THE ACTION-HESITANCY MODEL 25 On the other hand, the a-h model can explain these results reasonably well, though with less specificity than one would perhaps desire. First, it suggests that there should be a difference between judging the experiment from the inside and the outside. The physical environment is plainly different and it is quite reasonable that it should be perceived differently. Then, most of the subjects who decline to proceed in the experiment decline at the moment where the “victim” for the first time gives an unequivocal information that he wants the experiment to stop (“Ugh!!! Experimenter! That’s all. Get me out of here… My heart is starting to bother me now…I refuse to go on. Let me out” – this happens at 150V). Other subjects who decline to continue resign at other points where “the victim” brings additional information. Barely anybody declines to continue at 225, 240, 255 V – where “the victim” merely repeats what he already said (data according to Doris, 2002). The models (such as RAF) that put emphasis on vividness as opposed to informational content would have difficulty explaining why repeated screams help so much less than new information. Thus, interpreted within the a-h model, our subject starts with giving shocks because he unequivocally identifies his situation as the one in which instructions are to be followed – the ifthen relation in question might be something as simple lab experiment -> follow instructions. When “the victim” starts to grunt the situation somewhat changes, because it brings to attention new features of the environment that are difficult to integrate with those initially perceived as relevant. This new information induces hesitancy in some subjects, who, as the model predicts, seek information to make things obvious again. However, they have difficulty changing the course of action because information given to them by the experimenter reinforces their initial composite, making it hard for them to reinterpret their environment and discover new options. THE ACTION-HESITANCY MODEL 26 The obedience of Milgram’s subjects is ultimately a perceptual problem - it is not a problem of feelings or miscalculation. Learning The simple prediction of the a-h model is that, if a person faces exact same decision within the same context multiple times, her response time will become shorter. The reason for that is accessibility of the relevant composite. Accessibility refers to the relative ease by which a certain mental item can be retrieved. Studies in social cognition have showed (e.g. Bargh, 1997) that items that are made accessible can influence decision, judgments, and behaviors. This easiness is directly embedded into the a-h model. That model starts with the observation that the agent perceives the situation, and that that perception is mediated by his current state and existing relations of obviousness. If one needs to make the same decision multiple times in succession, the relevant composite will become easier to retrieve, and the response time will become shorter. None of the models discussed predicts that, when a person is exposed to the same problem multiple times – for example, if he faces the exact same gambling decision fifty times in a row, the course of action will be more readily recognized. It might be objected that this is a trivial problem – that agents want to appear consistent or that they are just responding “mechanically”. The first hypothesis does not explain why would response time be shorter (it only explains why would a person make the same decision each time), while the latter requires an explanation, not a dismissal. On the a-h model it is very clear why this happens – once the current composite has been successfully reduced to the one for which obvious course of action is available, the time needed to retrieve the relevant composite again will be shorter. But there is no reason why this should happen at all on any of the other models. THE ACTION-HESITANCY MODEL 27 That stems from the fact that those models don’t have answers to why decisions happen. Nothing in prospect theory suggests that, if one sees the same gamble a hundred times, he will hesitate less in giving his answer on his hundredth than on his first trial. The other two models have the same problem. None of them recognizes decisions as at the same time failures decisionmaking processes and opportunities for those processes to be improved. As we have already seen, risk-as-feelings hypothesis is, by the authors’ admission, incapable of meaningfully addressing decisions, and Kahneman’s view is that decisions are different from the rest of behavior because they involve intentionality. There is a lot of evidence in problem-solving literature that people do in fact improve their performance in the general direction of automaticity when repeatedly faced with similar problems. According to a popular view (Anderson, 1982; Anderson, 2000), this process goes through declarative and knowledge compilation phase before it reaches procedural stage characterized by the production rules of the ‘if-then’ form. The consistency of this account of cognitive skill-acquisition with the a-h model is clear: the pronounced role of verbal reasoning in the early stages can be interpreting as reflecting attempts to position the current composite within the already existing network of skills; further knowledge compilation refers to refining existing composites and correcting obviousness relations through feedback until a new production rule, that is, obviousness relation has been adopted. Unfamiliar situations According to the a-h model, crucial for the prompt recognition of an obvious course of action are agents’ abilities to properly interpret the situation as well as their ability to recognize similarity between the composite so computed and a class of composites associated with obvious THE ACTION-HESITANCY MODEL 28 actions. Since the composites are a product of interaction of the environment with the current state of the agent, it follows that in novel situations (as well as unusual states of mind) the agent will usually be more hesitant. Generally speaking, a new situation is harder to classify under obvious composites because an environmental part of the interpretative function has uncommon values on a great number of parameters (that is what makes it a novel situation). Although the notion of similarity cannot be reduced to physical similarity, a very different environment can serve as a special case of a novel situation in which we can look at predictions from different theories. The a-h model predicts, thus, that when an agent, say, an American, finds himself in a distant, unfamiliar country, or a novel environment, such as college or a summer camp, he will become more hesitant. In fact, it predicts that the more psychologically distant the country (for example, China as opposed to Great Britain) the more hesitant the agent will get. By becoming more hesitant, to reiterate, I mean that the agent will be more frequently considering alternative courses of action. For example, while he normally wouldn’t think about leaving his shoes in front of a church, he might now start to think about that possibility. It is no longer obvious to him that he shouldn’t do that. Likewise, if our agent is a freshman in college or even a student transferred to a new school, it might no longer be obvious to him that he should be speaking in a class when unclear on the point in the lecture. The a-h model thus predicts that when physical environment changes, the agent will be more hesitant and in turn spend more time in the learning phase – about the new country or new school. If this seems like a trivial observation – sure, what is more obvious than the fact that one should and does acquire new beliefs when in an unfamiliar environment - we should note that neither risk-as-feelings hypothesis nor prospect theory actually predict that any such thing should THE ACTION-HESITANCY MODEL 29 happen. Nor does Haidt’ model predict failures of moral intuitions in unfamiliar environments. Recall that TS hinges the distinction between decisions and no-decisions on intentionality (“judgments are always intentional and explicit even when they are not overtly expressed”). It is not at all clear that agents should become more intentional or explicit in a novel environment, or even that they actually do become more intentional. Two-system theory could resort, perhaps, to a claim that intuitive system fails to deliver answers in a novel environment (similar argument might be made on a behalf of Haidt’s model) but it is unclear that that is so and it is also unclear that system 2 (rule governed, controlled) should be able to compensate for this. As already discussed, seeking new information does not necessarily imply rule governed reasoning. Simple, ‘unreflective’ observation of others’ behavior (whether they leave their shoes in front of a church, for example) might be enough for a person to acquire new beliefs. Thus, it seems that there certainly exist situations where neither intuitive nor reflective system can compensate for lack of answers except by seeking new information outside of itself. Confirmation bias There exists abundant evidence (Ditto & Lopez, 1993) that people’s search for information is biased towards what they already suspect or desire to be true. Let’s call this effect, conceived most broadly, confirmation bias. In a typical demonstration of this effect, a person will be given a hypothesis, such as that another person is an introvert (or extrovert), and asked to determine whether the hypothesis is true. Invariably, subjects presented with the same evidence will lean towards the hypothesis which was suggested to them. Those to whom it was suggested that the person is an introvert will be more likely to conclude that the person is indeed an introvert, based on the evidence presented. THE ACTION-HESITANCY MODEL 30 This phenomenon can be easily explained within the a-h model. The kind of belief that the person is trying to acquire is a function of her current situation and the candidate situation for which the obvious course of action is available. Thus, if a new situation leans towards some composite, yet the decision is not yet obvious, a person will look for information that will make the decision more obvious, that is, confirming evidence that the new information is of the candidate type rather than objective evidence that might point to its unique status. None of the other three models can easily account for this phenomenon. Prospect theory says nothing about what kind of information will a person seek – as all other models its decisionmaking story begins with an assumption that the person has finished the search for information and at one fixed moment has to make a decision. Not only is such a set-up highly artificial – it positively precludes any such theory of saying anything about how information search precedes apart from reframing information search itself within the cost-benefit framework. Haidt’s model has some potential to account for information search because it does leave the space open for justificatory procedures that involve exchange of information. However, it can say little about the content of those procedures and their dependency on the content of existing moral intuitions. What allows the a-h model to predict confirmation bias is the fact that it has certainty embedded as a goal of information search. Status-quo bias There is a great amount of evidence that agents typically exhibit the status-quo bias. One of the most prominent examples in this area has been organ donor enrollment. For example, it was demonstrated (Johnson & Goldstein, 2003) that enrollments in organ-donation programs in countries with presumed-consent (in which the person needs to opt out of the program) is 86%- THE ACTION-HESITANCY MODEL 31 100% while in the countries with required-request (in which the person needs to opt into the program) is 4%-28%3. From the perspective of the a-h model, the main reason for this effect is that the feature of the current situation (such as enrollment in an organ donor program) that exhibits the status-quo bias is irrelevant for the psychological perception of the situation in a great variety of contexts. While person’s enrollment in an organ donor program is, technically speaking, indeed a feature of every situation he is in, it is irrelevant for the psychological nature of the person’s composite in most contexts. In other words, in most contexts, as they are actually perceived, people don’t care whether they are enrolled in an organ donation program or not, that is, they take whatever their status is for granted. It is other features of the context (combined with their psychological state) that define the composite. As a result, they see their situations as being about something entirely different. This of course does not mean that the feature in question will necessarily remain irrelevant. If it is brought to the agent’s attention then the agent might perceive a feature as relevant and reconsider his choice. People are, after all, not enslaved by default-bias. For example, one might not have thought a single moment about his enrollment in an organ donor program, but if his spouse gets a kidney disease he might suddenly start thinking (that is, acquiring new beliefs) regarding the organ donation enrollment. Less dramatically but more frequently, a background feature might be brought to attention by advertisements. One took his old car for granted, but now he saw an advertisement for a good deal on a new car. In a proper state of mind (say, if he just got a raise) he might find himself in a situation of hesitancy about 3 This observation has, as a rule, been followed by public policy recommendations. An agreement with those recommendations is not implied by my analysis. THE ACTION-HESITANCY MODEL 32 purchasing a new car. The decision to buy a new car might not be obvious to him (at that point or ever) but advertisement brought attention to the previously irrelevant feature of his situation. By their very nature, obviousness relations entail that the agent is not seriously thinking about the case as a case of a decision any more. Thus, the background parameter in question (for example, enrollment into an organ-donor program) has to come to the agent’s attention in the way of dilemma before he can at best, become hesitant. In other words, he needs to see his situation as one that is “about organ-donation enrollment” for hesitancy to be even possible to occur. If that does not happen, as it often doesn’t, then the agent will be doing whatever he has been doing regarding that aspect of his situation, that is, he will exhibit the status-quo bias. Status-quo bias is to an extent accounted for within the prospect theory by the notion of a reference point. Prospect theory starts with an observation that whatever options the person is considering, they will be compared to her reference point – his current situation as it is perceived at that moment. But this does not explain why some options are not considered at all – in this example, donor program enrollment. Status-quo bias is different and much stronger phenomenon than an endowment effect – the fact that we tend to over-value things that are ours. It is largely about the fact that we don’t think about many options that we could in principle think about. Theories that do not take into account a robust human ability to ignore most facets of their own lives will face difficulties trying to explain phenomena that follow from that ability. Change-blindness Studies of memory have shown a phenomenon called “change-blindness” – a general inability of people to register changes in their environment that are irrelevant to a task at hand. In one of the most striking demonstrations of this fact (Simons & Chabris, 1999), people are shown a video-clip of a group of students playing with a ball and asked to record a number of passes by THE ACTION-HESITANCY MODEL 33 one of the teams. At some point in the clip, a clearly visible person in a gorilla suit passes through this group of players. More than 50% of subjects fail to notice this unexpected event and deny it happened when questioned. This fact again corroborates the point that people see their environment through the lenses of a task at hand that is a central part of the a-h model. Boredom In a relatively recent dissertation (Johnston, 2003) it was shown that adding boring details to a story makes readers better grasp the essence of the story. This suggests that one of the functions of boredom might be to optimize information search by detracting from a pursuit of irrelevant information. Since on the a-h model the agent spends a lot of effort acquiring new beliefs, it seems that a signal that could help him sift relevant from irrelevant would be very useful. Feeling of boredom could be such signal. Being bored by something allows us to focus on what is relevant (interesting), as opposed to irrelevant (boring). According to the a-h model, it is not coincidence that people who are perceived as boring tend to offer less personal information and less edification (objective information) and ask more questions and offer more acknowledgments (Leary et al, 1987). Questions and acknowledgments are much less powerful in converting hesitancy and doubt into action and certainty than those of information provision. Zeigarnik effect Zeigarnik effect (Zeigarnik 1927; 1967) refers to an empirical finding that interrupted tasks are more likely to be remembered than those that are completed. The a-h model implies that the reason for that is that interrupted tasks can leave people in the state of doubt or hesitancy regarding the outcome of their activity. In essence, this sort of effect would work to “tag” questions that remained unanswered, making one more likely to seek information that would help resolve them. THE ACTION-HESITANCY MODEL 34 McGuire “inoculation theory” McGuire (1961) has demonstrated that previous exposure to counter-arguments makes one more resistant to persuasion by essentially “inoculating” him against contrary beliefs. In a typical procedure, the person is exposed to views contradicting some beliefs that he takes for granted (for example, that one should brush teeth every day) and then offered “refutational preemption” – a set of arguments that allow the agent to combat contradicting ideas. The evidence shows that such procedure makes the person more resistant to counter-arguments in comparison to those who have not gone through an inoculation process. The a-h model is consistent with this effect. According to the a-h model, the difference between those who were and who were not exposed to counter-arguments during refutational pre-emption is that the former have those counter-arguments more accessible in the hesitancy phase that begins after the person is exposed to contrary views. Thus, both groups of participants are likely to find themselves in the state of doubt when their views are contradicted and this doubt can be resolved either through reverting back to the previous view or in changing one’s views. If counter-arguments are not easily accessible, the process is more likely to end in reversal of previously held views than if they are easily accessible. Concluding remarks The expected utility theory was developed by economists for the purpose of predicting aggregate behavior. This theory allows for any behavioral preference over choices (assuming it satisfies certain axioms, such as transitivity) to be reinterpreted as involving probabilities and utilities of outcomes associated with those choices, with agents trying to maximize their expected utility. To my knowledge, no psychologist has ever seriously advocated a view that human THE ACTION-HESITANCY MODEL 35 decision-making consists of maximizing one’s utility. In fact, though the area of decision-making has been largely addressing decision-making through the lenses of this theory, other fields of psychology, such as moral psychology, developed without ever taking this model seriously, despite the fact that they also study humans trying to make decisions. Extensive psychological research has confirmed that human decision making cannot be adequately described by a simple consequentialist model. As a result, we have seen a proliferation of models trying to graft non-consequentialist appendages on the basic consequentialist framework. As a rule, however, these models have disproportionally directed themselves to explaining ever more subtle psychological phenomena from a conventional decision-making literature – a literature that, to a significant extent, has developed as a commentary on the expected utility theory. As a result, these theories have over-estimated importance of numeric cost-benefit analysis and, despite embracing feelings or intuitions, as a whole significantly underestimated heterogeneity of mental processes that people use to make decisions. The action-hesitancy model is an attempt to break this pattern by bringing under consideration a broader range of psychological phenomena and providing a fairly comprehensive model that can account for them. The model is more similar to social psychology models than to conventional decision-making models precisely because it pays much more attention to the extent to which decision-making is a social activity. THE ACTION-HESITANCY MODEL 36 References Abelson, R. P. (1988). Conviction. American Psychologist, 43, 267-275. Agnew, C.R, Carlston, C. E, Graziano, W. G, Kelly J. R. (2009). Then a miracle occurs: focusing on behavior in social psychological theory and research. Oxford University Press. Anderson, J. R. (1982). Acquisition of cognitive skill. Psychological Review, 89, 369-406 Anderson, J. (2000). Learning and memory. New York: Wiley Asch, S. E. (1946). Forming impressions of personality. Journal of Abnormal and Social Psychology, 41, 258-290. Asch, S. E. (1955). Opinions and social pressure. Scientific American, 31-35. Ayduk, O, Mischel, W. , Downey, G. (2002). Attentional mechanisms linking rejection to hostile reactivity: The role of "hot" versus "cool" focus. Psychological Science,13, 443-448. Barden, J., & Petty, R. E. (2008). The mere perception of elaboration creates attitude certainty: Exploring the thoughtfulness heuristic. Journal of Personality and Social Psychology, 95, 489-509 Bargh, J. (1997). The automaticity of everyday life. In R. S. Wyer, J. (Ed.), The automaticity of everyday life: Advances in social cognition. Mahwah, NJ: Erlbaum Bodenhausen, G. V., Gabriel, S., & Lineberger, M. (2000). Sadness and susceptibility to judgmental bias: The case of anchoring. Psychological Science, 11, 320-323. THE ACTION-HESITANCY MODEL 37 Burger, J. M. (2009). Replicating Milgram: Would people still obey today? American Psychologist, 64, 1-11 Buturovic-Bradaric, Z. (2009). Action-hesitancy: theoretical proposal and preliminary evidence. Dissertation Abstracts International, 70(11-B). Camerer, C. F. (2003). Behavioral game theory: Experiments in strategic interaction. Princeton, NJ: Princeton University Press. Chomsky, N. (1975). Reflections on Language. New York: Pantheon Books. Cialdini, R. (2001). Influence: Science and Practice. Allyn & Bacon, Needham Heights, MA Cross, J. G. (1983). A Theory of Adaptive Economic Behavior. New York: Cambridge University Press. Damasio, Antonio (1994). Descartes’ error. Harper Collins. de Groot, A. (1978). Thought and Choice in Chess. The Hague: Mouton. Dewey, J. (1924). Human Nature and Conduct. New York: Prometheus Books. Ditto, P. H., & Lopez, D. F. (1992). Motivated skepticism: Use of differential decision criteria for preferred and non-preferred conclusions. Journal of Personality and Social Psychology, 63, 568-584. Doris, J (2002). Lack of character: Personality and moral behavior. New York: Cambridge University Press. THE ACTION-HESITANCY MODEL 38 Gilboa, I., Schmeidler, D. (2001). A Theory of Case-Based Decisions. United Kingdom: Cambridge University Press. Gilovich, T. (1981). Seeing the past in the present: The effect of associations to familiar events on judgments and decisions. Journal of Personality and Social Psychology, 59, 623-634. Goldstein, W., Weber, E. (1997). Content and discontent: Indications and implications of domain specificity in preferential decision making. In W. M. Goldstein & R. M. Hogarth (Eds.) Research on Judgment and Decision making, Edited by Goldstein and Hogarth. Cambridge, UK: Cambridge University Press. Greene, J. D. (2007). The secret joke of Kant’s soul, in Moral Psychology, Vol. 3: The Neuroscience of Morality: Emotion, Disease, and Development, W. Sinnott-Armstrong, Ed., MIT Press, Cambridge, MA. Griffin, D. W. & Ross, L. (1991). Subjective construal, social inference, and human misunderstanding. Advances in Experimental Social Psychology, 24, 319-359. Gul, F., & Wolfgang Pesendorfer, W. (2005). The Case for Mindless Economics. www.princeton.edu/~pesendor/mindless.pdf. Haidt, J. (2001). The emotional dog and its rational tail: a social intuitionist approach to moral judgment. Psychological Review, 108, 814-834. Higgins, E. T. (1997). Beyond pleasure and pain. American Psychologist, 52, 1280-1300. Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives? Science, 302, 1338-1339. THE ACTION-HESITANCY MODEL 39 Johnston, G. S. (2003). Effects of seductive and boring details on readers' comprehension of explanatory texts. Dissertation Abstracts International, 63(9-B). Hume, D. (1739). A treatise of human nature. Oxford, UK: Oxford University Press. Kahneman, D., Tversky A. (2000). Choices, values, and frames. Cambridge, UK: Cambridge University Press. Kahneman, D. (2003). A Perspective on Judgment and Choice: Mapping Bounded Rationality. American Psychologist, 58, 697-720. Keynes, J. M. (1921). A Treatise on Probability. London: McMillan and Co. Koffka, K. (1935). Principles of Gestalt Psychology. London: Lund Humphries. Leary, M. R., Rogers, P. A., Canfield, R. W., & Coe, C (1986). Boredom in interpersonal encounters: Antecedents and social implications. Journal of Personality and Social Psychology, 51, 968-975. Loewenstein, G., Weber, E., Hsee, C., Welch, N. (2001). Risk as Feelings. Psychological Bulletin, 127, 267-286. Mas-Collel, A., Whinston, M., Green J. (1995). Microeconomic Theory. Oxford: Oxford University Press. McGuire, W. (1961). Resistance to persuasion conferred by active and passive prior refutation of the same and alternative counterarguments. Journal of Abnormal and Social Psychology, 63, 326-332. THE ACTION-HESITANCY MODEL 40 Milgram, S. (1969). Obedience to Authority. New York: Harper & Row. Mill, J. S. (1869/2006). On Liberty. New York: Filiquarian Publishing, LLC. Nisbett, R. E., & Ross, L. (1980). Human preferences: Strategies and shortcomings of social judgment. Englewood Cliffs, NJ: Prentice-Hall. Nisbett, R., & Wilson, T. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231-259. Pesendorfer, W. (2006). Behavioral Economics Comes of Age. www.princeton.edu/~pesendor/book-review.pdf. Ramsey, F. (1926). Truth and Probability. In Kyburg H., & Smokler, H. (Eds.) (1980). Studies in Subjective Probability, Huntington, New York: Robert E. Krieger Publishing Company. Reyna, V.F. (2004). How people make decisions that involve risk. A dual-processes approach. Current Directions in Psychological Science. 13, 60-66. Schultz, W. (2006). Behavioral theories and the neurophysiology of reward. Annual Review Psychology, 57, 87-115 Selten, R. (1978). The Chain-Store Paradox. Theory and Decision, 9, 127-58. Sheridan, C, King, R. (1972). Obedience to authority with an authentic victim. Proceedings of the Annual Convention of the American Psychological Association, 7, 165-166. Simons D.J., Chabris C.F. (1999). Gorillas in our midst: Sustained inattentional blindness for dynamic events. Perception, 28, 1059-1074. THE ACTION-HESITANCY MODEL 41 Sloman, S. A. (1996). The empirical case for two systems of reasoning. Psychological Bulletin, 119, 3-22. Smith, E. R. & DeCoster, J. (1999). Associative and rule-based processing: A connectionist interpretation of dual process models. In S. Chaiken & Y. Tropez (Eds.), Dual-process theories in social psychology (p. 323-336). New York, NY: The Guilford Press. Tetlock, P.E. (2003). Thinking the unthinkable: sacred values and taboo cognitions. Trends in Cognitive Sciences, 320-324. Tversky, A, & Kahneman, D. (2000). Advances in Prospect Theory: Cumulative Representation of Uncertainty. In Tversky, A., & Kahneman, D. (Eds), Choices, values, and frames. Cambridge, UK: Cambridge University Press. Tversky, A. & Kahneman, D. (1983). "Extension versus intuitive reasoning: The conjunction fallacy in probability judgment". Psychological Review, 293–315. Turiel, E. (1983). The Development of social knowledge: morality and convention. Cambridge: Cambridge University Press. Wegner, D. M., & Gilbert, D. T. (2000). Social psychology—The science of human experience. In H. Bless & J. P. Forgas (Eds.), The message within: The role of subjective experience in social cognition and behavior. Philadephia: Psychology Press. Wilson, T.D., Lisle, D., Schooler, J.W., Hodges, S.D, Klaaren, K.H., & LaFleur, S.J. (1993). Introspecting about reasons can reduce post-choice satisfaction. Personality and Social Psychology Bulleting, 19, 331-339. THE ACTION-HESITANCY MODEL 42 Wilson, T.D., & Schooler, J.W. Thinking too much: Introspection can reduce the quality of preferences and decisions. Journal of Personalitiy and Social Psychology, 60, 181-191. Zeigarnik, B. (1927). Das Behalten erledigter und unerledigter Handlungen. Psychologische Forschung, 9, 1-85. Zeigarnik, B. (1967). On finished and unfinished tasks. In W. D. Ellis (Ed.), A sourcebook of Gestalt psychology, New York: Humanities press.