Motivational Reasons for Biased Decisions: The Sunk-Cost Effect's Instrumental Rationality PDF

ORIGINAL RESEARCH published: 24 May...

ORIGINAL RESEARCH published: 24 May 2018 doi: 10.3389/fpsyg.2018.00815 Motivational Reasons for Biased Decisions: The Sunk-Cost Effect’s Instrumental Rationality Markus Domeier* , Pierre Sachse and Bernd Schäfer Institute of Psychology, University of Innsbruck, Innsbruck, Austria The present study describes the mechanism of need regulation, which accompanies the so-called “biased” decisions. We hypothesized an unconscious urge for psychological need satisfaction as the trigger for cognitive biases. In an experimental study (N = 106), participants had the opportunity to win money in a functionality test. In the test, they could either use the solution they had developed (sunk cost) or an alternative solution that offered a higher probability of winning. The selection of the sunk-cost option (SCO) was the most chosen option, supporting the hypothesis of this study. The reason behind the majority of participants choosing the SCO seemed to be the satisfaction of psychological needs, despite a reduced chance of winning money. An intervention, Edited by: Shira Elqayam, which aimed at triggering self-reflection, had no impact on the decision. The findings De Montfort University, of this study contribute to the discussion on the reasons for cognitive biases and United Kingdom their formation in the human mind. Moreover, it discusses the application of the label Reviewed by: Linden John Ball, “irrational” for biased decisions and proposes reasons for instrumental rationality, which University of Central Lancashire, exist at an unconscious, need-regulative level. United Kingdom C. Dominik Güss, Keywords: need regulation, PSI theory, cognitive bias, irrational decision, sunk-cost effect, decision-making, University of North Florida, debiasing United States *Correspondence: Markus Domeier INTRODUCTION [email protected] According to a dolphin, Michael Phelps is a lousy swimmer. According to a cheetah, Usain Bolt Specialty section: runs in slow motion. According to the homo oeconomicus model, people predominantly make This article was submitted to decisions irrationally. Do these judgments appear fair to you? In the first two cases, it seems rather Cognitive Science, clear that an inappropriate frame is being applied to assess these individuals’ ability in swimming a section of the journal or running. However, the third statement still comes up repeatedly when human decision-making Frontiers in Psychology abilities are being judged (Magrabi and Bach, 2013). The homo oeconomicus model (Simon, Received: 16 October 2017 1955) describes human beings as rational agents, who mostly follow monetary goals, have stable Accepted: 07 May 2018 preferences, maximize subjective utility, and ignore sunk-costs. Published: 24 May 2018 However, studies show that actual human performance in decision-making deviates from Citation: this model (Stanovich and West, 2000), influenced by so-called cognitive biases, departing Domeier M, Sachse P and systematically from the normative standards (Kerr et al., 1996). These biases have an effect on real- Schäfer B (2018) Motivational Reasons for Biased Decisions: life decision-making (for a taxonomy of cognitive biases see Carter et al., 2007; Stanovich et al., The Sunk-Cost Effect’s Instrumental 2008). Based on 20 years of past work analyzing 100s of business decisions, Nutt (2002) affirms Rationality. Front. Psychol. 9:815. that 50% of all decisions fail. This number could lead to the impression that the human mind is doi: 10.3389/fpsyg.2018.00815 hopelessly flawed, thereby confirming the accusation of “irrational.” Frontiers in Psychology | www.frontiersin.org 1 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality However, Cohen (1981) has a more positive view of the However, if you omit the normative approach, which reference human reasoning ability. He argues that errors should not would guide you in the evaluation of behavior? Elqayam (2012) be seen as proof of human irrationality. He assumes that proposes “grounded rationality” to address these concerns. This errors originate from three areas: the normative system, the framework aims to evaluate the rationality of human behavior interpretation problem, and the external validity problem (Evans, without referring to a universal standard. Elqayam (2012, p. 43) 1993). Errors are rated in relation to a normative system that proposed a first working definition of grounded rationality: depends on the understanding of logic and deductive reasoning. “Behavior B is rational for agent A, in epistemic context E, However, that system does not necessarily fit to behavior in if B facilitates achievement of A’s goals within the constraints the real world. The interpretation problem concerns the fact of E.” The approach focuses on the achievement of personal that participants do not necessarily understand the task in the goals. The epistemic context includes everything that affects the same way the instructor thinks they should. Finally, the external beliefs and desires of a person, and the constraints cover all validity problem describes the fact that many tasks in the lab are physical or mental limitations which human beings might have. somewhat artificial and have little in common with real-life tasks. Therefore, grounded rationality combines bounded rationality In Cohen’s view, errors can only be rated as such when, under with cognitive variability. ideal conditions, a person agrees that it is an error. He concludes Focusing on the rationality concerning cognitive biases, that no present or future findings in literature or research can Tetlock and Mellers (2002) provided another perspective, lead to the assumption of faulty competence in human reasoning. emphasizing the possibility of reinterpreting cognitive biases He instead points to patterns of illusions, which might be active and, thus, showing possible “upsides” to these apparently during reasoning and could lead to faulty conclusions. poor decisions. Looking at the example of entrepreneurs Evans and Pollard (1981) criticized this line of argumentation, demonstrating the overconfidence effect, defined as a systematic as it offers low practical relevance. Specifically, they consider that overestimation of probabilities (Camerer and Lovallo, 1999), the realistic tasks could also lead to biases, as personal experiences authors reflect that this bias could also be the reason behind and emotions can influence the participants’ behavior. Moreover, success because it allows them to overcome paralyzing loss they express that pointing to illusions, as Cohen (1981) mentions, aversion, taking more risks in business ventures. This interesting does not clarify the conditions when an individual is “rational,” point of view leads to the question of whether biases can involve or presents a “cognitive illusion,” and what the “ideal conditions” rationales outside the conventional frame of assessment. are. Although some of this criticism might apply, Evans (2005) We agree with the position that people sometimes rather disagrees with the assumption that task designs should be follow instrumental rationality (to achieve their goals) than considered artificial, as many well-researched effects in the normative rationality. In the present study, we sought to take laboratory have been successfully transferred to the real world. a closer look at these personal goals and which role they play One approach to deal with the problem of the normative when people present a “bias” in decision-making, which means system is distinguishing between two kinds of rationality. The that they engage in behaviors that are not expected from the first refers to personal/instrumental reasoning, which is used to viewpoint of the normative theory. As an example of an effect achieve one’s goals; the second refers to normative reasoning, that is typically referred to as a bias, we chose the sunk- which occurs in relation to a normative system (Evans and Over, cost effect. Our focus concerning personal goals lies in the 1996). These terms have been presented earlier as rationality1 role of motivational needs, especially the need for competence. and rationality2 , respectively (Evans, 1993; Evans et al., 1993). Following the descriptive approach from Elqayam and Evans This distinction might explain errors, especially when individuals (2011) and Elqayam (2012), we aimed to determine which violate the normative rationality and persist with their personal behavior is rational for individuals to achieve their personal goals rationality to achieve their goals. Other distinctions of rationality in a given context. (Stanovich and West, 2000) focus on the difference between evolutionarily developed rationality and individual rationality. These two distinctions might not always correspond, especially Understanding the Effects Called in our modern world. “Cognitive Biases” In general, Elqayam and Evans (2011) express doubts about The term “bias” (as well as “debiasing”) itself implies that the prevailing role of normativism; this approach assumes that there is an undistorted and normative standard from which human rationality should be evaluated to the degree that it the behavior deviates (Kerr et al., 1996). Thus, the term in corresponds to a normative standard. Therefore, they proposed a itself is directly judgmental and as it is widely used, shows descriptivist paradigm, which “favors an instrumental, pragmatic the dominance of the normative research tradition in the field conception of rationality, based on achievements of one’s goals” of decision-making. The term evaluates behavior only from (Elqayam, 2012, p. 628). More specifically, Elqayam (2012) one perspective (normative rationality) and does not sufficiently presents a descriptive approach that focuses more on the actual consider alternative perspectives (i.e., instrumental rationality). behavior but goes beyond mere description. Descriptivism is Although we do not follow this prejudgment, the term “bias” is based on instrumental rationality, that is, the achievement used in the following article for two reasons. First, to maintain of personal goals and casts doubt on the supremacy of a connection with the previous research tradition at least in normative rationality. The latter is not completely excluded wording (e.g., Heuristics and Biases Program, Tversky and but integrated as a subcategory of instrumental rationality. Kahneman, 1974), and second, to clarify that behavior can Frontiers in Psychology | www.frontiersin.org 2 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality be “biased” and “useful” at the same time, depending on the According to Bach (2012), three different kinds of competence perspective. can be distinguished: epistemic competence (coping with any If we want to find the instrumental rationality in relation to specific task), general competence (overall ability to cope with cognitive biases, we need to clearly understand their function, the environment), and effect-related competence (ability to and the purpose they might have. Therefore, the goal should not have an effective impact on the environment). Across all three only be to eliminate or reduce biases but also to understand them variations, the need for competence is met through the successful and identify their negative and possible positive effects. Despite satisfaction of other needs, when the person experiences success, the fast-growing literature in the field of decision-making, there especially in demanding situations, and in general when they is still a lack of understanding on how to overcome or how to perceive competence signals. The need for competence is understand biases (Milkman et al., 2009). According to Fischhoff frustrated through failed attempts to satisfy needs in general, the (1982), the best way to reduce biases would be to have a better loss of active control, signals arising from failing, and signals of understanding of mental processes in general. This would add incompetence (Dörner et al., 2002). motivational and emotional aspects to the focus on cognition Dörner (1999) compares the function of these needs to a in decision research. There are interesting developments in reservoir, which has an inherently limited capacity. Positive this field, for example, the concept of “actively open-minded information fills the reservoir, and negative information leads to thinking” (Baron, 1993; Haran et al., 2013). Besides other aspects, a drain. When an individual interacts with the environment, the it describes a tendency during the decision process to weigh new incoming information is assessed. Depending on the assessment evidence against a favored belief. As discussed before, if people of the situation concerning aspects, such as importance, urgency, follow an instrumental rationality when making a decision, threat, subjective competence to cope with the situation, past they pursue personal goals. Therefore, we need a task design, history of the information, relevance, and anticipation of further wherein the variety of personal goals—including motivational development (Detje, 1999), the situation is rated as positive aspects—can be observed. Research on the sunk-cost effect, for or negative in relation to the level of the reservoir. In the instance, shows that most of the tasks utilized mainly dealt interaction with the environment, the actual level of the reservoir with cognitive aspects, and therefore, present a low potential might differ from the target level. If the level drops below the for triggering motivational and emotional processes (e.g., van target level, a need becomes active. The further and faster this Putten et al., 2010; Hafenbrack et al., 2014). In this regard, after level drops, the higher the pressure to satisfy the need. If the conducting a comprehensive overview of judgment and decision- assessment of the situation is rated as an actual or potential making research, Weber and Johnson (2009) suggested that it future threat to the need, an urge arises, and the process of need- is important to connect more of the research to theories of regulation is initiated. Overall, this process follows a homeostatic motivational and emotional processes. principle, aiming at balancing motivational needs in the dynamic environment (Bach, 2009). Satisfaction of the “need pressure” Psi: A Theory of the Human Mind is followed by a pleasure signal; whereas, a high need pressure To gain a better understanding of the human mind regarding generates a displeasure signal. Most of these processes occur at the motivational, emotional, and cognitive processes, a unifying the unconscious level. However, the individual can experience the theory is called for. The Psi theory (Dörner, 1999; Dörner et al., result consciously when it feels “good” or “bad.” 2002; Bach, 2009; Dörner and Güss, 2013) presents a holistic According to the Psi theory, there is a difference between architecture of the human mind. Unlike other architectures, such goal-oriented and need-oriented behavior. A goal is defined as as State, Operator, and Result (SOAR, Newell, 1987) and the a situation associated with a motivational value (Bach, 2009). Of Adaptive Control Thought-Rational theory (ACT-R, Anderson, course, every behavior is somehow goal-oriented and motivated 1990), the Psi theory integrates the emotional and motivational in some way. However, the distinction between goals and needs aspects into the cognitive system. As the theory is particularly in the Psi theory has another underlying meaning. A goal, in extensive, we focused on the concept of motivational needs general, describes an entity that directs the behavior. A goal within the theory. according to the Psi theory is best compared to an “objective In the Psi theory, five different needs are distinguished: goal” (“Sachziel,” Strohschneider, 2003). On the other hand, (1) existential needs (thirst-, hunger-, and pain avoidance), (2) need-oriented behavior refers to a behavior focused on the the need for sexuality, (3) the need for affiliation (positive satisfaction of needs. For example, a student has to finish a signals from others), (4) the need for certainty (predictability), research paper (Dörner and Güss, 2013). The paper represents and (5) the need for competence (active control). Of these the objective goal. Completing and submitting the thesis would five needs, we will focus on the need for competence. Several give the student a strong feeling of competence. However, this researchers from various fields view competence as a central goal feels far away when the student is sitting at his/her desk and need and a drive for human behavior (Bischof, 1987; Adler, struggles with the content, therefore receiving a lot of inefficacy- 1912; Deci and Ryan, 2000). The need for competence refers signals. Frustrated, he/she stops working on the paper and starts to having active control over a situation. This includes the doing the dishes instead. Washing the dishes does not bring extent to which a person feels capable of handling the problems him/her closer to finishing the paper but regulates the need for presented by their environment. This need also encompasses competence by creating a feeling of being effective in the short the need for power, control, or autonomy, and is connected run. In the long run, the remaining time to finish the paper runs to status, self-competence, and self-worth (Dörner et al., 2002). out. The orientation of the behavior changed from a long-term Frontiers in Psychology | www.frontiersin.org 3 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality goal-orientated objective (finishing the paper) to a short-term a scenario is that an option is selected to satisfy motivational need-oriented activity (getting efficacy-signals from doing the needs, rather than best fulfill the requirements of the situation. dishes). This also implies a short-term advantage (i.e., the feeling of Another approach that also sheds light on the underlying competence being maintained) and long-term disadvantage (i.e., mental processes that occur during the reasoning processes the actual goal not being reached). However, there is also the is found in the Meta-Reasoning framework of Ackerman and possibility of satisfactorily meeting both needs and present goals Thompson (2017). Particularly, the framework refers to the with a single decision, especially in cases when the goal equals processes that monitor the progress of reasoning and how the fulfillment of needs. As these cases usually do not lead well a process has unfolded. The states of these processes are to difficult decisions, because it is a win-win situation, we experienced as feelings of certainty or uncertainty. The level of focused on decisions where the goal and fulfillment of needs certainty experienced is an important aspect of the reasoning differed. We assume that in these cases, regulation can take place process. If a certain level of certainty is reached, a judgment is over the formation of cognitive biases, as they could serve the made. However, the level is not permanently set. According to (unconscious) preservation of needs or increase the levels in the duration of the task, the level can sink, thereby leading to the need reservoirs. Ignoring counterfactual information, over- judgments with lower requirements for certainty (Diminishing or underestimating probabilities, and staying with the familiar Criterion Model, Ackerman, 2014). During the task, a “feeling of option are all possible ways to regulate the needs (Dörner et al., rightness” is experienced if a first solution feels right (Thompson 2002; Dörner and Güss, 2013). et al., 2011; Ackerman and Thompson, 2017). In the Psi theory, the “feeling of rightness” refers to the level of confidence a person The Sunk-Cost Effect ascribes to a solution option and its anticipated potential for One of the best-known effects, which is considered a cognitive need-satisfaction. This concept is similar to the “good” or “bad” bias, is the sunk-cost effect. It is defined as a “tendency to feeling in Psi, which is the result of the need-satisfaction itself. continue an endeavor once an investment in money, effort, or In the Meta-Reasoning framework, certainty plays a central role time has been made” (Arkes and Blumer, 1985, p. 124). This when implementing an action. The Psi theory also determines the means that when individuals exhibit sunk-cost effect behavior, action depending on the need for certainty but additionally, it they persist with the option which they have already invested includes the need for competence and describes action as a result in and resist changing to another option that might be more of the interaction between both of these needs (Dörner and Güss, suitable regarding the future requirements of the situation 2013). (Arkes and Ayton, 1999; Hastie and Dawes, 2001). Moon (2001) distinguishes two different situations in which the sunk-cost Influence of Motivation on effect occurs. The first is described as “utilization decision,” a Decision-Making decision wherein the decision maker has to choose between Decision-making is part of the problem-solving process (Güss two equal alternatives. Roth et al. (2015) illustrated this type of and Robinson, 2014). The act of decision-making can be decision with the example of a person who purchased a ticket to described as the ability to choose one of several alternatives and a play at the local theater but later gets invited by a good friend to to act accordingly (Güss, 2004). Mostly, the goal of a decision lies a special Italian dinner. Even when the person prefers attending in the future (Hsee and Hastie, 2006). That means, when making the dinner, he/she thinks about the sunk cost already paid for the a good decision, an individual has to pick an alternative in the ticket and decides to go to the play. present moment, which best fulfills the future requirements of The second situation is as a “progress decision,” that describes an upcoming situation (Pronin et al., 2008). Güss et al. (2017) a situation where the decision hinges upon whether or not a emphasize the role of motivation in complex problem-solving; chosen pathway should be continued or not. An example of this consequently, motivation also has an influence on decision- type of decision is found in the building of the supersonic plane making. Concorde. Already in the early development stages, the costs Decisions do not take place in a vacuum; the needs are rose higher than expected, and the financial success was unclear. influenced by specific characteristics of the situation (Dörner, Nevertheless, the project was not stopped but further funded 1999). Depending on the subjective assessment, the situation because of the amount of money that was already invested (Arkes is rated as positive or negative with respect to the needs, and and Ayton, 1999). Experiences in economics do not help to therefore, has an influence on the levels in the reservoirs. sustain the sunk-cost effect. A meta-analysis by Roth et al. (2015), Subsequently, the individual also looks for aspects in the options, which involved solely monetary sunk-cost decisions, revealed which could mean the fulfillment of their actual needs. Thus, that an economic background on the part of the participants does the available options are not only assessed with regard to their not have a significant influence on utilization decisions. goal-reaching potential but also by their need-fulfilling potential. There are different approaches to explain the occurrence of the However, this process does not necessarily have to be conscious. sunk-cost effect. van Putten et al. (2010) differentiated between Errors can arise when the logic of acting changes from goal- individuals with a “state orientation,” who struggle to let go oriented rationality to need-oriented rationality (Dörner, 1996); of past events, and individuals with an “action orientation,” that is, when the individual unconsciously shifts regarding the who seem relatively untroubled by past events. They found that decision, moving from the original objective to a sole focus on state-oriented decision-makers were more prone to exhibit the the satisfaction of one or several needs. The consequence of such sunk-cost effect. Other researchers bring in further aspects, such Frontiers in Psychology | www.frontiersin.org 4 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality as people not wanting to appear wasteful (Arkes and Blumer, deciding for others and to adopt this normative rationality when 1985), the effect of mental accounting (Thaler, 1985), or the deciding for themselves. Therefore, this change of perspective escalation of commitment (Schaubroeck and Davis, 1994; Carter should draw attention to aspects that are important for reaching et al., 2007). Most of the time, sunk-cost effect studies involve the objective goals of a situation, and less on subjective aspects hypothetical scenarios. In some cases, a situation is described which are only important from the view of the decision-maker. which places the participants in a position where they have to Thus, the intervention should ultimately lead to a more goal- make a decision (“Imagine you are the CEO of...”) (see van oriented decision (Strohschneider, 2003). As a general point, Putten et al., 2010; Hafenbrack et al., 2014) concerning whether Strough et al. (2016) highlight the importance of understanding a course of action or project they have already invested in should the mental models and existing beliefs of participants when be continued and, therefore, supported financially, even when a designing an intervention. competitor presents a more promising solution. In these cases, participants often have to make a decision involving millions Aims of the Present Study of theoretical dollars of investment. However, most of these The current theoretical background suggests that there is still cases seem rather artificial, as ‘normal’ participants suddenly are little understanding of how and why these effects, in our example required to imagine that they are a CEO with great responsibility. the sunk-cost effect, occur. However, a better understanding Moreover, most of these cases only address money (Roth et al., would have significant positive effects, leading to better decision- 2015), and focus less, if at all, on time or effort. Even when they making (up to 7% higher return after reducing the effect of do, it is only within a hypothetical frame. biases in business decisions, Kahneman et al., 2011). Therefore, the present study aims to address the following questions: which goals do people follow when they demonstrate instrumental Self-Reflection as a “Debiasing” rationality (Evans and Over, 1996), and what role does the Intervention unconscious need regulation play in the formation of these There are various approaches that could improve goals? What would a structured experimental design that aims decision-making, preventing decision-makers from committing to capture the complexity of a real-life situation look like? the so-called cognitive biases (Soll et al., 2015). These approaches (Dörner and Funke, 2017). Finally, can an intervention influence can be summed up as “debiasing” interventions. As discussed the rationalities that the decision-maker follows? We conducted before, the problem with the term “bias” also concerns the term a study designed to analyze participants’ decisions in a more “debiasing” which is used to describe interventions that aim to realistic sunk-cost scenario. Participants were asked to choose eliminate biases. This process should eventually lead to a decision between a sunk-cost option (SCO) and an alternative option outcome which corresponds to a normative rationality. However, (AO), whereby one group received an intervention with the aim despite the focus on normative rationality, biases can also make of interrupting the choice of the SCO. the decision-maker aware of the different rationalities he or she We derived the following hypotheses: biased decisions might follow in a given situation. (i.e., taking the SCO) have their root in unconscious need One of these debiasing interventions is to initiate a process of regulation, triggered by sunk costs. As proposed in the Psi self-reflection to unveil biases running at the unconscious level theory (Dörner, 1999; Bach, 2009; Dörner and Güss, 2013), we (Donovan et al., 2015; Phillips et al., 2016). Hafenbrack et al. assumed that need regulation is positively linked to choosing (2014) found that including a meditation-based intervention to the SCO (Hypothesis 1). Accordingly, we hypothesized that prevent the sunk-cost effect seems to be an effective approach to participants who choose the SCO show lower levels of self- lead participants to focus more on the actual situation, and less reflection (Hypothesis 2a), achievement motives, and self-control on the past, where the sunk costs took place. Another approach (Hypothesis 2b). Research indicates that biases can be prevented to initiate a process of self-reflection is to let decision-makers by inducing a process of self-reflection (Hsee and Weber, 1997; assess the situation from an external point of view. There is a Kahneman et al., 2011; Hong and Chang, 2015). Consequently, difference between making decisions for ourselves and for others. we hypothesized that an intervention fostering self-reflection Various studies on the concept of self-other decision-making have prevents decision-makers from choosing the SCO (Hypothesis 3). shown that individuals making a decision for others focus on fewer attributes and make the decision more readily (Kray and Gonzalez, 1999), focus less on feelings (Hsee and Weber, 1997; MATERIALS AND METHODS Hong and Chang, 2015), and make more risk-averse decisions when risk-aversion is pertinent to the situation (Stone et al., To motivate the participants, we created a more “involving” 2013). A debiasing intervention designed by Kahneman et al. experimental scenario to test the hypothesized mechanism, in (2011) aims at exhibiting this effect, as the individual is forced to which all three aspects of the sunk-cost effect were present. evaluate the decision from an external point of view (“How would Participants were required to work on their own project (effort), a new CEO decide?”). In the case of the sunk-cost effect, this into which they invested their own time and money, and had to should disclose unrelated motivational factors, as the decision- decide whether to persist with their own project or choose an maker has to analyze the decision from an external point of view. alternative and more promising option, in an attempt to win the In our study, the aim of the debiasing intervention was to make remaining sum of their investment. Following the understanding participants aware of the normative rationality they follow when of Greitemeyer et al. (2005), when participants chose their own Frontiers in Psychology | www.frontiersin.org 5 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality option, in which they had already invested, even when the AO that I’m successful with it;... generally a positive feeling); and offered a higher chance of winning money, their choice was rated “My solution/the alternative solution...” (...gives my self-esteem as a sunk-cost effect. a positive feeling;...gives me the feeling of being able to do Against the background of the Psi theory concerning something actively;...makes me feel self-determined). Items were psychological needs, we aimed to unveil the underlying scored on a seven-point Likert scale ranging from 1 (strongly mechanisms and reasons why people choose the SCO. The disagree) to 7 (strongly agree). question was: can we observe need-regulation through the The alternative construction was coded with a 50% chance actions of participants who choose the SCO (SCO-Selectors) of winning. Pre-tests showed a 30% chance of winning when reporting greater need-satisfaction than participants who choose the participants’ own constructions were used. The functionality the Alternative Option (AO-Selectors)? Moreover, we tested an of the constructions was physically tested at the end of the intervention (Kahneman et al., 2011) designed to induce a change experiment. of perspective regarding the decision and reduce the sunk-cost effect, with the aim of generating a more goal-oriented, and less need-oriented, decision. Procedures The researcher welcomed the participants and led them to Participants their seats. A maximum of three participants was tested at From a total of 138 adult participants who were recruited for the same time. The working spaces were divided by partitions the experiment, 13 were excluded due to missing data (n = 2), to prevent any communication. After a short introduction, in unfinished construction (n = 4), misunderstanding of the task, which participants were asked not to use their phones or talk error in the experimental flow (n = 3), and knowledge of the task to each other, they were required to wear earplugs. Afterward, (n = 4). Another 19 participants from the pre-testing phase were participants were randomly assigned to the group with or without also excluded, as the task was slightly adapted. The remaining 106 the debiasing intervention. participants (40 men and 66 women; mean age = 21.75 years; In both conditions, participants completed a demographic SD = 3.0; range = 18–37 years, 95% psychology students) were survey, the SRIS from Grant et al. (2002), the SCS-K-D from included in the analysis. Participants were recruited from the Bertrams and Dickhäuser (2009), and the Achievement Motives campus of the University of Innsbruck. Scale from Lang and Fries (2006). Then, participants were assigned a decision-making task, adapted from the cases of Arkes Materials and Blumer (1985) and van Putten et al. (2010). In this case, Participants completed the Self-Control Scale (SCS-K-D) participants were asked to take the role of the project leader in (α = 0.82) from Bertrams and Dickhäuser (2009) (13 items). a company that produced packing solutions. The goal was to They also completed the Self-Reflection and Insight Scale (SRIS) design and build a construction, within 30 min, which could (Grant et al., 2002), which is divided into the Self-Reflection protect a raw egg when dropped from a height of 3 m. For this Scale (α = 0.85) (12 items) and the Insight Scale (α = 0.84) (eight task, participants were provided with 15€ and presented with a items). The Achievement Motives Scale (Lang and Fries, 2006), selection of different materials that they could purchase [a straw (10 items), which is divided into two dimensions, hope of success (0.50€), 0.5 m of rope (2€), 4 cm of tape (0.50€), one napkin (five items, α = 0.77) and fear of failure (five items, α = 0.81), was (0.50€), a plastic cup with a volume of 0.2 l (2.50€), a DinA4 paper used to identify the motives of participants. The Construction (1€), a balloon (2€), a plastic bag (2.50€), a cotton pad (0.50€)]. Task was adapted from the classical sunk-cost cases provided They were allocated 10 min to think about their construction and by Arkes and Blumer (1985) and van Putten et al. (2010). In buy the materials they would need. There was no restriction on the intervention, participants had to decide for a fictional ‘other how many units of an item they could purchase. The incentive person’ to use the construction that the participant has made to build economic solutions was that participants could keep any or the alternative construction (“Please put yourself in the role funds remaining in their budget after purchasing the materials of a person who has not worked on your construction and but only if their solution worked (i.e., if the egg survived the has to make the same decision as you. In your opinion, which 3 m drop). For example, with an investment of 10€ in materials, construction should the person take?”). The intervention and they could earn 5€ if the egg survived the drop. Subsequently, the actual decision were binary-coded (adopting the self-built participants’ understanding of the conditions of the task, and the construction versus taking the alternative construction). possibility of winning money was checked using a short survey. The assessment of Competence Satisfaction (α = 0.90, seven After the 10-min planning and purchasing phase, the 20- Items) was adapted from Bach (2009) and Dörner and Güss min construction phase began. In this phase, participants once (2013), and the estimated probability of the success of the again had an opportunity to order new materials. At the end construction was rated on a number scale ranging from 0 to 100% of the construction phase, participants were notified that there (where 0% equaled no chance of winning the money with the was another type of construction available, which they could use construction, and 100% equaled a safe win). for the drop. They could not view the construction but were The items were preceded by the sentences: “With my informed that this alternative construction would perform better solution/the alternative solution, I associate...” (...the feeling than the average self-created construction. They then had to of being in control of this situation;... the feeling of being weigh up the possibility of success of their own construction set effective in my actions;... a positive feeling, because I think against this other alternative. Frontiers in Psychology | www.frontiersin.org 6 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality The control group had to decide which of the two constructions they would like to use for the drop: their own, on which they had planned and worked for 30 min and had invested money, or the alternative construction. The experimental group, however, received a short intervention (adapted from Kahneman et al., 2011) before their choice. They had to consider what a new project leader, who had not worked on either of the solutions, should choose. They noted down the reasons for their decision regarding this extra choice. After making this decision, the participants of the experimental group then had to make the choice for themselves. Subsequently, both groups assessed the two solutions for their Competence Satisfaction. After assessing the constructions and FIGURE 1 | Rating of competence satisfaction for the two options. choosing between them, the functionality test was conducted. If participants chose the AO, a generator with a 50% chance of winning was used to decide if they could keep the rest of their budget. If participants chose their own construction, the Concerning the success probability of the option, which was researcher dropped their construction from a height of 3 m. The assessed on a percentage scale (where 0% meant no chance remaining budget was given to them only if the egg survived the of protecting the egg, and 100% equaled an effective, safe drop. construction), in both groups, participants assessed the AO as significantly safer (M = 64.06, SD = 15.37) than the SCO Design (M = 52.38, SD = 21.90, t = −6.42, p < 0.001, N = 106, r = 0.53). Participants were assigned to conditions in a 2 × 2 design This effect could also be seen when the success probability was (experimental group/control group × sunk-cost option analyzed depending on the decision (see Figure 2). Following selectors/alternative option selectors). The experiment was expectations, a Wilcoxon signed-ranks test indicated that the designed to test whether an intervention inducing a self- AO-Selectors rated the AO significantly higher (median = 61.00) reflection process leads to a reduced selection of the SCO in than the SCO (median = 26.00, z = −3.30, p < 0.001, n = 14). comparison to the control group without an intervention. According to Cohen’s effect size (1992), r = 0.88 indicates a Moreover, after the choice was made, the reasons for the strong effect. Surprisingly, the SCO-Selectors also rated the AO decision were assessed on a need-regulative level. Responses significantly higher (M = 64.27, SD = 15.88) than their own to scales for self-control, self-reflection, and achievement chosen SCO (M = 56.13, SD = 20.67, t = −4.68, p < 0.001, n = 91, motivation were obtained to check for differences between the r = 0.19). groups. The SCO-Selectors assessed the SCO with a higher success probability (median = 59.50) than did the AO-Selectors (median = 26.00, U = 161.5, p < 0.001). According to Cohen RESULTS (1992), this indicates a medium effect (r = 0.44). There were no significant differences between the To check Hypothesis 1, which assumed that need regulation experimental and control groups with respect to the scores is positively linked to choosing the SCO, the assessment of on the SRIS [Levene-test: F(1,104) = 1.417, p = 0.237, N = 106] Competence Satisfaction was analyzed (Figure 1). Interestingly, (Hypothesis 2a). The Self-Control Scale [t(104) = 0.107, the SCO-Selectors assessed the SCO as more satisfactory in p = 0.915] and the Achievement Motives Scale for the terms of competence (median = 4.00) than did the AO-Selectors dimensions ‘hope of success’ [t(104) = 1.594, p = 0.114] (median = 2.86, U = 201.5, p < 0.001). According to Cohen and ‘fear’ [t(104) = 0.527, p = 0.599] also showed no differences (1992), this indicates a medium effect (r = 0.40). A significant between the groups (Hypothesis 2b). The amount of investment difference was also found in the assessment of the AO. The AO- in the construction and the decision revealed no significant Selectors rated the AO higher (median = 2.86) than did the findings (χ2 = 0.354, p = 0.552, df = 1). SCO-Selectors (median = 1.86, U = 244.5, p < 0.001). According Moreover, the analysis did not reveal any significant to Cohen (1992), this indicates a medium effect (r = 0.36). differences between the SCO-Selectors and the AO-Selectors This difference was not only found between the two groups regarding the SRIS (U = 461.0, p = 0.087, r = 0.17), the Self- but also within the group of SCO-Selectors, who rated the SCO Control Scale (U = 595.5, p = 0.651, r = 0.04), or the Achievement significantly higher (M = 4.01, SD = 0.86) than the AO (M = 2.07, Motives Scale in the dimensions ‘hope of success’ (U = 556.5, SD = 0.90, t = 15.46, p < 0.001, n = 92, r = 0.85). No such p = 0.410, r = 0.08) and ‘fear’ (U = 441.5, p = 0.058, r = 0.18). difference was found in the assessment of the AO-Selectors. To check for the effect of the intervention (Hypothesis 3), Running a Wilcoxon signed-rank test, no significant differences we took a closer look at participants in the experimental group. were found between the ratings of the SCO (median = 2.94) and Before deciding on whether to choose their own option (SCO) those of the AO (median = 2.95, z = −0.189, p = 0.850, n = 14, or the AO, they had to decide on behalf of an external person, r = 0.05). who was not involved in the process. The majority chose the Frontiers in Psychology | www.frontiersin.org 7 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality To take a deeper look into the reasons behind the decisions made, we analyzed the choice from a psychological needs perspective, based on the Psi theory (Dörner et al., 2002; Bach, 2009). The focus of this analysis was on the need for competence. The need for competence describes one’s perception of active control over the environment. Signals of effectiveness add a positive value in terms of competence; signals of ineffectiveness do the opposite, decreasing the levels in the competence “reservoir” (Dörner and Güss, 2013). Having invested in vain in a particular option would send a significant signal of ineffectiveness to the participant. Therefore, choosing the SCO would protect the individual from this FIGURE 2 | Subjective probability of success for the two options. negative feeling, in the short term (Dörner, 1996). Taking this into account, we analyzed if need-regulation can better explain why people make economically “irrational” decisions AO (86.5%, n = 45), rather than the SCO (13.5%, n = 7, exact (Simon, 1955), assuming a mechanism of need-regulation, which binominal test, two-sided, p < 0.001, n = 52). leads to biased decisions, from the normative point of view However, when making the actual decision (which option (Dörner, 1996, 1999; Strohschneider, 2003; Dörner and Güss, to use for the egg test), the groups showed similar results. In 2013). From this perspective, the decision-maker favors the the experimental group, 45 participants chose the SCO (86.5%) option which adds most value concerning their psychological and only seven, the AO (13.5%). Running a two-sided binomial needs, rather than the option which is likely to maximize test revealed a significant deviation from the expected 50% their chances of winning. From the viewpoint of rationality, distribution (p < 0.001, n = 52). In the control group, a similar participants followed an instrumental rationality when choosing picture was observed: 47 participants chose the SCO (87%) and an option. seven, the AO (13%). A two-sided binominal test also showed The most remarkable result was that when participants had a significant deviation from the expected 50% distribution to make the decision, it did not matter which option was (p < 0.001, n = 54). A binominal test was conducted taking 87% of more promising regarding likely returns. This directly violates the control group as a reference point; the analysis did not show the expectation that an individual will act according to the any significant difference between the two groups (p = 0.522, homo oeconomicus model (Simon, 1955). Participants rather n = 52). chose the option which potentially maintained or boosted Taken together, the results show that 33.7% of the 92 their feeling of competence. Results from the list of reasons SCO-Selectors were successful and won an average of 4.76€ for their choice indicated that the SCO-Selectors made their (median = 5.00); 71.4% of the 14 AO-Selectors were successful choice to check their effectiveness in building the construction. and won an average of 4.05€ (median = 4.00). The difference was However, the AO-Selectors lacked such confidence in their not significant (U = 107.00, p = 0.136). construction. When they expected that their construction would fail to protect the egg, using the SCO equaled a possible threat to their feelings of competence. The data showed that DISCUSSION in this case, participants chose the AO because this option gave them a higher chance of winning while also avoiding the In the present study, we placed participants in a situation potential signal of ineffectiveness from the failure of their own where they were confronted with two options; one of these construction. was the so-called SCO (Arkes and Blumer, 1985; Greitemeyer The intervention that aimed at changing participant’s et al., 2005). With the materials that could be purchased from rationality by inducing self-reflection (Donovan et al., 2015) a given budget, the aim was to build a construction to protect worked in the first stage (Kahneman et al., 2011); participants a raw egg from a 3-m drop. They could only “win” any reported that the other person should take the AO, given money left over from the budget if the egg survived the drop it was likely the “better” one. In the second stage, however, without damage. After building their own construction for the when they had to choose for themselves, this prior normative drop-test, participants were confronted with an AO, with a rational insight did not prevent them from choosing their own, reportedly higher possibility of securing the remaining money. reportedly less successful, option. According to Evans and Over Subsequently, they had to decide whether they wanted to use (1996), participants showed more normative rationality (acting their own construction or the alternative construction for the according to the homo oeconomicus model, selecting the option drop. With either option, participants had the chance to win the with the maximum chance of winning) when choosing for others remaining money from the budget. If the participants persisted but switched to an instrumental rationality (achieving personal with their own option, on which they had invested time, money, goals) when choosing for themselves. As participants reported a and effort, despite being informed that this would generate, on higher need satisfaction from their chosen option, we suppose an average, a lower chance of winning, it was rated a “sunk-cost that regulation of the need for competence might be an important effect.” aspect concerning the formation of personal goals. Frontiers in Psychology | www.frontiersin.org 8 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality Limitations and Outlook increase their feeling of competence. Is the sunk-cost effect a The intervention showed no impact on preventing participants bias? Yes and no. From a normative point of view, involving from choosing the SCO. Even when making the normative sunk costs into a decision is a deviation from the normative rational choice for someone else, they still picked the SCO when model, and therefore, a “bias.” However, this only applies from they made the decision for themselves. The reason could be the normative perspective. From the viewpoint of instrumental that the intervention did not involve the satisfaction of needs. rationality, the sunk-cost effect serves personal goals which can Even when participants were able to reflect on the situation differ from the normative standards. Therefore, reflecting back on consciously, they still had an (unconscious) urge to regulate the framework of grounded rationality (Elqayam, 2012), a “bias” their needs. An intervention designed to satisfy needs before the can be seen as a behavior that is rational to the decision maker decision is made regarding their own choice might lead to better in an epistemic context, if the “bias” facilitates the achievement results. of the decision maker’s goals within his or her physical or mental As proportionally so many participants chose their own limitations. option, we had to use some non-parametric tests. Our According to Strohschneider (2003), we can observe a shift explanation for this imbalance lies in the task itself. In artificial from a goal-oriented behavior to a need-regulation oriented cases, more people choose the rational option (for instance, behavior. Typically, this shift should not pose a problem but when between 29 and 44% resisted the sunk-cost bias in the control the decision-maker is not aware of it, he or she might gain a short- groups of Hafenbrack et al., 2014). However, in our more term regulation of the need but fails to maximize their chances of “realistic” scenario, persisting with their own SCO seemed to have attaining their objective in the long run. Additionally, the goals a stronger pull. It would be very interesting for future research to arising from the need-regulation do not necessarily have to be examine whether there is a financial threshold where participants conscious to the decision-maker. Therefore, the final question would be more influenced by the option with a higher possibility would be: even when the exhibited behavior is in some way of winning than the regulation of needs. Future research could rational, is it the way I want to or should act in the given situation? conduct a systematic research on various cognitive biases and the need-regulative function during the decision process. The presented findings are correlational and are based on self- ETHICS STATEMENT report measures. Therefore, it remains unclear if need-regulation leads to the sunk-cost effect, the sunk-cost effect triggers the This study was carried out in accordance with the need-regulation, or both. Future research should include these recommendations of “The Board for Ethical Questions in Science considerations. of the University of Innsbruck” with written informed consent from all participants. All participants gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the “The Board for Ethical Questions CONCLUSION in Science of the University of Innsbruck.” Is the observed behavior irrational? The answer depends on the point of view; from a normative view, some participants AUTHOR CONTRIBUTIONS acted irrationally when they did not choose the option which maximized their chance of winning. However, when taking the MD and BS conducted the data collection, data analysis, and the instrumental rationality (Evans and Over, 1996) into account, writing of the manuscript. All authors were substantially involved there could be a different interpretation. Instrumental rationality in the planning of the study, the interpretation of data, and states that a person acts to achieve his or her goals. In the revision of the article. given situation, participants worked on their solution, invested time, money, and effort. Being confronted with the thought of having done this in vain might be a strong negative signal to FUNDING their competence. Subsequently, along with the goal to choose the option with the highest chance of winning, the goal to This research was funded by the “Studienstiftung des Deutschen regulate the need for competence also arises. Therefore, choosing Volkes” [German Academic Scholarship Foundation], the their own option, even when it had a lower chance of winning “Austrian Marshall Plan Foundation,” and the University of the money, might serve a personal goal: to maintain or even Innsbruck. REFERENCES Adler, A. (1912). Über den nervösen Charakter. Grundzüge einer vergleichenden Individualpsychologie und Psychotherapie [The Neurotic Constitution. Outlines Ackerman, R. (2014). The diminishing criterion model for metacognitive of a Comparative Individualistic Psychology and Psychotherapy]. Wiesbaden: J. F. regulation of time investment. J. Exp. Psychol. Gen. 143, 1349–1368. Bergmann. doi: 10.1037/a0035098 Anderson, J. R. (1990). The Adaptive Character of Thought. Hillsdale, NJ: Erlbaum. Ackerman, R., and Thompson, V. A. (2017). Meta-reasoning: monitoring and Arkes, H. R., and Ayton, P. (1999). The sunk cost and concorde effects: are humans control of thinking and reasoning. Trends Cogn. Sci. 21, 607–617. doi: 10.1016/ less rational than lower animals? Psychol. Bull. 125, 591–600. doi: 10.1037/0033- j.tics.2017.05.004 2909.125.5.591 Frontiers in Psychology | www.frontiersin.org 9 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality Arkes, H. R., and Blumer, C. (1985). The psychology of sunk cost. Organ. Behav. Greitemeyer, T., Schulz-Hardt, S., Popien, G., and Frey, D. (2005). Der Einfluss Hum. Decis. Process. 35, 124–140. doi: 10.1016/0749-5978(85)90049-4 versunkener monetärer und zeitlicher Kosten auf Ressourcenallokationen: eine Bach, J. (2009). Principles of Synthetic Intelligence PSI: An Architecture of Studie zum Sunk-Cost-Effekt mit Experten [The impact of sunken costs on Motivated Cognition. Oxford: Oxford University Press. doi: 10.1093/acprof: resource allocations: a study on the sunk cost effect with experts]. Zeitschrift oso/9780195370676.001.0001 Arbeits und Organisationspsychologie A&O 49, 35–43. doi: 10.1026/0932-4089. Bach, J. (2012). A framework for emergent emotions, based on motivation and 49.1.35 cognitive modulators. Int. J. Synth. Emot. 3, 43–63. doi: 10.4018/jse.2012010104 Güss, C. D. (2004). Decision making in individualistic and collectivistic cultures. Baron, J. (1993). Why teach thinking? An essay. Appl. Psychol. 42, 191–214. Online Read. Psychol. Cult. 4:3. doi: 10.9707/2307-0919.1032 doi: 10.1111/j.1464-0597.1993.tb00731.x Güss, C. D., Burger, M. L., and Dörner, D. (2017). The role of motivation in Bertrams, A., and Dickhäuser, O. (2009). Messung dispositioneller Selbstkontroll- complex problem solving. Front. Psychol. 8:851. doi: 10.3389/fpsyg.2017.00851 Kapazität: eine deutsche Adaptation der Kurzform der Self-Control Scale (SCS- Güss, C. D., and Robinson, B. (2014). Predicted causality in decision KD) [Measurement of dispositional self-control capacity: a German adaptation making: the role of culture. Front. Psychol. 5:479. doi: 10.3389/fpsyg.2014. of the short form of the self-control scale (SCS-KD)]. Diagnostica 55, 2–10. 00479 doi: 10.1026/0012-1924.55.1.2 Hafenbrack, A. C., Kinias, Z., and Barsade, S. G. (2014). Debiasing the mind Bischof, N. (1987). Das Rätsel Ödipus [The Enigma Oedipus]. Munich: Piper. through meditation: mindfulness and the sunk-cost bias. Psychol. Sci. 25, Camerer, C., and Lovallo, D. (1999). Overconfidence and excess entry: an 369–376. doi: 10.1177/0956797613503853 experimental approach. Am. Econ. Rev. 89, 306–318. doi: 10.1257/aer.89.1.306 Haran, U., Ritov, I., and Mellers, B. A. (2013). The role of actively open-minded Carter, C. R., Kaufmann, L., and Michel, A. (2007). Behavioral supply management: thinking in information acquisition, accuracy, and calibration. Judgm. Decis. a taxonomy of judgment and decision-making biases. Int. J. Phys. Distrib. Logist. Mak. 8, 188–201. Manag. 37, 631–669. doi: 10.1108/09600030710825694 Hastie, R., and Dawes, R. M. (2001). Rational Choice in an Uncertain World. Cohen, J. (1992). A power primer. Psychol. Bull. 122, 155–159. doi: 10.1037/0033- London: Sage. 2909.112.1.155 Hong, J., and Chang, H. H. (2015). “I” follow my heart and “We” rely on reasons: Cohen, L. J. (1981). Can human irrationality be experimentally demonstrated? the impact of self-construal on reliance on feelings versus reasons in decision Behav. Brain Sci. 4, 317–370. doi: 10.1017/S0140525X00009092 making. J. Consumer Res. 41, 1392–1411. doi: 10.1086/680082 Deci, E. L., and Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Hsee, C. K., and Hastie, R. (2006). Decision and experience: why don’t we choose human needs and the self-determination of behavior. Psychol. Inq. 11, 227–268. what makes us happy? Trends Cogn. Sci. 10, 31–37. doi: 10.1016/j.tics.2005. doi: 10.1207/S15327965PLI1104_01 11.007 Detje, F. (1999). Handeln erklären: Vergleich von Theorien menschlichen Handelns Hsee, C. K., and Weber, E. U. (1997). A fundamental prediction error: self-other und Denkens [Explainig Action: Comparison of Theories of Human Action and discrepancies in risk preference. J. Exp. Psychol. 126, 45–53. doi: 10.1037/0096- Thought]. Wiesbaden: DUV. doi: 10.1007/978-3-663-08224-8 3445.126.1.45 Donovan, S. J., Güss, C. D., and Naslund, D. (2015). Improving dynamic decision Kahneman, D., Lovallo, D., and Sibony, O. (2011). Before you make that big making through training and self-reflection. Judgm. Decis. Mak. 10, 284–295. decision. Harvard Bus. Rev. 89, 50–60. Dörner, D. (1996). The Logic of Failure: Recognizing and Avoiding Error in Complex Kerr, N. L., MacCoun, R. J., and Kramer, G. P. (1996). Bias in judgment: comparing Situations. New York, NY: Basic Books. individuals and groups. Psychol. Rev. 103, 687–719. doi: 10.1037/0033-295X. Dörner, D. (1999). Bauplan für eine Seele [Blueprint for a Soul]. Reinbeck, IA: 103.4.687 Rowolt. Kray, L., and Gonzalez, R. (1999). Differential weighting in choice versus advice: Dörner, D., Bartl, C., Detje, F., Gerdes, J., Halcour, D., Schaub, H., et al. (2002). Die I’ll do this, you do that. J. Behav. Decis. Mak. 12, 207–217. doi: 10.1002/(SICI) Mechanik des Seelenwagens [The Mechanics of the Soul Car]. Bern: Huber. 1099-0771(199909)12:33.0.CO;2-P Dörner, D., and Funke, J. (2017). Complex problem solving: what it is and what it Lang, J. W., and Fries, S. (2006). A revised 10-item version of the achievement is not. Front. Psychol. 8:1153. doi: 10.3389/fpsyg.2017.01153 motives scale. Eur. J. Psychol. Assess. 22, 216–224. doi: 10.1027/1015-5759.22. Dörner, D., and Güss, C. D. (2013). PSI: a computational architecture of cognition, 3.216 motivation, and emotion. Rev. Gen. Psychol. 17, 297–317. doi: 10.1037/ Magrabi, A., and Bach, J. (2013). “Entscheidungsfindung [Decision Making],” in a0032947 Handbuch Kognitionswissenschaften [Handbook of Cognitive Science], eds A. Elqayam, S. (2012). Grounded rationality: descriptivism in epistemic context. Stephan and S. Walter (Stuttgart: Verlag J. B. Metzler), 274–288. Synthese 189, 39–49. doi: 10.1007/s11229-012-0153-4 Milkman, K. L., Chugh, D., and Bazerman, M. H. (2009). How can decision making Elqayam, S., and Evans, J. S. B. T. (2011). Subtracting “ought” from “is”: be improved? Perspect. Psychol. Sci. 4, 379–383. doi: 10.1111/j.1745-6924.2009. descriptivism versus normativism in the study of human thinking. Behav. Brain 01142.x Sci. 34, 233–248. doi: 10.1017/S0140525X1100001X Moon, H. (2001). Looking forward and looking back: integrating completion Evans, J. S. B. T. (1993). “Bias and rationality,” in Rationality: Psychological and and sunk-cost effects within an escalation-of-commitment progress decision. Philosophical Perspectives, eds K. I. Manktelow and D. E. Over (London: J. Appl. Psychol. 86, 104–113. doi: 10.1037/0021-9010.86.1.104 Routledge), 6–30. Newell, A. (1987). Unified Theories of Cognition. Cambridge, MA: Harvard Evans, J. S. B. T. (2005). “Deductive reasoning,” in The Cambridge Handbook of University Press. Thinking and Reasoning, eds K. J. Holyoak and R. G. Morrison (New York, NY: Nutt, P. C. (2002). Why Decisions Fail: Avoiding the Blunders and Traps that Lead Cambridge University Press), 169–184. to Debacles. San Francisco, CA: Berrett-Koehler Publishers. Evans, J. S. B. T., and Over, D. E. (1996). Rationality and Reasoning. Hove: Phillips, W. J., Fletcher, J. M., Marks, A. D. G., and Hine, D. W. (2016). Thinking Psychology Press. styles and decision making: a meta-analysis. Psychol. Bull. 142, 260–290. Evans, J. S. B. T., Over, D. E., and Manktelow, K. I. (1993). Reasoning, decision doi: 10.1037/bul0000027 making and rationality. Cognition 49, 165–187. doi: 10.1016/0010-0277(93) Pronin, E., Olivola, C. Y., and Kennedy, K. A. (2008). Doing unto future selves as 90039-X you would do unto others: psychological distance and decision making. Pers. Evans, J. S. B. T., and Pollard, P. (1981). On defining rationality unreasonably. Soc. Psychol. Bull. 34, 224–236. doi: 10.1177/0146167207310023 Behav. Brain Sci. 4, 335–336. doi: 10.1017/S0140525X00009146 Roth, S., Robbert, T., and Straus, L. (2015). On the sunk-cost effect in economic Fischhoff, B. (1982). “Debiasing,” in Judgment under Uncertainty: Heuristics decision-making: a meta-analytic review. Bus. Res. 8, 99–138. doi: 10.1007/ and Biases, eds D. Kahneman, P. Slovic, and A. Tversky (Cambridge: s40685-014-0014-8 Cambridge University Press), 422–444. doi: 10.1017/CBO97805118094 Schaubroeck, J., and Davis, E. (1994). Prospect theory predictions when escalation 77.032 is not the only chance to recover sunk cost. Organ. Behav. Hum. Decis. Process. Grant, A. M., Franklin, J., and Langford, P. (2002). The self-reflection and insight 57, 59–82. doi: 10.1006/obhd.1994.1004 scale: a new measure of private self-consciousness. Soc. Behav. Pers. 30, 821–835. Simon, H. A. (1955). A behavioral model of rational choice. Q. J. Econ. 69, 99–118. doi: 10.2224/sbp.2002.30.8.821 doi: 10.2307/1884852 Frontiers in Psychology | www.frontiersin.org 10 May 2018 | Volume 9 | Article 815 Domeier et al. Sunk-Cost Effect’s Instrumental Rationality Soll, J. B., Milkman, K. L., and Payne, J. W. (2015). “A user’s guide to debiasing,” in Thompson, V. A., Prowse Turner, J. A., and Pennycook, G. (2011). Intuition, The Wiley Blackwell Handbook of Judgment and Decision Making, eds G. Keren reason, and metacognition. Cogn. Psychol. 63, 107–140. doi: 10.1016/j.cogpsych. and G. Wu (Chichester: John Wiley & Sons, Ltd). 2011.06.001 Stanovich, K. E., Toplak, M. E., and West, R. F. (2008). The development of Tversky, A., and Kahneman, D. (1974). Judgment under uncertainty: heuristics and rational thought: a taxonomy of heuristics and biases. Adv. Child Dev. Behav. biases. Science 185, 1124–1131. doi: 10.1126/science.185.4157.1124 36, 251–285. doi: 10.1016/S0065-2407(08)00006-2 van Putten, M., Zeelenberg, M., and van Dijk, E. (2010). Who throws good money Stanovich, K. E., and West, R. F. (2000). Individual differences in reasoning: after bad? Action vs. state orientation moderates the sunk cost fallacy. Judgm. implications for the rationality debate? Behav. Brain Sci. 23, 645–665. Decis. Mak. 5, 33–36. doi: 10.1017/S0140525X00003435 Weber, E. U., and Johnson, E. J. (2009). Mindful judgment and decision making. Stone, E. R., Choi, Y., de Bruin, W. B., and Mandel, D. R. (2013). I can take the risk, Annu. Rev. Psychol. 60, 53–85. doi: 10.1146/annurev.psych.60.110707.163633 but you should be safe: self-other differences in situations involving physical safety. Judgm. Decis. Mak. 8, 250–267. Conflict of Interest Statement: The authors declare that the research was Strohschneider, S. (2003). “Ja mach ì nur einen Plan [Yes, just make a plan],” in conducted in the absence of any commercial or financial relationships that could Panne – Irrtum –Missgeschick [Hitch - Error - Misadventure], ed. B. Boothe be construed as a potential conflict of interest. (Bern: Huber), 127–144. Strough, J., Bruine de Bruin, W., Parker, A. M., Karns, T., Lemaster, P., Copyright © 2018 Domeier, Sachse and Schäfer. This is an open-access article Pichayayothin, N., et al. (2016). What were they thinking? Reducing sunk-cost distributed under the terms of the Creative Commons Attribution License (CC BY). bias in a life-span sample. Psychol. Aging 31, 724–736. doi: 10.1037/pag0000130 The use, distribution or reproduction in other forums is permitted, provided the Tetlock, P. E., and Mellers, B. A. (2002). The great rationality debate. Psychol. Sci. original author(s) and the copyright owner are credited and that the original 13, 94–99. doi: 10.1111/1467-9280.00418 publication in this journal is cited, in accordance with accepted academic practice. Thaler, R. H. (1985). Mental accounting and consumer choice. Mark. Sci. 4, No use, distribution or reproduction is permitted which does not comply with these 199–214. doi: 10.1287/mksc.4.3.199 terms. Frontiers in Psychology | www.frontiersin.org 11 May 2018 | Volume 9 | Article 815