Fighting the Illusion of Control: How to Make Use of Cue Competition and Alternative Explanations*

r e s u M e n Los errores en la percepción de la causalidad constituyen la base del pensamiento supersticioso y la pseudociencia. El principal objetivo del presente trabajo fue demostrar cómo puede utilizarse nuestro conocimiento sobre los mecanismos involucrados en la inducción causal para impedir o reducir el desarrollo de este tipo de creencias. La evidencia disponible mostró que a veces las personas perciben relaciones causales que en realidad no existen. La propuesta es que esto podría deberse, al menos en parte, a que las personas no suelen tener en cuenta factores alternativos que puedan estar jugando un papel causalmente relevante. El presente experimento demuestra que efectivamente se pueden reducir los errores en la atribución causal, tales como las ilusiones de control, proporcionando a los participantes información precisa y difícil de ignorar sobre otras causas posibles. Palabras clave autores Ilusión de control, superstición, pseudociencia, descarte de hipótesis, competición de claves. Palabras clave descriptores Epistemología de la Psicología, Psicología Experimental, ciencia, causalidad. SICI: 1657-9267(201303)12:1<261:FICUCC>2.0.Tx;2-A Para citar este artículo: Vadillo, M. A., Matute, H. & Blanco, F. (2013). Fighting the illusion of control: How to make use of cue competition and alternative explanations. Universitas Psychologica, 12(1), 261-270. * Support for this research was provided by Grant PSI2011-26965 from Ministerio de Ciencia e Innovación and Grant IT363-10 from Departamento de Educación, Universidades e Investigación of the Basque Government. ** Correspondence concerning this article should be addressed to either Miguel A. Vadillo, Division of Psychology and Language Sciences, University College London, 26 Bedford Way, London WC1H 0AH, UK, or Helena Matute, Departamento de Fundamentos y Métodos de la Psicología, Universidad de Deusto, Apartado 1, 48080 Bilbao, Spain. Fax: (34) (94) 4139089. E-mails: m.vadillo@ucl. ac.uk or [email protected] Miguel A. VAdillo, HelenA MAtute, FernAndo BlAnco 262 Un i v e r s i ta s Ps yc h o l o g i ca V. 12 No. 1 e n e r o-m a r z o 2013 Introduction Most articles on causal induction begin with a remark on our outstanding abilities to detect causal relations and the importance of these abilities for our survival. Human beings, and probably other animals as well, are indeed superb causality detection machines. But our highly evolved capacities are not without their failures. The cognitive abilities responsible for our otherwise precise causal inferences do not prevent us from often perceiving causal patterns where only mere chance and luck exist. Not surprisingly, the study of causal misattribution and related issues is becoming a major topic in the current psychological research and theory (McKay & Dennett, 2009; Wegner, 2002). These shortcomings are interesting not only because of the information they provide about the psychological processes underlying the perception of causality, but also because our systematic failures to correctly explain some events are a major source of suffering and superstition. Whenever we feel inclined to think that certain ethnic groups are more prone to crime than others, or that the gods will respond to our dancing around the fire with the long awaited rain, we are betrayed by the limitations of our causal reasoning abilities. No matter how much science and education have progressed, pseudoscience and superstition still wander at will through our developed societies (Davis, 2009; Dawkins, 2006; Goldacre, 2008; Lilienfeld, Ammirati & Landfield, 2009; Shermer, 1997). Fortunately, the abundant literature on causal learning and reasoning provides critical information about the mechanisms underlying these processes, and contains, either explicit or implicitly, many hints for reducing causal illusions (for reviews, see Gopnik & Schulz, 2007; Matute, Yarritu & Vadillo, 2011; Mitchell, De Houwer & Lovibond, 2009; Shanks, 2010; Sloman, 2005). However, to the present, researchers have dedicated little time and effort to exploring how their theories and empirical knowledge can be used to reduce superstitious beliefs. One of the main goals of the present work is to show how some of the most consistently replicated effects in causal induction studies can be used to impair the development of these beliefs and reduce their impact. One of the most prominent and best-studied characteristics of causal induction is that the inferences regarding the causal status of one event are usually influenced by previous or subsequent knowledge about the causal relation that other events hold with the same outcome. Imagine, for example, that you read a text in which a celebrity advocates for a given political party. No doubt, you will probably attribute his writing to his political preferences (Jones & Davis, 1965). However, if you are later told that the author was paid in exchange for writing the paper, will this not cast doubt on his true political inclinations? Even when we have good reasons to believe that there is a causal relation between a given candidate cause and an event to be explained, our belief is usually revised, or even dramatically altered, if we later get to know about other potential causes that might account for the effect. This phenomenon, known as discounting (Einhorn & Hogarth, 1986; Kelley, 1973; Morris & Larrick, 1995) or cue competition (Baker, Mercier, Vallée-Tourangeau, Frank & Pan, 1993; Dickinson, Shanks & Evenden, 1984; Wasserman, 1990) in different research areas, addresses the competitive nature of causal induction. Although discounting and cue competition have received extensive attention in current research on causal induction (e.g., Goedert & Spellman, 2005; Laux, Goedert & Markman, 2010; Van Overwalle & Van Rooy, 2001), to our best knowledge, no serious attempt has been made to relate these effects to the development of superstitions and to the perception of illusory causal relations. However, as will be shown below, the available evidence suggests that a failure to take into account the role of alternative causes of an event is one of the many factors that can give rise to superstitions and causal misattributions. The illusion of control is one of the most remarkable and puzzling instances of causal misattributions: People often tend to attribute actually uncontrollable events to their own behavior rather than to their real causes or to mere chance (Langer, 1975; Matute, 1995; Ono, 1987; Wright, 1962). Sometimes, the reason why people believe that Fighting the illusion oF Control Un i v e r s i ta s Ps yc h o l o g i ca V. 12 No. 1 e n e r o-m a r z o 2013 263 they can control some random event is just that they are more prone to expose themselves to the kind of evidence that would confirm these beliefs, than to alternative evidence that would prove the opposite (Nickerson, 1998; Wason, 1960). In other words, they are underexposed to the information that could be used to discount the role of their own behavior in producing these events. For example, students who carry lucky charms to their exams and obtain high grades usually fail to notice that they would have passed the exams anyway, even without the use of the amulet. In order to get that information, these students should keep their lucky charms at home at least in some exams, so that they could make the comparison between the grades they get when they use the charm and the ones they get when they do not. But, for obvious reasons, they are reluctant to do so. Indeed, recent research shows that leaving them at home increases anxiety and reduces self-esteem, which, in the end, can hinder performance (Damisch, Stoberock & Mussweiler, 2010). Given that most of their good grades are contiguous to the use of the lucky charm and that there are no or few instances of good grades in the absence of its use, it is just natural to conclude (erroneously) that the lucky charm had something to do with the results of the exam. As illustrated by this example, people trying to control an important outcome usually expose themselves to evidence that suggests a positive relationship between their behavior and the desired outcome (i.e., a high number of coincidences between the target response and the outcome). Consequently, they are also underexposed to the evidence contrary to this hypothesis (i.e., occurrences of the outcome in the absence of that response, which would be indicative of the presence of alternative causes). Experiments conducted in the laboratory confirm that this pattern of behavior is an important factor in the emergence and maintenance of the illusion of control. When participants are highly involved in their attempts to control an event and refuse to check what would happen if they did not act, their illusion of control is enhanced (Blanco, Matute & Vadillo, 2011; Matute, 1996). However, when they are forced to refrain from responding in some trials, their illusion is reduced (Hannah & Beneteau, 2009). This might explain why depressed people, who are usually more passive and less motivated to control life events, usually tend to show few or no illusions of control (Blanco, Matute & Vadillo, 2009). In fact, the role of activity in the development of illusions of control is a straight-forward prediction of some formal models of causal learning (e.g., Rescorla & Wagner, 1972; see Matute, Vadillo, Blanco & Musca, 2007 for a computer simulation): The participants who are highly involved in trying to obtain an outcome experience more coincidences between their own actions and the outcome. Therefore, they have less opportunities to experience what would happen in the absence of their actions. As a consequence, they develop a stronger illusion of control. In light of this evidence, one might think that the illusion of control should be easily reduced by alerting people about alternative, potential explanations of the events they are trying to control. However, both common experience and experimental findings suggest otherwise. Even when people are alerted about other potential explanations for the apparent success of alternative medicine or charms of any kind, they are still unwilling to reconsider the basis of their beliefs and usually prefer to consult other “experts” who will confirm their illusory beliefs. Just as an example, popular science writers and journalists who alert the general public against pseudoscientists and propose science-grounded explanations for their claims, have been sued in several countries (Gámez, 2007; Sense about Science, 2009). This illustrates quite well how, at least sometimes, our society places more value in the right to keep irrational beliefs than on the attempts to discover and teach the truth. Experimental data gathered in the laboratory shows a similar pattern of behavior. For instance, in a recent experiment, conducted both in the laboratory and through the Internet (Matute, Vadillo, Vegas & Blanco, 2007), participants were asked to try to control, by pressing the space bar, the onset of a series of (actually uncontrollable) flashes appearing on the computer screen. Half of the participants were explicitly warned that they might have no Miguel A. VAdillo, HelenA MAtute, FernAndo BlAnco 264 Un i v e r s i ta s Ps yc h o l o g i ca V. 12 No. 1 e n e r o-m a r z o 2013 real control over the flashes. However, this piece of information made no difference in their illusion of control, as measured by the subjective ratings of control provided at the end of the experiment. An important fact that can explain why people refuse to consider alternative causes, even when they are informed about their potential role, is that the format in which this information about alternative explanations is provided might not be optimal for the revision of erroneous beliefs. Many researchers have argued that the psychological processes responsible for causal learning might vary depending on the way information is presented. For example, when people are given information about the covariation between a cause and an effect, whether this information is provided in a summary table or directly experienced in a series of trials, makes a difference in their ability to detect the cause-effect contingency (Shanks, 1991; ValléeTourangeau, Payton & Murphy, 2008; Waldmann & Hagmayer, 2001; Ward & Jenkins, 1965; but see Van Hamme & Wasserman, 1993). Thus, previous failures to reduce the illusion of control by alerting participants about the potential role of alternative causes (e.g., Matute, Vadillo, Vegas et al., 2007) might have been due to the fact that participants were not provided with direct experience on the presence and absence of the alternative cause during their attempts to control the target event. In order to test the prophylactic effects of discounting upon the development of illusions of control, it is important to use a procedure that is known to produce strong illusions. Therefore, in the present experiment, the participants were exposed to a standard preparation for the study of the illusion that has been used extensively in previous studies (Blanco et al., 2009; Matute, 1996; Matute, Vadillo, Vegas, et al., 2007; see also Matute, Vadillo & Bárcena, 2007). Participants were instructed to try to turn off some blue and white flashes appearing on the computer’s screen by using a joystick. These instructions, aimed at framing the task as a skill-based situation (Langer, 1975), are expected to promote the illusion of control. Moreover, this procedure is similar to other experimental tasks that are known to produce strong illusions of control and that have been properly validated in previous research (e.g., Fenton-O’Creevy, Nicholson, Soane & Willman, 2003). For all the participants, the termination of the flashes was preprogrammed following a random sequence and was, therefore, uncontrollable. In one of the conditions these flashes were the only stimuli appearing on the screen. However, in the other condition, the offset of the flashes was always preceded by a signal that acted as an alternative cause that might account for the termination of the flashes. According to our hypothesis, this signal should compete with the participant’s response as a potential explanation for the offset of the flashes, thereby reducing the illusion of control, in comparison to participants in the no-signal condition.