The role of strategi es in deciding advantageously in ambiguous and risky

RESEARCH REPORT The role of strategies in deciding advantageously in ambiguous and risky situations Matthias Brand Æ Katharina Heinze Æ Kirsten Labudda Æ Hans J. Markowitsch Received: 1 June 2007 / Revised: 25 September 2007 / Accepted: 11 January 2008 / Published online: 30 January 2008 � Marta Olivetti Belardinelli and Springer-Verlag 2008 Abstract In decision situations of everyday life, the potential positive or negative consequences of a decision are often specified and the associated probabilities are known or they are principally calculable (‘‘decisions under risk’’). On the basis of correlations reported in patient studies, it has been recently proposed that decisions under risk involve strategic components, i.e. calculation of the risk, as well as emotional processes, i.e. processing feed- back from previous decisions. However, the potential impact of calculative strategies on decision-making under risk has not been investigated systematically, so far. In the current study, we examined 42 healthy subjects (21 females) with the Game of Dice Task measuring decisions under risk, and a questionnaire assessing strategy applica- tion in items comparable to the choices in the Game of Dice Task. In addition, the subjects performed the Iowa Gambling Task, examining decision-making under ambi- guity, and a neuropsychological test battery focusing on executive functions. Results indicate that deciding advan- tageously in a decision-making task with explicit and stable rules is linked to applying calculative strategies. In con- trast, individuals who decide intuitively prefer risky or disadvantageous choices in the Game of Dice Task. Applying calculative strategies was correlated with exe- cutive functions but not with performance on the Iowa Gambling Task. The results support the view that calcu- lative processes and strategies may improve decision- making under explicit risk conditions. Keywords Executive functions � Game of Dice Task � Iowa Gambling Task � Calculation � Intuition Introduction Making decisions in ambiguous or risky situations is a key function of everyday life. Whenever a situation offers more than one option to choose, we are in a conflict between direct or indirect consequences usually associated with the different options. In some situations, the consequences of the decision are completely undefined and we do not have any information about how likely positive or negative consequences will appear. These kinds of decisions are typically called ‘‘decisions under ambiguity’’ (Bechara 2004). On the other hand, in some decision situations, the consequences are specified and the associated probabilities are known or they are principally calculable (cf. Lee 2005). These types of decisions are commonly referred to as ‘‘decisions under risk’’ (Brand et al. 2006). Both, making decisions under ambiguity and under risk conditions, received great attention in neuropsychological research, as it was demonstrated in a series of studies that decision- making dysfunctions could be a core symptom in various patient populations (see below). However, the fundamental processes underlying decisions under ambiguity and risk are still a topic of debate. In this paper, we focus on the potential role of strategy application and on the perfor- mance of different types of deciders in gambling situations with and without explicit rules and probabilities. One frequently used task to investigate decision-making under ambiguous conditions is the Iowa Gambling Task (IGT), shown to be sensitive for decision-making dys- functions following ventromedial prefrontal cortex lesions (e.g., Bechara et al. 1994, 2000a, b). In addition, patients M. Brand (&) � K. Heinze � K. Labudda � H. J. Markowitsch Physiological Psychology, University of Bielefeld, P.O. Box: 10 01 31, 33501 Bielefeld, Germany e-mail: m.brand@uni-bielefeld.de 123 Cogn Process (2008) 9:159–173 DOI 10.1007/s10339-008-0204-4 with lesions restricted to limbic structures, primarily sub- jects with amygdala damage, also perform inferior to healthy individuals on the IGT (Bechara et al. 1999; Brand et al. 2007b), as patients with other neurological diseases or psychological disorders do, for example patients with substance dependencies (e.g., Bechara 2005; Bechara and Damasio 2002; Bechara et al. 2001, 2002; Bolla et al. 2003, 2005; Fishbein et al. 2005; Verdejo et al. 2004; Whitlow et al. 2004), and persons with obsessive-compulsive dis- order, schizophrenia, pathological gambling, anorexia nervosa, suicide attempters, and other individuals suffering from neuropsychiatric symptoms (e.g., Bark et al. 2005; Cavedini et al. 2002a, b, 2004; Goudriaan et al. 2005; Jollant et al. 2005) (see the review by Dunn et al. 2006). Although discussed controversially (e.g., Maia and McClelland 2004, 2005), one main interpretation of IGT deficiencies is that of reduced functioning of somatic markers, which can normally bias individuals against a disadvantageous decision (Damasio 1996; Damasio et al. 2002). In accordance to the assumption that emotional processes (e.g., processing the feedback of previous trials and associated emotional reactions) may significantly contribute to intact decision-making in the IGT are the results of reduced skin conductance responses in patients with ventromedial or amygdala damage while performing the task. Also, correlations between emotional reactions and task performance emphasize the view of the impor- tance of emotional processes in deciding advantageously (e.g., Bechara and Damasio 2002; Bechara et al. 1997, 2002; Brand et al. 2007b; Crone et al. 2004; Denburg et al. 2006; Hinson et al. 2002). By contrast, the impact of executive functions and strategic components on IGT performance, although also a topic of debate, was seen as rather minimal, given that in most of the aforementioned studies no correlations between IGT and executive func- tions were found (cf. Brand et al. 2007c for a critical discussion of this point). Nevertheless, the fundamental processes underlying IGT performance and the factors leading to deficient performance in the IGT are still under controversial discussion. There are plausible arguments that IGT performance can be disrupted due to various cognitive reductions. For instance, as shown by Busemeyer and Stout (2002), in patients with ventromedial prefrontal cortex damage performance on the IGT may be guided by the most recent outcome rather than by the outcome of all previous trials. The authors also found that the decision- makers’ impulsiveness may additionally influence task performance (and also principle learning abilities, such as to build long-term contingencies). Furthermore, difficulties in reversal learning and inhibition can also compromise IGT performance given that patients with prefrontal lesions perform normally on the task when the initial reward/punishment schedule is changed such that the disadvantageous decks (A and B) are not advantageous (associated with higher gains than decks C and D) in the opening trials (before the first punishment is displayed) (Fellows and Farah 2005). In addition, the potential influ- ence of working memory, general risk-taking preference or apathy on IGT performance is also discussed controver- sially (see the detailed discussion of these topics in Dunn et al. 2006). We developed a decision-making task aiming to mea- sure decisions under risk conditions, the Game of Dice Task (GDT). In this task, the information about the amounts of gains and losses associated with the different alternatives and their winning probabilities are explicitly provided from the very beginning of the task and are stable over the entire task duration. Therefore, it was hypothe- sized that the GDT allows the individuals to apply strategies in order to decide advantageously in the long run. It was also assumed that executive functions may correlate with decision-making in this task, meaning that high executive abilities are linked to good performance on the GDT. And in fact, in a sum of studies with patients suf- fering from Korsakoff’s syndrome, Parkinsonism, Urbach- Wiethe disease, pathological gambling or bulimia nervosa, we found clear hints for the association between GDT performance and measures of executive functioning, such as categorization, set-shifting and cognitive flexibility (Brand et al. 2004a, b, 2005a, b, 2007a, b). Deficits in this task were also described by other research groups for patients with Alzheimer’s disease or schizophrenia (Del- azer et al. 2007; Lee et al. 2007) and in preadolescents with attention-deficit/hyperactivity disorder (Drechsler et al. 2007). Beyond the prominent role of executive functions in decisions under explicit risk conditions, processing feed- back from previous trials also impacts GDT performance. In our studies mentioned above, we found correlations between GDT performance and the frequency of using negative feedback following a risky decision to shift to a non-risky alternative in the upcoming trial. In accordance with the correlations found in patient studies, Bechara et al. (1997) proposed a model of deci- sion-making under ambiguity, which focuses on the role of somatic markers and related emotional processes in deciding advantageously. Inspired by this model and based upon the correlations found in neuropsychological studies on decision-making under risk, we have recently formu- lated ideas on the potential relationship between strategies, executive functions and making non-risky or advantageous choices in tasks assessing decisions under risk conditions (Brand et al. 2006). The model is illustrated in Fig. 1. This model on decisions under risk was able to retro- spectively fit previous results and to deduce hypotheses on how patients with specific cognitive decline should per- form on tasks measuring decisions under risk in contrast to 160 Cogn Process (2008) 9:159–173 123 decisions under ambiguity. However, the single steps of the proposed decision-making process have not been investi- gated experimentally. We have only started to examine the potential impact of feedback processing on decisions under risk by excluding the feedback components of the GDT and compared performance of healthy subjects in both the original GDT and the modified version (without feedback) (Brand 2008). Results demonstrated that subjects profit from feedback after each trial, as their performance in the modified version was lower than in the original GDT. However, subjects were nevertheless able to perform advantageously as they chose more non-risky alternatives than risky options. This result indirectly shows that the explicit information provided in the GDT is sufficient to generally perform well, but that feedback after each trial can optimize decisions in the long run, potentially by monitoring or revising current decision strategies. Beyond the neuropsychological perspective on decision- making, as summarized above, psychological research on decision-making and related issues (thinking, reasoning) has a long tradition. One of the most influential theories on decision-making is the dual process account, i.e. the assumption of two different modes of information pro- cessing influencing thinking and reasoning (see Kahneman 2003 for a recent review). The one mode, often referred to as intuitive, natural or heuristic (e.g., Tversky and Kahn- eman 1983), intuitive-experiential (Epstein et al. 1996) or simply system 1 (cf. Evans 2003; Stanovich and West 2000), is associated with fast, parallel, automatic, emo- tional and effortless information processing (e.g., Kahneman 2003). The second mode, referred to as exten- sional (Tversky and Kahneman 1983), analytic-rational (Epstein et al. 1996) or system 2 (cf. Evans 2003; Stano- vich and West 2000), is linked to slow but controlled, flexible, neutral, rule-governed and effortful information processing (e.g., Kahneman 2003). It was demonstrated in multiple series of studies that in complex situations indi- viduals often make their decisions intuitively depending on the accessibility of information or thoughts. However, in accordance with dual process theories, intuition and rea- soning often interact in concert. It seems, nevertheless, also very likely that individuals differ regarding their suscepti- bility to solve problems more intuitively or analytically. As recent studies demonstrated, subjects with high cognitive functioning more reliably use the analytical-rational mode (system 2) and are better in solving statistical problems and decision-making tasks dealing with explicit knowledge (e.g., Stanovich and West 2000). Although both fields of decision-making research, (1) the neuropsychological perspective dealing with Fig. 1 The figure indicates the steps of decision-making under risk as proposed in the model described in Brand et al. (2006). It is suggested that deciding advantageously in decisions under risk conditions is principally possible on the basis of rational or cognitive strategies alone, i.e. feedback is not necessary for developing or applying a decision strategy in such situations. However, it is also proposed that the optimal way of decision-making in risky situations is to use both, cognitive strategies as well as biasing emotional signals. The claim of the figure is not to develop a new psychological decision-making theory. Instead, it is aimed to visualize the steps of decision-making believed to be involved in neuropsychological tasks measuring decision-making under risk conditions (as the GDT) Cogn Process (2008) 9:159–173 161 123 decision-making dysfunctions of patients with brain dam- age or abnormalities measured for example with the IGT, and (2) the dual process perspective dealing with different cognitive systems involved in reasoning, intuition and making decisions under uncertainty, extensively contrib- uted to a better understanding of how subjects make decisions, the two research perspectives were rarely com- bined. For instance, potential individual differences in decision-making as measured by neuropsychological tasks, such as the IGT and GDT, were less extensively studied in healthy subjects so far. Only a few studies exist which systematically examined personality traits or cognitive styles potentially influencing IGT or GDT performance, e.g., sensation seeking (Donohew et al. 2000; Harmsen et al. 2006; Reavis and Overman 2001), impulsivity (Brand et al. 2007a; Zermatten et al. 2005) or the influence of the ‘‘Behavioral Inhibition System’’ and the ‘‘Behavioral Approach System’’ (Franken and Muris 2005), with mixed results. In addition, it has not been investigated systematically, whether or not a tendency towards making decisions in an intuitive or analytical–rational way influences decision- making in IGT and GDT. In a recent study, Turnbull et al. (2005) showed that performing the IGT may rely on system 1 or in other words, the IGT can be solved by intuition rather than by reasoning. Nonetheless, it has not been shown so far whether or not neuropsychological gambling tasks with explicit information are solved pri- marily by intuition or reasoning and whether or not individuals who preferably use either intuition or reasoning perform superior to others. The aim of the current study was to investigate the potential role of decision strategies in performing neuro- psychological decision-making tasks, i.e. the GDT and the IGT. We hypothesized that in healthy subjects the prefer- ence of applying calculative strategies (e.g., calculating the probabilities with respect to the potential gains and losses) co-varies with deciding advantageously in the GDT. Although the exact expectancy value of the choices is not displayed, the information about gains and losses and the winning probabilities is principally manageable. Therefore, individuals who apply a calculative strategy (or in other words, who use the analytical-rational mode, system 2) on the basis of the information explicitly displayed should perform superior compared to the individuals who do not use calculative processes but decide intuitively (i.e., on the basis of the intuitive-experiential mode, system 1). As information about gains and losses and probabilities is not displayed explicitly in the IGT, a preference towards cal- culative or intuitive strategies should not influence IGT performance, as the IGT may be successfully solved by intuition (system 1) as previously shown by Turnbull et al. (2005). Subjects and methods Subjects Forty-two healthy individuals (21 males) participated in the study (age: M = 44.45, SD = 17.46, range 19–80 years). The volunteers were students or staff members of the cafeteria of the University of Bielefeld and their relatives and friends. In addition, elderly people were recruited by local advertisements and placards (e.g., in a local chorus for older adults). Inclusion criteria required that individuals have no history of neurological or psychiatric disease, as determined by a telephone screening. Any substance-rela- ted disorder was also a reason for exclusion from participation. Furthermore, individuals were excluded if they had taken part in previous studies on decision-making in the Department of Physiological Psychology, University of Bielefeld. In addition, subjects with signs of cognitive decline, measured with a cognitive screening instrument (see description of the neuropsychological background test battery) were excluded. All participants gave written informed consent prior to the investigation. They did not receive financial compensation for participation. Instruments The Game of Dice Task The participants performed the original computerized Game of Dice Task (GDT) (Brand et al. 2005a). The GDT is a decision-making task that explicitly provides information about the gains and losses associated with a given choice. The information about the amount of gains and losses is told to the participant at the very beginning of the task and is visualized on the screen during the entire task duration. The winning probabilities of the alternatives are obvious and can be calculated easily. The task is described in more detail in the paper mentioned above (Brand et al. 2005a) in which the similarities and differences between the GDT and the Iowa Gambling Task are also explained. A color figure of the task’s screen can also be found on the following website: http://dx.doi.org/10.1037/0894-4105.19.3.267.supp. Participants were instructed to maximize their fictitious starting capital of 1,000 € within 18 throws of a single virtual dice, meaning that they were told that the goal of the game was to win as much money as possible and to lose as little money as possible. Before each throw, they had to decide which number or combination of numbers may include the number that will be thrown. In more detail, before the dice-box is shaken in the GDT, participants may choose one single number (e.g., the ‘‘six’’) having a 162 Cogn Process (2008) 9:159–173 123 winning probability of 1:6 (16.6%) and resulting in a high fictitious gain of 1,000 € when the chosen number (e.g., the ‘‘six’’) is thrown, but also a fictitious loss of 1,000 € when one of the five numbers not chosen is thrown (e.g., the ‘‘one’’, ‘‘two’’, ‘‘three’’, ‘‘four’’, or ‘‘five’’). The category of one single dice options contains six different alternatives (all six numbers that can be thrown). Participants also have the choice of selecting a combination of two numbers (e.g., the ‘‘five’’ and the ‘‘six’’ together), which has a winning probability of 1:3 (33.3%) and which produces a gain of 500 € wh