This article was downloaded by: [University of Southern Queensland] On: 12 October 2014, At: 14:07 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Social Neuroscience Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/psns20
State- and trait-greed, its impact on risky decisionmaking and underlying neural mechanisms a
b
a
a
Patrick Mussel , Andrea M. F. Reiter , Roman Osinsky & Johannes Hewig a
Department of Psychology I, Differential Psychology, Personality Psychology, and Psychological Diagnostics, Julius Maximilians University Würzburg, Würzburg, Germany b
Max Planck Fellow Group “Cognitive and Affective Control of Behavioural Adaptation”, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany Published online: 10 Oct 2014.
To cite this article: Patrick Mussel, Andrea M. F. Reiter, Roman Osinsky & Johannes Hewig (2014): State- and trait-greed, its impact on risky decision-making and underlying neural mechanisms, Social Neuroscience, DOI: 10.1080/17470919.2014.965340 To link to this article: http://dx.doi.org/10.1080/17470919.2014.965340
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
SOCIAL NEUROSCIENCE, 2014 http://dx.doi.org/10.1080/17470919.2014.965340
State- and trait-greed, its impact on risky decision-making and underlying neural mechanisms Patrick Mussel1, Andrea M. F. Reiter2, Roman Osinsky1, and Johannes Hewig1
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
1
Department of Psychology I, Differential Psychology, Personality Psychology, and Psychological Diagnostics, Julius Maximilians University Würzburg, Würzburg, Germany 2 Max Planck Fellow Group “Cognitive and Affective Control of Behavioural Adaptation”, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
We investigated whether greed would predict risky decision-making and recorded neural responses during a monetary gambling task using the electroencephalogram. We found that individuals high in trait-greed took higher risks to maximize monetary outcome. Furthermore, this relation was moderated by state-greed; specifically, traitgreed had a stronger impact on risky decision-making when activated by situational characteristics. On the neural level, greedy individuals showed a specific response to favorable and unfavorable outcomes. Specifically, they had a reduced feedback-related negativity-difference score to these events, indicating that they might have difficulty in learning from experience, especially from mistakes and negative feedback. It is concluded that greed may explain risky and reckless behavior in diverse settings, such as investment banking, and may account for phenomena such as stock market bubbles.
Keywords: Cognitive neuroscience; Personality; Risk taking; Feedback-related negativity; Psychopathy.
The financial crisis in the recent past, initiated by the subprime mortgage crisis in the United States and the debt crisis in Europe, evolved into a global economy crisis, affecting the real economy with profit shrinkage, downsizing, unemployment, and insolvency. As a consequence, governments were forced to spend vast sums of money for banking bailout and economic growth packages. The causes for this development lay not only in structural problems, such as rise of national debts or failures in financial regulation, but also in excessive risk taking by agents operating in investment departments of financial firms or stock exchanges. We investigated whether greed is a predictor of risky decision-making. Greed can be defined as desire to get more at all costs, including the excessive striving for desired goods and the willingness to accept that such
striving may be at the expense of others (Balot, 2001). As such, greed is associated with concepts of money, economic values such as wealth, and striving for power (Rokeach, 1973; Schwartz, 1992; Tang, 2007). Additionally, greed can be linked to antisocial, unethical, and deviant behavior (Krueger, Markon, Patrick, Benning, & Kramer, 2007; Patrick, Hicks, Krueger, & Lang, 2005): Individuals with high values on greed may justify the rightness of their actions by its instrumental value for increasing targeted goods, thereby qualifying socially shared norms, moral standards, rules, and laws which ultimately justifies unethical and deviant behavior (Aquino, Freeman, Reed, Lim, & Felps, 2009; Bosse, Siddiqui, & Treur, 2010; Eek & Biel, 2003; Piff, Stancato, Cote, Mendoza-Denton, & Keltner, 2012; Vohs, Mead, & Goode, 2006). Based on this definition,
Correspondence should be addressed to: Patrick Mussel, Department of Psychology I, Differential Psychology, Personality Psychology, and Psychological Diagnostics, Julius Maximilians University Würzburg, Marcusstr. 9-11, 97070 Würzburg, Germany. E-mail: [email protected] The research was conducted at the Julius Maximilians University Würzburg. This research was supported by a Schumpeter-Fellowship by the VolkswagenStiftung [85650; II/86486].
© 2014 Taylor & Francis
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
2
MUSSEL ET AL.
it is conceivable that greedy individuals will take excessive risks to increase profits, thereby accepting the possibility of enormous losses for their department, their company, or even the society. In addition to the influence of personality, the behavior of agents operating in investment departments of financial firms or stock exchanges is likely to be influenced by situational demands or incentives, like excessive yield on turnover associated with personal bonuses (Lo, Repin, & Steenbarger, 2005). From the perspective of interactionism (Mischel, 1973), situational attributes may activate personality traits and thus increase their impact on behavior (Ekehamma, 1974; Franken & Muris, 2005). Therefore, we propose that greed as a stable personality trait predicts risky decision-making, especially when this personality trait is activated by situational characteristics. When taking risky decisions, as well as in other situations, individuals try to adjust their behavior according to positive and negative feedback provided by their environment in order to optimize desirable outcomes. A theoretical framework that accounts for such processes is reinforcement learning theory (Sutton & Barto, 1998), which posits that individuals have expectations regarding upcoming events in the future, and that deviations from these expectations are used to learn from experience and, subsequently, adapt behavior. Specifically, events that are “worse than expected”, such as punishment or absence of reward, evoke a negative temporal difference error, whereas events that are “better than expected”, such as reward, evoke a positive temporal difference error. Neuropsychological indicators have been linked to these processes and allow their investigation during decisionmaking. Specifically, the negative temporal difference error has been related to a phasic decrease in dopaminergic signaling in basal ganglia, followed by a disinhibition of apical dendrites of the motor neurons of the anterior cingulate cortex, eliciting a feedback-related negativity (FRN) that can be measured at the scalp using electroencephalography (EEG) (Debener et al., 2005; Fiorillo, Tobler, & Schultz, 2003; Gehring & Willoughby, 2002; Holroyd & Coles, 2002; Montague, Hyman, & Cohen, 2004; Pessiglione, Seymour, Flandin, Dolan, & Frith, 2006; Schultz, 2002). The FRN is a negative deflection with a maximum at fronto-central electrode positions approximately 250–350 ms after the onset of negative, compared to positive stimuli (Miltner, Braun, & Coles, 1997). We assessed this component during a risky decision-making task to investigate whether state- and traitgreed would moderate neural processes associated with adapting behavior and learning from experience. In the following, we report results from a risktaking paradigm, the Balloon-Analogue-Risk-Task
(BART) (Lejuez et al., 2002). We predicted that risky decision-making would be related to traitgreed. To investigate whether the relation between trait-greed and risk taking is moderated by situational characteristics, we used cues which induced either state-modesty or state-greed. We predicted that the relation between trait-greed and risk-taking should be stronger in the state-greed condition, compared to the state-modesty condition. During the risk-taking task, we recorded event-related potentials to investigate the neural mechanisms underlying greedy decision-making.
METHOD Participants Twenty participants were recruited from the student population of the Julius Maximilians University Würzburg. All participants studied economics and were between 20 and 31 years (on average 24.5). As gender is known to impact risk-taking behavior (Byrnes, Miller, & Schafer, 1999; Daghofer, 2007), only male participants were recruited for the study. For their participation, they were paid € 20; additionally, they could win € 100 if they obtained the highest fictive account balance in the decision-making task.
Task Risky decision-making was assessed using the BART (Lejuez et al., 2002). Decision-making in the BART has been found to be related to various real-world risk-taking behaviors that include alcohol and drug use, cigarette smoking, gambling, theft, aggression, psychopathy, and unprotected sexual intercourse (Lejuez, Aklin, Zvolensky, & Pedulla, 2003; Lejuez et al., 2002). Participants played two blocks of 150 trials, each block under 1 of 2 conditions (see below). In each trial, a picture of a balloon was presented on a screen representing a value of € 1000,– (see Figure S1 in the Supplemental material available online). Subsequently, individuals had to decide whether they wanted to inflate the balloon or not. If they rejected to inflate the balloon, the money was collected on a fictive account. If they decided to blow up the balloon (risky decision), it either burst, resulting in a total loss of the money for the current trial, or inflated, resulting in double value (i.e., € 2000,–). In the latter case, participants had to decide again whether they wanted to inflate the balloon or not. As the balloon inflated, the risk of bursting increased from 15% to 100% by 7
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
GREED AND RISKY DECISION-MAKING
percentage points per level (i.e., 15%; 22%; 29%; . . .). Participant had been informed that the player with the highest fictive account balance in the BART won € 100 as financial incentive. Risky choices were defined as the number of decisions to pump up the balloon, compared to all decisions when the balloon had not yet burst. Therefore, on level 1, risky choice was computed as the number of decisions to pump up the balloon, compared to all 300 trials. As the balloon had a 15% probability of bursting on level 1, the maximum number of risky choice in level 2 is 85% * 300 trials = 255 trials; therefore, risky choice in level 2 was computed as the number of risky decisions to pump up the balloon in level 2, compared to 255 trials, and so forth.
Measures We developed a self-report measure for assessing greed as a personality trait. Based on the definition provided earlier, a pool of personality items was developed and applied to several convenience samples of students (total N = 640). Based on item analyses, seven items were chosen (see additional Table S1 available online). An example item is “When I think about all the things I have, my first thought is about what I would like to have next”. The trait-greed measure had a one-factorial structure (eigenvalues 3.2; 0.92; 0.73; 0.59; 0.55; 0.52; 0.46) and acceptable internal consistency (.79). Additional evidence regarding the validity of the trait-greed measure can be found in the Supplemental material available online.
Procedure After receiving verbal instructions about the experiment, participants gave written informed consent for participation (see Figure 1). Next, participants were given the
3
biography of either a greedy or a modest character (randomized across participants) to induce state-greed and state-modesty, respectively. Each biography was fictitious however based upon famous personae from financial management. In the state-greed condition, greed was activated by a fictitious biography of a successful and greedy person. The biography contained traits of the person (e.g. “uncompromising when he really wants something”), behaviors (e.g. “developed a highly profitable business idea”), life events (“investors plastered him with money”), and affective reactions (e.g. “it’s an exciting thrill, like ecstasy”). In the state-modest condition, a paralleled biography of a successful, but more modest person was used (e.g., “is conscientious”, “developed a sustainable management”). Participants were told that the first part of the assessment aimed to test their skills to put themselves in someone else’s position. Therefore, they had to read the biography and subsequently complete a three-item situational judgment test (SJT) from the perspective of the character described in the biography, which was also intended to intensify the induction of state-greed and state-modesty. Each item contained an item stem, describing a work situation (counseling a customer regarding an investment decision; negotiate with a supervisor regarding pay rise; deciding upon a stock investment, based on an insider tip). Followed by the item stem, five behavioral options were presented that differed in terms of greediness. Participants had to rank the options from 1 (behavior that the character would most likely exert) to 5 (behavior that the character would least likely exert). Each option was rated for greediness by three experts that allowed for calculating a “greed-score” for each situational judgment item. Across the three items, we found large differences between the two conditions (t = 15.7, p ≤ .001; d = 5.4); in the state-greedy conditions, greedy behavior options were much more likely chosen, compared to the state-modesty condition, indicating that participants read the biography thoroughly, understood it fully and were capable to adopt the perspective of the described
Figure 1. Time line and experimental design. Biography 1 and 2 were fictive biographies of either a greedy or a modest person (assigned randomized to either Biography 1 or 2 across participants). The SJT includes three scenarios with five behavioral options each which were filled out from the perspective of the person described in the biography. Each BART consisted of 150 trials. During the BART, we recorded neural responses to favorable (non-burst of the balloon) and unfavorable (burst of the balloon) outcomes via EEG. The questionnaire at the end included the seven-item trait-greed measure.
4
MUSSEL ET AL.
character. To control for confounding effects of the state induction, participants filled out rating scales for valence and arousal using 5-point Likert scales. We found no effect of greed induction on either valence (t = 0.8, p = .45) or arousal (t = −0.3, p = .77). After the BART task, participants filled out the trait-greed measure, as described above. They were informed about the purpose of the study and paid their allowance.
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
EEG recording and quantification While participants performed the BART, EEG (analog bandpass: 0.1–80 Hz, sampling rate: 250 Hz) was recorded from 31 scalp sites according to the 10–20 system, using Ag/AgCl electrodes and a BrainAmpDC amplifier (Brain Products GmbH, Gilching, Germany). Impedances were kept below 10 kΩ and electrodes were referenced to the vertex (Cz). For the detection of blinks and eye movements the horizontal and vertical electrooculogram was recorded. Data were processed offline, using MATLAB R2011 b (MathWorks, Natick, MA) and the Toolbox EEGLAB 12.0.1 (Delorme & Makeig, 2004). First, data were filtered, using a 30 Hz low-pass filter and a 0.5 Hz high-pass filter. Subsequently, the EEG was segmented into feedback-locked epochs of 1000 ms (−200 to 800 ms), baseline-corrected (−200 to 0 ms), and re-referenced to averaged mastoid electrodes. For artifact rejection, trials in which the amplitude exceeded the criterion of 3.5 standard deviations were excluded from further analyses (7.8% of the trials). At least 49 artifact-free trials were available per participant and condition. Next, we used independent component analysis decomposition for the detection of eye blink and movement artifacts. Components representing artifacts were detected using a semi-automated procedure and were subsequently removed from the dataset. Finally, data were averaged for each participant and each of four conditions, defined by the two factors state-greed (greedy vs. modest) and outcome (burst vs. non-burst of the balloon) (see Figure 5A). We used principal component analysis (PCA) to identify the FRN component, using the MATLAB Toolbox ERP PCA 2.32 (Dien, 2010). A temporal PCA across the 250 time points of the epoched and averaged data was computed, following Promax rotation of the factors. The covariance matrix and Kaiser normalization were used for the PCA. Upon inspection of the scree plot we decided to extract 26 factors. Based on visual inspection, we identified a factor that readily responded to the FRN (see Figure 5B). The factor is characterized by a fronto-central maximum, a typical negativity between 270 and 310 ms with a
preceding positivity, and a frequency in the upper theta–lower alpha range (8 Hz). The PCA waveform reveals a difference between outcomes, driven by a negative response to the unfavorable event (burst of the balloon) that is considerably reduced on favorable events (non-burst of the balloon). We also performed a separate spatial PCA on the 26 temporal factors with Infomax rotation. Two factors were extracted per temporal factor. Inspection of the results regarding the factor representing the FRN revealed two factors with very similar topography (both fronto-central). Presumably, there were no separable activities which overlapped in time at different scalp locations. Therefore, we stuck to the factor obtained via temporal PCA and reconstructed the waveforms by multiplying the factor pattern matrix with the standard deviations to convert the signal to microvolts. Subsequently, we used a Hilbert transformation to adequately quantify the amplitude of the P2–N2 complex (Cohen, Elger, & Fell, 2009) (see Figure 5C). The Hilbert transformation provides the complex analytic signal, z(t). The squared modulus of the resulting complex signal represents the power at each time point. Finally, data were log-transformed and averaged across 150–350 ms for each participant and each of the four conditions and subsequently normalized.
RESULTS The maximum number of times that a balloon was pumped up was eight (see Supplementary Figure S1 for details regarding the task). As depicted in Figure 2, participants almost always inflated the balloon in levels 1 and 2. In 56% of the trials, the balloon was pumped up at least four times. Using linear mixed
Figure 2. Behavioral results for the BART task. The maximum number of times that a balloon was pumped up was eight. For each of the eight levels, the relative frequency of pumping up the balloon (i.e., risky decision) is given, computed as the absolute number of “pump up” decisions divided by the number of trials in which the balloon had not yet burst.
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
GREED AND RISKY DECISION-MAKING
5
Figure 3. Effect of greed on risky decision-making. Greed was assessed on the trait-level, using a personality self-report measure. Risk was operationalized as the relative frequency of pumping up the balloon, compared to the number of trials in which the balloon had not yet burst. (A) Main effect of trait-greed on risk-taking across all eight levels. (B) Correlation between trait-greed and risk-taking separately for the eight levels and for the induction of state-greed (upper row) and state-modesty. State-greed and state-modesty were induced by reading a biography of either a greedy or a modest person, followed by an additional task that had to be solved from the perspective of the respective character. *p < .05.
models with unstructured repeated covariance matrices with the decision to pump up the balloon as dependent variable and standardized trait-greed as covariate, we found a significant main effect of traitgreed on risky decision-making (F1,8 = 6.3; p = .04), which was qualified by an interaction with level (F7,5 = 5.0; p = .04). As illustrated in Figure 3A, across all levels, greedy individuals more often choose to inflate the balloon. Separate correlations for each level revealed that the correlation was significant on levels 4–7, but not on levels 1, 2, 3, and 8, presumably due to variance restriction on these levels. Therefore, we found that individuals with high levels of trait-greed are more likely to make risky decisions. Figure 4A depicts this pattern of results for individuals high and low in trait-greed, respectively (via median-split). While we found no interaction between state-greed and trait-greed across levels (F1,17 = 0.9; p = .35), a significant three way interaction indicated that the interaction between state-greed and trait-greed was qualified by level (F7,18 = 4.3; p = .01). Specifically, after the
induction of state-modesty, trait-greed was not significantly correlated with risk-taking on any level. However, after the induction of state-greed, the correlation between trait-greed and risk-taking was significant on the levels 4, 5, 6, and 7 (see Figure 3B). The difference between the correlation coefficients in the state-greed and state-modesty condition approached significance for level 4 (p = .08) and level 5 (p = .05). Therefore, in line with interactionism, the correlation between trait-greed and risky decision-making was stronger when greed was activated. In line with previous research on the FRN, we found a stronger negativity for unfavorable outcomes (i.e., the balloon bursts) compared to favorable outcomes (i.e., the balloon inflated without bursting; F1,19 = 15.9, p < .01; see Figure 5). We subsequently computed a difference in FRN response for unfavorable compared to favorable outcomes and regressed this difference score on trait-greed. In general, we found a negative relation (r = −.49; p = .03) between trait-greed and the FRN-difference score (see Figure 6A). Specifically, while individuals with low
Figure 4. (A) Risk-taking, operationalized as the relative frequency of pumping up the balloon, for participants high and low in trait-greed (via median-split). (B) Feedback-related negativity (FRN) for favorable (no-burst of the balloon) and non-favorable (burst of the balloon) events, separately for participants high and low in trait-greed (via median-split). High power values indicate a stronger FRN (i.e. more positive amplitudes for the P2 and/or more negative amplitudes for the N2 component).
MUSSEL ET AL.
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
6
Figure 5. (A) Event-related potential for four conditions at medial–frontal position (Fz) as change in voltage (in μV, upward deflections indicating negative deflections) as a function of time (in ms); conditions are defined by outcome (burst of the balloon, black line vs. non-bursts, gray line) and state induction (stategreed, solid line vs. state-modesty, dashed line). (B) Component identified as representing the FRN, based on principal component analysis. (C) Hilbert transformation of the component shown in (B) to adequately quantify the amplitude of the P2–N2 complex. (D) Topography for the component shown in (B) at 284 ms (maximum negative deflection) across all four conditions. (E) Topography for the component shown in (B) at 284 ms for the difference wave between unfavorable events (burst of the balloon) and favorable events (non-bursts).
levels on trait-greed showed the typical FRN effect (i.e., a negative deflection for losses compared to wins), the FRN effect diminished for individuals high on trait-greed (see also Figure 4B). Additionally, we tested separate regressions for stategreed and state-modesty which are depicted in Figure 6B and C. In the state-greed condition, a significant correlation of r = −.54 (p = .01) between traitgreed and the FRN-difference score was found; in the state-modesty condition, the correlation was not significant (r = −.36; p = .12). However, the difference between these correlations was not significant (p = .25). Therefore, we conclude that in a situation where greed is activated, neural responses to negative compared to positive events diminish for individuals high in trait-greed, whereas individuals low in traitgreed show the typical response of stronger negativity to unfavorable compared to favorable outcomes. Finally, we investigated whether neural processes underlying the FRN would account for decision-making. We repeated the mixed model for risk-taking, as described earlier, and additionally included the FRN-difference score between unfavorable and favorable events as covariate. Indeed, we found an additional significant interaction between the FRN-difference score and state-greed (F1,525 = 4.4; p = .04), which was further qualified by level (three-way interaction: F1,21 = 2.5; p = .05). Post hoc correlations revealed a positive correlation (r = .51, p = .02) between the FRN-difference score and risk-taking on level 3 after the induction of state-greed. The positive correlation suggests that, in line with the results reported earlier, a reduced difference between favorable and unfavorable results (resulting in less negative values on the FRN-difference score) predicts higher risk-taking. Corrected for multiple comparisons via the Bonferroni method, this correlation was significantly different form the correlation between FRN-difference score and risk-taking on level 5 after the induction of state-modesty (Δr = .89, p = .003). However, it should be noted that despite the inclusion of the FRN-difference score as covariate, the effects of trait-greed on risk-taking, as reported earlier, remained significant. Therefore, the FRN does not seem to account for all processes underlying the effect of traitgreed on risk-taking.
DISCUSSION In the present study, we investigated whether stateand trait-greed would be related to risky decision-
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
GREED AND RISKY DECISION-MAKING
7
Figure 6. Correlation between trait-greed and feedback-related negativity (FRN) for bursts, compared to non-bursts (computed as difference between the FRN for bursts minus the FRN for non-bursts). A positive value for the FRN effect indicates the typical reaction of an individual to respond with a stronger negative deflection to unfavorable outcomes compared to favorable ones. Trait-greed was assessed by the greed personality self-report measure. (A) Main effect for the correlation between trait-greed and the FRN effect. (B) Correlation between trait-greed and the FRN effect, separately for state-greed. (C) Correlation between trait-greed and the FRN effect, separately for state-modesty. State-greed and state-modesty were induced by reading a biography of either a greedy or a modest person, followed by an additional task that had to be solved from the perspective of the respective character.
making and underlying neural mechanisms. We found support for our hypothesis that greedy individuals take higher risks, compared to less greedy individuals. Additionally, in line with interactionism, the correlation between trait-greed and risky decision-making was stronger when greed was activated. These results suggest that situational properties, such as incentives, bonuses, or organizational values, may moderate the impact of a greedy personality on risky decision-making as they may activate greed, similar to the biographies in the present study. We found neural mechanisms accompanying the feedback processing during risky decision-making to be moderated by trait-greed. Specifically, we found that the typical FRN (Miltner et al., 1997) diminished for individuals high in trait-greed, especially after greed was activated. Additionally, diminished difference scores between favorable and unfavorable events predicted risky decision-making after greed was activated. The FRN that we observed for unfavorable compared to favorable outcomes in individuals low on trait-greed can be interpreted as a stronger negative temporal difference error that indicates events that are “worse than expected”, compared to a positive temporal difference error that indicates events that are “better than expected” (Holroyd & Coles, 2002). This temporal difference error allows for learning, i.e. adjusting expectations as well as future behavior (Sutton & Barto, 1998). In contrast, the diminished FRN that we found for individuals high in trait-greed might indicate difficulties in learning from experience, especially from mistakes, punishment, or negative events. Interestingly, a reduced FRN effect and deficits in conditioning have previously been reported for psychopaths (Birbaumer et al., 2005; Dikman & Allen, 2000). Also, investment bankers and psychopaths
behave similarly in social dilemma situations, where both have been found to more strongly defect than a comparison group (Noll et al., 2012). Therefore, according to our results, the neural pattern of greedy individuals, as reflected by the FRN to favorable compared to unfavorable events, is similar to the neural pattern found for psychopathy. Probably, our results might also explain the occurrence of stock market bubbles. Such bubbles typically emerge when, in a bull market, investors run with their profit too long, ignoring signs of a potential burst of the bubble, like negative corporate data; raising prime rates; or warnings from analysts (Coates, 2011).
Supplemental material Supplementary material is available via the ‘Supplementary’ tab on the article’s online page (http:// dx.doi.org/10.1080/17470919.2014.965340) Original manuscript received 7 April 2014 Revised manuscript accepted 9 September 2014 First published online 9 October 2014
REFERENCES Aquino, K., Freeman, D., Reed, A., Lim, V. K. G., & Felps, W. (2009). Testing a social-cognitive model of moral behavior: The interactive influence of situations and moral identity centrality. Journal of Personality and Social Psychology, 97(1), 123–141. doi:10.1037/A0015406 Balot, R. K. (2001). Greed and injustice in classic Athens. Princeton, NJ: Princeton University Press. Birbaumer, N., Veit, R., Lotze, M., Erb, M., Hermann, C., Grodd, W., & Flor, H. (2005). Deficient fear conditioning in psychopathy: A functional magnetic resonance
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
8
MUSSEL ET AL.
imaging study. Arch Gen Psychiatry, 62(7), 799–805. doi:10.1001/archpsyc.62.7.799 Bosse, T., Siddiqui, G. F., & Treur, J. (2010, August). Supporting financial decision making by an intelligent agent estimating greed and risk. Paper presented at the Fourth International Workshop on Human Aspects in Ambient Intelligence, Toronto. Byrnes, J. P., Miller, D. C., & Schafer, W. D. (1999). Gender differences in risk taking: A meta-analysis. Psychological Bulletin, 125(3), 367–383. doi:10.1037/ 0033-2909.125.3.367 Coates, J. (2011). The hour between dog and wolf: Risktaking, gut feelings and the biology of boom and bust. London: Harper Collins. Cohen, M. X., Elger, C. E., & Fell, J. (2009). Oscillatory activity and phase-amplitude coupling in the human medial frontal cortex during decision making. Journal of Cognitive Neuroscience, Research Support, N.I.H., Extramural Research Support, U.S. Gov’t, Non-P.H.S. 21(2), 390–402. 10.1162/jocn.2008.21020 Daghofer, F. (2007). Financial risk-taking on “Who Wants to Be a Millionaire”: A comparison between Austria, Germany, and Slovenia. International Journal of Psychology, 42(5), 317–330. doi:10.1080/ 00207590600852389 Debener, S., Ullsperger, M., Siegel, M., Fiehler, K., Von Cramon, D. Y., & Engel, A. K. (2005). Trial-by-trial coupling of concurrent electroencephalogram and functional magnetic resonance imaging identifies the dynamics of performance monitoring. Journal of Neuroscience, 25(50), 11730–11737. doi:10.1523/ Jneurosci.3286-05.2005 Delorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134(1), 9–21. doi:10.1016/j. jneumeth.2003.10.009 Dien, J. (2010). The ERP PCA Toolkit: An open source program for advanced statistical analysis of event-related potential data. [Research Support, U.S. Gov’t, Non-P.H. S.] Journal of Neuroscience Methods, 187(1), 138–145. doi:10.1016/j.jneumeth.2009.12.009 Dikman, Z. V., & Allen, J. J. B. (2000). Error monitoring during reward and avoidance learning in high- and lowsocialized individuals. Psychophysiology, 37(1), 43–54. doi:10.1111/1469-8986.3710043 Eek, D., & Biel, A. (2003). The interplay between greed, efficiency, and fairness in public-goods dilemmas. Social Justice Research, 16(3), 195–215. doi:10.1023/ A:1025984611796 Ekehamma, B. (1974). Interactionism in personality from a historical perspective. Psychological Bulletin, 81(12), 1026–1048. doi:10.1037/h0037457 Fiorillo, C. D., Tobler, P. N., & Schultz, W. (2003). Discrete coding of reward probability and uncertainty by dopamine neurons. Science, 299, 1898–1902. doi:10.1126/ science.1077349 Franken, I. A., & Muris, P. (2005). Individual differences in decision-making. Personality and Individual Differences, 39(5), 991–998. doi:10.1016/j.paid.2005.04.004 Gehring, W. J., & Willoughby, A. R. (2002). The medial frontal cortex and the rapid processing of monetary gains and losses. Science, 295, 2279–2282. doi:10.1126/ science.1066893
Holroyd, C. B., & Coles, M. G. H. (2002). The neural basis of human error processing: Reinforcement learning, dopamine, and the error-related negativity. Psychological Review, 109(4), 679–709. doi:10.1037/ 0033-295X.109.4.679 Krueger, R. F., Markon, K. E., Patrick, C. J., Benning, S. D., & Kramer, M. D. (2007). Linking antisocial behavior, substance use, and personality: An integrative quantitative model of the adult externalizing spectrum. Journal of Abnormal Psychology, 116(4), 645–666. doi:10.1037/ 0021-843x.116.4.645 Lejuez, C. W., Aklin, W. M., Zvolensky, M. J., & Pedulla, C. M. (2003). Evaluation of the Balloon Analogue Risk Task (BART) as a predictor of adolescent real-world risk-taking behaviours. Journal of Adolescence, 26(4), 475–479. doi:10.1016/S01401971(03)00036-8 Lejuez, C. W., Read, J. P., Kahler, C. W., Richards, J. B., Ramsey, S. E., Stuart, G. L., & Brown, R. A. (2002). Evaluation of a behavioral measure of risk taking: The Balloon Analogue Risk Task (BART). Journal of Experimental Psychology-Applied, 8(2), 75–84. doi:10.1037/1076-898X.8.2.75 Lo, A. W., Repin, D. V., & Steenbarger, B. N. (2005). Fear and greed in financial markets: A clinical study of daytraders. American Economic Review, 95(2), 352–359. doi:10.1257/000282805774670095 Miltner, W. H. R., Braun, C. H., & Coles, M. G. H. (1997). Event-related brain potentials following incorrect feedback in a time-estimation task: Evidence for a “generic” neural system for error detection. Journal of Cognitive Neuroscience, 9(6), 788–798. doi:10.1162/jocn. 1997.9.6.788 Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological Review, 80(4), 252–283. doi:10.1037/h0035002 Montague, P. R., Hyman, S. E., & Cohen, J. D. (2004). Computational roles for dopamine in behavioural control. Nature, 431(7010), 760–767. doi:10.1038/ Nature03015 Noll, T., Endrass, J., Scherrer, P., Rossegger, A., Urbaniok, F., & Mokros, A. (2012). A comparison of professional traders and psychopaths in a simulated non-zero sum game. Catalyst: A Social Justice Forum, 2, 1–13. Patrick, C. J., Hicks, B. M., Krueger, R. F., & Lang, A. R. (2005). Relations between psychopathy facets and externalizing in a criminal offender sample. Journal of Personality Disorders, 19(4), 339–356. doi:10.1521/ pedi.2005.19.4.339 Pessiglione, M., Seymour, B., Flandin, G., Dolan, R. J., & Frith, C. D. (2006). Dopamine-dependent prediction errors underpin reward-seeking behaviour in humans. Nature, 442(7106), 1042–1045. doi:10.1038/ Nature05051 Piff, P. K., Stancato, D. M., Cote, S., Mendoza-Denton, R., & Keltner, D. (2012). Higher social class predicts increased unethical behavior. Proceedings of the National Academy of Sciences of the United States of America, 109(11), 4086–4091. doi:10.1073/pnas.1118373109 Rokeach, M. (1973). The nature of human values. New York, NY: Free Press. Schultz, W. (2002). Getting formal with dopamine and reward. Neuron, 36(2), 241–263. doi:10.1016/S08966273(02)00967-4
GREED AND RISKY DECISION-MAKING
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
Schwartz, S. H. (1992). Universals in the content and structure of values: Theory and empirical tests in 20 countries. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 25, pp. 1–65). New York, NY: Academic Press. doi:10.1016/S0065-2601(08)60281-6 Sutton, R. S., & Barto, A. G. (1998). Reinforcement learning: An introduction. Cambridge, MA: MIT Press.
9
Tang, T. L. P. (2007). Income and quality of life: Does the love of money make a difference? Journal of Business Ethics, 72(4), 375–393. doi:10.1007/s10551006-9176-4 Vohs, K. D., Mead, N. L., & Goode, M. R. (2006). The psychological consequences of money. Science, 314 (5802), 1154–1156. doi:10.1126/science.1132491
Social Neuroscience Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/psns20
State- and trait-greed, its impact on risky decisionmaking and underlying neural mechanisms a
b
a
a
Patrick Mussel , Andrea M. F. Reiter , Roman Osinsky & Johannes Hewig a
Department of Psychology I, Differential Psychology, Personality Psychology, and Psychological Diagnostics, Julius Maximilians University Würzburg, Würzburg, Germany b
Max Planck Fellow Group “Cognitive and Affective Control of Behavioural Adaptation”, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany Published online: 10 Oct 2014.
To cite this article: Patrick Mussel, Andrea M. F. Reiter, Roman Osinsky & Johannes Hewig (2014): State- and trait-greed, its impact on risky decision-making and underlying neural mechanisms, Social Neuroscience, DOI: 10.1080/17470919.2014.965340 To link to this article: http://dx.doi.org/10.1080/17470919.2014.965340
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
SOCIAL NEUROSCIENCE, 2014 http://dx.doi.org/10.1080/17470919.2014.965340
State- and trait-greed, its impact on risky decision-making and underlying neural mechanisms Patrick Mussel1, Andrea M. F. Reiter2, Roman Osinsky1, and Johannes Hewig1
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
1
Department of Psychology I, Differential Psychology, Personality Psychology, and Psychological Diagnostics, Julius Maximilians University Würzburg, Würzburg, Germany 2 Max Planck Fellow Group “Cognitive and Affective Control of Behavioural Adaptation”, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
We investigated whether greed would predict risky decision-making and recorded neural responses during a monetary gambling task using the electroencephalogram. We found that individuals high in trait-greed took higher risks to maximize monetary outcome. Furthermore, this relation was moderated by state-greed; specifically, traitgreed had a stronger impact on risky decision-making when activated by situational characteristics. On the neural level, greedy individuals showed a specific response to favorable and unfavorable outcomes. Specifically, they had a reduced feedback-related negativity-difference score to these events, indicating that they might have difficulty in learning from experience, especially from mistakes and negative feedback. It is concluded that greed may explain risky and reckless behavior in diverse settings, such as investment banking, and may account for phenomena such as stock market bubbles.
Keywords: Cognitive neuroscience; Personality; Risk taking; Feedback-related negativity; Psychopathy.
The financial crisis in the recent past, initiated by the subprime mortgage crisis in the United States and the debt crisis in Europe, evolved into a global economy crisis, affecting the real economy with profit shrinkage, downsizing, unemployment, and insolvency. As a consequence, governments were forced to spend vast sums of money for banking bailout and economic growth packages. The causes for this development lay not only in structural problems, such as rise of national debts or failures in financial regulation, but also in excessive risk taking by agents operating in investment departments of financial firms or stock exchanges. We investigated whether greed is a predictor of risky decision-making. Greed can be defined as desire to get more at all costs, including the excessive striving for desired goods and the willingness to accept that such
striving may be at the expense of others (Balot, 2001). As such, greed is associated with concepts of money, economic values such as wealth, and striving for power (Rokeach, 1973; Schwartz, 1992; Tang, 2007). Additionally, greed can be linked to antisocial, unethical, and deviant behavior (Krueger, Markon, Patrick, Benning, & Kramer, 2007; Patrick, Hicks, Krueger, & Lang, 2005): Individuals with high values on greed may justify the rightness of their actions by its instrumental value for increasing targeted goods, thereby qualifying socially shared norms, moral standards, rules, and laws which ultimately justifies unethical and deviant behavior (Aquino, Freeman, Reed, Lim, & Felps, 2009; Bosse, Siddiqui, & Treur, 2010; Eek & Biel, 2003; Piff, Stancato, Cote, Mendoza-Denton, & Keltner, 2012; Vohs, Mead, & Goode, 2006). Based on this definition,
Correspondence should be addressed to: Patrick Mussel, Department of Psychology I, Differential Psychology, Personality Psychology, and Psychological Diagnostics, Julius Maximilians University Würzburg, Marcusstr. 9-11, 97070 Würzburg, Germany. E-mail: [email protected] The research was conducted at the Julius Maximilians University Würzburg. This research was supported by a Schumpeter-Fellowship by the VolkswagenStiftung [85650; II/86486].
© 2014 Taylor & Francis
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
2
MUSSEL ET AL.
it is conceivable that greedy individuals will take excessive risks to increase profits, thereby accepting the possibility of enormous losses for their department, their company, or even the society. In addition to the influence of personality, the behavior of agents operating in investment departments of financial firms or stock exchanges is likely to be influenced by situational demands or incentives, like excessive yield on turnover associated with personal bonuses (Lo, Repin, & Steenbarger, 2005). From the perspective of interactionism (Mischel, 1973), situational attributes may activate personality traits and thus increase their impact on behavior (Ekehamma, 1974; Franken & Muris, 2005). Therefore, we propose that greed as a stable personality trait predicts risky decision-making, especially when this personality trait is activated by situational characteristics. When taking risky decisions, as well as in other situations, individuals try to adjust their behavior according to positive and negative feedback provided by their environment in order to optimize desirable outcomes. A theoretical framework that accounts for such processes is reinforcement learning theory (Sutton & Barto, 1998), which posits that individuals have expectations regarding upcoming events in the future, and that deviations from these expectations are used to learn from experience and, subsequently, adapt behavior. Specifically, events that are “worse than expected”, such as punishment or absence of reward, evoke a negative temporal difference error, whereas events that are “better than expected”, such as reward, evoke a positive temporal difference error. Neuropsychological indicators have been linked to these processes and allow their investigation during decisionmaking. Specifically, the negative temporal difference error has been related to a phasic decrease in dopaminergic signaling in basal ganglia, followed by a disinhibition of apical dendrites of the motor neurons of the anterior cingulate cortex, eliciting a feedback-related negativity (FRN) that can be measured at the scalp using electroencephalography (EEG) (Debener et al., 2005; Fiorillo, Tobler, & Schultz, 2003; Gehring & Willoughby, 2002; Holroyd & Coles, 2002; Montague, Hyman, & Cohen, 2004; Pessiglione, Seymour, Flandin, Dolan, & Frith, 2006; Schultz, 2002). The FRN is a negative deflection with a maximum at fronto-central electrode positions approximately 250–350 ms after the onset of negative, compared to positive stimuli (Miltner, Braun, & Coles, 1997). We assessed this component during a risky decision-making task to investigate whether state- and traitgreed would moderate neural processes associated with adapting behavior and learning from experience. In the following, we report results from a risktaking paradigm, the Balloon-Analogue-Risk-Task
(BART) (Lejuez et al., 2002). We predicted that risky decision-making would be related to traitgreed. To investigate whether the relation between trait-greed and risk taking is moderated by situational characteristics, we used cues which induced either state-modesty or state-greed. We predicted that the relation between trait-greed and risk-taking should be stronger in the state-greed condition, compared to the state-modesty condition. During the risk-taking task, we recorded event-related potentials to investigate the neural mechanisms underlying greedy decision-making.
METHOD Participants Twenty participants were recruited from the student population of the Julius Maximilians University Würzburg. All participants studied economics and were between 20 and 31 years (on average 24.5). As gender is known to impact risk-taking behavior (Byrnes, Miller, & Schafer, 1999; Daghofer, 2007), only male participants were recruited for the study. For their participation, they were paid € 20; additionally, they could win € 100 if they obtained the highest fictive account balance in the decision-making task.
Task Risky decision-making was assessed using the BART (Lejuez et al., 2002). Decision-making in the BART has been found to be related to various real-world risk-taking behaviors that include alcohol and drug use, cigarette smoking, gambling, theft, aggression, psychopathy, and unprotected sexual intercourse (Lejuez, Aklin, Zvolensky, & Pedulla, 2003; Lejuez et al., 2002). Participants played two blocks of 150 trials, each block under 1 of 2 conditions (see below). In each trial, a picture of a balloon was presented on a screen representing a value of € 1000,– (see Figure S1 in the Supplemental material available online). Subsequently, individuals had to decide whether they wanted to inflate the balloon or not. If they rejected to inflate the balloon, the money was collected on a fictive account. If they decided to blow up the balloon (risky decision), it either burst, resulting in a total loss of the money for the current trial, or inflated, resulting in double value (i.e., € 2000,–). In the latter case, participants had to decide again whether they wanted to inflate the balloon or not. As the balloon inflated, the risk of bursting increased from 15% to 100% by 7
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
GREED AND RISKY DECISION-MAKING
percentage points per level (i.e., 15%; 22%; 29%; . . .). Participant had been informed that the player with the highest fictive account balance in the BART won € 100 as financial incentive. Risky choices were defined as the number of decisions to pump up the balloon, compared to all decisions when the balloon had not yet burst. Therefore, on level 1, risky choice was computed as the number of decisions to pump up the balloon, compared to all 300 trials. As the balloon had a 15% probability of bursting on level 1, the maximum number of risky choice in level 2 is 85% * 300 trials = 255 trials; therefore, risky choice in level 2 was computed as the number of risky decisions to pump up the balloon in level 2, compared to 255 trials, and so forth.
Measures We developed a self-report measure for assessing greed as a personality trait. Based on the definition provided earlier, a pool of personality items was developed and applied to several convenience samples of students (total N = 640). Based on item analyses, seven items were chosen (see additional Table S1 available online). An example item is “When I think about all the things I have, my first thought is about what I would like to have next”. The trait-greed measure had a one-factorial structure (eigenvalues 3.2; 0.92; 0.73; 0.59; 0.55; 0.52; 0.46) and acceptable internal consistency (.79). Additional evidence regarding the validity of the trait-greed measure can be found in the Supplemental material available online.
Procedure After receiving verbal instructions about the experiment, participants gave written informed consent for participation (see Figure 1). Next, participants were given the
3
biography of either a greedy or a modest character (randomized across participants) to induce state-greed and state-modesty, respectively. Each biography was fictitious however based upon famous personae from financial management. In the state-greed condition, greed was activated by a fictitious biography of a successful and greedy person. The biography contained traits of the person (e.g. “uncompromising when he really wants something”), behaviors (e.g. “developed a highly profitable business idea”), life events (“investors plastered him with money”), and affective reactions (e.g. “it’s an exciting thrill, like ecstasy”). In the state-modest condition, a paralleled biography of a successful, but more modest person was used (e.g., “is conscientious”, “developed a sustainable management”). Participants were told that the first part of the assessment aimed to test their skills to put themselves in someone else’s position. Therefore, they had to read the biography and subsequently complete a three-item situational judgment test (SJT) from the perspective of the character described in the biography, which was also intended to intensify the induction of state-greed and state-modesty. Each item contained an item stem, describing a work situation (counseling a customer regarding an investment decision; negotiate with a supervisor regarding pay rise; deciding upon a stock investment, based on an insider tip). Followed by the item stem, five behavioral options were presented that differed in terms of greediness. Participants had to rank the options from 1 (behavior that the character would most likely exert) to 5 (behavior that the character would least likely exert). Each option was rated for greediness by three experts that allowed for calculating a “greed-score” for each situational judgment item. Across the three items, we found large differences between the two conditions (t = 15.7, p ≤ .001; d = 5.4); in the state-greedy conditions, greedy behavior options were much more likely chosen, compared to the state-modesty condition, indicating that participants read the biography thoroughly, understood it fully and were capable to adopt the perspective of the described
Figure 1. Time line and experimental design. Biography 1 and 2 were fictive biographies of either a greedy or a modest person (assigned randomized to either Biography 1 or 2 across participants). The SJT includes three scenarios with five behavioral options each which were filled out from the perspective of the person described in the biography. Each BART consisted of 150 trials. During the BART, we recorded neural responses to favorable (non-burst of the balloon) and unfavorable (burst of the balloon) outcomes via EEG. The questionnaire at the end included the seven-item trait-greed measure.
4
MUSSEL ET AL.
character. To control for confounding effects of the state induction, participants filled out rating scales for valence and arousal using 5-point Likert scales. We found no effect of greed induction on either valence (t = 0.8, p = .45) or arousal (t = −0.3, p = .77). After the BART task, participants filled out the trait-greed measure, as described above. They were informed about the purpose of the study and paid their allowance.
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
EEG recording and quantification While participants performed the BART, EEG (analog bandpass: 0.1–80 Hz, sampling rate: 250 Hz) was recorded from 31 scalp sites according to the 10–20 system, using Ag/AgCl electrodes and a BrainAmpDC amplifier (Brain Products GmbH, Gilching, Germany). Impedances were kept below 10 kΩ and electrodes were referenced to the vertex (Cz). For the detection of blinks and eye movements the horizontal and vertical electrooculogram was recorded. Data were processed offline, using MATLAB R2011 b (MathWorks, Natick, MA) and the Toolbox EEGLAB 12.0.1 (Delorme & Makeig, 2004). First, data were filtered, using a 30 Hz low-pass filter and a 0.5 Hz high-pass filter. Subsequently, the EEG was segmented into feedback-locked epochs of 1000 ms (−200 to 800 ms), baseline-corrected (−200 to 0 ms), and re-referenced to averaged mastoid electrodes. For artifact rejection, trials in which the amplitude exceeded the criterion of 3.5 standard deviations were excluded from further analyses (7.8% of the trials). At least 49 artifact-free trials were available per participant and condition. Next, we used independent component analysis decomposition for the detection of eye blink and movement artifacts. Components representing artifacts were detected using a semi-automated procedure and were subsequently removed from the dataset. Finally, data were averaged for each participant and each of four conditions, defined by the two factors state-greed (greedy vs. modest) and outcome (burst vs. non-burst of the balloon) (see Figure 5A). We used principal component analysis (PCA) to identify the FRN component, using the MATLAB Toolbox ERP PCA 2.32 (Dien, 2010). A temporal PCA across the 250 time points of the epoched and averaged data was computed, following Promax rotation of the factors. The covariance matrix and Kaiser normalization were used for the PCA. Upon inspection of the scree plot we decided to extract 26 factors. Based on visual inspection, we identified a factor that readily responded to the FRN (see Figure 5B). The factor is characterized by a fronto-central maximum, a typical negativity between 270 and 310 ms with a
preceding positivity, and a frequency in the upper theta–lower alpha range (8 Hz). The PCA waveform reveals a difference between outcomes, driven by a negative response to the unfavorable event (burst of the balloon) that is considerably reduced on favorable events (non-burst of the balloon). We also performed a separate spatial PCA on the 26 temporal factors with Infomax rotation. Two factors were extracted per temporal factor. Inspection of the results regarding the factor representing the FRN revealed two factors with very similar topography (both fronto-central). Presumably, there were no separable activities which overlapped in time at different scalp locations. Therefore, we stuck to the factor obtained via temporal PCA and reconstructed the waveforms by multiplying the factor pattern matrix with the standard deviations to convert the signal to microvolts. Subsequently, we used a Hilbert transformation to adequately quantify the amplitude of the P2–N2 complex (Cohen, Elger, & Fell, 2009) (see Figure 5C). The Hilbert transformation provides the complex analytic signal, z(t). The squared modulus of the resulting complex signal represents the power at each time point. Finally, data were log-transformed and averaged across 150–350 ms for each participant and each of the four conditions and subsequently normalized.
RESULTS The maximum number of times that a balloon was pumped up was eight (see Supplementary Figure S1 for details regarding the task). As depicted in Figure 2, participants almost always inflated the balloon in levels 1 and 2. In 56% of the trials, the balloon was pumped up at least four times. Using linear mixed
Figure 2. Behavioral results for the BART task. The maximum number of times that a balloon was pumped up was eight. For each of the eight levels, the relative frequency of pumping up the balloon (i.e., risky decision) is given, computed as the absolute number of “pump up” decisions divided by the number of trials in which the balloon had not yet burst.
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
GREED AND RISKY DECISION-MAKING
5
Figure 3. Effect of greed on risky decision-making. Greed was assessed on the trait-level, using a personality self-report measure. Risk was operationalized as the relative frequency of pumping up the balloon, compared to the number of trials in which the balloon had not yet burst. (A) Main effect of trait-greed on risk-taking across all eight levels. (B) Correlation between trait-greed and risk-taking separately for the eight levels and for the induction of state-greed (upper row) and state-modesty. State-greed and state-modesty were induced by reading a biography of either a greedy or a modest person, followed by an additional task that had to be solved from the perspective of the respective character. *p < .05.
models with unstructured repeated covariance matrices with the decision to pump up the balloon as dependent variable and standardized trait-greed as covariate, we found a significant main effect of traitgreed on risky decision-making (F1,8 = 6.3; p = .04), which was qualified by an interaction with level (F7,5 = 5.0; p = .04). As illustrated in Figure 3A, across all levels, greedy individuals more often choose to inflate the balloon. Separate correlations for each level revealed that the correlation was significant on levels 4–7, but not on levels 1, 2, 3, and 8, presumably due to variance restriction on these levels. Therefore, we found that individuals with high levels of trait-greed are more likely to make risky decisions. Figure 4A depicts this pattern of results for individuals high and low in trait-greed, respectively (via median-split). While we found no interaction between state-greed and trait-greed across levels (F1,17 = 0.9; p = .35), a significant three way interaction indicated that the interaction between state-greed and trait-greed was qualified by level (F7,18 = 4.3; p = .01). Specifically, after the
induction of state-modesty, trait-greed was not significantly correlated with risk-taking on any level. However, after the induction of state-greed, the correlation between trait-greed and risk-taking was significant on the levels 4, 5, 6, and 7 (see Figure 3B). The difference between the correlation coefficients in the state-greed and state-modesty condition approached significance for level 4 (p = .08) and level 5 (p = .05). Therefore, in line with interactionism, the correlation between trait-greed and risky decision-making was stronger when greed was activated. In line with previous research on the FRN, we found a stronger negativity for unfavorable outcomes (i.e., the balloon bursts) compared to favorable outcomes (i.e., the balloon inflated without bursting; F1,19 = 15.9, p < .01; see Figure 5). We subsequently computed a difference in FRN response for unfavorable compared to favorable outcomes and regressed this difference score on trait-greed. In general, we found a negative relation (r = −.49; p = .03) between trait-greed and the FRN-difference score (see Figure 6A). Specifically, while individuals with low
Figure 4. (A) Risk-taking, operationalized as the relative frequency of pumping up the balloon, for participants high and low in trait-greed (via median-split). (B) Feedback-related negativity (FRN) for favorable (no-burst of the balloon) and non-favorable (burst of the balloon) events, separately for participants high and low in trait-greed (via median-split). High power values indicate a stronger FRN (i.e. more positive amplitudes for the P2 and/or more negative amplitudes for the N2 component).
MUSSEL ET AL.
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
6
Figure 5. (A) Event-related potential for four conditions at medial–frontal position (Fz) as change in voltage (in μV, upward deflections indicating negative deflections) as a function of time (in ms); conditions are defined by outcome (burst of the balloon, black line vs. non-bursts, gray line) and state induction (stategreed, solid line vs. state-modesty, dashed line). (B) Component identified as representing the FRN, based on principal component analysis. (C) Hilbert transformation of the component shown in (B) to adequately quantify the amplitude of the P2–N2 complex. (D) Topography for the component shown in (B) at 284 ms (maximum negative deflection) across all four conditions. (E) Topography for the component shown in (B) at 284 ms for the difference wave between unfavorable events (burst of the balloon) and favorable events (non-bursts).
levels on trait-greed showed the typical FRN effect (i.e., a negative deflection for losses compared to wins), the FRN effect diminished for individuals high on trait-greed (see also Figure 4B). Additionally, we tested separate regressions for stategreed and state-modesty which are depicted in Figure 6B and C. In the state-greed condition, a significant correlation of r = −.54 (p = .01) between traitgreed and the FRN-difference score was found; in the state-modesty condition, the correlation was not significant (r = −.36; p = .12). However, the difference between these correlations was not significant (p = .25). Therefore, we conclude that in a situation where greed is activated, neural responses to negative compared to positive events diminish for individuals high in trait-greed, whereas individuals low in traitgreed show the typical response of stronger negativity to unfavorable compared to favorable outcomes. Finally, we investigated whether neural processes underlying the FRN would account for decision-making. We repeated the mixed model for risk-taking, as described earlier, and additionally included the FRN-difference score between unfavorable and favorable events as covariate. Indeed, we found an additional significant interaction between the FRN-difference score and state-greed (F1,525 = 4.4; p = .04), which was further qualified by level (three-way interaction: F1,21 = 2.5; p = .05). Post hoc correlations revealed a positive correlation (r = .51, p = .02) between the FRN-difference score and risk-taking on level 3 after the induction of state-greed. The positive correlation suggests that, in line with the results reported earlier, a reduced difference between favorable and unfavorable results (resulting in less negative values on the FRN-difference score) predicts higher risk-taking. Corrected for multiple comparisons via the Bonferroni method, this correlation was significantly different form the correlation between FRN-difference score and risk-taking on level 5 after the induction of state-modesty (Δr = .89, p = .003). However, it should be noted that despite the inclusion of the FRN-difference score as covariate, the effects of trait-greed on risk-taking, as reported earlier, remained significant. Therefore, the FRN does not seem to account for all processes underlying the effect of traitgreed on risk-taking.
DISCUSSION In the present study, we investigated whether stateand trait-greed would be related to risky decision-
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
GREED AND RISKY DECISION-MAKING
7
Figure 6. Correlation between trait-greed and feedback-related negativity (FRN) for bursts, compared to non-bursts (computed as difference between the FRN for bursts minus the FRN for non-bursts). A positive value for the FRN effect indicates the typical reaction of an individual to respond with a stronger negative deflection to unfavorable outcomes compared to favorable ones. Trait-greed was assessed by the greed personality self-report measure. (A) Main effect for the correlation between trait-greed and the FRN effect. (B) Correlation between trait-greed and the FRN effect, separately for state-greed. (C) Correlation between trait-greed and the FRN effect, separately for state-modesty. State-greed and state-modesty were induced by reading a biography of either a greedy or a modest person, followed by an additional task that had to be solved from the perspective of the respective character.
making and underlying neural mechanisms. We found support for our hypothesis that greedy individuals take higher risks, compared to less greedy individuals. Additionally, in line with interactionism, the correlation between trait-greed and risky decision-making was stronger when greed was activated. These results suggest that situational properties, such as incentives, bonuses, or organizational values, may moderate the impact of a greedy personality on risky decision-making as they may activate greed, similar to the biographies in the present study. We found neural mechanisms accompanying the feedback processing during risky decision-making to be moderated by trait-greed. Specifically, we found that the typical FRN (Miltner et al., 1997) diminished for individuals high in trait-greed, especially after greed was activated. Additionally, diminished difference scores between favorable and unfavorable events predicted risky decision-making after greed was activated. The FRN that we observed for unfavorable compared to favorable outcomes in individuals low on trait-greed can be interpreted as a stronger negative temporal difference error that indicates events that are “worse than expected”, compared to a positive temporal difference error that indicates events that are “better than expected” (Holroyd & Coles, 2002). This temporal difference error allows for learning, i.e. adjusting expectations as well as future behavior (Sutton & Barto, 1998). In contrast, the diminished FRN that we found for individuals high in trait-greed might indicate difficulties in learning from experience, especially from mistakes, punishment, or negative events. Interestingly, a reduced FRN effect and deficits in conditioning have previously been reported for psychopaths (Birbaumer et al., 2005; Dikman & Allen, 2000). Also, investment bankers and psychopaths
behave similarly in social dilemma situations, where both have been found to more strongly defect than a comparison group (Noll et al., 2012). Therefore, according to our results, the neural pattern of greedy individuals, as reflected by the FRN to favorable compared to unfavorable events, is similar to the neural pattern found for psychopathy. Probably, our results might also explain the occurrence of stock market bubbles. Such bubbles typically emerge when, in a bull market, investors run with their profit too long, ignoring signs of a potential burst of the bubble, like negative corporate data; raising prime rates; or warnings from analysts (Coates, 2011).
Supplemental material Supplementary material is available via the ‘Supplementary’ tab on the article’s online page (http:// dx.doi.org/10.1080/17470919.2014.965340) Original manuscript received 7 April 2014 Revised manuscript accepted 9 September 2014 First published online 9 October 2014
REFERENCES Aquino, K., Freeman, D., Reed, A., Lim, V. K. G., & Felps, W. (2009). Testing a social-cognitive model of moral behavior: The interactive influence of situations and moral identity centrality. Journal of Personality and Social Psychology, 97(1), 123–141. doi:10.1037/A0015406 Balot, R. K. (2001). Greed and injustice in classic Athens. Princeton, NJ: Princeton University Press. Birbaumer, N., Veit, R., Lotze, M., Erb, M., Hermann, C., Grodd, W., & Flor, H. (2005). Deficient fear conditioning in psychopathy: A functional magnetic resonance
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
8
MUSSEL ET AL.
imaging study. Arch Gen Psychiatry, 62(7), 799–805. doi:10.1001/archpsyc.62.7.799 Bosse, T., Siddiqui, G. F., & Treur, J. (2010, August). Supporting financial decision making by an intelligent agent estimating greed and risk. Paper presented at the Fourth International Workshop on Human Aspects in Ambient Intelligence, Toronto. Byrnes, J. P., Miller, D. C., & Schafer, W. D. (1999). Gender differences in risk taking: A meta-analysis. Psychological Bulletin, 125(3), 367–383. doi:10.1037/ 0033-2909.125.3.367 Coates, J. (2011). The hour between dog and wolf: Risktaking, gut feelings and the biology of boom and bust. London: Harper Collins. Cohen, M. X., Elger, C. E., & Fell, J. (2009). Oscillatory activity and phase-amplitude coupling in the human medial frontal cortex during decision making. Journal of Cognitive Neuroscience, Research Support, N.I.H., Extramural Research Support, U.S. Gov’t, Non-P.H.S. 21(2), 390–402. 10.1162/jocn.2008.21020 Daghofer, F. (2007). Financial risk-taking on “Who Wants to Be a Millionaire”: A comparison between Austria, Germany, and Slovenia. International Journal of Psychology, 42(5), 317–330. doi:10.1080/ 00207590600852389 Debener, S., Ullsperger, M., Siegel, M., Fiehler, K., Von Cramon, D. Y., & Engel, A. K. (2005). Trial-by-trial coupling of concurrent electroencephalogram and functional magnetic resonance imaging identifies the dynamics of performance monitoring. Journal of Neuroscience, 25(50), 11730–11737. doi:10.1523/ Jneurosci.3286-05.2005 Delorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134(1), 9–21. doi:10.1016/j. jneumeth.2003.10.009 Dien, J. (2010). The ERP PCA Toolkit: An open source program for advanced statistical analysis of event-related potential data. [Research Support, U.S. Gov’t, Non-P.H. S.] Journal of Neuroscience Methods, 187(1), 138–145. doi:10.1016/j.jneumeth.2009.12.009 Dikman, Z. V., & Allen, J. J. B. (2000). Error monitoring during reward and avoidance learning in high- and lowsocialized individuals. Psychophysiology, 37(1), 43–54. doi:10.1111/1469-8986.3710043 Eek, D., & Biel, A. (2003). The interplay between greed, efficiency, and fairness in public-goods dilemmas. Social Justice Research, 16(3), 195–215. doi:10.1023/ A:1025984611796 Ekehamma, B. (1974). Interactionism in personality from a historical perspective. Psychological Bulletin, 81(12), 1026–1048. doi:10.1037/h0037457 Fiorillo, C. D., Tobler, P. N., & Schultz, W. (2003). Discrete coding of reward probability and uncertainty by dopamine neurons. Science, 299, 1898–1902. doi:10.1126/ science.1077349 Franken, I. A., & Muris, P. (2005). Individual differences in decision-making. Personality and Individual Differences, 39(5), 991–998. doi:10.1016/j.paid.2005.04.004 Gehring, W. J., & Willoughby, A. R. (2002). The medial frontal cortex and the rapid processing of monetary gains and losses. Science, 295, 2279–2282. doi:10.1126/ science.1066893
Holroyd, C. B., & Coles, M. G. H. (2002). The neural basis of human error processing: Reinforcement learning, dopamine, and the error-related negativity. Psychological Review, 109(4), 679–709. doi:10.1037/ 0033-295X.109.4.679 Krueger, R. F., Markon, K. E., Patrick, C. J., Benning, S. D., & Kramer, M. D. (2007). Linking antisocial behavior, substance use, and personality: An integrative quantitative model of the adult externalizing spectrum. Journal of Abnormal Psychology, 116(4), 645–666. doi:10.1037/ 0021-843x.116.4.645 Lejuez, C. W., Aklin, W. M., Zvolensky, M. J., & Pedulla, C. M. (2003). Evaluation of the Balloon Analogue Risk Task (BART) as a predictor of adolescent real-world risk-taking behaviours. Journal of Adolescence, 26(4), 475–479. doi:10.1016/S01401971(03)00036-8 Lejuez, C. W., Read, J. P., Kahler, C. W., Richards, J. B., Ramsey, S. E., Stuart, G. L., & Brown, R. A. (2002). Evaluation of a behavioral measure of risk taking: The Balloon Analogue Risk Task (BART). Journal of Experimental Psychology-Applied, 8(2), 75–84. doi:10.1037/1076-898X.8.2.75 Lo, A. W., Repin, D. V., & Steenbarger, B. N. (2005). Fear and greed in financial markets: A clinical study of daytraders. American Economic Review, 95(2), 352–359. doi:10.1257/000282805774670095 Miltner, W. H. R., Braun, C. H., & Coles, M. G. H. (1997). Event-related brain potentials following incorrect feedback in a time-estimation task: Evidence for a “generic” neural system for error detection. Journal of Cognitive Neuroscience, 9(6), 788–798. doi:10.1162/jocn. 1997.9.6.788 Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological Review, 80(4), 252–283. doi:10.1037/h0035002 Montague, P. R., Hyman, S. E., & Cohen, J. D. (2004). Computational roles for dopamine in behavioural control. Nature, 431(7010), 760–767. doi:10.1038/ Nature03015 Noll, T., Endrass, J., Scherrer, P., Rossegger, A., Urbaniok, F., & Mokros, A. (2012). A comparison of professional traders and psychopaths in a simulated non-zero sum game. Catalyst: A Social Justice Forum, 2, 1–13. Patrick, C. J., Hicks, B. M., Krueger, R. F., & Lang, A. R. (2005). Relations between psychopathy facets and externalizing in a criminal offender sample. Journal of Personality Disorders, 19(4), 339–356. doi:10.1521/ pedi.2005.19.4.339 Pessiglione, M., Seymour, B., Flandin, G., Dolan, R. J., & Frith, C. D. (2006). Dopamine-dependent prediction errors underpin reward-seeking behaviour in humans. Nature, 442(7106), 1042–1045. doi:10.1038/ Nature05051 Piff, P. K., Stancato, D. M., Cote, S., Mendoza-Denton, R., & Keltner, D. (2012). Higher social class predicts increased unethical behavior. Proceedings of the National Academy of Sciences of the United States of America, 109(11), 4086–4091. doi:10.1073/pnas.1118373109 Rokeach, M. (1973). The nature of human values. New York, NY: Free Press. Schultz, W. (2002). Getting formal with dopamine and reward. Neuron, 36(2), 241–263. doi:10.1016/S08966273(02)00967-4
GREED AND RISKY DECISION-MAKING
Downloaded by [University of Southern Queensland] at 14:07 12 October 2014
Schwartz, S. H. (1992). Universals in the content and structure of values: Theory and empirical tests in 20 countries. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 25, pp. 1–65). New York, NY: Academic Press. doi:10.1016/S0065-2601(08)60281-6 Sutton, R. S., & Barto, A. G. (1998). Reinforcement learning: An introduction. Cambridge, MA: MIT Press.
9
Tang, T. L. P. (2007). Income and quality of life: Does the love of money make a difference? Journal of Business Ethics, 72(4), 375–393. doi:10.1007/s10551006-9176-4 Vohs, K. D., Mead, N. L., & Goode, M. R. (2006). The psychological consequences of money. Science, 314 (5802), 1154–1156. doi:10.1126/science.1132491
