Running Head: ESCALATION of COMMITMENT, EMOTIONAL PERCEPTION, VISUOSPATIAL INFLUENCES

1

Running Head: ESCALATION OF COMMITMENT, EMOTIONAL PERCEPTION, VISUOSPATIAL INFLUENCES, HEMISPHERIC SPECIALIZATION

Emotional Enhancement of Escalation of Commitment: Hemispheric Differences in Auctions

Yuan Hang Li

Ming-Hong Tsai

Eran Zaidel

University of California, Los Angeles

ABSTRACT

The current study examined the conjunctive effects of emotional content (angry or neutral faces) and visuospatial presentation (visual fields stimulatingthe cerebral hemispheres) of the opponent bidder’s face on escalation of commitment. When the opponent displays an angry face compared with a neutral face to the Right Hemisphere (RH), participants are more likely to have increased anxiety and therefore increased escalation. However, when the opponent’s face was shown to the right visual hemifield, the difference in escalation disappears. This study suggests that the RH provides a strong connection between emotional content and escalation of commitment in an auction task.

Keywords:

Decision making; Emotions; Hemispheric Specialization

INTRODUCTION

Decisions are often made with relative uncertainty of the ensuing outcomes. However, in certain circumstances, feedback is given and the decision maker has a choice of continuing along with his original path or changing his course. In fact, when negative feedback was provided to decision makers regarding their investments, they often brashly continue contributing time and resources to their failing projects (Brockner, 1992; Conlon & Parks, 1987). This effort spent on attempting to turn an investment around is described as an escalation of commitment on prior choices (see Staw, 1997 for review).

Situations of escalating commitment can vary from investing more in a failing company, paying more to complete projects with unforeseen costs, or doubling the bet at the gambling table when one is on a losing streak. However, they all share the common features of 1) initial investments of resources (i.e., sunken costs), 2) negative feedback on the current course of action (e.g., stocks down, unforeseen costs, losing bets, etc.), and 3) a decision to be made between continued investment or abandonment of current failing path (Staw, 1976, 1997; Staw & Ross, 1987). Auctions are prime examples of escalation of commitment. The winner’s curse (Thaler, 1992) describes the ambivalence of the winner who achieved bidding victory but overpaid for his prize. In many instances, auctioneers are fully cognizant that they are overpaying for a particular item and have several chances to withdraw, but they choose to continue.

Escalation of commitment occurs across different levels of organizations. Individuals, companies, and even governments regularly engage in auctions over contracts, projects, and resources. For example, in 2007, the China Metallurgical Group Corporation bid $3.4 billion, $1 billion more than competitors, to win mining and development rights over the Aynak region in Afghanistan, and thus secured precious copper resources for China’s voracious manufacturing industries. In 1999, a series of fiberglass cows crafted by local artisans in Chicago were auctioned off on the internet and by Sotheby’s. The estimated price of each item ranged from $2000-$4000, but competition of bids became so fierce that final prices reached 7 times more than the intended price, greatly benefiting the charities who received the proceeds. These examples suggest that studying auction behavior is a relevant way of understanding decision makingbiases.

Hemispheric Specialization and Emotion

The left side of the world, that is, all the visual information within the Left Visual Field (LVF), goes to the Right Cerebral Hemisphere (RH) to be processed (Purves, Augustine, Fitzpatrick, Hall, LaMantia, & et al., 2004, chapter 12), and vice versa. While information can be shared between the two cerebral halves via fiber connections called the corpus callosum, previous research shows in a number of circumstances information do not transfer(Zaidel, 1986). The Direct Access Model states that each cerebral hemisphere will process the information it receives independently from the other in many situations. This has been demonstrated in normal participants as well as in “split-brain” patients who have undergone complete callosotomies in which all fibers connecting the two hemispheres are severed to treat intractable epilepsy (Zaidel, 1983; Zaidel, 1986).

Visuospatial position directly affects which hemisphere receives incoming stimuli, and each hemisphere shows considerable specialization. For example, the Left Hemisphere (LH) is the center for language in most individuals (Purves et al., 2004, chapter 27), while face processing and emotion processing are more focused in the RH (Etcoff, 1984; Kanwisher, 2000; Ley Bryden, 1979). The RH is also recognized to play a special role in decoding prosody in speech (Carmon & Nachshon, 1973; Ley and Bryden, 1982; Ross & Mesulam, 1979; Ross, 1981), and is considered dominant for negative emotions (Hellige, 1993). Therefore, visuospatial position may be important for emotional processing –emotional content (e.g., an emotional face) shown to the LVF may have a stronger impact than the Right Visual Field (RVF).Past research suggested that an enhanced automatic processing of negative facial expressions (angry and fearful) was associated with social anxiety (Mu¨hlberger, Wieser, Herrmann, Weyers, Tro¨ger, & et al., 2009). Thus, we assume that an angry face shown to the LVF may elicit a stronger anxiety than an angry face shown to the RVF.

Previous research has shown that decision making is keenly affected by emotions (Green & Haidt, 2002; Loewenstein, Weber, Hsee, & Welch, 2001). In particular, Ku and colleagues (2005) examined how arousal (e.g., anxiety and excitement) affects participants’ self-rating of their likelihood to overbid (i.e., escalation of commitment). They required participants to use a high or low amount of time and energy to research a prize item for an upcoming auction, and they manipulated the number of competitors (eight or one) in the auction itself.Results showed that arousal predicted overbidding behavior. However, there is little or no previousresearch that directly examines the relationship between visuospatial position, emotional content, and escalation of commitment.

Based on previous findings of RH dominance of negative emotion processing (Hellige, 1993) and the Direct Access Model of hemispheric processing (Zaidel, 1986), we expect greater escalation of commitment exposed to an angry face compared with a neutral face in the RH because the former leads to more anxiety than the latter. However, the above differences disappear in the LH. The stimuli reaches the left or right cerebral hemispheres via presentation to the right or left visual hemifields (i.e., RVF, LVF), respectively.

Our explicit hypotheses are as follows:

Hypothesis 1. The visuospatial position and the emotional content of the opponent bidder’s face jointly affect escalation of commitment in an auction task. When an angry face compared with a neutral is shown in the left visual hemifield (i.e., RH), people are more likely to escalate their bid decisions. However, the above effect will not be significant for right visual hemifield.

Hypothesis 2.Anxiety mediates the effect of emotional content on escalation of commitment in the left visual hemifield. When an angry face compared with a neutral face is presented to the left visual hemifield (i.e., RH), people have greater anxiety, and therefore escalate their commitment.

METHODS

The Overview of the Present Study

In the present study participants engage in a modified version an auction task by Ku and colleagues (2008) where they make bids against a computerized opponent. However, each time the opponent bids, an angry or neutral face is flashed to the left or right hemifield, depending on the condition. The participant receives constant negative feedback from being outbid by the opponent, in concert with the escalation of commitment situation defined by Brockner (1992). Following the auction, the participant answers a set of questionnaires, is thanked, debriefed, and dismissed.

Participants

Nineteen adults (84% female; mean age = 19.63)froma university in the west coast participated for course credit. Each participant was screened to rule out factors that may impact their ability to perform the tasks required (i.e., substance abuse, mental disorders, and immediate medical complications). All experimental procedures were approved of by the university’s Institutional Review Board.

Design and Procedures

Our paradigm employed a 2 x 2 within-subjects design, manipulating the variables of Emotion (Angry, Neutral) and Hemisphere (Left, Right). Our dependent measure was the amount of credits the participant spent in a modified version of Ku’s auction task (Ku, 2008).

Emotions were conveyed through the presentation of human face photographs from the NimStim Face Stimulus Set* (Tottenham et al., in press). Normal participants are highly skilled at recognizing facial expressions, and can decipher the emotional content rapidly and accurately (Ekman & Friesen, 1971; Calder, Young, Perrett, Etcoff, & Rowland, 1996).

Hemispheric stimulation was achieved through tachistoscopic (200ms, to prevent involuntary saccades) presentation stimuli to the left or right visual hemifield with the participant focusing on a central crosshair (Zaidel, 1979). Facial stimuli subtended 4.5 by 6 visual degrees, and were 2 visual degrees away from central fixation. Participants were instructed to fixate on a central cross that was displayed during the entirety of the experiment, and were monitored by an experimenter. Participants were tested using E-Prime version 2 Professional (Psychology Software Tools, Inc., 2002) on PC desktops running Windows XP. The LCD monitor resolution was 1024 x 768 with standard refresh rate of 60Hz, and participants were placed in chinrests 57cm away.

The computerized experiment proceeds as follows: Introductory instructions, prompting to begin auction, computer bids, sees lateralized emotional feedback, participant decides (sees statistics) and bids. Then participants may see lateralized emotional feedback again if they chose to bid.

The bidding task is between the participant and a computer simulation of another participant, and the currency and prize were both credits that have no intrinsic value outside of the experiment. Participants were given an allotment of 800 credits to bid against an opponent who had an unknown amount of credits. Bidding always started with the computer, and in increments of 40 point bids, without the ability ‘pass’ or make bids greater than 40 credits. The auction prize was 400 credits; both the winner and the loser paid their last bids. The goal was to win the bid and to pay as little as possible in credits.

Participants were given those instructions explicitly, and were told to try their best to behave as if they were using real currency to play. Furthermore, participants were told that their performances were going to be compared with other subject averages; however, this was to provide participants with performance feedback to raise their motivation, and only ‘above average performance’ evaluations were given.

In each round of bidding in the experiment, participants viewed statistics on the current bid, the prize, and the seconds remaining for themto make their bid that round. If the participant failed to bid within 10 seconds, then the prize is forfeit.Then an image of an emotional face was flashed left or right visual hemifield. Thereafter the computer bid, and thus the decision wentback to the participant once again. The computer had unlimited credits and the task ended when the participant ran out of credits, refused to bid, or ran out of time to bid in a given round.

Each participant was exposed to the four conditions (Angry LH, Neutral LH, Angry RH, Neutral LH) in a random sequence, a subsequent analysis of order will access for carryover, order, and practice effects.

After each session, two additional anxiety measures were assessed:“I got nervous when I bid with my opponents in the auction.” and “When engaging in the auction task, I felt uncomfortable.”Participants responded on 7 point scales with endpoint anchors of 1 (strongly disagree) and 7 (stronglyagree). (α = 0.94).

RESULTS

The effect of Condition of escalation of commitment.Before we explore the effect of Condition (i.e., the Angry LH condition; Neutral LH;Angry RH;Neutral RH) on escalation of commitment, we examine the effect of the order of the four conditions on escalation of commitment. We did not find any associations between the order and escalation of commitment, F (3, 16) = 0.76, p > .10, η2 =.12. Thus, the ordering of the four conditions did not affect the experimental outcome.

To examine the effect of the Condition on escalation of commitment, we ran a repeated measures model, where the Condition revealed a significant difference in escalation of commitment, F (3, 16) = 1.21, p .05, η2 =.43. Whena bidder’s angry face (M = 450.53, SD = 261.57) compared with neutral face (M = 496.84, SD = 280.42) was presented to participants’ left hemisphere, participants behaved consistently in their auction decision, t(18) = - 0.53, p > .10. However, when an bidder’s angry face (M = 513.68, SD = 273.66) compared with neutral face (M = 311.58, SD = 259.77) was presented to participants’ right hemisphere, they are more likely to escalate their auction decision to the bidder’s angry face, t(18) = 2.72, p .05. Thus, our results supported Hypothesis 1.

Mediation analysis.Because there was no difference between the Angry LH condition and the Neutral LH condition for escalation of commitment, we ruled out these two conditions in our subsequent mediation analysis.Instead, we examined the mediating effect of anxiety on the effect of f the RHCondition (i.e., Angry RH vs. Neutral RH) and escalation of commitment.

Participants in the Angry RH Condition (M = 513.68, SD = 273.66) reported having a greater tendency of escalation of commitment than those in the Neutral RH Condition (M = 311.58, SD = 259.77), F (1, 18) =7.41, p .05, η2 =.29. Participants in the Angry RH Condition (M = 4.34, SD = 1.56) reported having a greater degree of anxiety about the auction task than those in the Neutral RH Condition (M = 3.18, SD = 1.71), F (1, 18) =9.55, p < .01, η2 =.35. Furthermore, results of the RH Condition (Angry RH vs. Neutral RH) ANCOVA revealed that when we controlled the degree of anxiety, the previously observed effect ofthe RH Condition on escalation of commitmentwas reduced to insignificance, F (1, 17) =0.71, p > .10, η2 =.04. Therefore, in support of our Hypothesis 2, the degree of anxiety mediated the effect of the RH Condition on escalation of commitment. In other words, when participants were presented a bidder’s angry face compared with neutral face to their right hemisphere, these participants have a higher tendency of anxiety about the task and therefore escalated their auction decisions.

DISCUSSION

In the auction task, the results showed that when people saw an angry opponent’s face compared with a neutral opponent’s face in the left visual hemifield, they were more significantly more likely to escalate their commitment in an auction task. Specifically, when angry face is presented to the RH via left visual hemifield presentations, people increased their anxiety, and therefore significantly escalated their commitment. However, the above effects did not appear when the opponent’s faces were given to the LH.

There are two ways in which the present study differs from past research. First, the present study also offers insight on a previously unexplored topic of how the visuospatial location of information presented can affect decision making in an auction task. The perception of the world is divided by the visual input from the left and right sides go to the right and left hemispheres, respectively (Purves, et al., 2004, chapter 12). Importantly, this is not to be confused with left and right eye, as each eye sends information to both hemispheres. This design suggests that hemispheric specialization and visuospatial presentation are both important components that should be considered in understanding the complexities of decision making.

Second, past studies on escalation of commitment focused on participants’ psychological fixation on sunk costs (Wong, 2005). The current study employs the competitive arouse model (Ku, et al., 2005) to explain the mediating effect of emotion on the relationships between emotional contents and escalation of commitment in an auction task.Furthermore, we have identified anxiety as the mediator of greater escalation, which is consistent with the predictions of the competitive arouse model.

These results also contribute to a growing body of work that illuminates the effects of negative emotions on escalation of commitment. Ku (2008) found that regret was negatively related to escalation of commitment. However, Tsai and Young (2009) discovered that anger was positively associated with escalation of commitment. Our results have demonstrated a positive relationship between anxiety and escalation tendency.

Future studies of the present paper may consider a broader array of emotional faces and their influence on escalation of commitment and their mediators. For example, if fearful faces were presented to participants, would fear be interpreted as a sign of weakness in their opponents and cause participants to escalate? And what individual differences would exist for risk taking after seeing weakness in your opponent? Also, if the opponent is happy instead, then would participants perceive the happiness as a sign of gloating or confrontation, or rather as something benign and neutral? Thus, future studies ought to employ different emotional faces and measure different emotional mediators.

Further studies examining the emotional influence on decision making should consider the context of hemispheres because hemispheric differences play a strong role in emotion processing (Hellige, 1993). Specifically, the amygdala in the right hemisphere is responsible for regulating emotion. Thus, it is worthwhile to study how hemispheric specialization interferes with the relationship between emotions and decision-making.

*Development of the MacBrain Face Stimulus Set was overseen by Nim Tottenham and supported by the John D. and Catherine T. MacArthur Foundation Research Network on Early Experience and Brain Development. Please contact Nim Tottenham at for more information concerning the stimulusset.

REFERENCES

Borod, J. C. & Caron, H. S. 1980. Facedness and emotion related to lateral dominance, sex and expressiontype. Neuropsychologia, 18: 237-241.

Calder, A. J., Young, A. W., Perrett, D. I., Etcoff, N. L., & Rowland, D. 1996. Categorical perception of morphed facial expressions. Visual Cognition, 3: 81-117.

Carmon, A., &Nachshon, I.1973. Ear asymmetry in the perception of emotional non-verbal stimuli. ActaPsychologica, 37: 351-357.

Ekman, P., & Friesen, W. V. 1971. Constants across cultures in the face and emotion. Journal of Personality and Social Psychology,17(2): 124-129.

Etcoff, N. L. 1984. Perceptual and conceptual organization of facial emotions: Hemispheric differences.Brain and Cognition, 3: 385-412.

Greene, J., & Haidt, J.2002. How (and where) does moral judgment work?Trends in Cognitive Science.6: 517-523.