Emotion, Decision and Risk: Betting on Gambles vs. Betting on People TA M A R K U G L E R , L I S A O R D E Z , T E R R Y C O N N O L LY I C S D, A U G U S T 2 0 0 9 Decisions Making Involves Powerful Emotions Two ways emotions can affect decisions: 1. Integral emotions can arise from the decision itself Angry about the salary raise your boss offers you 2. Incidental emotions from other events can spillover to unrelated decisions Sad jilted lover makes some rash decisions Wait until a boss/friend/parent is happy to ask for a favor We will investigate how incidental emotions affect subsequent risk-related decisions in social dilemmas

Decision Making & Emotions Slow to develop Early work: Valence approach (good vs. bad moods) Good moods = heuristic processing, more collaboration, lower perceived risks Bad moods = analytic processing, more competition, higher perceived risks (Isen & Colleagues; Gasper& Clore; Johnson & Tversky, Slovic & Peters) Notable exception: Appraisal Tendency Framework (ATF) (Lerner & Keltner, 2000, 2001) Appraisal Tendency Framework Draws on cognitive appraisal theories of emotions (Smith & Ellsworth, 1985; Lazarus, 1991) Dimensional structure of emotions (e.g., pleasantness, control, certainty, anticipated effort) Distinct emotions cognitive appraisal

distinct behavioral tendencies For example Example Anger: certain and individual control. A demeaning offense occurred with certainty and situation is under control of human agency Fear: uncertain and situational control. A sense that even basic needs such as safety are uncertain and situational factors beyond ones control shape outcomes Prediction: Fear should increase perceptions of risk and then lead to risk averse choices compared to Anger Findings (Lerner and Keltner 2001): fearful people perceive risk as higher than angry people, and make hypothetical risk averse choices However Judgments or hypothetical choices only No consequences to choice The source of risk is always in the environment and not in the behavior of other people (lottery risk vs. interactive risk) Interactive risk is the type of risk that is relevant to behavior in Social Dilemmas Our Research Goals 1. Focus on specific, discrete emotions Fear, anger and happiness 2. Measure effect of emotions on incentive

compatible behavior rather than on attitudes, opinions, expected decisions 3. Differentiate the impact of incidental emotions on lottery risk and interactive risk Summary of dimensional structure Emotion Dimension Anger Happiness Fear Core theme Demeaning offence against the individual Progress towards realization of goal Uncertain threat

Negative Positive Negative Certain Certain Very uncertain Human (other) Human (self) Situation Valence Certainty Control Experiment 1: Lottery Risk Goals: Replicate Lerner & Keltners results regarding risky choices (most of their studies examine risk perception)

Real monetary consequences (all of their results were hypothetical) Does Fear increase risk aversion compared to Anger? How about happiness? (happiness is high in certainty and individual control like anger, but positive in valence) Experiment 1 Design Emotion (3) x Emotion Check (2) BS design Approximately n=20 per group useable data About 5 additional subjects per group had to be eliminated for not following directions Emotion: incidental emotion evoked thru writing task fear, anger, happiness Emotion Check: whether or not emotion checklist given after emotion induction Procedure Induce emotion in Study 1 ( answer emotion checklist,

dont) Risk attitude assessed in Study 2paid based on decisions & chance Emotion Induction Writing Task Please try to remember an experience in the past 2 years that made you feel extremely fearful [angry, happy]. Put yourself back into the situation as though you were just now experiencing it. Then prepare to imagine telling the experience to a best friend or relative. Remember that it is very important that your friend understand exactly how you feel about the incident and why you felt that way. Please write what you would tell your best friend or relative. Use as much detail as possible. Emotion Manipulation Checklist PANAS (Watson & Clark, 1994) emotion checklist Please rate very carefully the degree to which you are currently experiencing each of the following feelings (circle a number): Do not feel at all 1 2 3 4 5

6 7 Feel stronger than ever have 8 9 Measured Emotion Individual Items Fear Afraid Scared Nervous Anger Mad Irritated Angry Happiness Joyful Elated Delighted Happy Distractors Hopeful

Self-fulfilled Wishful Thankful Ashamed Upset Regretful Emotion Manipulation Checks Measured Emotion Fear Manipulated Emotion Anger Condition Happiness * Fear Anger Happiness n 3.37 b 2.22a

3.81b 20 2.15a 4.45 b 3.9b 20 2.73a 2.08a 5.98 d 20 different subscripts indicate significant difference at p <.05 High Cronbachs alphas (.92, .88, .94) Manipulation check good (but not perfect) * Main problem = students too happy in AZ! Assessing Individual Risk Attitude

Risk attitude questionnaire (Holt and Laury, 2002). 10 consecutive choices between 2 lotteries (A and B) with two outcomes each Lottery A outcomes: $20 and $16 (low variance) Lottery B outcomes: $38.50 and $1 (high variance) The probabilities are systematically manipulated so that for small Ps A has a higher expected value than B, but as P increases the difference gradually decreases, until B has a higher expected value than A. Assessing Individual Risk Attitude Decision # Option A Option B Your Choice 1 1/10 chance of winning $20

9/10 chance of winning $16 1/10 chance of winning $38.50 9/10 chance of winning $1 A B 2 2/10 chance of winning $20 8/10 chance of winning $16 2/10 chance of winning $38.50 8/10 chance of winning $1 A B 3 3/10 chance of winning $20 7/10 chance of winning $16 3/10 chance of winning $38.50

7/10 chance of winning $1 A B Risk Seeking6/10 Option chance of winning $1 A B A B A B A B A B A

B A B 4 5 6 7 8 9 10 Risk Averse6/10 Option chance of winning $16 1/10 chance of 5/10 chance of winning $20 5/10 winning chance of winning $16 6/10 chance of winning $20$204/10 chance of winning $16 9/10 chance of winning 7/10 chance

of winning $20 3/10 chance of winning $16 Risk Averse Option chance 8/10 4/10 of winning chance $20$16of 2/10 winning chance of winning $16 9/10 chance of winning $20$201/10 chance of winning $16 chance 6/10 chance of chance winning 10/10 of winning $20 0/10 of winning $16 4/10 chance of winning $20 $16 4/10 chance of winning $38.50

1/10 chance of 5/10winning chance of winning $1 $38.50 6/10 chance of winning $38.50 4/10 chance of winning $1 5/10 chance of winning $38.50 9/10 chance of winning 7/10 chance of winning $38.50 3/10 chance of winning $1 Risk Seeking Option $1 of 8/10chance of winning $38.50 2/10winning chance of winning $1 4/10 chance $38.50 1/10 chance of winning $1

9/10 chance of winning $38.50 6/10 chance of 0/10 winning chance of winning $1 $1 10/10 chance of winning $38.50 A rationalRisk Averse Option player should switch from Option A to B only once Risk Seeking Option 10/10 chance of winning Risk Aversion score = switching point (higher values more risk averse) 10/10 chance of= winning $20 Risk neutral switching point 4-5 (when EV is approx. $38.50 matched)

0/10 chance of winning 0/10 of winning $1 out One lottery randomly chosen, played, and chance cash payments given $16 R is k A v e r s io n S c o r e Risk Aversion Scores ME of Emotion Highly Risk Averse9 (F 2,110=5.31, p=.006) Risk aversion scores ordered as predicted 8 7 6

Risk 5 Neutral Uncerta in Fear Happiness Anger Manipulated Emotion Certain R is k A v e r s io n S c o r e Risk Aversion Scores Interesting Aside: No significant differences if compare positive (happiness) vs. negative (fear or anger) emotions Highly Risk Averse 9

8 7 6 Risk 5 Neutral Positive Negative Manipulated Emotion Summary of Experiment 1 Lerner & Keltner results extended Fear > Happiness>Anger wrt risk aversion Strong result Based on actual decisions with large consequences (mean payout = $23.81) Interesting that happiness is somewhere in between (similar to Lerner & Keltner results) Thus, incidental emotion appears to impact

risk attitude in predictable manner Not simply emotion-valence since two negative emotions lead to different risk attitudes Experiment 2: Interactive Risk What happens when the source of risk is the behavior of another person instead of the environment? Risk is not a simple construct Lottery risk (Exp 1) Interactive risk (Exp 2) Predictions In interactive risk anger (negative, in individual control) will lead to more risk aversion than fear. Happiness is also in individual control, but positive. The Stag-Hunt Game Study disclaimer: No animals will be harmed in the course of this experiment. We apologize for the animal hunting cover story. The Stag-Hunt Game

Equilibriu m Solution 1 Hunter 2 Rabbit Hunter 1 Rabbit Stag $10,$10 Stag $0,$10 $10, $0 $20,$20 Risk averse option = hunt rabbits Risk seeking option = hunt stags Equilibriu m Solution 2 Experiment 2 Design Emotion (3) BS design Approximately n=15 per cell Emotion: incidental emotion evoked thru writing task fear, anger, happiness

Emotion Check: none conducted for this study Procedure Induce emotion in Study 1 Interactive game in Study 2 Single decision No cover story simple Option A ($10) vs. Option B ($0 or $20) After decision, estimated probability of others choices % S e le c tin g R A O p tio n Experiment 2 Summary Emotions do not seem Risk Averse 50 Estimated % Actual % 40 30

20 10 Risk 0 Fear Happiness Anger Seekin g Uncertain/ Manipulated EmotionCertain/ Human Situationl Control Control to impact our perceived risk (other persons behavior) Emotions do impact interactive behavior Angry and Happy people are more risk averse than fearful people. Opposite results from Exp 1

Conclusion Distinct, differentiated, incidental emotions can affect unrelated decisions Exp 1: Aversion to Lottery risk Fear > Happiness > Anger Exp 2: Aversion to Interactive risk Fear < Happiness < Anger Emotions are complex. So is risk. We need to consider the interaction between the emotional dimensions and the risk dimensions. Thank you! Come to AZ and get happy! Questions? Thoughts?