Bargaining and Psychology Lecture One: The Structure & Evolution of Preferences Keith Chen, Nov 3rd 2004 Plan of Action Overall: • Broadly: Whirlwind tour of Biology, Psychology and Experimental Economics. • Provide a toolbox of insights from these disciplines: – a conceptual framework through which to understand the interpersonal / cultural aspects of negotiations. Plan of Action Today: • Basic departure from the self-interested model: Fairness, Altruism & Spite. • How do we begin to conceptualize and measure notions of Fairness / Blame / Blameworthiness? Economic Experiments. • Framing, Cultural & SES Effects • Stability and Strategies: my work on the evolution of Altruism and Fairness Preferences. Economic Experiments • Rules of the Road for all Economics Experiments: – NO DECEPTION (on my part)! – Real stakes!! • Ask that you play within the rules: – Do your best and think hard about the game. Ultimatum Game: The rules of the game: 1. You will be randomly paired with someone else in the class; this pairing will remain completely anonymous. 2. One of you will be chosen (by coin flip) to be either the Proposer or the Responder in this experiment, and one pair will be randomly chosen to play for $100. 3. The Proposer gets to make an offer to split $100 in some proportion with the Responder. So the proposer can offer $x to the responder, proposing to keep $100-x for themselves. 4. The Responder must decide what is the lowest amount offered by the proposer that he / she will accept; i.e. “I will accept any offer which is greater than or equal to $y.” 5. If the responder accepts the offer made by the proposer, they split the sum according to the proposal. If the responder rejects, both parties lose their shares. Ultimatum Game: Cross-Cultural Variation that tracks social structure: Henrich et al. Variation in predictions for first and second answer: Different levels of rationality The beauty-contest game. When self-interest becomes less stable (weaker consequences,) fairness has more room to emerge? Fairness in the Market • Constraints are based more on notion of what is normal rather than “objective” justice; KaKnTh • Reference Transactions – History of previous transactions (Q2) – Market (nominal) Prices (Q4) – Posted (list, nominal) Prices (Q5) • Occasions for Pricing Decisions / Reference Profits – Protecting Profits (Q9) – Profit Increases (Q11) – Increases in Market Power • Punishment Economic Fairness: Surveys & Framing Effects Reference Points, Regret & Framing. Effect of Knowing Alternate Outcomes • Larrick & Bowles: References Affect Bargaining • People are risk-averse when they won’t know the outcome of an alternate gamble • People are more risk-neutral when they will be told the alternate gamble outcome • Coin flip +$0/+$200 vs. $90 guaranteed • People are willing to pay a “regret premium” to avoid discovering the outcome • Negotiators who won’t get feedback on a foregone alternative are more likely to reach agreement Surveys Say: • There is a temporal pattern to regret (Bowles) – In the short term, people regret actions taken more than inaction – In the long term, people regret inaction more than action • Train-Track surveys, Glaeser-Sacerdote work on sentencing (VHS +56% women, -53% black) • Takeaway: reference points strongly effects their perceptions of Fairness. Culture, Blame & Attribution • South vs. North • East vs. West (collectivist vs. individualist) • Rich vs. Poor (the illusion of control.) • Economic Position and in/out-group – • Jews, Chinese & Copts Ultimatum Games in Other Cultures. The Theory of Reciprocal Altruism / Cooperation • Why study reciprocal play in animals? 1. Examining Animals gives us: • Perspective on cross-cultural validity of game-theory. • Way of thinking about how fairness is either cultural or biological, socially constructed or evolutionary. Subjects: Cotton-top Tamarins: a cooperative breeder New World monkeys Old World monkeys Apes Experimental Setup: barrier receiver actor tool food tray Training & Solo Conditions Condition 1 Condition 2 Condition 3 & 4 Condition 5 no barrier w. barrier Intentionality? A Simple Reciprocation Game S2 S1 First Game S2 Experiment 1 tests for: (i) Contingent cooperation (ii) Recognition of reputation Setup: trained vs untrained actors S1 SH: 100% Altruist JG: 100% Defector 0.6 Frequency of Pulls 0.5 0.4 Pulls for Alt. 0.3 Pulls for Def. 0.2 0.1 0 1 2 3 Session 4 Frequency of Pulls Pulling for the Altruist by Trial 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Pulls for Alt. Odd Trials Even Trials 1 3 5 7 9 11 13 Trial # 15 17 19 21 23 Reciprocation Game 2a: S2 Untrained Actors S1 S2 Experiment 2a tests for: Reciprocal strategies between untrained actors S1 Description of the Model Predictions of a linear fit. Pull then imitate. Pull if last 2 trials were pull. Pull if last 3 trials were pull. Play what opponent has done most. Play what opponent did most over last 3 moves. Play what opponent did most over last 5 moves. Play the historic dist. of your opponent's play. history only goes back 3 moves. history only goes back 5 moves. Play a Beta=1/2 weighted historic distribution. Optimally fit Beta to the data. Play a cut off rule at 1/2. Play a cut off rule at 1/2 using the fit Beta. Play what has the highest historic payoff. Mix in proportion to historic payoffs. See discussion section for governing dynamic. An optimally fit Beta stimulus-response model. Chi Sqr. Er. Rate Post. Prob. Param. Adj. Cond.Prob. 37.66 10.82 16.56 14.16 11.98 22.37 19.84 8.59 11.48 11.15 15.11 14.88 10.57 26.08 2.63 3 6.41 6.55 0% 41% 34% 33% 39% 39% 38% 49% 37% 36% 37% 36% 43% 22% 44% 41% 29% 27% 2.0295E-204 1.0452E-205 9.1842E-200 1.0454E-200 4.3737E-203 1.9999E-202 3.4967E-201 5.6628E-206 1.1650E-202 4.5700E-203 1.6596E-203 2.1978E-203 2.9351E-207 5.9163E-204 1.5320E-207 1.2272E-206 4.0656E-201 4.2224E-201 3 2 2 2 2 2 2 2 2 2 2 3 2 3 2 2 2 3 0.000 0.000 83.168 9.467 0.040 0.181 3.166 0.000 0.105 0.041 0.015 0.001 0.000 0.000 0.000 0.000 3.682 0.134 Reciprocation Game 2b: One fixed session 3 sessions of: 1 session: 1 final session Unconditioned Play with Altered 4th Session Payoffs Frequency of Pulls 0.5 Altered payoff session, subjects always pulled 0.4 0.3 0.2 0.1 0 1 2 3 Session 4 5 Reciprocation Game 3: S2 Selfish Altruism S1 S2 Experiment 3 tests for: Recognition of byproduct altruism, or “selfish” altruism S1 Frequency of Pulls Pulling with a "Selfish" Partner 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Selfish Partner w. Selfish New Experiment: 3 Year Old Kids Subjects: Cotton-top Tamarins: a cooperative breeder New World monkeys Old World monkeys Apes Experimental Setup: barrier receiver actor tool food tray Training & Solo Conditions Condition 1 Condition 2 Condition 3 & 4 Condition 5 no barrier w. barrier Intentionality? A Simple Reciprocation Game S2 S1 First Game S2 Experiment 1 tests for: (i) Contingent cooperation (ii) Recognition of reputation Setup: trained vs untrained actors S1 SH: 100% Altruist JG: 100% Defector 0.6 Frequency of Pulls 0.5 0.4 Pulls for Alt. 0.3 Pulls for Def. 0.2 0.1 0 1 2 3 Session 4 Frequency of Pulls Pulling for the Altruist by Trial 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Pulls for Alt. Odd Trials Even Trials 1 3 5 7 9 11 13 Trial # 15 17 19 21 23 Reciprocation Game 2: S2 Untrained Actors S1 S2 Experiment 2 tests for: Reciprocal strategies between untrained actors S1 Description of the Model Predictions of a linear fit. Pull then imitate. Pull if last 2 trials were pull. Pull if last 3 trials were pull. Play what opponent has done most. Play what opponent did most over last 3 moves. Play what opponent did most over last 5 moves. Play the historic dist. of your opponent's play. history only goes back 3 moves. history only goes back 5 moves. Play a Beta=1/2 weighted historic distribution. Optimally fit Beta to the data. Play a cut off rule at 1/2. Play a cut off rule at 1/2 using the fit Beta. Play what has the highest historic payoff. Mix in proportion to historic payoffs. See discussion section for governing dynamic. An optimally fit Beta stimulus-response model. Chi Sqr. Er. Rate Post. Prob. Param. Adj. Cond.Prob. 37.66 10.82 16.56 14.16 11.98 22.37 19.84 8.59 11.48 11.15 15.11 14.88 10.57 26.08 2.63 3 6.41 6.55 0% 41% 34% 33% 39% 39% 38% 49% 37% 36% 37% 36% 43% 22% 44% 41% 29% 27% 2.0295E-204 1.0452E-205 9.1842E-200 1.0454E-200 4.3737E-203 1.9999E-202 3.4967E-201 5.6628E-206 1.1650E-202 4.5700E-203 1.6596E-203 2.1978E-203 2.9351E-207 5.9163E-204 1.5320E-207 1.2272E-206 4.0656E-201 4.2224E-201 3 2 2 2 2 2 2 2 2 2 2 3 2 3 2 2 2 3 0.000 0.000 83.168 9.467 0.040 0.181 3.166 0.000 0.105 0.041 0.015 0.001 0.000 0.000 0.000 0.000 3.682 0.134 Reciprocation Game 3: S2 Selfish Altruism S1 S2 Experiment 3 tests for: Recognition of byproduct altruism, or “selfish” altruism S1 Frequency of Pulls Pulling with a "Selfish" Partner 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Selfish Partner w. Selfish Reciprocation Game 4: S2 Play with Humans S1 S2 human Experiment 4 tests for: Recognition of unintentional altruism, when a human pulls. S1 Pulling for the Human Compared to Pulling for the Defector Frequency of Pulls 0.6 0.5 0.4 Pulls for Alt. 0.3 Pulls for Def. Pulls with human 0.2 0.1 0 1 2 3 Session 4 Pulling for the Human Compared to Pulling for the Defector Frequency of Pulls 0.25 0.2 0.15 Pulls for Def. Pulls with human 0.1 0.05 0 1 2 3 Session 4 Tamarin Conclusions: Biology: Economics: 1. Strong evidence of Reciprocal Altruism. 1. Sophisticated play not confined to humans. 2. Evidence Tamarins are quite sophisticated about reciprocal play. 2. Informs parsimony, suggests wide scope for theory. 3. Preferences such as Altruism seem universal and shaped by evolved cognitive structures. Capuchin Experiments: • More complex division tasks: Ultimatum, dictator and public goods games • The use of fiat money: – Recognition of Common Value – Basic tests of Rational Preferences • Risk preferences (and biases) Capuchin Money and Trading • Trained Capuchin monkeys to know that “tokens” can be traded for different foods, at different prices with different RAs. • New: The Payoffs can be risky. Capuchin Money and Trading • Three Conditions: – Searched for foods for which each actor consumed a relatively interior bundle. – A compensated halving of one good’s price – A money-illusion condition Capuchin Money and Trading Baseline and Compensated Price Shift 12 11 10 9 8 7 6 5 4 3 2 1 0 0 2 4 6 8 10 12 14 16 18 20 22 Capuchin Money and Trading Baseline and Money Illusion Condition 12 10 8 6 4 2 0 0 2 4 6 8 10 Broader Tasks: Now have subjects trained on a five-arm mechanism, allows two subjects to play either an ultimatum or dictator setup. Conclusions: Biology: Economics: 1. Strong evidence of Reciprocal Altruism. 1. Sophisticated and interactive behavior not confined to humans. 2. Evidence Tamarins are quite sophisticated about reciprocal play. 2. Informs parsimony, suggests wide scope for theory of fairness.