Moving Viewpoint what makes human subjects different from

Moving Viewpoint: what makes human subjects different from computer agents? Sobei H. Oda Kyoto Sangyo University The Sixth International Workshop on Agent-based Approaches in Economic and Social Complex Systems National Chengchi University, Taipei 14 November 2009

Computer Agents Program Action Dynamics Human Subjects Strategy known observable unknown

Allais’ Paradox TWD 400 (20 per cent) > TWD 300 (25 per cent) × 4 reversed × 4 TWD 400 (80 per cent) < TWD 300 (100 per cent) 100% is special, while difference between 99% and 98% is a matter of degree.

Hyperbolic Discounting 22 Feb. 2010 TWD 100000 < (in 100 98 99 97 96 80 70 60 50 40 30 20 10 10 9 days) 8 7 6 5 4 3 2 day < -10 99 98 97 96 80 70 60 50 40 30 20 0 9 day 8 7 6 5 4 3 2 1 - -100 days 23 Feb. 2010 TWD 100100 (in 101 days) 12 10 99 98 97 81 71 61 51 41 31 21 11 9 day 8 7 6 5 4 3 0 0 -10 99 98 97 96 80 70 60 50 40 30 20 9 day 8 7 6 5 4 3 2 1 - -100 days reversed TWD 100000 (today) > TWD 100100 (tomorrow) Nov. 2009 14 15 16 Feb. 2010 17 ・ ・ ・ 21 22

Similarity and Difference between Allais’ Paradox and Hyperbolic Discounting TWD 4000 (20 %) > TWD 3000 (25 %) can be rational but TWD 4000 (80 %) < TWD 3000 (100 %) TWD 100000 (in 100 d. ) < TWD 100100 (in 101 d. ) but cannot be rational (maintained) TWD 100000 (today) > TWD 100100 (tomorrow)

Kyoto Experimental Economics Laboratory (KEEL) f. MRI at Brain Activity Imaging Centre

Combination of alternatives today 100% 80% 40% 1 week 2 weeks later

+ A B today 2 weeks 4000 yen 5000 yen 100 % time + A B today 4000 yen 80 % 40 % choice presentation 1 -5 seconds decision making 12 seconds choice presentation 1 -5 seconds decision making

Results (Green-Blue) today 2 1 weeks later 100% BA 39 80% 40% • BA 39 is involved in calculation (? ) OFC

Results (Blue-Green) today 2 1 weeks later Self-projection 100% 80% parahippocampal gyrus 40% neural activity in these regions tracks the revealed subjective value of delayed rewards. Kable & Glimcher (Nature Neuroscience 2007) striatum PCC precuneus PFC

Self Projection reflects the workings of the same core brain network. OFC parahippocampal gyrus Remember past to imagine future. Why have we memory? Because with memory we can make better decisions and have greater chance to survive.

Self-Projection (Bucker and Carroll, TRENDS in Cognitive Science 2006) Remembering Theory of Mind Prospection Navigation

Self Projection as Moving viewpoint Narrator navigation Past Self remembering Present Self Future Self prospection Another Person theory of mind

today 1 week later 2 weeks later Self-projection 100% 80% 40% PFC The regions contain precuneus and parahippocampal gyrus, which are considered to be activated when people are involved in complicated decision-making. Together with other observations, it seems to support self-projection, suggesting also why people reveal such intertemporal preference that does not allow a simple explanation. precuneus parahippocampal gyrus

annual discounting factors No convergence is observed. Why? テキスト Frederick, Loewenstein and O'Donghue (JEL 2002)

Questions; Intertemporal choice Risky choice (Environmant) Brain f. MRI Self-projection Calculation (? ) more complicated Observation Answers; (Behaviors) lab field Instable Stable

stimulus condition insider f. MRI outsider behaviour conscious thinking unconscious process

At least one of you have a white hat on your head. Can you tell whether your hat is white or not? Yasugi and Oda (2002, 2003)

Girl B’s inference She (A) knows that my hat or her hat or both hats are white; she does not know whether hat is white or not; she knows inference whether my hat is white or not. I (B) know: My hat is white.

Girl A’s inference I (A) know: She (B) has a white hat on her head; at least one of us wears a white hat. inference I (A) does not know: where My hat is white or not.

I (A) know: She (B) has a white hat on her head; at least one of us wears a white hat. My hat is white I (A) know: My hat is not white I (A) do not know whether my hat is white or not. (Yasugi and Oda 2002, 2003)

Human behaviour Jumping out of the system (Hofstader 1979) Preference: what they prefer Information: what they know Options: inferenc e Action: what they do what they can do emotion, instinct, experience, etc. : what they feel consciously or unconsciously

Jumping out controllable Jumping out uncontrollable

Bertrand Duopoly with product differentiation Nash Equilibrium Player A’s Reaction Curve • Participants: Class: 140 LAB: 58 Player B’s Reaction Curve • All pairs are reshuffled randomly • All players’ decisions are posted simultaneously

Perticipants/experiments: 120 - Participants/experiments: Max 28

Percentage of students who chose Price = 3: Nash Equilibrium strategy %

Percentage of students who chose Price = 7: Pareto optimal strategy %

Percentage of students who chose Price = 2: What strategy? %

　 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 0 6 10 47 73 104 183 282 403 544 707 890 1095 1320 2 6 0 16 52 78 110 188 288 408 550 712 896 1100 1326 3 10 16 0 37 63 94 173 272 534 697 880 1085 1310 Player 393 A’s profit >Player B’s profit 4 47 52 37 0 26 58 136 236 356 498 660 844 1048 1274 5 73 78 63 26 0 31 110 209 330 471 634 817 1022 1247 6 104 110 94 58 31 0 79 178 299 440 603 786 991 1216 7 183 188 173 136 110 79 0 100 220 362 524 708 912 1138 8 282 288 272 236 209 178 100 0 121 262 425 608 813 1038 9 403 408 393 356 330 299 220 121 0 142 304 488 692 918 10 544 550 534 498 471 440 362 262 142 0 163 346 551 776 11 707 712 697 660 634 603 524 425 304 163 0 184 388 614 12 890 896 880 844 817 786 708 608 488 346 184 0 205 430 13 1095 1100 1085 1048 1022 991 912 813 692 551 388 205 0 226 14 1320 1326 1310 1274 1247 1216 1138 1038 918 776 614 430 226 0 Player A’s profit <Player B’s profit Player A’s profit =Player B’s profit • Choosing 2 is the unique Dominant strategy for each player if they maximises not their profit but the difference between their profits and their opponents’.

Q=2 Q=3 P=2 363, 363 388, 372 P=3 372, 388 402, 402 Without Monetary rewards students played not P=3 to maximize their profits but P=2 to beat their opponents, which are also confirmed in the debriefing questionnaires.

Cournot-Stackelberg Duopoly • All pairs are fixed. • First 1 st movers’ decisions are shown and then 2 nd movers make decisions (8, 8) 1 st mover profit =2624 2 nd mover profit =2624 Maximum 1 st mover’s profit on the 2 nd mover’s reaction Curve: SPNE 1 st mover’s Reaction Cuarve 2 nd mover’s Reaction Curve: 2 nd mover’s SPNE strategy 1 st mover’s SPNE strategy

Percentage of pairs who realised Nash Equilibrium: (13, 5) • Without Money are students more “rational”?

2 nd player’s strategy 1 st player’s strategy (3, 10) 1 st mover Profit 1362 2 nd mover Profit 4540 Nash Equilibrium most advantageous for the 2 nd mover 2 nd 1 st mover’s profit increases (13, 5) 1 st mover Profit 3172 2 nd mover Profit 1220 Nash Equilibria

2 nd mover’s choice 2 nd movers are more Submissive in Classroom Laboratory Experiment 3 Profit table Pareto optimal 2 nd mover’s Reaction Curve Practically Ultimatum game Cooperation seeking (? ) No-conditional accept Conditional accept Reject

Dynamics of st 1 mover’s choice Classroom Period 1 3 8 Lab Period 1 13 3 Classroom Period 10 3 8 6 8 13 Lab Period 11 13 3 6 8 13

In the classroom the second movers were ready for accepting the SPNE, which is the Nash Equilibrium least favorable to them. In the debriefing Questionnaires they were only too happy to give a rational explanation why they had not earned more. They seemed to have changed their objective: from maximising their profit to explaining why they couldn’t, which change was not observed in the laboratory. Monetary rewards prevented subjects from setting their own goal by themselves and made them play seriously; though it may not be the case in every

Differences between the classroom and the laboratory Laboratory Classroom With Monetary Rewards Perticipants/experiments: Max 28 Participants/experiments: 120 -142 3 experiments/1 day 1 experiment/1 day No debriefing for each experiment Debriefing for each experiment … … Other Differences

Human subjects can (cannot but) jump out of the system. • They move their viewpoint to make decisions so that they can make better decisions; which process is realised by dynamic brain activities. • They do not limit their inference within the system; they do meta-thinking to make decisions. • They change (find or create) new objectives if they think the original ones are not interesting or too difficult to be realised.

references [1] Allais, Maurice (1953): ''Le comportement de l’homme rationel devant le risque, critique des postulates et axiomes de l’ecole americaine'', in conometrica, vol. 21, pp. 503 -546. [2] Barron, Greg & Ido Erev (2003): ''Small feedback-based decisions and their limited correspondence to description-based decisions'', in Journal of Behavioral Decision Making, vol. 16, pp. 215 -233. [3] Buckner, Randy L. & Daniel C. Carroll (2007): ''Self-projection and the Brain'', in Trends in Cognitive Sciences, vol. 11, pp. 49 -57. [4] Camerer, Colin F. & George Loewenstein & Drazen Prelec (2005): ''Neuroeconomics: How neuroscience can inform economics'', in Journal of Economic Literature, vol. 43 (no. 1), pp. 9 -64. [5] Frederick, Shane & George Loewenstein & Ted O'Donoghue (2002): ''Time Discounting and Time Preference: A Critical Review'', in Journal of Economic Literature, vol. 40 (June), pp. 351 -401. [6] Glimcher, Paul W. (2003): ''Decisions, Uncertainty, and the Brain : The Science of Neuroeconomics'', MIT Press, Cambridge, Massachusetts, USA; 宮下英三【訳】(2008): 『神経経済学入門--不確実な状況で脳はどう意思決定するのか』, 生産性出版.

references [7] Gul, Faruk & Wolfgang Pesendorfer (2005): http: //www. princeton. edu/˜pesendor/mindless. pdf. “The Case for Mindless Economics”, [8] Hertwig, R. & Greg Barron & E. Weber & Ido Erev (2004): “Decisions from experience and the weighting of rare events”, in Psychological Science vol. 15, pp. 534 -539. [9] Hofstadter, Douglas R. (1979, Escher, Bach: an Eternal Golden Braid, Basic Books, New York, USA; 野 崎昭弘・はやしはじめ・柳瀬尚紀【訳】(1985): 『ゲーデル, エッシャー, バッハ—あるいは不思議の環』, 白揚 社. [10] Hsu, Ming & Meghana Bhatt & Ralph Adolphs & Daniel Tranel & Colin F. Camerer (2005): “Neural Systems Responding to Degrees of Uncertainty in Human Decision-Making”, in Science, vol. 310, pp. 1680 -1683. [11] Kable, Joseph W. & Paul W. Glimcher (2007): “The Neural Correlates of Subjective Value During Intertemporal Choice”, in Nature Neuro. Science, vol. 10(12), pp. 1625 -1633. [12] Kelman, Mark & Yuval Rottenstreich & Amos Tversky (1996): “Context-Dependence in Legal Decision Making”, in Journal of Legal Studies, vol. 25, issue 2, pp. 287 -318. [13] Laibson, David (1997): “Golden Eggs and Hyperbolic Discounting”, in Quarterly Journal of Economics, vol. 112, pp. 443 -477. [14] Logothetis, Nikos K. (2008): “What we can do and what we cannot do with f. MRI”, in Nature, vol. 453, pp. 869 -878.

references [15] Mc. Clure, Samuel M & David I. Laibson & George Loewenstein & Jonathan D. Cohen (2004): “Separate Neural Systems Value Immediate and Delayed Monetary Rewards”, in Science, vol. 306, pp. 503 -507. [16] Moreno, Jonathan D. (2005): Mind Wars: Brain Research and National Defense, Dana Press, New York, USA; 久保田競【監訳】(2008): 『マインド・ウォーズ　操作される脳』, アスキー・メディアワークス角川 グループパブリッシング. [17] 奥田次郎(2008): 「未来への予見に携わる脳神経ネットワーク」, 『玉川大学脳科学研究所紀要』, 第 1 号, pp. 13 -23. [18] Pasinetti, Luigi, L. (1981): Structural Change and Economic Growth: a theoretical essay on the dynamics of the wealth of nations, Cambridge University Press, Cambridge, UK; 大塚勇一郎・渡会勝義【 訳】(1983): 『構造変化と経済成長—諸国民の富の動学に関する理論的エッセイ』, 日本評論社. [19] Rubinstein, Ariel (2006): “ Discussion of “Behavioral Economics”: “Behavioral Economics” (Colin Camerer) and “Incentives and Self-Control” (Ted O’Donoghue and Matthew Rubin)”, in Advances in Economics and Econometrics: Theory and Applications, Ninth World Congress. Volume 2 (Econometric Society Monographs, no. 42) edited by Richard Blundell & Whitney K. Newey & Torsten Persson, Cambridge University Press, Cambridge and New York: pp. 246 -54. [20] Samuelson, Paul, A. (1947): Foundation of Economic Analysis, Harvard University Press, Cambridge, Mass. ; 佐藤隆三【訳】(1967): 『経済分析の基礎』, 勁草書房.

references [21] Shafir, Sharoni & Taly Reich & Erez Tsur & Ido. Erev & Arnon Lotem (2008): “Perceptual accuracy and conflicting effects of certainty on risk-taking behaviour”, in Nature, vol. 453, pp. 917 -921. [22] Smith, Vernon L. (1982): “Microeconomic Systems as an Experimental Science”, in American Economic Review, vol. 66, pp. 274 -279. [23] Weber, Bethany J. & Scott A. Huettel (2008): ‘The neural substrates of probabilistic and intertemporal decision making”, in Brain Research, doi: 10. 1016/j. brainres. 2008. 07. 105. [24] Yasugi, Mariko & Sobei H. Oda (2002): “A Note on the Wise Girls Puzzle”, in Economic Theory, vol. 19 (no. 1), pp. 145 -156. [25] Yasugi, Mariko & Sobei H. Oda (2003): “Notes on Bounded Rationality”, in Scientiae Mathematicae Japonicae, vol. 57 (no. 1), pp. 83 -92.

International Conference How and why economists and philosophers do experiments: dialogue between Experimental economics and experimental philosophy Kyoto Sangyo University, Kyoto, Japan
 27 -28 March 2010