Abstract
In this paper we provide a summary of results concerning two series of experiments we ran based on a modified signalling game, which was presented graphically to subjects on a screen. The game for the initial experiment was selected by Reinhard Selten in coordination with the first named author. It has the interesting property that the strategically stable outcome (Kohlberg and Mertens 1986) does not coincide with the outcome of the Harsanyi-Selten solution (Harsanyi and Selten 1988). However, it is a complex game insofar as standard refinement concepts like the intuitive criterion, or the never-a-weak-best-response criterion, do not help to refine among the equilibria. The second motive for the design was to analyse, how the change in the reward at a particular terminal node would affect behaviour.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
We would like to thank Reinhard Selten for the design of the game used in the initial set of experiments; the staff working at the Bonn Laboratory of Experimental Economics and the Finance and Economics Experimental Laboratory in Exeter (FEELE), for their support while conducting the experiments; and Karim Sadrieh for the organisation of the “Scientific Excursions with Reinhard Selten” which motivated us to conduct the new experiments.
- 2.
The term is due to Reinhard Selten.
- 3.
Detailed proofs are available on request.
- 4.
The 2 × 2 game was added following the strategically safe choice of type 1b, but then moves were coalesced.
- 5.
These experiments were designed and conducted under the supervision of Reinhard Selten.
- 6.
However, we did not see any indication that this difference of information mattered.
- 7.
We calculated the average and standard deviation of the percentage of times Player 1 chose left at information set 1a, relative to the number of times this information set was reached for each session. The following tables are calculated in a similar manner.
- 8.
The test is highly significant for Rounds 1–25, but not for Rounds 26–50. Thus, the original stronger incentive for Player 1 to take the strategically risky option gets somewhat dampened by experience.
- 9.
Except for Rounds 51–55 in the new set of experiments which just misses the 5% level of significance.
- 10.
We disregarded the sessions of the old experiments since information set 2α was not reached often enough.
- 11.
The fractions are calculated relative to the number of times Nature chose right and Player 1 did not choose left.
- 12.
In the impulse balance equilibrium, right is chosen with probability ½ by Player 1 and with probability 2/3 by Player 2. In the action sampling equilibrium, right is chosen with probability 0.56 by Player 1 and with probability 0.66 by Player 2.
- 13.
Our programme initially highlights each choice at the relevant information set with equal probability. Subject can then change which choice is highlighted with the left and right cursor keys. Once the desired choice is highlighted, subjects decide on it by pressing the Enter key.
References
Banks J, Camerer C, Porter D (1994) An experimental analysis of Nash refinements in signalling games. Games Econ Behav 6:1–31
Brandts J, Holt CA (1992) An experimental test of equilibrium dominance in signalling games. Am Econ Rev 82(5):1350–1365
Brandts J, Holt C (1993) Adjustment patterns and equilibrium selection in experimental signalling games. Int J Game Theory 22(3):279–302
Camerer CF (2003) Behavioral game theory: experiments in strategic interaction. Princeton University Press, Princeton, NJ, pp 408–473
Cooper DJ, Kagel JH (2003) The impact of meaningful context on strategic play in signalling games. J Econ Behav Organ 50:311–337
Cooper DJ, Kagel JH (2005) Are two head better than one? Team versus individual play in signalling games. Am Econ Rev 95:477–509
Cooper R, DeJong D, Forsythe B, Ross T (1990) Selection criteria and coordination games: some experimental results. Am Econ Rev 80:218–233
Eichberger J, Kelsey D (1996) Uncertainty aversion and preference for randomisation. J Econ Theory 71:31–43
Eyster E, Rabin M (2005) Cursed equilibrium. Econometrica 73(5):1623–1672
Harsanyi JC, Selten R (1988) A general theory of equilibrium selection in games. MIT Press, Cambridge, MA
Kohlberg E, Mertens J-F (1986) On the strategic stability of equilibria. Econometrica 54:1003–1039
Selten R (1967) Die Strategiemethode zur Erforschung des eingeschränkt rationalen Verhaltens im Rahmen eines Oligopolexperiments. In: Sauermann H (ed) Beiträge zur experimentellen Wirtschaftsforschung. J. C. B. Mohr (Paul Siebeck), Tübingen, pp 136–168
Selten R, Chmura T (2008) The American Economic Review 98(3): pp 938–966
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Appendix
Appendix
Old Experiment, Terminal Nodes Reached in Rounds 1–25a
A | B | C | D | E | F | G | H | I | |
1 (x = 4) | 75 | 0 | 4 | 0 | 1 | 11 | 22 | 10 | 27 |
2 (x = 4) | 15 | 59 | 5 | 8 | 0 | 12 | 20 | 6 | 25 |
3 (x = 4) | 30 | 44 | 8 | 4 | 1 | 7 | 19 | 9 | 28 |
4 (x = 5) | 66 | 3 | 10 | 0 | 2 | 9 | 24 | 17 | 19 |
5 (x = 5) | 46 | 13 | 14 | 1 | 1 | 10 | 22 | 11 | 32 |
6 (x = 5) | 53 | 9 | 20 | 3 | 3 | 10 | 19 | 11 | 22 |
7 (x = 6) | 21 | 37 | 21 | 1 | 0 | 4 | 16 | 22 | 28 |
8 (x = 6) | 38 | 26 | 18 | 6 | 2 | 4 | 34 | 2 | 20 |
9 (x = 6) | 24 | 37 | 21 | 0 | 2 | 9 | 27 | 9 | 21 |
a“x” denotes the payoff of Player 1 at terminal node B
Old Experiment, Terminal Nodes Reached in Rounds 26–50
A | B | C | D | E | F | G | H | I | |
1 (x = 4) | 72 | 2 | 3 | 0 | 0 | 13 | 26 | 10 | 24 |
2 (x = 4) | 51 | 13 | 13 | 4 | 2 | 2 | 21 | 17 | 27 |
3 (x = 4) | 27 | 32 | 14 | 3 | 1 | 13 | 15 | 19 | 26 |
4 (x = 5) | 32 | 38 | 7 | 1 | 0 | 13 | 16 | 13 | 30 |
5 (x = 5) | 28 | 34 | 14 | 3 | 0 | 8 | 24 | 7 | 32 |
6 (x = 5) | 50 | 10 | 13 | 1 | 3 | 15 | 17 | 10 | 31 |
7 (x = 6) | 3 | 54 | 20 | 1 | 1 | 11 | 24 | 18 | 18 |
8 (x = 6) | 27 | 33 | 13 | 5 | 0 | 9 | 14 | 19 | 30 |
9 (x = 6) | 7 | 55 | 11 | 5 | 4 | 8 | 21 | 11 | 28 |
Old Experiment, Terminal Nodes Reached in Rounds 51–55
A | B | C | D | E | F | G | H | I | |
1 (x = 4) | 91 | 0 | 0 | 0 | 0 | 10 | 26 | 15 | 38 |
2 (x = 4) | 72 | 16 | 3 | 4 | 2 | 10 | 51 | 6 | 16 |
3 (x = 4) | 69 | 12 | 14 | 3 | 1 | 22 | 25 | 20 | 14 |
4 (x = 5) | 15 | 68 | 8 | 0 | 0 | 16 | 26 | 17 | 30 |
5 (x = 5) | 50 | 31 | 11 | 1 | 0 | 5 | 20 | 12 | 50 |
6 (x = 5) | 62 | 13 | 20 | 2 | 5 | 13 | 20 | 13 | 32 |
7 (x = 6) | 0 | 72 | 19 | 3 | 1 | 3 | 24 | 12 | 46 |
8 (x = 6) | 58 | 17 | 20 | 1 | 8 | 8 | 32 | 10 | 26 |
9 (x = 6) | 16 | 44 | 35 | 7 | 6 | 13 | 25 | 10 | 24 |
New Experiment, Terminal Nodes Reached in Rounds 1–25a
A | B | C | D | E | F | G | H | I | |
1 | 17 | 45 | 15 | 7 | 5 | 61 | 0 | 0 | 0 |
2 | 4 | 53 | 27 | 7 | 2 | 57 | 0 | 0 | 0 |
3 | 10 | 32 | 28 | 15 | 7 | 58 | 0 | 0 | 0 |
4 | 26 | 16 | 24 | 7 | 8 | 69 | 0 | 0 | 0 |
5 | 4 | 43 | 25 | 6 | 5 | 67 | 0 | 0 | 0 |
6* | 28 | 31 | 23 | 5 | 5 | 58 | 0 | 0 | 0 |
7* | 15 | 37 | 17 | 17 | 5 | 59 | 0 | 0 | 0 |
8* | 14 | 40 | 26 | 3 | 8 | 59 | 0 | 0 | 0 |
9* | 9 | 48 | 25 | 15 | 8 | 45 | 0 | 0 | 0 |
10* | 30 | 23 | 26 | 17 | 11 | 43 | 0 | 0 | 0 |
∑ | 157 | 368 | 236 | 99 | 64 | 576 | 0 | 0 | 0 |
aThe * refers to experimental sessions where information on all simultaneous plays was not given to the subjects
New Experiment, Terminal Nodes Reached in Rounds 26–50
A | B | C | D | E | F | G | H | I | |
1 | 2 | 49 | 14 | 5 | 1 | 2 | 16 | 6 | 55 |
2 | 1 | 61 | 12 | 13 | 3 | 0 | 2 | 2 | 56 |
3 | 24 | 29 | 34 | 2 | 0 | 2 | 3 | 5 | 51 |
4 | 33 | 21 | 26 | 0 | 1 | 0 | 1 | 4 | 64 |
5 | 17 | 38 | 23 | 6 | 6 | 0 | 14 | 2 | 44 |
6* | 35 | 24 | 17 | 8 | 2 | 1 | 9 | 7 | 47 |
7* | 36 | 7 | 33 | 4 | 5 | 0 | 2 | 1 | 62 |
8* | 10 | 50 | 19 | 2 | 2 | 0 | 1 | 3 | 63 |
9* | 5 | 48 | 16 | 5 | 0 | 1 | 7 | 5 | 63 |
10* | 43 | 9 | 23 | 5 | 7 | 0 | 7 | 1 | 55 |
∑ | 206 | 336 | 217 | 50 | 27 | 6 | 62 | 36 | 560 |
New Experiment, Terminal Nodes Reached in Rounds 51–55
A | B | C | D | E | F | G | H | I | |
1 | 7 | 59 | 23 | 11 | 3 | 2 | 17 | 1 | 57 |
2 | 0 | 64 | 32 | 10 | 6 | 0 | 2 | 3 | 63 |
3 | 19 | 55 | 20 | 0 | 0 | 1 | 9 | 19 | 57 |
4 | 31 | 22 | 27 | 0 | 0 | 0 | 5 | 6 | 89 |
5 | 16 | 38 | 35 | 7 | 2 | 0 | 26 | 2 | 54 |
6* | 39 | 35 | 30 | 2 | 5 | 0 | 15 | 3 | 51 |
7* | 25 | 24 | 27 | 0 | 0 | 0 | 0 | 7 | 97 |
8* | 7 | 45 | 41 | 4 | 3 | 0 | 2 | 11 | 67 |
9* | 0 | 59 | 25 | 7 | 4 | 1 | 7 | 2 | 75 |
10* | 32 | 28 | 36 | 13 | 9 | 0 | 15 | 6 | 41 |
∑ | 176 | 429 | 296 | 54 | 32 | 4 | 98 | 60 | 651 |
Rights and permissions
Copyright information
© 2010 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Balkenborg, D., Talloo, S. (2010). Deviations from Equilibrium in an Experiment on Signaling Games: First Results . In: Sadrieh, A., Ockenfels, A. (eds) The Selten School of Behavioral Economics. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-13983-3_7
Download citation
DOI: https://doi.org/10.1007/978-3-642-13983-3_7
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-13982-6
Online ISBN: 978-3-642-13983-3
eBook Packages: Business and EconomicsEconomics and Finance (R0)