Signaling game

(Redirected from Signaling games)
An extensive form representation of a signaling game

In game theory, a signaling game is a simple type of a dynamic Bayesian game. [1]

It is a game with two players, called the sender (S) and the receiver (R):

• The sender can have one of several types. The sender's type, t, determines the payoff function of the sender. It is the private information of the sender - it is not known to the receiver.
• The receiver has only a single type, so his payoff function is known to both players.

The game has two steps:

• The sender plays in the first step. He can play one of several actions, which are called "messages". The set of possible messages is M = {m1, m2, m3,..., mj}.
• The receiver plays in the second step, after viewing the sender's message. The set of possible actions is A = {a1, a2, a3,...., ak}.

The two players receive payoffs dependent on the sender's type, the message chosen by the sender and the action chosen by the receiver.[2][3]

Perfect Bayesian equilibrium

The equilibrium concept that is relevant for signaling games is Perfect Bayesian equilibrium - a refinement of both Bayesian Nash equilibrium and subgame-perfect equilibrium.

A sender of type ${\displaystyle t_{j}}$ sends a message ${\displaystyle m^{*}(t_{j})}$ in the set of probability distributions over M. (${\displaystyle m(t_{j})}$ represents the probabilities that type ${\displaystyle t_{j}}$ will take any of the messages in M.) The receiver observing the message m takes an action ${\displaystyle a^{*}(m)}$ in the space of probability distributions over A.

A game is in perfect Bayesian equilibrium if it meets all four of the following requirements:

• The receiver must have a belief about which types can have sent message m. These beliefs can be described as a probability distribution ${\displaystyle \mu (t_{i}|m)}$, the probability that the sender has type ${\displaystyle t_{i}}$ if he chooses message ${\displaystyle m}$. The sum over all types ${\displaystyle t_{i}}$ of these probabilities has to be 1 conditional on any message m.
• The action the receiver chooses must maximize the expected utility of the receiver given his beliefs about which type could have sent message ${\displaystyle m}$, ${\displaystyle \mu (t|m)}$. This means that the sum ${\displaystyle \sum _{t_{i}}\mu (t_{i}|m)U_{R}(t_{i},m,a)}$ is maximized. The action ${\displaystyle a}$ that maximizes this sum is ${\displaystyle a^{*}(m)}$.
• For each type, ${\displaystyle t}$, the sender chooses to send the message ${\displaystyle m^{*}}$ that maximizes the sender's utility ${\displaystyle U_{S}(t,m,a^{*}(m))}$ given the strategy chosen by the receiver, ${\displaystyle a^{*}}$.
• For each message ${\displaystyle m}$ the sender can send, if there exists a type ${\displaystyle t}$ such that ${\displaystyle m^{*}(t)}$ assigns strictly positive probability to ${\displaystyle m}$ (i.e. for each message which is sent with positive probability), the belief the receiver has about the type of the sender if he observes message ${\displaystyle m}$, ${\displaystyle \mu (t|m)}$ satisfies Bayes' rule: ${\displaystyle \mu (t|m)=p(t)/\sum _{t_{i}}p(t_{i})}$

The perfect Bayesian equilibria in such a game can be divided in three different categories: pooling equilibria, separating equilibria and semi-separating

• A pooling equilibrium is an equilibrium where senders with different types all choose the same message. This means that the sender's message does not give any information to the receiver, so the receiver's beliefs are not updated after seeing the message.
• A separating equilibrium is an equilibrium where senders with different types always choose different messages. This means that the sender's message always reveals the sender's type, so the receiver's beliefs become deterministic after seeing the message.
• A semi-separating (also called partial-pooling) equilibrium is an equilibrium where some types of senders choose the same message and other types choose different messages.

Note that, if there are more types of senders than there are messages, the equilibrium can never be a separating equilibrium (but may be semi-separating equilibria). There are also hybrid equilibria, in which the sender randomizes between pooling and separating.

Examples

Reputation game

Sane, Prey P1+P1, D2 P1+M1, 0
Sane, Accommodate D1+D1, D2 D1+M1, 0
Crazy, Prey X1, P2 X1, 0

In this game,[1]:326-329[4] the sender and the receiver are firms. The sender is an incumbent firm and the receiver is an entrant firm.

• The sender can be one of two types: Sane or Crazy. A sane sender can send one of two messages: Prey and Accommodate. A crazy sender can only Prey.
• The receiver can do one of two actions: Stay or Exit.

The payoffs are given by the table at the right. We assume that:

• M1>D1>P1, i.e, a sane sender prefers to be a monopoly (M1), but if it is not monopoly, it prefers to accommodate (D1) than to prey (P1). Note that the value of X1 is irrelevant since a Crazy firm has only one possible action.
• D2>0>P2, i.e, the receiver prefers to stay in a market with a sane competitor (D2) than to exit the market (0), but prefers to exit than to be stay in a market with a crazy competitor (P2).
• Apriori, the sender has probability p to be sane and 1-p to be crazy.

We now look for perfect Bayesian equilibria. It is convenient to differentiate between separating equilibria and pooling equilibria.

• A separating equilibrium, in our case, is one in which the sane sender always accommodates. This separates it from a crazy sender. In the second period, the receiver has complete information, and his best strategy is: "If Accommodate then Stay, if Prey then Exit". The payoff of the sender when he accommodates is D1+D1, but if he deviates to Prey his payoff changes to P1+M1; therefore, a necessary condition for a separating equilibrium is D1+D1≥P1+M1. It is possible to show that this condition is also sufficient.
• A pooling equilibrium is one in which the sane sender always preys. In the second period, the receiver has no new information: he still believes that the sender is sane with probability p and crazy with probability 1-p. Therefore, the receiver's expected payoff if he stays is: [p D2 + (1-p) P2]; the receiver stays if-and-only-if this expression is positive. The sender can gain from preying, only if the receiver exits. Therefore, a necessary condition for a pooling equilibrium is p D2 + (1-p) P2 ≤ 0. It is possible to show that, if D1+D1<P1+M1 (so there is no separating equilibrium), this condition is also sufficient.

To summarize:

• If preying is costly for a sane sender (D1+D1≥P1+M1), he will accommodate and there will be a separating PBE.
• If preying is not too costly for a sane sender but harmful for the receiver (p D2 + (1-p) P2 ≤ 0), the sender will prey and there will be a pooling PBE. The sender is willing to lose some value by preying in the first period, in order to build a reputation of a predatory firm, and convince the receiver to exit.
• If preying is not costly for the sender nor harmful for the receiver, there will be a unique hybrid PBE - both the sender and the receiver will randomize between their two actions.

Education game

This game was first presented by Michael Spence.[5][1]:329-331 In this game, the sender is a worker and the receiver is an employer.

• The worker can be one of two types: Wise (with probability p) or Dumb (with probability 1-p). Each type can select his level of education, e.g. GoToCollege or StayAtHome. Going to college has a cost; the cost is lower for a wise worker than for a dumb one.
• The employer has to decide how much salary to offer the worker. The goal of the employer is to offer a high salary to a Wise worker and a low salary to a Dumb worker. However, the employer does not know the true talent of the worker - only his level of education.

In this model it is assumed that the level of education does not influence the productivity of the worker; it is used only as a signal regarding the worker's talent.

To summarize: only workers with high ability are able to attain a specific level of education without it being more costly than their increase in wage. In other words, the benefits of education are only greater than the costs for workers with a high level of ability, so only workers with a high ability will get an education.

Applications of signaling games

Signaling games describe situations where one player has information the other player does not have. These situations of asymmetric information are very common in economics and behavioral biology.

Philosophy

The first known use of signaling games occurs in David K. Lewis' Ph. D. dissertation (and later book) Convention.[6] Replying to W.V.O. Quine,[7][8] Lewis attempts to develop a theory of convention and meaning using signaling games. In his most extreme comments, he suggests that understanding the equilibrium properties of the appropriate signaling game captures all there is to know about meaning:

I have now described the character of a case of signaling without mentioning the meaning of the signals: that two lanterns meant that the redcoats were coming by sea, or whatever. But nothing important seems to have been left unsaid, so what has been said must somehow imply that the signals have their meanings.[9]

The use of signaling games has been continued in the philosophical literature. Others have used evolutionary models of signaling games to describe the emergence of language. Work on the emergence of language in simple signaling games includes models by Huttegger,[10] Grim, et al.,[11] Skyrms,[12][13] and Zollman.[14] Harms,[15][16] and Huttegger,[17] have attempted to extend the study to include the distinction between normative and descriptive language.

Economics

The first application of signaling games to economic problems was Michael Spence's #Education game. A second application was the #Reputation game.

Biology

Valuable advances have been made by applying signaling games to a number of biological questions. Most notably, Alan Grafen's (1990) handicap model of mate attraction displays.[18] The antlers of stags, the elaborate plumage of peacocks and bird-of-paradise, and the song of the nightingale are all such signals. Grafen’s analysis of biological signaling is formally similar to the classic monograph on economic market signaling by Michael Spence.[19] More recently, a series of papers by Getty[20][21][22][23] shows that Grafen’s analysis, like that of Spence, is based on the critical simplifying assumption that signalers trade off costs for benefits in an additive fashion, the way humans invest money to increase income in the same currency. This assumption that costs and benefits trade off in an additive fashion might be valid for some biological signaling systems, but is not valid for multiplicative tradeoffs, such as the survival cost – reproduction benefit tradeoff that is assumed to mediate the evolution of sexually selected signals.

Charles Godfray (1991) modeled the begging behavior of nestling birds as a signaling game.[24] The nestlings begging not only informs the parents that the nestling is hungry, but also attracts predators to the nest. The parents and nestlings are in conflict. The nestlings benefit if the parents work harder to feed them than the parents ultimate benefit level of investment. The parents are trading off investment in the current nestlings against investment in future offspring.

Pursuit deterrent signals have been modeled as signaling games.[25] Thompson's gazelles are known sometimes to perform a 'stott', a jump into the air of several feet with the white tail showing, when they detect a predator. Alcock and others have suggested that this action is a signal of the gazelle's speed to the predator. This action successfully distinguishes types because it would be impossible or too costly for a sick creature to perform and hence the predator is deterred from chasing a stotting gazelle because it is obviously very agile and would prove hard to catch.

The concept of information asymmetry in molecular biology has long been apparent.[26] Although molecules are not rational agents, simulations have shown that through replication, selection, and genetic drift, molecules can behave according to signaling game dynamics. Such models have been proposed to explain, for example, the emergence of the genetic code from an RNA and amino acid world.[27]

Costly versus cost-free signaling

One of the major uses of signaling games both in economics and biology has been to determine under what conditions honest signaling can be an equilibrium of the game. That is, under what conditions can we expect rational people or animals subject to natural selection to reveal information about their types?

If both parties have coinciding interest, that is they both prefer the same outcomes in all situations, then honesty is an equilibrium. (Although in most of these cases non-communicative equilbria exist as well.) However, if the parties' interests do not perfectly overlap, then the maintenance of informative signaling systems raises an important problem.

Consider a circumstance described by John Maynard Smith regarding transfer between related individuals. Suppose a signaler can be either starving or just hungry, and she can signal that fact to another individual which has food. Suppose that she would like more food regardless of her state, but that the individual with food only wants to give her the food if she is starving. While both players have identical interests when the signaler is starving, they have opposing interests when she is only hungry. When the signaler is hungry she has an incentive to lie about her need in order to obtain the food. And if the signaler regularly lies, then the receiver should ignore the signal and do whatever he thinks best.

Determining how signaling is stable in these situations has concerned both economists and biologists, and both have independently suggested that signal cost might play a role. If sending one signal is costly, it might only be worth the cost for the starving person to signal. The analysis of when costs are necessary to sustain honesty has been a significant area of research in both these fields.

References

1. ^ a b c Subsection 8.2.2 in
2. ^ Gibbons, Robert (1992). A Primer in Game Theory. New York: Harvester Wheatsheaf. ISBN 0-7450-1159-4.
3. ^ Osborne, M. J. & Rubenstein, A. (1994). A Course in Game Theory. Cambridge: MIT Press. ISBN 0-262-65040-1.
4. ^ which is a simplified version of a reputation model suggested in 1982 by Kreps, Wilson, Milgrom and Roberts
5. ^ Spence, A. M. (1973). "Job Market Signaling". Quarterly Journal of Economics. 87 (3): 355–374. doi:10.2307/1882010.
6. ^ Lewis, D. (1969). Convention. A Philosophical Study. Cambridge: Harvard University Press.
7. ^ Quine, W. V. O. (1936). "Truth by Convention". Philosophical Essays for Alfred North Whitehead. London: Longmans, Green & Co. pp. 90–124. ISBN 0-8462-0970-5. (Reprinting)
8. ^ Quine, W. V. O. (1960). "Carnap and Logical Truth". Synthese. 12 (4): 350–374. doi:10.1007/BF00485423.
9. ^ Lewis (1969), p. 124.
10. ^ Huttegger, S. M. (2007). "Evolution and the Explanation of Meaning". Philosophy of Science. 74 (1): 1–24. doi:10.1086/519477.
11. ^ Grim, P.; Kokalis, T.; Alai-Tafti, A.; Kilb, N.; St. Denis, Paul (2001). "Making Meaning Happen". Technical Report #01-02. Stony Brook: Group for Logic and Formal Semantics SUNY, Stony Brook.
12. ^ Skyrms, B. (1996). Evolution of the Social Contract. Cambridge: Cambridge University Press. ISBN 0-521-55471-3.
13. ^ Skyrms, B. (2010). Signals Evolution, Learning & Information. New York: Oxford University Press. ISBN 978-0-19-958082-8.
14. ^ Zollman, K. J. S. (2005). "Talking to Neighbors: The Evolution of Regional Meaning". Philosophy of Science. 72 (1): 69–85. doi:10.1086/428390.
15. ^ Harms, W. F. (2000). "Adaption and Moral Realism". Biology and Philosophy. 15 (5): 699–712. doi:10.1023/A:1006661726993.
16. ^ Harms, W. F. (2004). Information and Meaning in Evolutionary Processes. Cambridge: Cambridge University Press. ISBN 0-521-81514-2.
17. ^ Huttegger, S. M. (2005). "Evolutionary Explanations of Indicatives and Imperatives". Erkenntnis. 66 (3): 409–436. doi:10.1007/s10670-006-9022-1.
18. ^ Grafen, A. (1990). "Biological signals as handicaps". Journal of Theoretical Biology. 144 (4): 517–546. doi:10.1016/S0022-5193(05)80088-8. PMID 2402153.
19. ^ Spence, A. M. (1974). Market Signaling: Information Transfer in Hiring and Related Processes. Cambridge: Harvard University Press.
20. ^ Getty, T. (1998). "Handicap signalling: when fecundity and viability do not add up". Animal Behaviour. 56 (1): 127–130. doi:10.1006/anbe.1998.0744. PMID 9710469.
21. ^ Getty, T. (1998). "Reliable signalling need not be a handicap". Animal Behaviour. 56 (1): 253–255. doi:10.1006/anbe.1998.0748. PMID 9710484.
22. ^ Getty, T. (2002). "Signaling health versus parasites". The American Naturalist. 159 (4): 363–371. doi:10.1086/338992. PMID 18707421.
23. ^ Getty, T. (2006). "Sexually selected signals are not similar to sports handicaps". Trends in Ecology & Evolution. 21 (2): 83–88. doi:10.1016/j.tree.2005.10.016.
24. ^ Godfray, H. C. J. (1991). "Signalling of need by offspring to their parents". Nature. 352 (6333): 328–330. doi:10.1038/352328a0.
25. ^ Yachi, S. (1995). "How can honest signalling evolve? The role of the handicap principle". Proceedings of the Royal Society of London B. 262 (1365): 283–288. doi:10.1098/rspb.1995.0207.
26. ^ John Maynard Smith. (2000) The Concept of Information in Biology. Philosophy of Science. 67(2):177-194
27. ^ Jee, J.; Sundstrom, A.; Massey, S.E.; Mishra, B. (2013). "What can information-asymmetric games tell us about the context of Crick's 'Frozen Accident'?". Journal of the Royal Society Interface. 10 (88): 20130614. doi:10.1098/rsif.2013.0614. PMC 3785830. PMID 23985735.