« California Legislature Passes Driver Licenses for Illegal Aliens | Gene Expression Front Page | SES and IQ Heritability »
September 04, 2003

HONEST SIGNALS

Last month I commented (here) on the Handicap Principle of Amotz Zahavi. The Handicap Principle maintains that the high cost of producing a signal (such as a mating display) may contribute to its success, and may even be necessary for ‘honest’ signals to evolve.

Since my earlier post I find that there is a flourishing field of research on signaling theory in both biology and economics. A good non-technical introduction is here.

It seems now to be generally accepted that the Handicap Principle is valid, in the sense that costly signals can be selected for in some circumstances.

There is more debate about whether honest signals must be costly. (Of course, all signals require some expenditure of energy or materials - the point at issue is whether they must be costly over and above what is necessary for effective transmission of information.)

There is a theoretical proof by Maynard Smith that honest signals need not be costly where there is no danger of conflict of interest between sender and receiver. Examples would be communication between clonal organisms (including parts of the same body), or species recognition markings which help different species avoid infertile matings with each other.

But more usually there is a danger of conflict of interest. In sexual selection, for example, it is in the interests of a fit male to inform females of his fitness, but it is also in the interests of unfit males to pretend that they are fit. To convey reliable information the signal must therefore be of a kind that costs more to a dishonest signaler than to an honest one.

There remains an argument about whether it is ncessary for the cost to fall on both honest and dishonest signalers, or whether it is sufficient if dishonest signalers pay a penalty when they are found out. The latest view seems to be that the ‘penalty’ system can be sufficient if the danger of discovery is high enough, or if there is collective punishment of the offender.

All of this has obvious implications for human communication. On the face of it human speech is a very low-cost system of signals. This may be an oversimplification - it takes time to talk, and a con-man, seducer, preacher or politician may have to invest a great deal of time and effort in convincing his victims.

But no doubt a lot of human speech is genuinely low-cost. Much routine speech falls into the category of ‘no conflict of interest’. In all kinds of team effort, individuals stand to gain from the success of the team, so they have no interest in disrupting its efforts.

This still leaves a wide range of communication, especially between strangers, where there could be conflicts of interest. So it is an interesting question what mechanisms exist to avoid deception, and how effective they are. One key mechanism is that of reputation. In many societies a man’s good name is his most precious possession. (There is a quote from Shakespeare somewhere, if only I could remember it!) A single instance of dishonesty destroys a reputation, and it is virtually impossible to rebuild it.

But different societies attach different degrees of importance to honesty, or they may apply it selectively. In 18th century Europe an aristocrat was expected to keep his word to his equals, but could cheerfully cheat his social inferiors. There are also implications for communications between different cultural or ethnic groups, with possibly different standards of honesty.

Not much of the recent work on signal theory has yet reached the popular domain, but I predict that it will before long. A book on Animal Signals by John Maynard Smith and David Harper is scheduled for publication later this year.

DAVID BURBRIDGE

Posted by David B at 04:57 AM




In the game theory branch of economics, there is much discussion of "cheap talk", ie, costless signals. These affect equilibria in various ways. Usually, these cheap signals can generate equilibria where the signals are honest if other people believe them to be so. This may seem odd, but there are many game theoretic equilibia that depend on the participant's 'prior' beliefs, so that in effect, if they believe it to be true, it will be.

Check out "cheap talk" and "game theory" in search engines, it's a lifetime labor of love for some people, and the games can be very elegant, and very rich.

Posted by: eric f at September 4, 2003 07:48 AM