On the Emergence of Lies

lie -- pronunciation [lahy] noun,
verb: lied, ly·ing.

noun:
1. a false statement made with deliberate intent to deceive; an intentional untruth; a falsehood.
2. something intended or serving to convey a false impression.
3. an inaccurate or false statement.
4. the charge or accusation of lying.

verb:
5. to speak falsely or utter untruth knowingly, as with intent to deceive.
6. to express what is false; convey a false impression.

[Source: Dictionary.com]

---

I've been browsing recently through the references in the previously reviewed book "Complex Adaptive Systems" by Miller and Page about the use of agent based computational models for social interactions. While doing so, I came across a paper that I found quite interesting:

Communication and cooperation
By John H. Miller, Carter T. Butts and David Rode
Journal of Economic Behavior & Organization
Volume 47, Issue 2, February 2002, p. 179-195

In this paper the authors examine the role that communication plays in the development of strategies. They use a very specific model, but the results they find have the potential to be more general. And since this is a blog, I want to speculate somewhat about it.

---

The Model

In the model examined in the paper the agents play in the prisoner's dilemma. This is a fairly simple game, in which the players receive payoffs depending on whether they cooperate or defect. Wikipedia summarizes the classical prisoner's dilemma as follows:

Two suspects are arrested by the police. The police have insufficient evidence for a conviction, and, having separated both prisoners, visit each of them to offer the same deal: if one testifies for the prosecution against the other and the other remains silent, the betrayer goes free and the silent accomplice receives the full 10-year sentence. If both remain silent, both prisoners are sentenced to only six months in jail for a minor charge. If each betrays the other, each receives a five-year sentence. Each prisoner must make the choice of whether to betray the other or to remain silent. Each one is assured that the other would not know about the betrayal before the end of the investigation. How should the prisoners act?

Or, in table form:

	Prisoner B Stays Silent	Prisoner B Betrays
Prisoner A Stays Silent	Each serves 6 months	Prisoner A: 10 years Prisoner B: goes free
Prisoner A Betrays	Prisoner A: goes free Prisoner B: 10 years	Each serves 5 years

In this game, regardless of what the opponent chooses, each player always receives a higher payoff (lesser sentence) by betraying, and thus betraying is the strictly dominant strategy.

In the model examined in the paper, the players can now in addition exchange communication tokens, where one of the tokens signals that the player selected a move. Their exchange continues until either both players indicate they have made a decision, or until the communication exceeds some chat limit. The additional payoffs from this possibility are that a player who has not chosen a move before reaching the chat limit obtains a punishment (a negative payoff), a player who picks a move but his opponent fails to does receive a payoff between that of mutual defection and mutual cooperation.

As far as I understand it, in each round each player plays with every other player. Payoffs are summed up, and then the player's strategies undergo a selection and mutation process, in which the best strategies have a survival advantage, plus some amount of randomness. And then the next round starts. I think for the following results it is crucial that too long communication without outcome has a negative payoff, whether or not that has to be included in exactly this form I don't know. I would have thought e.g. that those players who talk too much just get to play less in each round, which would also amount to a disadvantage. Either way, interpret it as you wish, the point is that blahblah without outcome sucks.

---

The Results

So here is in a nutshell the result of running this model a lot of times; I summarized it in the figure below. It sketches the hypothesis the authors put forward to interpret the data that they have collected.


The rounded boxes indicate the dominant strategy, and the arrows are some learning processes.

Suppose we start at the top, in a world in which there is no communication and the players in the prisoner's dilemma thus mutually defect. There might be the occasional mutant that tries to communicate, but if the other player doesn't listen, or doesn't understand this doesn't have any effect. But worse than not having an effect: if the talkative players get trapped into chatting to much, they receive a punishment. The authors further point out that the reason why no communication and mutual defection is a stable strategy is that communicating players are more vulnerable to mutations, which in addition with chatting too much being a disadvantage leads them to suspect this reduces the survivability of the communicating players.

The situation changes if two players meet each other who communicate and understand each other. They can then choose to cooperate, receive a higher payoff, and have a survival advantage. This leads to a rather sudden increase in communication and cooperation.

However, this emergence of communication and cooperation sows the seeds of its own destruction: It doesn't take a large mutation from the cooperative players to those that pretend to cooperate, but then defeat - which results in a higher payoff for them, to the disadvantage of the cooperative ones. Now one could suspect that the cooperative players will try to use some code to identify each other as being of the same type, and the others as being mimics. But whatever the code is they come up with, again it takes only a small mutation to turn it into a mimic.

This then leads to a lot of communication with decreasing cooperation. In the course of this the players will come to notice that talking without outcome is a disadvantage, so to improve the strategy it is beneficial to not talk at all. This leads to a gradual decay of communication back to the initial state.

These outbreaks of communication and cooperation with following decay are neither periodic, not are the outbreaks of equal size.

---

And We

Now I find this kind of interesting, as I think the development of sophisticated communication among humans, and the possibility to exchange information efficiently, is one of the most important evolutionary advantages. Of course the investigated model is a very simplistic one, and there is no good reason to believe it tells us something about beings playing such complex games as The-Game-Of-Human-Life. Maybe most importantly, in the examined case the players are not able to consider long-term effects of their actions, neither can they learn over the course of various cycles. But it's intriguing to speculate about the analogy.

I am completely convinced the amount of advertisement and commercials is an indicator for the certain decline of civilization. Above all other things it signals a culture of betrayal that we get more or less used to. Thus, we learn to some extend to mistrust information we receive. How many of the pills that you can buy on the internet will actually hold their promise? How many of these lotions will actually make you look younger? How much of what is 'guaranteed' is actually 'guaranteed'? How much of the stuff they try to convince you you can't live without is actually a completely unnecessary waste of resources?

Can you trust your used car dealer? Will the candidate hold his promises after the election? Do you believe what they write in the newspaper, or do they just sex up stories to obtain more attention, higher payoffs? (13 years old boy corrects NASA! - Fact or Fiction?). Are these boobs real?

What can we do to deal with this emergence of deceit, originating in the larger individual advantage? Well, we make up laws to punish lies* that can lead to damage. And make up religions to scare those who lie. In this way, we essentially incorporate the long-term effects of our actions.

However, dishonesty for the own advantage, and the resulting mistrust is a serious political problem on the global scale that corrupts our efforts to address the challenges we are facing in the 21st Century. Jeffrey Sachs said that very aptly:

"Despite the vast stores of energy, including nonconventional fuels, solar power, geothermal power, nuclear power, and more, there is a pervasive fear of an imminent energy crisis resulting from the depletion of oil. The scramble of powerful countries to control Middle East oil or newly discovered reserves in other parts of the world, such as West Africa and the Arctic, has surely intensified, while investments in alternative and sustainable energy sources have been woefully insufficient. This is an example of a vicious cycle of distrust. The world could adopt a cooperative approach to develop sustainable energy supplies, with sustainability in the dual sense of low greenhouse gas emissions and long-term, low-cost availability. Alternatively, we can scramble for the depleting conventional gas and oil resources. The scramble, very much under way today, reduces global cooperation, spills over into violence and risks great power confrontations, and makes even more distant the good-faith cooperation to pool R&D investments to develop alternative fuels and alternative ways to use nonconventional fossil fuels.
The Bush administration has been more consumed by the scramble rather than by cooperative global investments in a long-term future [...]"

~Jeffrey D. Sachs, in "Common Wealth", p. 45, typos and emphasis mine.

So, where are we in the diagram?

---

* I here refer to lie as with being of the intention to deceit the communicating partner to the own advantage. In other instances, lies serve various social purposes, as e.g. politeness, simplification or to cover lack of knowledge.

---

TAGS: AGENT BASED MODELS, LIES, EMERGENCE
See also: Communication