Hidden Games: The Surprising Power of Game Theory to Explain Irrational Human Behavior
These notes are old and were written while reading — they don’t necessarily reflect my current views.
This book tackles seemingly irrational behaviour from a game-theoretic perspective.
They start off with an example of the Europeans learning how to domesticate corn from American tribes. One of the steps, which softened the shell and made it easier to crush seemed unnecessary to them (they had powerful crushing machines), so they skipped it. It turns out that this step is crucial to extract some vitamin out of the corn. As a result many Europeans died due to a lack of this vitamin.
Similarly, if you show how to open a box to both a chimp and a human child and include some unnecessary extra steps, the child will copy the extra steps while the chimp wont. This might lead people to thing that the chimps are smarter due to their ability to recognise the causal links. Still, the approach of the humans may be wiser, as there might always be hidden mechanisms we don’t understand.
This leads to a fundamentally conservative argument.
One concept they introduce is the distinction between primary and secondary rewards; primary rewards are things we are had wired to like (food, sex, human interaction). Secondary rewards are things we are trained to like (often because they are in our culture instrumental to get primary rewards)
They critique the notion of irrational behaviour by pointing out the example of the Israeli daycare. This behaviour seems irrational at first glance but is very rational if we also consider the social cost in our model.
They show the power of game theory by tackling the question of sex ratio among species (see Dawkins). Further, they show how shared expectations lead to clear outcomes in dove-hawk scenarios.
They show that signaling is useful when the signal is easier to send for fit individuals than unfit ones. Humility is a form of advanced signaling. Discovering a positive trait that an individual tried to hide is impressive. A less fit individual might not be in a position to hide these positive attributes, as it is less like some of them will come up when you have fewer.
Games work differently depending on the (expected) motive of an agent. Optimal behaviour is different in relation to an individual trying to inform than trying to persuade.
Humans seem to suffer from overconfidence due to biased updating of data. This is the dominant strategy in situations with information asymmetry. if you have more information than the other party, they will expect you to release it selectively to favour you (as this would increase your expected payoff). You know that it is rational for them to suspect that behaviour from you, hence releasing the information in an unbiased way would be bad for you (as they would treat it as biased information). Hence your optimal strategy is to release biased information.
Further, they describe the effects of higher order punishment. Values don’t have to be internalised, as other actors might punish you for misbehaviour. Similarly they will be motivated by the fear of being punished by other if they do not sufficiently punish the misbehaviour.
In chapter 12 they focus on signals. The problem is often that one wants to give a response A if another party will act in a certain way but B if not. Therefore, it is crucial to assess whether the other party will act in a certain way. This might be contingent on the information the other party has. Continous public signals are less efficient than discrete ones as they are more open to interpretation. For more details reread chapter 12.
One seemingly irrational behaviour is that people give to charity but it does not make a difference how effective the charity is. This is due to the fact that it is not about the actual effect but the signaling. It not about the continous variable but the discrete one if you gave something. The results are very different if the lifes one could save are not abstract but that of family members. Here higher effectiveness is valued.