Two prisoners are being interrogated. The matrix below displays the outcomes of their strategies.

Which one should they choose?

In game theory, **rational players** will always choose a **“dominant strategy”**—the best thing to do for oneself *no matter what* the other prisoner does—if there is one.

This reasoning suggests that both prisoners in the example above will confess. This is also a **Nash equilibrium:** each prisoner’s strategy yields the highest possible payoff *given* the other prisoner’s strategy.

In the prisoners’ dilemma, if both prisoners confess and thereby arrive at a Nash equilibrium, both are worse off than if they kept quiet.

Sending both prisoners to jail for ten years is not a satisfactory outcome.

**Strategic thinking,**

however,

is not the only

valid form of **rationality.**

Other concepts can be invoked to allow an individual to make **rational decisions.**

**Bounded rationality** is one of them.

Both prisoners could also assume that their decisions will **mirror each other** and thus decide to keep quiet, spending only one year in jail.

Given their options, this is the **best possible outcome.**

For example, **Kantian optimisation** posits that players **modify their strategy** assuming that their **opponent will do the same.**

Such reasoning leads to both prisoners choosing to “keep quiet". Each prisoner thus spends a single year in jail: the **collectively rational solution** to the game.

Jean Dreze’s article, “The Real Insights of Game Theory” discusses the nature of strategic reasoning and argues that in strategic situations, self-interested behaviour need not be well defined, or let alone compelling.