Animals and Rational Behaviour

Having introduced a notion of Instrumental Rationality, perhaps a good question to ask is "What is an example of a rational agent which is not human?"

It might seem that animals behave rationally at times; for example, worker bees give up their own production in favor of other offspring of the queen, surely this must give some benefit to the donor. This idea (generalizing the bee situation) is known as Hamilton's Rule, and it doesn't really work. For example, it is neither testable nor does it make predictions.

With the bees, the altruist (i.e., worker bee) cooperates by giving a benefit b to the recipient (another offspring) at a cost c to itself. Both b and c are measured in terms of fitness, specifically the expected number of offspring. Naively one might expect b > c to suffice, but Hamilton's major insight was that relatedness ("degree of kinship") r between donor and recipient must enter into the equation, giving us Hamilton's rule br > c.

Game theorists are overjoyed to hear this can be derived from utility maximization, and one might expect it to have a status similar to Newton's laws in physics. However, Nowak, Tarnita, and Wilson have argued that Hamilton's rule almost never holds. In short, simple game theoretic models here fail to describe the biological situation.

Decades ago, evolutionary biologists would have treated Hamilton's rule as an "iron law". That no longer seems to be the case. For more on this, see de Vladar and Szathmáry's "Beyond Hamilton's Rule".