The emphasis on choice as the basic building block of explanation in the social sciences goes together with an emphasis on the intended consequences of action. In this chapter, I discuss explanations of actions in terms of their objective consequences. This might seem like an unpromising idea. All explanation is causal explanation. We explain an event by citing its cause. Causes precede their effects in time. It follows that we cannot explain an event, e.g. an action, by its consequences.

If, however, the explanandum is a pattern of recurrent behavior, the consequences of that behavior on one occasion can enter into the causes that make its occurrence on a later occasion more likely. There are two main ways in which this can happen: by reinforcement and by selection. I shall focus on the second, which is the more important for my purposes, but begin with some words about the first.

Reinforcement

If the consequences of given behavior are pleasant or rewarding, we tend to engage in it more often; if they are unpleasant or punishing it will occur less often. The underlying mechanism could simply be conscious rational choice, if we notice the pleasant or unpleasant consequences and decide to act in the future so as to repeat or avoid repeating the experience. Often, however, the reinforcement can happen without intentional choice. When infants learn to cry because the parents reward them by picking them up when they do, there is no reason to think that they first consciously note the benefits from crying and then later cry at will to get them. When older children throw a tantrum to get their way, parents can usually tell that it is not a genuine one.

Reinforcement learning has been extensively studied in laboratory experiments on animals. One typically offers the animal the opportunity to press a lever, or one among several levers, and rewards the presses either as a function of the number of lever presses since the last reward or as a function of the time passed since the last reward. In either case, the function can be deterministic or probabilistic. In fixed-ratio schedules, the animal receives a reward after it has pressed a lever a fixed number of times, whereas in variable-ratio schedules the number of presses needed to produce a reward varies randomly. In either case, each press produces a “reward point” that is added to previous points.