Reasoning about agents

Rational choice theory usually advances a highly abstract theory of decision-making -- utility-maximizing choice among discrete options -- and then draws deductive conclusions. But actual human reasoners don't look much like this abstract ideal. It is interesting to consider how much one can explain while weakening the heroic assumptions about agent rationality. It turns out that it is possible to explain quite a bit of social behavior on the basis of a theory of agency that incorporates only a few qualitative assumptions about practical agency: purposiveness, deliberation, comparison, and choice -- without the specific mathematics of utility theory. And this gives a basis for a “practical agency” type of social explanation that doesn’t fall directly under the many criticisms that are offered of formal rational choice theory. And it represents a framework of analysis that corresponds fairly directly to Aristotle’s conception of deliberation.

First, purposiveness. We assume that agents have purposes -- states of affairs that they want to bring about. Their actions are intended to bring about their goals. (This is the basic assumption of means-end rationality.)

Second, deliberation. We assume that agents collect information about the courses of action that are feasible in the moment. And they collect information about the likely consequences of the various actions that they are considering. Within a deliberative process they choose an action to pursue.

Third, comparison. We assume that the agent deliberates by considering the advantages and disadvantages of each action (costs and benefits) and the degree to which the possible or probable effects of the choice serve the set of purposes the agent pursues. Further, the agent may compare choices on the basis of the nature of the action itself rather than solely on the basis of consequences -- that is, the agent may combine deontological considerations with utilitarian considerations (this represents an element of Kant’s theory of moral decision-making). That is, the agent chooses on the basis of full comparison of the choices.

So far we have described purposive deliberation without invoking utilities. We have invoked the idea of preference in this account, because we have assumed that the agent prefers outcomes that better fulfill his/her purposes. And we have likewise invoked comparison and the idea of “more and less” of something in ranking outcomes. The apparatus of utility theory is one way of articulating or modeling these features -- but we are not forced to attribute a full apparatus of utility measurement and aggregation in order to attribute purposiveness and comparison. It will suffice if the agent can judge "X is better than Y in fulfilling my purposes."

So let's refer to this less abstract description as a “broad theory of purposive agency”. The question here is a simple one: does this description give us enough to get a social explanation going? I believe that it does, and that this demonstrates that the conception of agents as purposive decision-makers is more durable than any particular formal theory of decision-making.

I support this observation with a single inconclusive illustration. We can reproduce the public goods problem, and the derivation of the rationality of free-riding, using only the comparative resources of the reduced theory of agency. And we can simultaneously prepare a theoretical location for future empirical research for those instances where free-riding does not materialize as expected.

First the derivation of the tendency toward free riding in the presence of public goods. The agent has preferences among the possible outcomes of possible collective action. He/she recognizes that the probability of realizing the collective good is not significantly altered by participation. Agent further recognizes that there is a cost to participation. Agent decides to not participate, based on costs and outcomes: the benefit of participation and non-participation is equal and the cost of participation is greater, so non-participation is preferred.

But now consider what happens when we turn on a deontological component of the comparison. Agent notes that non-participation involves taking unfair advantage of the actions of others. Agent prefers actions that are fair. Agent therefore chooses participation.

Finally, we might further complicate things by allowing that agent allows tradeoffs between the cost-benefit comparison and the deontological comparison. In some instances the weight of fairness prevails and agent chooses the less-good alternative. In other cases the situation reverses. Agent sacrifices minor unfairness and chooses the better outcome.

There are many, many examples of explanations in sociology, anthropology, and history that account for an outcome based on analysis of the situation of deliberation confronting a hypothetical set of purposive agents; and substantial explanations ensue. (As one example, consider Jean Ensminger’s explanation of Kenyan cattle-tending and bride price practices as a solution to a principal-agent problem between cattle owners and cattle tenders; Making a Market: The Institutional Transformation of an African Society.)