2.2. Cognitive costs

We have so far assumed that there are ‘‘resources’’ available for solving

voluntary reasoning problems and that these resources can be moved

around easily from one reasoning strategy to another. (In economic parlance,

we’re assuming these resources are ‘‘fungible.’’) But that is not so.

There are a number of quite different kinds of resources required for

solving voluntary reasoning problems—time, attention, and short-term

and long-term cognitive capacity. Different reasoning strategies will place

more pressure on some of these resources than others. And some of these

resources handle the strain better than others. One reasoning strategy might

be easy to use because the effort is spread out across various capacities that

work together easily. Another strategy might be just as easy to use (measured,

say, in terms of how many basic operations it employs), but it

might be very hard to implement because its demands overwhelm a single

capacity (e.g., short term memory) or are spread out across capacities that

interfere with each other’s effective operation. It is doubtful that there is

any way to sensibly reduce these various demands on our various capacities

to a single notion of cognitive cost that can be represented by units

along a single dimension.

As was the case with cognitive benefits, the fact that we can’t accurately

assign units of epistemic cost to a reasoning strategy is not a particularly

serious problem for our cost-benefit approach to epistemology.

The notion of epistemic cost does map onto a notion we are able to use to

good effect—the notion of some strategies being harder or easier to use

than others. Indeed, given that people are less likely to use a reasoning

strategy as it gets more complicated (Nisbett 1993), there is good reason to

The Costs and Benefits of Excellent Judgment 87

stick to as simple a strategy as possible. We can measure the time it takes

people to implement the strategy; and this tracks reasonably well (in most

cases) the real costs of reasoning. So this measurable quantity can serve as

an imperfect stand-in for the notion of cognitive costs.

We have so far assumed that there are ‘‘resources’’ available for solving

voluntary reasoning problems and that these resources can be moved

around easily from one reasoning strategy to another. (In economic parlance,

we’re assuming these resources are ‘‘fungible.’’) But that is not so.

There are a number of quite different kinds of resources required for

solving voluntary reasoning problems—time, attention, and short-term

and long-term cognitive capacity. Different reasoning strategies will place

more pressure on some of these resources than others. And some of these

resources handle the strain better than others. One reasoning strategy might

be easy to use because the effort is spread out across various capacities that

work together easily. Another strategy might be just as easy to use (measured,

say, in terms of how many basic operations it employs), but it

might be very hard to implement because its demands overwhelm a single

capacity (e.g., short term memory) or are spread out across capacities that

interfere with each other’s effective operation. It is doubtful that there is

any way to sensibly reduce these various demands on our various capacities

to a single notion of cognitive cost that can be represented by units

along a single dimension.

As was the case with cognitive benefits, the fact that we can’t accurately

assign units of epistemic cost to a reasoning strategy is not a particularly

serious problem for our cost-benefit approach to epistemology.

The notion of epistemic cost does map onto a notion we are able to use to

good effect—the notion of some strategies being harder or easier to use

than others. Indeed, given that people are less likely to use a reasoning

strategy as it gets more complicated (Nisbett 1993), there is good reason to

The Costs and Benefits of Excellent Judgment 87

stick to as simple a strategy as possible. We can measure the time it takes

people to implement the strategy; and this tracks reasonably well (in most

cases) the real costs of reasoning. So this measurable quantity can serve as

an imperfect stand-in for the notion of cognitive costs.