2. A cost-benefit approach to epistemology

Before tackling the difficulties associated with a cost-benefit approach to

epistemology, it might be valuable to note that cost-benefit considerations

are a familiar, even banal, feature of perceptual and cognitive psychology.

From early psychophysics to contemporary cognitive psychology, psychologists

often explain successes, as well as routine breakdowns, in terms

of specific allocations of attention and memory across perceptual modalities

and cognitive capacities. For example, many of our reflexes are best

understood as cognitive instantiations of a complex cost-benefit analysis.

The ducking reflex occurs whenever you see a rigid object translating

toward your head. Those who don’t duck, to quote Quine, ‘‘have a pathetic

but praiseworthy tendency to die before reproducing their kind’’

(1969, 126). This reflex has accommodated a complex payoff matrix.

When a rigid object is moving toward your head, the potentially serious

consequences of not avoiding it call for a fast and mandatory reflex. So

you duck, even when the trusted friend who threw the object tells you that

it won’t hurt you. Low risk and low benefit outcomes do not produce

comparably reliable behavior. We are skeptical that this kind of adaptationist

story will work for all of our cognitive mechanisms. But adaptationist

explanations are quite plausible for reflexes, and such explanations

tend to rely explicitly on cost-benefit considerations (e.g., Parker 1974;

Maynard Smith 1978).

Processes of greater cognitive depth also reveal tradeoffs between costs

and benefits. Consider memory. Our recall performance suffers terribly when resources get allocated to competing tasks. In the classic experiments

on divided attention, people were presented with two spoken messages.

The participants could not identify the contents of both messages, recalling

one well and only the most basic characteristics of the other (e.g., that

it changed from speech to tone, etc.). Consider a timely example. Talking

on a cellphone (either handheld or hands-free) while driving resulted in

twice as many failures to detect a simulated traffic signal and slower reactions

to those signals when they were detected (Strayer and Johnston,

2001). These effects of divided attention exert a powerful influence on our

daily activities, which often involve doing two or more things at once. In

short, strategies involving divided attention bring predictable costs. (This

point is embodied in the wise advice of a parent: ‘‘Don’t smooch while

driving—you’ll do both badly.’’) We can drive safely or talk on the cellphone.

We can read or listen to someone addressing us. The choice to do

one, the other, or both should be based on an analysis of the costs and

benefits of the various distributions of cognitive resources (Payne, Bettman,

and Johnson 1993).

Now it’s time to pay the piper. If applied epistemology involves a kind

of cost-benefit analysis, then we need to clearly identify the costs and

benefits of reasoning. But this is hard to do for three reasons. First, it is not

clear how to reduce the myriad benefits of reasoning to countable units.

Second, it is not clear how to reduce all the various costs of reasoning to

countable units. And third, it’s doubtful that we can reasonably interpret a

smooth cost-benefit curve for a reasoning strategy. Let’s consider these

issues in turn.

Before tackling the difficulties associated with a cost-benefit approach to

epistemology, it might be valuable to note that cost-benefit considerations

are a familiar, even banal, feature of perceptual and cognitive psychology.

From early psychophysics to contemporary cognitive psychology, psychologists

often explain successes, as well as routine breakdowns, in terms

of specific allocations of attention and memory across perceptual modalities

and cognitive capacities. For example, many of our reflexes are best

understood as cognitive instantiations of a complex cost-benefit analysis.

The ducking reflex occurs whenever you see a rigid object translating

toward your head. Those who don’t duck, to quote Quine, ‘‘have a pathetic

but praiseworthy tendency to die before reproducing their kind’’

(1969, 126). This reflex has accommodated a complex payoff matrix.

When a rigid object is moving toward your head, the potentially serious

consequences of not avoiding it call for a fast and mandatory reflex. So

you duck, even when the trusted friend who threw the object tells you that

it won’t hurt you. Low risk and low benefit outcomes do not produce

comparably reliable behavior. We are skeptical that this kind of adaptationist

story will work for all of our cognitive mechanisms. But adaptationist

explanations are quite plausible for reflexes, and such explanations

tend to rely explicitly on cost-benefit considerations (e.g., Parker 1974;

Maynard Smith 1978).

Processes of greater cognitive depth also reveal tradeoffs between costs

and benefits. Consider memory. Our recall performance suffers terribly when resources get allocated to competing tasks. In the classic experiments

on divided attention, people were presented with two spoken messages.

The participants could not identify the contents of both messages, recalling

one well and only the most basic characteristics of the other (e.g., that

it changed from speech to tone, etc.). Consider a timely example. Talking

on a cellphone (either handheld or hands-free) while driving resulted in

twice as many failures to detect a simulated traffic signal and slower reactions

to those signals when they were detected (Strayer and Johnston,

2001). These effects of divided attention exert a powerful influence on our

daily activities, which often involve doing two or more things at once. In

short, strategies involving divided attention bring predictable costs. (This

point is embodied in the wise advice of a parent: ‘‘Don’t smooch while

driving—you’ll do both badly.’’) We can drive safely or talk on the cellphone.

We can read or listen to someone addressing us. The choice to do

one, the other, or both should be based on an analysis of the costs and

benefits of the various distributions of cognitive resources (Payne, Bettman,

and Johnson 1993).

Now it’s time to pay the piper. If applied epistemology involves a kind

of cost-benefit analysis, then we need to clearly identify the costs and

benefits of reasoning. But this is hard to do for three reasons. First, it is not

clear how to reduce the myriad benefits of reasoning to countable units.

Second, it is not clear how to reduce all the various costs of reasoning to

countable units. And third, it’s doubtful that we can reasonably interpret a

smooth cost-benefit curve for a reasoning strategy. Let’s consider these

issues in turn.