J.F. AAS Sept. 10, 96

Bounded Rationality

Bounded rationality recognizes that it is impossible to comprehend and analyze all of the potentially relevant information in making choices. The only possible way of coping with the complexity of the world is to develop techniques, habits and standard operating procedures (SOP) to facilitate decision making. Widely shared SOP's are institutions.

Habit--Routine responses and behaviors based on reinforcement. e.g. brushing your teeth in the morn.

Technique--Ways to deal with generalized situations. eg. check Consumer Reports when buying an appliance.

SOP--group and organizational rules for decisions. eg. our firm reorders when inventories reach one month of recent sales.

The point of bounded rationality is not that people might decide differently if they had more or different information, or differently with different items in the utility function. Rather it is that they can't process all the information even if they had it. Thus if we are to predict people's actions, it will not be enough to know the amount or quality of available information. Rather we must know the cognitive process of selecting among information or choosing rules of thumb or slogans. Selective perception is part cognitive (learned, conscious or not) and part geneticly programed (e.g. we can't hear certain frequencies).

Institutionalists do argue for additional items in the utility function, but this is not the point of bounded rationality. And they grant that all theories are simplified--in fact people must themselves simplify to account for bounded rationality. People can't handle a lot of items and the items considered in a particular decision may differ from the items in the next situation which may look similar to the analyst (so that we observe non-transitivity, apparently inconsistent choices.) This is related to the subjectivity argument of Austrian economics.

If the problem were simply to discover which items were in people's utility functions, our research would just ask what items are valued and what are the relative valuation weights on each item. But, if the problem is bounded rationality, we need some other information, i.e., what cues the consideration of item A in one case and item B in another. What cues whether we calculate at all but rather go directly to action (rule of thumb perhaps) that is seen to fit a perceived environment.

The amount of information is not simply a constraint. If the only problem is limited information, then the analyst can objectively measure the information the decision makers actually have and thus predict choice. But, if the decision makers necessarily select from available information, then analysts must inquire how that selection is made.

Some try to save rationality theory by postulating a preference for cooperation or voting or altruism. So a person cooperates in a prisoner's dilemma game because they have a preference for cooperation. This is convenient, but worthless if persons cooperate in some games and not others. We can't tell ahead of time whether the cooperation will happen. If we are going to predict, we will need to know what cues the cooperative item in the utility function to be operative. And that may not be relevant if the cues go directly to behavior in some contexts rather than being calculated and traded off in some total aggregation process.

What difference does theory make? Lots, because it suggests where to look. For example, many of you are interested in developing natural resource management plans for public lands. We would like to have a plan that supplies what the public wants. After you get around the problem that people are not homogeneous and have conflicting preferences, what are the implications of Simon?

So, many researchers conduct surveys asking such questions as,

Do you want to preserve more wetlands?

Should wetland owners be compensated for building restrictions?

Or in the private sector, Would you buy Charmin Toilet Paper for 50 cents?

These questions presume the existence of stable preferences and that preferences and behaviors are independent of frames (cues). But we know that frames make a difference. We get a different choice and ranking of alternatives if we say a health program results in 80 % survival or 20 % death. Kenrick Pierre tested the effect of frames on the first two questions above. He varied the frame with two different introductions to the questionnaire. One reaffirms the well known notion that owners should be compensated when the value of their property is reduced by regulation. The second points out that the public can be an owner too and entitled to compensation. Or one private owner may be harmed by another private owner and both use the compensation argument to different ends. If people have stable prior preferences, then variation in frame should not affect their ranking of alternatives.

If your theory assumes stability, the researcher will never bother to look to see if frames make a difference.


A perceptive student wrote:

"I was nine and a half years old when my best friend, Kelly, informed me that there was no Santa Claus. I remember distinctly that I was not so upset by Santa not being "real," but by the realization that Kelly had figured the situation out before myself. I thought about the assumptions I had made over the years and the questions I asked myself (How does Santa make it around the world in one night? Why do some gifts have price tags: How do all the toys fit in the sled?) and wondered what caused me to cling to the particular notion of Santa while others turned to more plausible theories for receiving gifts at Christmas?"

Answer: Selective perception (bounded rationality necessarily selecting from available information). We might discover that both children had access to the same information about the size of chimneys and bulky presents. Both necessarily selected from available information, but for Kelly it did not lead to this particular error. We know that cultural learning (informal institutions) influences what we see and do not see. By the way, if we did see everything, we would still be calculating whether to get out of bed in the morning, even on Christmas!



AAS\810\bond-rat.htm