[“Adaptive Thinking: Rationality in the Real World”, Gerd Gigerenzer, Oxford University Press, 2000]
In part III (chapters 7-8), Professor Gigerenzer tells us: “To understand the power of human intelligence, one needs to analyze the match between cognitive strategies and the structure of environments. Together they are like a pair of scissors, each blade of little use on its own but effective in concert with the other.” [page 125]
A little later, he states the question he’s dealing with in Part III:
Models of bounded rationality address the following question: How do people make decisions in the real world, where time is short, knowledge lacking, and other resources limited?”
This is a good question to ask in an era where we’ve misunderstood rationality, thinking a man to be an autonomous agent whose major relationships to his environments and even his fellow-men are mediated through channels of information. The information taken in by that autonomous agent is then processed and used in logical processes which were mostly developed in recent centuries. Artificial intelligence researchers learned the error in this view of real-world intelligence from a basic robotic experiment conducted decades ago, I believe at Stanford. When they tried to ‘teach’ a robot to walk across a room by finding ways around each obstacle through brute-force logic, they found the robot froze, not even able to ‘prove’ it possible to move around the chair immediately in front of it. More importantly, we human beings tend to freeze or react with emotionally induced blindness when we’re constantly presented with moral and social and political problems which we’re supposed to be solving from scratch as if we were moral philosophers developing a new approach to human moral life. This is the reason that all of us should be interested in this sorts of ‘bounded rationality’. We’re not autonomous agents, moral or rational creatures in a free-standing way. Our morality is a result of our interaction with our fellow-creatures, our physical environments, and our Creator.
In Chapter 7, Probabilistic Mental Models, Gigerenzer speaks of situations where test-takers are overconfident about their answers to individual questions but have good estimates of the frequency of their correct answers over the entire set of questions. Further analysis indicated that the problem is due to the deliberate selection of questions to correspond to a desired ratio of easy and hard questions rather than using representative sampling from the possible questions. The test-takers were giving their confidence in the correctness of each question as if they were facing a representative sample of questions rather than a set selected to have a higher number of hard questions. By the time they were through the entire test, they gave a better estimate of their percentage of correct answers, presumably because they’d had a chance to evaluate the mixture of hard and easy questions on the test. As Gigerenzer notes, this implies that a selection of more easy questions than would be expected would result in an underconfidence problem. Other possibilities can arise but this should give a good idea of the type of misunderstanding that can arise when there is a mismatch between an experimental test and the expectations of the test-takers. We think in interaction with our environment and artificial tests which ‘lie’ in some sense about that environment can confuse us in one way or another.
At the same time, we should realize the environment can lie as well. It can generate possible but unlikely situations. It can change dramatically in fundamental ways as it does when a cosmopolitan society develops from a rural society. This raises two possibilities in my mind:
- We human beings are born already being particular creatures and members of a species well-developed to a specific environment, most likely that of the Neolithic era (approximately 100,000BC to 10,000BC).
- We human beings are born as a host of potential beings and members of a species which carries in its flesh a lot of possible responses to a lot of possible environments.
In fact, the truth lies in some combination of the two possibilities. We’re flexible to the extent that the human mind is the type of entity capable — in principle — of encapsulating the universe, but no actual human mind can do that though most of us can ‘expand’ our minds somewhat to imagine possibilities not found explicitly in our environments. There is a problem with not only smallness of mind but also limited flexibility and limited ability to overcome prejudices, some warranted in local circumstances. I intend only to plant seeds and won’t further pursue this more general issue for now.
In Chapter 8, Reasoning the Fast and Frugal Way, Gigerenzer tells us of situations where such effective knowledge is contained in the environment that too much explicit knowledge can actually drop the scores of test-takers. I’ll give one simple example to show how this happens. Suppose German students are given a simple test where each question gives a pair of American cities drawn from, say, the largest 100 such cities and the task is to say which has the larger population. Those students are not likely to know enough about American cities to be able to rank those cities in strict order of population size. It turns out that there is something called the ‘recognition heuristic’ which tends to increase the number of correct answers.
If you recognize the name of one city and not the other, guess that the recognized city is larger in population.
Suppose you have a few different ‘fast and frugal’ ways of making possible decisions. You can use them to make the final decision in ways far simpler than the techniques of ‘classical reasoning’ which came out of Enlightenment thought. For example, you can simply make the decision on the basis of the first ‘strong’ answer you get. Simulations indicate that such ‘fast and frugal’ methods can equal or exceed the performance of the fancier methods which are taught at Harvard and MIT and are used to manage General Motors and the Defense Department of the United States.
Gigerenzer’s way of viewing human thinking processes is consistent with the Thomistic, empirically-based attitudes I’ve been advocating. It tells us we can be frugal in this encapsulation by letting the universe itself be our partner in thinking and remembering.