Learning is an example of an activity carried out by a complex adaptive system (to use the terminology employed by some of us at the Santa Fe Institute). Other examples are biological evolution; the operation of the immune system in various kinds of vertebrate animals, including human beings; the functioning of the human scientific enterprise; and the generation by computers of new strategies for playing games. What characterizes all of these processes?
In each case, the complex adaptive system receives a stream of data about itself and about the rest of the world. Those data are necessarily incomplete and often approximate. The system then identifies perceived regularities in the data and compresses those regularities into a concise "schema" such as a set of ideas in the learning process, a theory in science , or a genotype in biological evolution. In general, a schema can be modified or another schema substituted for it, so that there is competition among schemata.
A given schema, filled out with some more data, may be used to describe the world or to predict its future course, or else to prescribe behavior for the complex adaptive system. Such incursions into the real world have real world consequences. They are subject to selection pressures, to use the expression common in evolutionary biology. A description or prediction may be more successful or less, according to criteria set by the selection pressures. Of those criteria, agreement with observation is only one. For scientific theories it is supposed to be the principal one, accompanied by consistency and generality, but of course human characteristics such as greed and prejudice can supply others. Similarly, the consequences of behavior according to a schema may be favorable or disastrous or somewhere in between, again according to criteria set by the selection pressures.
In any case, real world consequences feed back to influence the competition among schemata. There is a general tendency for schemata that are "selected for" to be promoted or to survive and for those that are "selected against" to be demoted or to disappear. While that does not happen according to some simple mechanical rule, the fate of the various schemata is certainly shaped in a general sense by the selection pressures.
The selection pressures in biological evolution favor survival, sexual selection (where that is relevant), having offspring that survive, and so forth. In science, when it is working properly, they favor, as already indicated, theories that not only possess coherence and generality but agree with observation. What are the selection pressures on human learning and thinking? In natural situations, ideas may be rewarded by acceptance because they are confirmed by observation, as in science. But they can also be rewarded because they make us feel good, or because they form a superficially reasonable pattern, as in many superstitions.
So the process of learning, adaptation, or evolution in a complex adaptive system can turn out maladaptive. We have seen one way that can happen, in the case of superstition and also in the case of the distortions that occasionally occur in the evaluation of scientific theories when selection pressures from human failings predominate over those that are characteristically scientific. In those examples the results of adaptation can end up being labeled maladaptive because the real selection pressures do not all conform to our ideas of what they ought to be.
Perhaps we should reflect deeply on the kinds of selection pressures we are applying to students' ideas in our schools and universities. Are we really encouraging learning and thinking, or are we mainly honing skills in passing examinations that require the regurgitation of material fed to the students in class?
Another common source of maladaptation is a mismatch of time scales. Circumstances may change much more rapidly than the rate at which adaptation takes place. For example, ecological communities have trouble adjusting to rapid drastic changes such as those proded in recent times by human activity or those produced about sixty-five million years ago by the impact of a large meteorite on the coast of the Yucatan peninsula. Human communities and institutions may also adapt too slowly to changing circumstances.
How easily do our great universities adapt? In medieval Europe, books were produced in a scriptorium, where a lector read a manuscript, say a treatise in theology, to a room full of scriptores, who copied it down. At a university, many of the students were too poor to be able to afford books published in that way, and so a professor, say of theology, would read his own book to the students, who acted as their own scriptores, and copied it. Then printing was invented, and that method of publishing became obsolete, but in American universities today we see the same activity carried on, even in scientific subjects. For example, at each institution, some professor is busy giving a complete course of lectures on electromagnetic theory. We may imagine that at ten o'clock in the morning on a certain Monday during the school year, at every college and university in the eastern United States, a lecturer is describing Maxwell's equations to a class. Yet perfectly good textbooks exist that the students could read. For those who prefer to learn by listening and watching, a more recent technical advance has provided the possibility of showing complete sets of videotapes made by some of the finest lecturers in the world. Just as in the case of textbooks, many alternative sets of videotapes can compete in the market for the professors' choice. There is no need for uniformity. Nor would professors become obsolete.
Why are the local professors not reserved for functions that only they can perform, like answering questions when students get stuck, recommending individual programs of reading and study, setting challenging problems and discussing the solutions, and giving occasional lectures as dramatic performances, in which they can present some of their own ideas or points of view and exhibit themselves as role models? Why are there not more opportunities for undergraduates to have experiences like those of apprentices?
Our students are accustomed, all the way up to the start of the doctoral dissertation, to spoon feeding of material in classrooms. Even advanced students ask, "What pages do we have to study for Monday's quiz? Are we responsible for the material in today's lecture?" A sudden transition to a different system would be traumatic. I believe we need to wean students gradually away from their dependence on routine lectures and exams. Eventually, even juniors and seniors in high school might adjust to more adult habits of learning. Of course, we have to bear in mind the social function of classes. The students like to be brought together on a regular basis. But the showing of videotapes, along with occasional inspirational lectures by local teachers, can provide that experience.
Recently, the computer revolution has made it possible for our schools to use well-designed software to assist a great deal of learning, especially of comparatively routine material. Some excellent "teaching machine" programs have been available for many years. More are being created all the time. They allow pupils to engage in discovery, making errors and correcting them, exploring pathways to see where they lead, without the embarrassment of doing all that in public. Moreover, where it is desirable, the computer can track what the pupil is doing right and wrong and offer specific suggestions for improvement, all in private. Yet in many schools adaptation has been slow, with computers still largely restricted to a few uses and even, in some cases, still kept together in special computer rooms when they should be made constantly available to help with learning.
At the Santa Fe Institute, where nearly all our research is transdisciplinary, we spend a great deal of our time making connections among different subjects. Also, when it is appropriate, we try to put together different ideas, different ways of looking at problems. At our great universities and institutes of technology, however, and in our society as a whole, the disciplines are separated by fairly rigid barriers. Our system of measuring excellence is based on the separate disciplines, through the apparatus of departments, degrees, journals, professional societies, and sections of granting agencies. In fact, that system protects us against some real dangers. Incompetents and charlatans are often found crouching in the crevices between disciplines. We are all familiar with the kind of person who is thought by some physicists to be a great mathematician and by some mathematicians to be a great physicist. Yet today many of the most challenging issues involve five or ten subjects or even more. Our institutions would do well to step up the pace of adaptation to those new circumstances. Specialization, even increasing specialization, is necessary and important but it needs to be supplemented by integrative thinking and learning.
Cooperation among different ideas is in many cases of critical importance as well. Sometimes, of course, one idea is right and the others wrong. But often, especially in the life sciences and the social or behavioral sciences, that is not the case. Several points of view may have merit, and a reasonable picture would include a synthesis of all of them. Once, in a meeting at our Institute, five of the participants happened to be from the same department at the same university. At home, their professional conversations tended to be about the relative merits of their different approaches to their subject. But in our building the atmosphere was somewhat different, encouraging cooperation, and they began to work on combining their ideas and constructing a more general framework that could accommodate aspects of all of them.
In academic life and in many other places in our society, however, the selection pressures favor singling out a point of view, developing it for all it is worth, and disparaging other approaches, rather than trying to construct a synthesis. Moreover, making a little discovery on the frontier of a specialty, even if it later turns out to be wrong, can be worth a tenured professorship, while putting together the pieces of a puzzle can go largely unrewarded.
I believe we need to reexamine, in all these areas and in many others, the character of our academic institutions and the incentives they provide for students and for teachers, to see how and whether those institutions can adapt more promptly and effectively to the increasingly rapid pace of change around us.
[When citing this speech, please refer to it as M. Gell-Mann, Han-Feng
SciTech Online, Number 08, April 15, 1996]