obvious objection. Finally in this process of trial and rejection we more or less accidentally stumble upon an answer to which the objection is not so obvious. The smooth course of trial and rejection is brought to a halt. Our attention is arrested.” [Platt and Baker, 1931]
“Discovery is something a computer (if constructed and programmed well enough) could do, and do as well (even better) than any human who ever lived.” [Jason, 1989]
Insight spans, I suspect, a continuum from conscious to unconscious. Platt and Baker [1931] may be partly right, if much of the filtering of ideas occurs either subconsciously, on the fringe of consciousness, or in such a brief conscious flash that we are barely aware of it. The less absurd ideas require a little more conscious focus before they can be discarded. Yet Poincaré grasps a central point, missed by Platt and Baker, that the successful scientist owes as much to excluding broad regions from trial-and-error evaluation as to the evaluation itself. A better-trained computer is not the solution.
The approach of considering all possible data permutations is hopeless. What is needed is a leap of insight to the crux. As in chess, the best player does not simply examine all permutations methodically; instead the master visualizes patterns and focuses in on a small subset of the possibilities. I think that the pitfall of conditioned thinking offers a useful perspective: rather than systematically scanning a huge number of possible explanations, scientific thoughts get trapped among a few patterns, like a song that you cannot get out of your head. Relaxation and temporary abandonment may work because the problem continues to pop unbidden into the fringe of consciousness, interspersed with seemingly unrelated thoughts, until suddenly the mind sees the problem in the context of a potentially explanatory pattern.
From a neurobiologist’s perspective, the brain has a vast number of schemata-templates [Calvin, 1986]. Each schema is a neural pattern or pathway, formed and reinforced by electrical current flow. Each schemata-template is triggered whenever we see or experience something that seems to fit the pattern. Boyd [1985] describes hypothesis creation as “finding new combinations of previously understood ideas and concepts.” If schemata are reinforced via electrical current flow in the brain, could insight be a sudden flow in parallel of schemata that had never previously flowed simultaneously?
“It is a wondrous thing to have the random facts in one’s head suddenly fall into the slots of an orderly framework. It is like an explosion inside. . . I think that I spend half my time just talking and listening to people from many fields, searching together for how [plate tectonics] might all fit together. And when something does fall into place, there is that mental explosion and the wondrous excitement. I think the human brain must love order.” [marine geologist Tanya Atwater in 1981, during the period in which the new paradigm of plate tectonics was revolutionizing geology; cited by Calvin, 1986]
Even if the idea of insight as schema generation is correct, its practical usefulness may be small. It does, however, reveal a potential problem: those insights are favored that are similar to ideas and concepts that are already established. Breakthroughs to a radically new perspective are not fostered. Contrast these incremental advances with the following:
“In each of the 1905 papers, Einstein has totally transcended the Machian view that scientific theory is simply the ‘economical description of observed facts.’ None of these theories, strictly speaking, begins with ‘observed facts’. Rather, the theory tells us what we should expect to observe.” [Bernstein, 1982]
