And, for at least a century11, it has been recognized that even the highest-known biological function, human thought, involves random generation of many alternatives and is only shaped up into something of quality by a series of selections. Like the elegant eyes and ears produced by biological randomness, the Darwin Machine's final product (whether sentence or scenario, algorithm or allegory) no longer appears random because of many millisecond-long generations of selection shaping up alternative sequences off-line. (Calvin, 1987, p. 34).
The early comparative biologists and evolutionists assembled impressive evidence of the adaptive fit between organismic structure and environmental possibilities. To explain such fit, three principal alternatives were available. The first involved the detailed a priori planning of a prescient deity. The second involved appropriate or corrective structural modifications based on experience with the environment in question. But this model also involves prescience, however modest and distributed, in that the organism somehow foresees which modifications will fit better. Where the Lamarckian notion of inheritable habituation could be applied, some plausibility might be gained. But most instances of adaptive fit could not thus be explained. The third model was the Darwinian theory of natural selection. For this model, unlike the second, the modifications or variations are blind, are random, are individually nonappropriate, are not of the order of corrections. But by chance there do occur those which provide better fit, and these survive and are duplicated. While Darwinian theory of evolution has undergone considerable elaboration and modification, and while there has been disagreement as to the mechanism and magnitude of the variations involved, his basic model of natural selection is uniformly accepted today, and stands as one of the great conceptual achievements of the 19th century. In its abstract or formal aspects, it is a model which may be applied to other adaptive processes, or other apparently teleological series of events in which modifications seem guided by outcome. (Campbell, 1956, p. 330)
In the course of evolution, there have been tremendous gains in adaptive adequacy, in stored templates modeling the useful stabilities of the environment, in memory and innate wisdom. Still more dramatic have been the great gains in mechanisms for knowing, in visual perception, learning, imitation, language and science. At no stage has there been any transfusion of knowledge from the outside, nor of mechanisms of knowing, nor of fundamental certainties. (Campbell, 1974b, p. 413)
A blind-variation-and-selective-retention process is fundamental to all inductive achievements, to all genuine increases in knowledge, to all increases in the fit of system to environment. (Campbell, 1974b, p. 421)
In going beyond what is already known, one cannot but go blindly. If one can go wisely, this indicates already achieved wisdom of some general sort. (Campbell, 1974b, p. 422)
. . . increasing knowledge or adaptation of necessity involves exploring the unknown, going beyond existing knowledge and adaptive recipes. This of necessity involves unknowing, non-preadapted fumbling in the dark. (Campbell, 1974c, p. 147)
Rather than foresighted variation, hindsighted selection is the secret of rational innovation. (Campbell, 1977, p. 506)
It turns out that all of the organismic processes we would call learning involve vicarious selectors rather than a direct encounter with reality itself. (Campbell, 1987b, p. 186)
For some scholars, the success of science is both obvious and not at all puzzling. They have no need for naturalistic-epistemological theories as to how this success might be possible. For other scholars (especially the social constructivist, ontologically relativist, strong programme sociologists and historians of science, plus some philosophers) the fact that there is no apodictic proof of improved competence of reference in temporal sequences of conceptual change in science also removes any puzzle. Since there is no completely demonstrable fit between scientific beliefs and any supposed independent referents of these beliefs, there is no puzzle needing explanation. These two groups do not need selection theory. (Campbell, 1988b, p. 172)
Being a purposeful problem solver, or a sub-cultural tradition of purposeful problem solvers, does not make one clairvoyant or prescient. Therefore, a specific application of general selection theory is needed by those puzzled by success of intentional problem solving, individual or group, synchroncally or with historical continuity as in science. (Campbell, 1988b, p. 173)
Intelligent variations require an explanation for how these variations or hypotheses came to be wise-in-advance. That most hypotheses are wise, I have no doubt. As such, they reflect already achieved knowledge or, at very least, wise restrictions on the search space. Such wisdom does not, however, explain further advances in knowledge. That hypotheses, even if not wise, are far from random, I agree. But wise or stupid, restraints on the search space do not explain novel solutions. (Campbell, 1990b, p. 9)
Let us today limit our use of selection-theory to what we contingently judge to be impressive examples of fit. For these instances, let us use selection theory rather than Paley's or Descartes' God as an explanation. (It is important to note that these puzzles of fit will be based, at best, on contemporary scientific consensus, not ontologically entailing proof. The puzzles, as well as the selection-theory explanations being offered for them, will be "incomplete inductions.") (Campbell, 1990b, p. 12)
. . . the basic mode of operation is one of hypothesis testing. The input pattern is passed to the upper layer, which attempts to recognize it. The upper layer makes a guess about the category this bottom-up pattern belongs in and sends it, in the guise of the top-down pattern, to the lower layer. The result is then compared to the original pattern; if the guess is correct . . . the bottom-up trial pattern and the top-down guess mutually reinforce each other and all is well. If the guess is incorrect . . . the upper layer will make another guess. . . . thus, the upper layer forms a "hypothesis" of the correct category for each input pattern; this hypothesis is then tested by sending it back down to the lower layer to see if a correct match has been made. A good match results in a validated hypothesis; a poor match results in a new hypothesis. (Caudill, & Butler 1990, p. 208)
The 10,000 or so synapses per cortical neuron are not established immediately. On the contrary, they proliferate in successive waves from birth to puberty in man. . . . One has the impression that the system becomes more and more ordered as it receives "instructions" from the environment. If the theory proposed here is correct, spontaneous or evoked activity is effective only if neurons and their connections already exist before interaction with the outside world takes place. Epigenetic selection acts on preformed synaptic substrates. To learn is to stabilize preestablished synaptic combinations, and to eliminate the surplus. (Changeux, 1985, p. 248)
According to this scheme, culture makes its impression progressively. The 10,000 or so synapses per cortical neuron are not established immediately. On the contrary, they proliferate in successive waves from birth to puberty in man. With each wave, there is transient redundancy and selective stabilization. This causes a series of critical periods when activity exercises its regulatory effect. (Changeux, 1985, p. 248)
It is nevertheless worth noting that in the history of ideas "instructive" hypotheses have most often preceded selective hypotheses. When Jean-Baptiste Lamarck tried to found his theory of "descendence" on a plausible biological mechanism, he proposed the "heredity of acquired characteristics," a tenet that advances in genetics would eventually destroy. One had to wait almost half a century before the idea of selection was proposed by Charles Darwin and Alfred Wallace and validated the principle, if not all the details of its application. In the same way the first theories about the production of antibodies were originally based on instructive models before selective mechanisms replaced them. It could conceivably be the same for theories of learning. To understand the reasons for this temporal succession, we must obviously examine the functioning of the scientist's brain. An instructive concept consists of only one step. It is the simplest possible approach. Moreover, whether we like it or not, it contains an "egocentric" component. "Nature directs forms" much as the sculptor models clay into a statue. . . . The concept of selection, on the other hand, implies further reflection. It involves two steps, and it satisfies the quest for a material mechanism totally devoid of "intentional" aspects. It is natural that this more complicated procedure, more difficult to execute, should have systematically appeared in second place throughout the history of scientific thought. (Changeux, 1985, pp. 279-81)
There is a place for functional explanation, but it is on the level of evolution. It is possible that a heart develops in the course of evolution in order to satisfy a certain function. . . . But this is a point that is useful to keep in mind: functional explanation does not relate to the way organs develop in the individual. (Chomsky, 1977, p. 86)
I don't think that the notion of selection from preexisting materials is rich enough to provide an analysis for the large-scale interactions that are loosely called "learning," but it may be a step along the way. (Chomsky, 1980, p. 137)
In particular, neuroscience holds out the best hope for understanding the individual evolutionary process we call learnin, since the elements of variation, and the mechanisms of selection, whatever they are, are there under the skull, awaiting our exploration. A truly informed story of how the human cognitive system hooks up to the world must await their discovery and examination. (Churchland & Churchland, 1990, p. 310)
When times get tough, true novelty is needed--novelty whose important features cannot be preplanned--and for this we must rely on chance. Chance is the only source of true novelty. (Crick, 1981, p. 58)
Where previous providential and instructionist theories of adaptive change and knowledge growth have been found to be inadequate, selection theory provides a truly naturalistic and nonmiraculous account of puzzles of fit, whether this fit occurred with or without the assistance of humankind. The evolution of theories in many different disciplines from providential through instructionist to selectionist is a provocative suggestion of the superiority of selectionism. And this recent movement in so many different fields of inquiry constitutes what may be considered a second Darwinian revolution. (Cziko, 1995, p. 58)
[HOME] [NEXT]. . . a strong case can be made that universal selection theory provides the best explanation for both naturally and artificially produced puzzles of fit. It relies on patient, iterative cycles of blind variation and selection that over the course of time can result in biological adaptations and new species, functional human cultures, technological breakthroughs, and scientific revolutions. And what is perhaps most appealing from the naturalist perspective of modern science, it provides this explanation without miracles--except for the illusory miracle of how such an inherently blind, stupid, wasteful, and sluggish process can be found at the very foundation of life, its marvelous design, and all the subsequent knowledge that life in its human form has generated.(Cziko, 1995, p. 326)