The Limited Plasticity Of Human Intelligence

by Arthur R. Jensen

Originally published in The Eugenics Bulletin, Fall 1982

As societies become increasingly technological, the demand for superior intelligence begins to exceed the supply, and the demand for sheer physical labor begins to decline Increased leisure, early retirement, and a lengthened life-span all raise the premium on intelligence for the social and moral well-being of society. With the eradication of malnutrition and infectious childhood diseases, and as universal public education and the amenities of our technological civilization become more widespread, the improvement of human intelligence, if it is to come about at all, will depend increasingly upon eugenic means.

We are now gradually emerging from a period of over-optimism regarding the supposed plasticity of intelligence, and the hope of appreciably raising the IQ of those with below-average intelligence through strictly psychological and educational methods. This hope is probably as old as humanity itself. Widespread faith in its practical implementation originated in the 1920's with the radical behaviorism espoused by John B. Watson. Watson's behavioristic conception of intelligence has pervaded psychology even to this day, although it has lost favor among the new generation of researchers in experimental cognitive psychology and psychometrics.

In the behavioristic view, intelligence became equated with learning. Man's "original nature", psychologically, consisted only of an undifferentiated, general capability for learning. All that developed throughout the course of evolution was an ever-increasing plasticity of the brain for being shaped by the physical and cultural environment. Human mental capabilities were viewed as wholly a product of learning. The wide range of individual differences (except those resulting from some form of brain damage) was attributed to differences in opportunities for learning, or to differences in the content of learning. It was believed that these differences became socially salient merely due to the fact that some forms of knowledge and skills are more highly valued than others in a particular society. Accordingly, what Western industrial societies recognize as "intelligence" and measure by means of standard IQ tests was viewed only as a specialized collection of particular bits of acquired knowledge and skills which happen to be valued within a specific cultural context.

Given the view of intelligence as essentially a product of learning, it was reasonable to expect that intelligence itself could be taught much the same way one teaches reading or arithmetic. It led to the optimistic expectation that the intelligence of children in the bottom half of the IQ distribution could be dramatically raised by providing them with early learning opportunities like those enjoyed by children in the top half of the distribution. The well-established correlation between children's IQs and their parents' socioeconomic status (SES) was accorded an erroneous causal significance: Low SES children were believed to have lower IQ's and to achieve less well in school because they lacked the cultural advantages and learning opportunities enjoyed by children from higher SES backgrounds.

Over the past three decades, hundreds of experiments, many carried out on a massive scale, have sought to prove that intelligence can be substantially raised. In a few studies, subjects were given intensive training over a period of several years. No other field of psychological or educational research has commanded such vast funds nor marshalled such concerted efforts on such a grand scale. The truly remarkable finding is not the few points gain in IQ or scholastic achievement occasionally reported, but the fact that gains are so seldom found, and, when they are found, that they are so very small. The theoretical implication of this finding is that the behaviorist view of intelligence as synonymous with learning (or the products of learning) is seriously in error. Predictions based on this view have repeatedly failed to materialize under the prescribed conditions.

When gains in test performance have occurred as a result of educational treatments, they have displayed one or more of the following characteristics: (1) they have been small, rarely more than five or ten IQ points; (2) they have been of short duration, fading out within a year or so after the training has been completed; (3) they have been restricted to tasks or tests which closely resemble the actual training procedures themselves, and have failed to generalize to a broader range of mental tests.

Although I have scoured the research literature, I have yet to find a bona fide empirical demonstration that any psychological or educational techniques have succeeded in significantly raising children intelligence. Scores on one particular test or another, or achievement in particular scholastic subjects, may have been raised, usually only temporarily. But these gains are not reflected across a wide variety of tests or school subjects, as would be the case if it were g itself (the general intelligence factor) that had been improved. This conclusion is reinforced by evidence reported in a recent book which summarizes much of the best research and thinking in this field (Detterman and Sternberg, 1982).

The limited plasticity of intelligence can be more easily understood in terms of the newly ascending view of intelligence as comprising a small number of elementary information-processing capabilities which are closely dependent upon properties of the central nervous system. Learning itself is only one of many manifestations of these elemental processes involving stimulus encoding, discrimination, comparison, short-term memory capacity, speed of transfer of information from short- and long-term memory, and the like. The fact that ordinary IQ tests measure something more fundamental than acquired knowledge is demonstrated by the correlation of IQ with performance on laboratory tacks, such as reaction time, which have have virtually no intellectual content whatsoever, but which directly measure elemental information-processing capacities (Jensen, 1980, 1982a, 1982b). That these information-processing capabilities are closely linked to brain functions is shown by correlation of both IQ and reaction time measures with brain-wave measurements (termed average evoked potentials) (Hendrickson and Hendrickson, 1980; Jensen, Schafer, and Crinella, 1981).

It is now generally accepted that individual differences in IQ and information-processing capacity are strongly influences by hereditary factors, with genetic variance constituting about 70% of the total population variance in IQ (Jensen, 1981). There is also evidence that the genes for superior intelligence tend to be dominant, which is what would be theoretically expected if intelligence is a fitness character in the Darwinian sense, and if it had been subject to natural selection through the course of human evolution (Jensen, 1983).

The genetic and evolutionary view of human intelligence affords a possible explanation for its quite limited plasticity. If intelligence has evolved as an instrumentality for the survival of Homo Sapiens, it could well be that its biological basis has a built-in stabilizing mechanism, such an that of a gyroscope. Some degree of homeostatic autonomy in the ontogeny of mental ability would safeguard the individual's capacity for coping with the exigencies of survival. Mental development then would not be wholly at the mercy of often-erratic environmental happenstance. A too-plastic malleability would give the organism little protection against the vagaries of its environment. Hence, there may have evolved homeostatic processes to buffer the semi-autonomous ontogeny of human intelligence, protecting it from being pushed too far in one direction or the other, either by adventitiously harmful or by intentionally benevolent environmental forces.

Arthur R. Jensen is Professor of Educational Psychology at the University of California, Berkeley, California 94720. Reprints of any of his articles listed below may be obtained from Dr. Jensen.


Detterman, D.K., and Sternberg, R.J. (Eds.) 1982, How and How Much Can Intelligence be Increased? Norwood, NJ: ABLEX Publishing Corporation

Hendrickson, D.A. and Hendrickson, A.E. 1980, The biological basis of individual differences in intelligence, Personality and Individual Differences, 1: 3-33

Jensen, Arthur R. 1980, Chronometric analysis of intelligence, Journal of Social and Biological Structures, 3: 103-122

Jensen, Arthur R. 1981, Straight Talk About Mental Tests, New York: The Free Press

Jensen, Arthur R. 1982a, The chronometry of intelligence, in R.J. Sternberg (Ed.) Advances in the Psychology of Human Intelligence (vol. 1) Hillsdale, NJ: Erlbau.

Jensen, Arthur R. 1982b Reaction time and psychometric A, in Hans J. Eysenck (Ed., A Model for Intelligence New York: Springer-Verlag

Jensen, Arthur R 1983, The effects of inbreeding on mental ability factors, Personality and Individual Differences, 4: 71-87

Jensen, A.R., Schafer, E.W. and Crinella, F.M. 1981, Reaction time, evoked brain potentials, and psychometric in the severely retarded, Intelligence, 5: 179-197