Arthur R. Jensen (2000) The Heritability of g Proves Both its Biological Relevance. Psycoloquy: 11(085) Intelligence g Factor (46)

Volume: 11 (next, prev) Issue: 085 (next, prev) Article: 46 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 11(085): The Heritability of g Proves Both its Biological Relevance

Reply to Bub on Jensen on Intelligence-g-Factor

Arthur R. Jensen
Educational Psychology
School of Education
University of California
Berkeley, CA


Bub (2000) takes a wrong turn in his negative critique of my approach to exploring the nature and causal basis of psychometric g (Jensen, 1998, 2000). The fact that g is highly heritable proves that it has a physical basis that reflects more than learned knowledge and skills. Individual differences in heritable behavioral traits necessarily have some physical basis. Because heritable individual differences depend on some properties of the brain, the fundamental research problem with respect to g is to discover the inherited physical properties of the brain that can explain the g phenomenon. Progress toward this goal calls for the procedures of normal science: formulating models or hypotheses and testing them empirically. The first steps already taken in this process have led to the discovery of a number of behavioral correlates of g, such as reaction time (RT) and inspection time (IT) -- for which individual variation is related only slightly, if at all, to prior learning -- as well as certain anatomical and physiological brain measurements. Though we are at the beginning of this exploration, a number of promising correlates are already in evidence which suggest hypotheses worthy of further investigation. These are all indicated in THE G FACTOR (Jensen, 1998) and in many other publications cited therein. They surely amount to considerably more than Bub's characterization as mere "surmise".


behavior genetics, cognitive modelling, evoked potentials, evolutionary psychology, factor analysis, g factor, heritability, individual differences, intelligence, IQ, neurometrics, psychometrics, psychophyiology, skills, Spearman, statistics
1. I disagree in the first place with Bub's claim that the g factor "urgently requires a plausible link to underlying neurophysiological principles if it is not to remain forever a statistical concept" (Bub, 2000, Abstract). Aside from the fact that neurophysiological correlates of g have already been empirically discovered and are reviewed in The g Factor (Jensen, 1998, 1999), g does not depend on these findings or on any other demonstrations of a physical substrate to be considered more than just a "statistical concept". The G of physics, gravitation, does not require a physical explanation to be a valid and important construct that applies to many seemingly disparate phenomena from the proverbial apple falling on Newton's head to the relationship between the moon and the tides and the motions of the planets. The involvement of g in a great many disparate behavioral and social phenomena establishes it as an important construct in its own right, quite aside from a demonstration of its biological basis (Gottfredson, 1997).

2. The now well established fact that the g factor is the most heritable variance component of the various abilities assessed in IQ tests is a sufficient indication that some substantial part of g (the distillate or common factor in all the diverse abilities measured by IQ and other cognitive tests) resides in certain inborn physical aspects of the brain, not just in certain knowledge or skills acquired in individuals' environments. So we should search for an understanding of g (i.e., correlated individual differences in a host of diverse cognitive abilities) in mainly two spheres: (1) Cognitive tasks that either minimize dependence on prior learned knowledge and skills or minimize individual differences in any specific knowledge/skill requirements, reflecting mainly speed (RT or IT) and consistency, measured as trial-to-trial (intra-individual) variability, and (2) measurements of certain physical properties of the brain, such as size, features of the average evoked potential, brain glucose metabolic rate, nerve conduction velocity in a brain tract, and brain intracellular pH level, all of which have shown significant correlations with g or highly-g-loaded tests. Should all this be ignored, as Bub seems to recommend, just because it has not yet been developed to the stage of a theoretically coherent explanation of g that has been proved beyond any doubt? With such a philosophy, how would any science ever advance?

3. Elementary cognitive tasks (ECTs) are not my invention, but have been widely used in experimental cognitive psychology, with RT as the usual dependent variable. No one has claimed that the RT measurements derived from these ECTs reflect some inert property of the brain. They reflect neural activation, alertness, attentional resources, and the like, and their individual differences appear to reflect the same g factor involved in complex intellectual tasks such as a vocabulary test, number series completion, sentence completion, and Raven's Progressive Matrices. Cognitive tasks fall on a continuum of complexity in cognitive demands or operations; many ECTs are specially devised to fall at the lower end of this continuum. Probably the largest part of their individual variance is attributable to properties of the brain. These consist of (i) a constant factor that reflects both genetically inherited and constitutional properties of the brain and (ii) a variable factor attributable to momentary variations in physiological state at the time of taking the test.

4. The fact that RT measures (both the individual's median RT over n trials and the standard deviation of the individual's RTs over n trials) are g loaded and significantly correlated with scores on untimed complex knowledge-based cognitive tests is of great scientific interest. One and the same g factor can be extracted from a battery of ECTs (with RT as the dependent variable) or from a battery of diverse complex tests administered without time limits; this shows that g does "transcend" specific cognitive abilities. And the fact that this g is correlated with a number of distinct, independently measured brain variables does means that g does reflect something very general about the nervous system. This is already indicated by the substantial positive correlation (averaging about +.40) between measures of g and measures of brain size.

5. Libet's (1985) well established finding that some finite amount of time must elapse between an external stimulus and an individual's conscious recognition of that stimulus is not, as Bub claims, "outlandish" as applied to the measurement of RT in ECT paradigms that show RT-IQ correlations. It was Libet himself who pointed out to me the relevance to my own RT studies of his finding that conscious recognition of a peripheral stimulus takes, on average, about 500 milliseconds. Simple RT is between 200 and 300 msec. for college students. The nervous system does not act with infinite speed, as was once thought. Considerable time is required for the sensory transduction of the stimulus and the afferent neural conduction of the impulse to arrive at the relevant area in the brain for it to reach conscious awareness.

6. This preconscious time lapse helps to explain subjects' difficulty in 'faking' slower RTs than those that fall within their own normal distribution of RT or even within the broader distribution of RT for all persons within the normal range of IQ. When subjects are instructed to 'fake' slower RTs than they typically produce when doing their best, they far overshoot the normal range of RT; these extreme outliers are easily detectable as 'faked'. The subject's conscious intention not to respond as quickly as they can results in an atypically long RT, presumably because faking requires conscious awareness of the whole time course of the events between stimulus presentation and the subject's response. When subjects are doing their best on every trial, there is some normal amount of intra-individual variation in RT, measured as the SD of RT over a given number of trials. It is interesting that the magnitude of this normal variation in RT is even more highly correlated with IQ and g than is the median RT over the same number of trials (Jensen, 1992).

7. It is also noteworthy that virtually all of the subjects who have been tested in my Chronometric Laboratory in Berkeley believe that their RT is much shorter than their MT (movement time), when RT is measured as the interval between the onset of the reaction stimulus (a light going 'on') and subjects' removing their finger from a "home" button, and MT is the interval between leaving the "home" button and reaching six inches to touch the light, turning it 'off'. In fact, for all subjects (except the severely mentally retarded), the RT is about twice as long as the MT. Within intervals of less than 500 msec, subjects typically have very poor subjective judgement about how much time has elapsed between events.


Bub, D. N. (2000). Reflections on g and the brain. PSYCOLOQUY 11(043) psyc.00.11.043.intelligence-g-factor.42.bub

Gottfredson, L. S. (Ed.) (1997). Intelligence and social policy (special issue). Intelligence, 24 (1).

Jensen, A. R. (1992). The importance of intraindividual variability in reaction time. Personality and Individual Differences, 13, 869-882.

Jensen, A.R. (1998). The g factor: The science of mental ability. Westport, CT: Praeger.

Jensen, A.R. (1999). Precis of: "The g Factor: The Science of Mental Ability" PSYCOLOQUY 10(023). psyc.99.10.023.intelligence-g-factor.1.jensen

Libet, B (1985). Unconscious cerebral initiative and the role of conscious will in voluntary action. Behavioral and Brain Sciences, 8, 529-566.

Volume: 11 (next, prev) Issue: 085 (next, prev) Article: 46 (next prev first) Alternate versions: ASCII Summary