Arthur R. Jensen (2000) Artificial Intelligence and g Theory Concern Different Phenomena. Psycoloquy: 11(086) Intelligence g Factor (47)

Volume: 11 (next, prev) Issue: 086 (next, prev) Article: 47 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 11(086): Artificial Intelligence and g Theory Concern Different Phenomena

Reply to Bringsjord on Jensen on Intelligence-g-Factor

Arthur R. Jensen
Educational Psychology
School of Education
University of California
Berkeley, CA 94720-1670


Although I took pains in my book and its Precis (Jensen, 1998, 1999) to distinguish between the meaning of the term 'intelligence' and the meaning of the g factor, Bringsjord's (2000) critique has unfortunately viewed 'intelligence' only in the conceptual form adopted by those who work in the field of artificial intelligence, which is something very different from what my book is about. Bringsjord also confuses 'genius' with 'intelligence', although I clearly distinguished between these concepts. Bringsjord's critique accordingly seems to lose all relevance.


behavior genetics, cognitive modelling, evoked potentials, evolutionary psychology, factor analysis, g factor, heritability, individual differences, intelligence, IQ, neurometrics, psychometrics, psychophyiology, skills, Spearman, statistics
1. Artificial intelligence (AI) attempts to characterize certain generic aspects of what I have termed "intelligence" in abstract logical terms. This can usually be cast in some computer language to test whether the computer program can actually do what it was designed to do (e.g., play chess or solve problems). To avoid confusion between 'intelligence' and the g factor, I have used the word 'intelligence' in an open-ended, generic sense to include various behavioral and cognitive phenomena recognized by psychologists as stimulus apprehension, perception, discrimination, generalization, learning, memory, insight, reasoning, problem solving, and the like. These capacities are the subject matter of both experimental cognitive psychology and artificial intelligence, which can work hand-in-hand in their attempts to model these phenomena, without reference to species differences (i.e., comparative psychology), individual differences (i.e., differential psychology), or neural underpinnings (i.e., cognitive neuropsychology).

2. The capacities for displaying these 'intelligence' phenomena and the neural mechanisms involved differ between species. All biologically normal members of a given species, however, have the same behavioral capacities, although individuals may differ widely in the degree of development or the level of performance attained in these capacities. G-theory addresses the nature of intraspecies variation, or individual differences, in all the behavioral aspects of 'intelligence'. It is based on Spearman's discovery that, at least in humans, individual differences in proficiency in the diverse phenomena that we categorize as 'intelligence' are all positively correlated. G is the highest-order common factor among all these correlations. The major questions arising from g theory concern the psychometric and psychological nature of g, its causal basis in genetics and its associated brain physiology. That is what THE G FACTOR (Jensen, 1998) is all about, as far as the science of this subject had developed by 1998.

3. Bringsjord (2000) suggests that there are definitional quandaries in my work, despite the fact that Chapter 3 ("The Trouble with 'Intelligence'") defines the key terms of my formulation of g explicitly and objectively. It should be clear that I haven't merely substituted the words 'mental ability' for the word 'intelligence'. An ability is an empirically and objectively observable and quantifiable behavioral act that meets some objective standard of performance. An ability is 'mental' when virtually none of the variance in measures of it depends on, or covaries with, individual differences in sheer sensory acuity or motor strength in the general population. To qualify as an ability, the performance as measured must have some specified degree of temporal reliability or repeatability . Also, it must be voluntary. Bringsford seems to believe that the requirement of 'voluntary' presents insurmountable problems. I have made the meaning of this term clear by presenting a long list of criteria that rule out virtually all of the forms of involuntary behavior that would not qualify as an 'ability' (Jensen, 1998, pp. 50-51).

4. Bringsjord unfortunately exaggerates and misrepresents the significance I attach to Raven's Progressive Matrices test. Although the Raven is highly g loaded when factor analyzed with any battery of diverse mental tests, neither it nor any other single test consisting of homogeneous item content is the definitional criterion of g. An individual (or computer program) who could invariably perform 100 percent perfectly on the Raven (or any other tests in the matrices format) but who scored poorly on all other types of tests (a sure indicator of low g) would be considered an 'idiot savant'. The fact that a computer program can be written to solve Raven matrices items doesn't imply that the computer's performance has anything resembling g, i.e., the common factor in a great many diverse kinds of cognitive performance. Certain types of purely computational performance that can be executed by a computer but seem incredible when performed by a human, such as doing enormous numerical calculations, can be performed by rare 'savants' or prodigies who have unexceptional scores on the Raven (or on IQ tests and most other g-loaded tasks), as shown in my experimental study of Shakuntala Devi, probably the world's most remarkable mental calculator (Jensen, 1990). In a variety of numerical calculations, Mrs. Devi's brain, in its speed and accuracy, is practically on a par with a modern mainframe computer. Yet she doesn't operate like a computer program, any more than Kasparov play chess like a computer.

5. Bringsford confuses high 'intelligence' with 'genius', mentioning such famous figures as Shakespeare, Michelangelo, and Einstein as exemplars of exceptional 'intelligence'. But the sine qua non of genius is not characterized on a scale of intelligence, or IQ, or g. A highly exceptional, original, creative, and socially valued achievement is the defining characteristic of genius. As I have explained in my book (Jensen 1998, pp. 577-578) and spelled out in greater detail elsewhere (Jensen, 1996), genius, and even lesser forms of outstanding achievement, depend on other psychological traits besides those in the domain of 'intelligence' or g (Jensen, 1990). A relatively high level of g, however, seems to be one of the several necessary though not sufficient conditions for genius.


Bringsjord, S. (2000). In light of artificial intelligence, the science of mental ability is either silly or pointless. PSYCOLOQUY 11(043) psyc.00.11.044.intelligence-g-factor.43.bringsjord

Jensen, A.R. (1990). Speed of information processing in a calculating prodigy. Intelligence, 14, 259-274.

Jensen, A.R. (1996). Giftedness and genius: Crucial differences, pp. 393-411. In C.P.Benbow & D. Lubinski (Eds.), Intellectual talent: Psychometric and social issues. Baltimore: Johns Hopkins University Press.

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

Volume: 11 (next, prev) Issue: 086 (next, prev) Article: 47 (next prev first) Alternate versions: ASCII Summary