Perception, 1976, volume 5, pages 375-376

Editorial

Potted IQ—Pot One The Intelligence Quotient (IQ) scale was devised by Binet and Simon to "show that it is possible to determine in a precise and truly scientific way the mental level of an intelligence, to compare the level with a normal level, and consequently to determine by how many years a child is retarded". To some eyes IQ tests are the best justification of psychology as a useful science; to others they are a scientific and social disaster. It is even suggested—sometimes strongly—that science has no right to judge people in this way. To generalise the concepts involved (which are hidden rather than explicit) I shall, in later issues, go on to see whether IQ can be applied to subhuman species and to 'intelligent machines'. Readers' comments, corrections, or suggestions will be welcome. The Intelligence Quotient is defined as Mental age x 100/Chronological age. Mental age is multiplied by a hundred to get rid of the decimal point. IQ is measured with a battery of tests of 'mental' ability, the proportion of tests passed being used to assess individual IQ. Different tests are used for different age groups, to avoid, for example, young children being penalised for not being able to read. The average of the population across the age range is given, at least conventionally, as 100 IQ. The scores are normally distributed: a markedly non-Gaussian distribution might promote a change of the tests, as would a marked deviation from a mean of 100 for the population. Children of say five, and adults of twenty-five having the same score will of course have very different tested abilities. It follows that IQ is not & straightforward measure of abilities. It is, however, supposed to be a measure of intelligence —so intelligence is not being defined simply in terms of abilities. One complexity is the introduction of chronological age. This is introduced as a weighting constant— which seems to imply at least roughly the same slope of 'mental' maturation or development for all individuals. The supposed predictive power of IQ measures when taken at one age to predict to a later age, depends on this assumption of similar rates of development, as there is no predictive test for rate of development. Development of intelligence is supposed to slow down after the age of 15, so the quotient notion is only applicable for children. Where there are physical handicaps (such as colour anomalies, blindness, or deafness) there is a temptation to say that the handicap should not affect the IQ measure, even though performance (either in general or for specific tasks, such as recognising the colour codes of electronic components) is impaired. To maintain IQ measures in spite of such handicaps, intelligence must conceptually be distinguished from many kinds of performance (hence the notion of intelligence as 'mental' ability?). There are two ways of preserving IQ measures through handicaps. On one alternative, weighting constants may be applied. This would be logically similar to chronological-age weighting, applied to preserve the average IQ at 100 though performance changes with age. This alternative is not, however, adopted: instead, tests are chosen which are not affected by these common (and socially acceptable) handicaps. This fits our intuitive notion that it is unfair and wrong to think of physically handicapped people as having impaired intelligence, even though they cannot tackle all tasks normally. This in turn implies that we do commonly distinguish intelligence from many kinds of ability—sometimes with nightmare doubts about impairment through brain damage. Evidently the IQ tests are designed to

376

Editorial

preserve our usual view of these tricky matters, by not introducing explicit weighting functions to compensate for socially acceptable handicaps. This would, however, seem to have two dangers: one social, the other philosophical. By separating intelligence from what amounts to almost all the behaviour and performance of normal living, intelligence becomes mysterious, highly special—even Academic Essence of Mind. This is reflected in Binet's 'Mental age' which distinguishes mind from behaviour; which may be justified, but it puts a golden egg in the basket of abilities, and it is this which is attacked by the tests. Perhaps there is no such special egg. Perhaps (and this is the philosophical danger) the selection of socially, sexually, and racially neutral tests—which are then used for selecting individuals and even deciding their fate—has created this notion of an entity giving and limiting our abilities and our understanding. If, indeed, IQ tests are designed to minimise individual differences which may be associated with different education, race, or sex, what should we say of IQ differences which are found between human groups? It is, for example, currently found that human females have a smaller spread of IQ scores than males, and no attempt is made to devise tests to give the same spread. So it is regarded as a fact that the variation in IQ in girls and women is less than the variation in boys and men, though the mean scores may be the same. It would be interesting to know whether the IQ test designers change their tests when socially awkward differences, for example between the sexes, are found. The IQ scale has no bottom. There are other scales where we cannot attain the bottom; but at least they are defined. This is so for absolute temperature: we cannot attain 0 K but we can define it as the death of matter. Human IQs traditionally reach an upper value of perhaps 190, but the scale seems not to be applied below about IQ 50, although there are, sad to think, people who would not score 50 points. They do not get points for being able to feed themselves, or recognise what is after all a variety of situations well past the current achievements of Artificial Intelligence. IQ tests are not applied to people with extreme mental handicap, or to subhuman species. It is, however, allowed that other animals (monkeys, rats, and even worms) have intelligence. Why, then, is IQ not applied to other species? There is no simple answer; but by seeking an answer we might come to understand, with dispassion and more compassion, the problems of IQ. We will invoke Machine Intelligence to give a context for humanity next time. Richard L Gregory

Potted IQ--pot one.

Perception, 1976, volume 5, pages 375-376 Editorial Potted IQ—Pot One The Intelligence Quotient (IQ) scale was devised by Binet and Simon to "show t...
272KB Sizes 0 Downloads 0 Views