The Nature of General Intelligence and Ability

L. L. Thurstone

I SHALL present for your consideration a brief statement of what I consider to be the fundamental nature of intelligence and the implications of my definition with reference to the other psychological categories. For several years I have been working with intelligence tests and other varieties of mental test and I have often attempted to formulate a definition of intelligence. My definitions have been modified from time to time and I shall present here their present form.

When we define intelligence we start naturally with the products of intelligence. By this I mean that since we can not perceive intelligence or intellect as a tangible thing, we look at the things that are accomplished by means of intelligence in order to frame some hypothesis regarding its nature. One's first attempt is then to say that intelligence is the ability to get along well in society, to succeed in a competitive existence, to keep out of trouble, to be able to take care of one's own affairs with a minimum of guidance. All statements of this kind refer directly to the products of intelligence, but they do not constitute in any sense hypotheses regarding the nature of intelligence itself. Another variant of this kind of definition is to say that intelligence is the ability to learn by trial and error, to profit by experience, to have memory, to reason. All of these definitions are stated so that intelligence becomes the ability to do something and that something is defined. When we describe the products of intelligence we do not necessarily describe what intelligence itself really is. I shall venture a statement regarding the possible intrinsic nature of intelligent action as contrasted with unintelligent action.

I want to show that the degree of intelligence in behaviour can be judged by the degree of incompleteness of the alternatives in the trial and error life of the actor and that the higher cognitive categories constitute incomplete conduct in the process of being formed.

Overt trial and error without foresight is the most unintelligent kind of conduct. If it is a ring puzzle that we want to master we may pick up the rings blindfolded and idly allow every impulse to complete itself overtly. By such fumbling or overt trial and error the rings may come apart without the aid of intelligent guidance. If we want to draw a circle

( 244) tangent to two given arcs we may shut our eyes and stick the compass into the paper repeatedly and draw circles at random until we are told that the task is done. In overt trial and error we complete all our impulses into overt conduct and we record the failures on the environment, and sometimes also on ourselves.

Let us now consider perceptual intelligence of the simplest order. The distance receptors give the organism some foresight because the distance percept has as its meaning the equivalent contact experience. If we perceive a puddle in front of us on the sidewalk, the visual percept is the expectation of the contact experience in the puddle, but that contact experience is only partly specified or anticipated in the visual percept. If we were moving by overt trial and error we should complete the impulse and step right into the puddle. By perceptual intelligence we have transferred the trial and error process out of the puddle to the point where we now are. We also move the trial and error process back in time. Psychologically we transfer the trial and error selection from among overtly completed impulses to the stage at which these same impulses are as yet only partly formulated. Perception, or perceptual intelligence, is a device of biological value in that the organism gains some control over future time and space. To live is, in this sense, to telescope time.

If we use perceptual intelligence, or common sense, for the geometrical construction, we shall eliminate many of the trials while they are only partly expressed and before they reach particular form on the paper. Similarly with the ring puzzle we can by looking at it eliminate as foolish many of the fumblings while they are only percepts and before they reach overt and particular form. The psychological basis for this early elimination of failures while the failures are still incomplete and perceptual is the experiential identity of the sense impression and its completed equivalent contact experience. By perceptual intelligence we move the trial and error process from among overt alternatives to the realm of the incomplete and approximate alternatives which constitute perception.

Now let us move the trial and error point still farther in the same direction and we shall come to alternatives that are still more incomplete and loosely specified. The greater part of a percept is imaginal. It is only the insignificant cue of the percept that is sensorial. The percept of the puddle contains a small retinal image and its corresponding sense impression, but the most important part of the percept is its meaning, which is the expectation of getting one's feet wet, of splashing mud on one's self and on others, and possibly of falling into it. If the relatively insignificant sensorial cue be dropped from the percept we have the

( 245) corresponding idea which is a more tentative form of action than the percept.

An idea is an incomplete expected experience which is accepted or rejected in terms of its potential completion. Ideas are loosely formulated approximate conduct. They differ from percepts mainly in their degree of incompleteness. 'Suppose that we avoid a certain street because we recall that it is being paved. We take another route, and in so doing we are eliminating a failure before it becomes overtly experienced. The impulse to walk along the bad street is only roughly specified when we recall that the street is torn up. This is ideational intelligence when the trial and error selection takes place with alternatives that have not yet found their stimulus. If we were limited to perceptual intelligence we should walk up to the reconstruction work and eliminate this route perceptually in front of the sign `street closed.' In doing so we would have specified our impulse to action much more closely than if the same impulse were eliminated while it was still in ideational form before its stimulus was found. Ideational and perceptual intelligence differ in degree only, in their relative remoteness from the determining stimuli, in the degree of completeness of the act at which its overt fulfilment is anticipated. Ideational trial and error is more intelligent than perceptual trial and error. The amount of intelligence is measured by the degree of incompleteness of the alternatives in the effective trial and error life of the actor.

The highest type of intelligence, conceptual intelligence, consists in the capacity to carry on trial and error among the crude, loosely organised and incomplete actions that we know as concepts. A concept is, in fact, expected conduct, so loosely organised, so tentative and incomplete, that our bodily relations in the expected experience are not yet specified or anticipated in the concept moment.

We teach that a concept is the cognitive symbol of that which is common to many experiences. That is true. In order to have the concept `dog' it is necessary to have seen many dogs. But that concerns only the derivation of the concept, which is a secondary matter. It is more important to realise what the concept actually is and does when it is naturally focal. Let us consider an example. When you come to a convention your first purpose is probably to get to the headquarters. As you step from the train into the station, the concept `headquarters' may be focal in your mind. The concept constitutes an unfinished approximate act. You start to hunt for the stimuli that might by some route particularise your concept into the expected experience. Your trial and

( 246) error process is at that moment of a higher order than if you were limited to the possibilities of attaining your ends by that only which happened to be perceptually given. In conceptual intelligence we have the trial and error process of purposive conduct transferred to the stage of our impulses at which they are exceedingly vague, at which our bodily attitudes are still to be specified, and at which we have not yet found the stimuli by which our impulses are finally to precipitate into overt experience.

I have outlined what I consider to be the main characteristic of intelligent conduct, the transfer of the trial and error point from overt alternatives to percepts, from percepts to the still more tentative ideas, and from ideas to the still more approximate actions that we know as concepts. This development means that the alternatives in the trial and error process have less and less intension but more and more extension as we proceed from unintelligent conduct to more intelligent conduct. Let us define intelligent conduct as a trial and error process which takes place at that stage of our life impulses at which they constitute incomplete and approximate action. Conceptual intelligence is this same process of verification and selection among alternatives that are not even bodily or spatially defined and that have not yet been particularised by the expected environment.

What I have called our life impulses might also be called conative tendencies, desires, cravings or drives; I think of them as originating in the organism itself. Their complete expression constitutes seen or overt conduct. Their conflicts while partly formed constitute perception and ideation. They become focal at different stages of completion. If the organism can carry on effective trial and error selection among these impulses while they are only partly expressed, it is to that extent intelligent. It has a wider range of possible conduct, it saves time in finding the successful act, and it saves itself from actually experiencing the failures. Abstract thinking is this trial and error process carried on with impulses that are as yet only symbolic of types of experience, unfinished, loosely specified and flexible indices of expected conduct. The impulses probably exist in some form before they become sufficiently defined to constitute thinking. The subconscious is probably the realm of impulses that are not yet sufficiently defined to be cognitive and focal. That person is intelligent whose impulses take cognitive form or meaning while the impulses are still very incomplete. To do that is to think.

Several inferences can be made regarding the construction of tests for the measurement of intelligence if we accept this definition. Obviously

( 247) those questions would be the best measurement of intelligence which indicate the degree of abstraction of which the subject is capable. Experimental studies of the Binet test questions have indicated that the questions which are relatively abstract for their respective ages are the best measures of intelligence.

When we turn to the non-language tests we find many tasks which require very little abstraction for their solution, and I question whether such tasks can legitimately be called tests of intelligence. It is possible to construct age scales with test material that requires of the subject mechanical manipulation, but unless the scale shows an increasing degree of abstraction as the essential element of the measurement, I should not consider the test to be, strictly speaking, a measure of intelligence itself.

I should differentiate between two types of tests for intelligence. One type of test measures intelligence at work during the test, and the other type of test measures the product of intelligence. If we find that a test of general information places the subjects in the same rank order as an intelligence test, we may assume that the information test measures the product of intelligence which has been at work in the past. To answer information questions in a test does not require intellectual work during the test. On the other hand, a test of ingenuity or a test in which a problem of some degree of abstraction is to be solved is a more direct measure of intelligence, because the test question requires intellectual work and some degree of abstraction at the time the test is being taken.

If we define intelligence in this way, it should be apparent that the training of intelligence can not profitably be limited to particulars. A liberal education consists in the power to strip situations of their immediate, and distracting details, to reduce problematic situations to their universals, and to put them together again better than they were found. The study of particulars, no matter how practical or interesting, does not train intelligence. To educate native intelligence requires drill in the methods of dealing with a great variety of universals. The habit of reducing problems to their universals and solving them with freedom from distracting particulars is the mark of a trained intelligence.


  1. A paper read at the 7th International Congress of Psychology, Oxford, July 1923.

Valid HTML 4.01 Strict Valid CSS2