Quarterly Journal of Experimental Psychology

In remembering, items are used in the form in which they are alleged to have been perceived at some time in the past. In thinking, however, evidence is given and the essential character of the process is to move beyond this evidence, but in line with it, to something for which so far, no exact or complete perceptual basis is claimed.

From an experimental point of view the best approach is to consider thinking as a high level skill, and to design experiments in the light of what has been discovered during recent years about the learning, practice and achievements of skill on a psycho-motor level.

When this is done, it turns out that the thinking process can usefully be considered as one in which gaps are filled up in accord with specific evidence provided. The fundamental gap-filling processes are either of interpolation or of extrapolation.

In the present paper the methodology proposed is developed and illustrated in a preliminary manner.

Bartlett (1932) observed in one of his experiments that his subjects, when presented with a series of figures, tended to name them with reference to their general trend of development, sometimes anticipating the outcome rather than making an immediate identification. Two recent experiments have studied the effect of verbal “set” upon perception, and the present study was designed to show both the effect of the objective frame of reference and of verbal suggestions upon the naming of a series of figures. A series of 10 figures was presented to one group with instructions to name them. A second group was shown the same figures in exactly the reverse order. The distribution of responses was significantly different for the two groups, showing the effect of the different settings. Other groups were given the same figures, but with preliminary suggestions as to what they were to see, and their responses show some modifications. Individual variations in quality of response are also discussed, and the general implications of the method and materials for research on conceptual thinking are briefly surveyed.

An experiment was designed to show how immediate recall may affect recognition. A number of subjects were shown a picture. Some were asked to recall it and were then given a recognition test. Others were given only the recognition test after the same interval. Only 4 of the 16 subjects who had recalled it identified it; whereas 14 of the 16 others did so. A second experiment gave similar results.

Recall was constructed round dominant items of the picture. This distribution of emphasis together with the acceptance of an invented detail as genuine were the common causes of errors in subsequent recognition. Both the dominant and invented items in recall were those which became most obviously merged into an organization of related experiences and in consequence those which militated against subsequent recognition.

The recognition test was applied in two further groups of 16 subjects with a change in one of the dominant details (i.e. the wording). The number of subjects who now correctly identified the remainder of the material was 9 when there was no intermediate recall, and nil when immediate recall of the original material was interposed.

The case of a healthy normal subject is described in which negative after-images were reported to follow fixation of an “imaginary” pattern. The size of the after-image was measured at various distances and found to agree closely with the size of an after-image induced by fixation of a “real” visual pattern, and to conform roughly with Emmert's Law. No evidence for chromatic after-images was obtained. Two controls were conducted, one indicating that the results could not be attributed to after-images resulting from objective aids used to maintain an “imaginary” pattern of constant size; the other discounting the possibility of the subject's having guessed the correct values. Similar phenomena reported in the early literature are briefly mentioned, and some theoretical implications discussed.

An experiment is described to test the relation between angle of display and performance in prolonged visual tasks, using the Clock Test in three positions of display surface, vertical, at an angle of 45 degrees to the horizontal, and horizontal. In every case the line of regard was kept at right-angles to the display surface. The results indicate that significantly fewer stimuli are missed in the vertical position of the display surface than in the other two.

The deterioration observed in the second half-hour of the original Clock Test experiments was not found in the present experiment, and two possible reasons for this are discussed briefly.

This is the third of a series of papers dealing with the effects of repeated retesting on intelligence test scores. It comprises an account of two further experiments, and a discussion of the four so far performed, since each throws light on the results of the others and it is their joint interpretation which is thought to be of value.

The two earlier experiments consisted in repeated testing of a group of W.E.A. students and a group of mentally defective boys by an intelligence test (AH 4) designed for an unselected population. In the two later experiments, a group of Technical College students and a group of Naval Ratings were repeatedly tested on another intelligence test (AH 5), designed for a population of high-grade intelligence, such as university students. Examination of all the results confirms the artificiality of the “ceiling effect” which was suspected in the first experiment; it suggests that, given sufficient scope, subjects of all levels of intelligence would, with repeated trials on the same test, continue to improve their score up to and beyond 10 testing; and that, given comparable opportunity, subjects with initially higher scores will tend to improve more than those with initially lower scores. All these results concern short-term experiments with testing at weekly intervals.

The four sets of results considered jointly indicate that conclusions on these problems should be drawn only in terms of the relation between the level of the test and that of the group tested: this offers some resolution of the conflicting findings of earlier investigators.