Quarterly Journal of Experimental Psychology

An experiment was made to determine the extent to which the performance of a tracking task was adversely influenced by simulated tropical environmental conditions, by comparing this effect with that of loading the control handle with weights. The pursuitmeter used was similar to that used by Mackworth (1948) when he showed that the accuracy of performance was reduced under tropical conditions. The score of error was the total angular separation between the target and follower pointers expressed in arbitrary units. It was found that as the room temperature was raised from approximately 80° to 90° on the Effective Temperature scale, this score increased in the ratio 1–13, an increase which was also obtained by increasing the load on the control handle from 8 to 30 lb. There were also indications that the effect of room temperature was greater with higher handle loads. The effect on performance was not correlated with rectal temperature or with the amount of weight lost.

It is pointed out that the fate of generic images has been studied little, if at all, in amnesias of organic origin. Observations are described in a case of alcoholic Korsakoff psychosis which appear to indicate that generic images are subject to Ribot's Law of Regression. The patient, a spinster of 57, with some talent for drawing, was asked to represent a number of objects, e.g. a woman dressed in the latest fashions, which might reasonably be regarded as, “generic” in nature. These drawings suggested that the patient's generic images were, in some cases at least, appropriate to an earlier period of her personal history. In general, this ante-dating effect was in good correspondence with the scope of the retrograde amnesia. Some implications of these findings for the psychological theory of memory are briefly discussed.

An experiment on paired-associate learning was carried out to test Hovland's prediction that spaced practice should be more advantageous than massed if the order of presenting the pairs of nonsense syllables were fixed.

In the results, spaced and massed practice showed little difference in learning efficiency. Hovland's prediction was not confirmed.

In a previous experiment of this series of studies it had been found that spaced practice was more efficient than massed in serial learning when syllables of “low-meaningfulness” value were used; but that there was little difference in efficiency between the two forms of distribution when “high-meaningfulness” syllables were used. The results of the two experiments naturally suggested that, as regards the relative efficiency of spaced and massed practice, paired-associate learning of low-meaningfulness syllables differed from serial learning of similar syllables, but matched serial learning of high-meaningfulness syllables. It seems justifiable to draw the conclusion that the relative efficiency of spaced and massed practice is different in paired-associate and serial learning, not because remote associations can be formed in the one case and not in the other, as Hovland (1939a, 1939b) assumed; but because the organization of the material and the learning processes are different in the two forms of learning.

Twelve mentally-defective schoolboys between 14 and 16 years of age took an intelligence test (AH 4) once a week, ten times. They took also Progressive Matrices once, after they had finished the AH 4 testing. Subjects of low intelligence were chosen in order to throw some light on the findings of a previous experiment on repeated retesting, in which the subjects were intelligent adults. The main aims of the schoolboy enquiry were to ascertain how far the flattening of the progress curves observed in the earlier enquiry was due to the artificial “ceiling” imposed and how far the part played by variation within the individuals varies with the level of intelligence of the group.

The mentally-defective subjects showed a gradual improvement in test performance with little sign of flattening towards the end; their individual progress curves were far less smooth than had been those of the more intelligent subjects. Other findings included relatively high proficiency on the diagrammatic part of the test, as opposed to the verbal and numerical, and some transfer of training from AH 4 to Progressive Matrices.

The difference threshold for the velocity of a seen object was measured by the method of constant stimuli, using two categories. An approximate correspondence with Weber's law was found, the divergence from it appearing, in general, as an increase of the threshold at both ends of the range of initial velocities. The Mean Threshold (0·5 probability of perception, corrected for guessing) was, in favourable conditions, about 12 per cent, of the initial velocity. Whether the stimulus was an increase or a decrease of velocity made no marked difference. With two moving objects, which converged, crossed, and then diverged, both suffering the same change of velocity, the threshold was higher.

Velocity changes as low as 2 · 5 per cent, elicited a significant proportion of correct responses. Some theoretical points in connection with this are discussed. Responses to blank stimuli showed a strong tendency to guess “slower,” which tendency differed significantly in degree between most of the experimental conditions. Tests with reduced exposure times showed that exposures could be as short as 0 · 5 second (the velocity change occurring in the middle of the exposure) without appreciable detriment.