If you are taking a child's temperature with a "minute thermometer", it is best to use your watch to tell you when the minute is up, for the minute, when you are simply waiting for it to pass, seems very long. But if you are "working against time", a minute seems short. The professor is shocked when the closing bell rings, and thinks that certainly the hour cannot be up; but some of the students have been consulting their watches for quite a long while, being sure the hour must be nearly over. These are scarcely errors of sense, but they are errors of perception.
Where we tend to err in one certain direction from the truth, as in the examples just cited, psychology speaks of a "constant error", and evidently the knowledge of such constant errors is of importance wherever the facts are of importance. In a court of law, a witness often has to testify regarding the length of time occupied by some event, and a knowledge of the constant errors in time perception would therefore be of considerable legal importance. They would need to be worked out in considerable detail, since they differ according to the desires and attitude of the witness at the time of the event.
Besides constant errors, there are accidental or variable errors, due to slight momentary causes. Both constant and variable errors can be illustrated by a series of shots at a target. The variable error is illustrated by the scatter of [{448}] the hits, and the constant error by the excess of hits above the bull's-eye, or below, or to the right or left. The constant error can be corrected, once you know what it is; if results show that you tend to shoot too high, you can deliberately aim lower. But the variability of any performance cannot be eliminated except by long practice, and not altogether even then.
Fig. 66.--Constant error and scatter in hitting at a target. The little circle was the target, but the center of the actual distribution of the attempts lies at the cross, which was drawn in afterwards. The constant error could be stated by saying that the center of distribution was so far from the target, and in such and such a direction. The scattering of the attempts can be measured also.
Experimental psychology has taken great pains in measuring the accuracy of different sorts of perception. How small a difference in length can be perceived by the eye, how small a difference of weight by the hand--these are sample problems in this line.
For example, to measure the fineness with which weights can be perceived when "hefted" in the hand, you take two objects that are alike in size and appearance but differing slightly in weight, and endeavor to decide which is the heavier just by lifting them. You try repeatedly and keep track of the number of errors, using this number as a measure of the accuracy of perception. Now, if one weight were twice as heavy as the other (one, for example, weighing 100 grams [{449}] and the other 200), you would never make an error except through carelessness; but if one were 100 and the other 120 grams, you would make an occasional error, and the number of errors would increase as the difference was decreased; finally, comparing 100 and 101 grams, you would get almost as many wrong as right, so that your perception of that small difference would be extremely unreliable.
ERRORS IN PERCEIVING SMALL DIFFERENCES
OF WEIGHT (From Warner Brown)
Difference 20 16 12 8 4 8 2 1 grams
Errors 1 2 5 18 28 81 89 44 per hundred trials
The weights were in the neighborhood of 100 grams; each
weight was compared with the 100-gram weight, and each
such pair was lifted and judged 1400 times. Notice that
the per cent of errors gradually increases as the
difference becomes smaller.
The smaller the difference between two stimuli, the more numerous the errors in perceiving it, or, the less perceptible it is, and there is no sharp line between a difference that can be perceived and one that is too small to be perceived. That is the first great result from the study of the perception of small differences.