opinion holds that attention has an opposite effect, and that "all stimuli appear relatively less when the attention is directed to them from the outset." This view has been advanced by Prof. Münsterberg as the outcome of a series of experiments made in the Harvard laboratory and published in the Psychological Review, Vol. I, January, 1894, pp. 39-44. Prof. Münsterberg speaks of his conclusion as "an unexpected result." It seemed to us not only an "unexpected," but also an unexplained result. The published account raised many most important questions, to which it gave no answer. The conclusions drawn by the author could not be accepted on the evidence given. There seemed to be a good opportunity for other interpretations of the numerical results. But no one would be competent to judge between different possible interpretations without repeating the experiments. Since the question was one of great significance for any theory of attention, it clearly demanded further investigation. And it was evident that the first step in our investigation must be a critical examination of Prof. Münsterberg's results, on the basis of a careful analysis of the same and similar experiments. Should our results coincide with his, we should then be in a position to decide independently whether they confirmed his conclusions or seemed to be more reasonably explained through factors not noticed in the original experiments. And if our results should materially differ from those obtained by Prof. Münsterberg, this fact would only emphasize the need of a more careful and complete analysis of the factors included in the experiments.

We may here anticipate our account of our results far enough to state that the greatest value of the experiments seemed to us to lie in their bearing upon a problem not mentioned by Prof. Münsterberg,-the problem of the nature and degree of the distraction employed in order to secure inattention. No one can answer the question of the intensifying effect of attention, on the basis of these experiments, until he has first secured some reliable source of distraction, or some method of measuring the degree of distraction obtained. We think that these particular experiments have no conclusive answer to give to the problem of "the intensifying effect of attention"; but as material for the analysis of so-called "distraction" they have been valuable and suggestive. Accordingly our criticisms and conclusions have been grouped together under the heading of this chapter, "Attention and Distraction."

We shall give (1) a brief account of our own experimen (2) a comparison of these results with Prof. Münsterberg together with a statement of our criticisms and conclusion

and (3) a discussion of the sources of error discovered, and of the precautions suggested by our study of these errors.

Account of our own experiments.

In general, Prof. Münsterberg's directions were followed out. The one important exception was a difference in the relative intensities of stimulus used. It seemed to us that the difference judged should be somewhere near the limen of differences; for the object of the experiments was the discovery of a tendency to overestimate or underestimate stimuli under certain conditions, and unless the differences given are comparatively near the limen of difference, the tendency, though it may be present, may give no evidence of its presence in the judgment. We found that in the series with weights, distances and light stimuli, the larger differences used by Prof. Münsterberg were correctly judged by our subjects with such ease and certainty, both when their attention was free and when it was distracted, that these differences were eliminated from the series. The change would inevitably have made any tendency to overestimate or underestimate stimuli more clearly evident in our numerical results than in Prof. Münsterberg's, had any such tendency been present.

Minor variations in the experiments, such as slight changes in the intervals of time or in the manipulation of the apparatus, will be mentioned in connection with the special series in which they occurred. The general directions for the experiments we give in Prof. Münsterberg's words :

"Our problem was to arrange the experiments in such a manner that the intensities of two impressions of moderate strength could be compared, and at the same time the attention be directed toward one and away from the other. In this way we examined intensities produced by light, sound, and the lifting of weights, and also the distances between visible points, the distances serving as measures for the intensity of the sensations produced by the movement of the eyes. The method always employed for diverting the attention was as follows: The subject was directed to give his attention fully to the adding of numbers, which in the case of optical impressions were read to him, and in the case of the auditory impressions were read by him. The adding took place before and during the time the stimulus was present. Since the order of the stimuli to be compared is of great influence upon the judgment, two sorts of experiment were arranged for each series. In one case, the attention was directed to the first stimulus, while the second was perceived with diverted attention; and in the other case, the attention was directed to the second stimulus, while the first was perceived with diverted attention. In order to discover from these series the influence of the attention, independently of other conditions, both series must be compared with the results of experiments. in which the attention was either directed to both stimuli, or turned away from both. If we designate attention to the first stimulus by A, and that to the secon 60 4 an correspondingly, the inatten

tion by I, I, we have then for each sense and for the same magnitude experiments with A-A', A-I, I-A', and I-I1.""

We have, then, four sets of results: (1) for visual distances, (2) for light, (3) for weights, and (4) for sounds.

Visual Distances. "The optical distances were given by an apparatus consisting of a black cloth surface, 80 cm. square, upon which were two white points. The vertical distance between these points could be changed by a screw upon the back of the screen and the exact distance moved could be accurately read."2 The subject opened his eyes at a signal; looked at the points for 3 sec.; at a second signal closed his eyes for 5 sec., while the distance between the points was being changed; at a third signal opened his eyes again, and looked at the points for 3 sec. In every instance a distance of 30 cm. was compared with 28.5, 29, 29.5, 30, 30.5, 31, or 31.5 cm. In half of the experiments it preceded the other member, and in half followed it. It was most frequently compared with 29 and 31 cm., and least frequently with 28.5 and 31.5 cm. The series were not precisely the same for the different subjects in any one of the four groups of experiments, since their ability to discriminate varied considerably; but the same series were used for each subject under all four sets of conditions. We had four subjects and made 400 experiments with each. The following table gives the numerical results :

We postpone the discussion of these numerical results until we have given the results of the experiments with the other stimuli.

Light. For the light stimuli two rotating disks were used with variable black and white sectors. The time intervals given were the same as in the preceding group. The darkest gray was produced by a black disk with a white sector of 65°, the lightest by a disk with a white sector of 115°. In each instance the subject was given differences of 5°-15° to judge, and care was taken that the number of cases in which the second disk was lighter should equal those in which it was darker. We have results from four subjects, and from 320 experiments given to each subject.

Weights. "The weight was given by lifting a funnel-shaped vessel, held between the thumb and first finger. The elbow rested upon the table, and the weight was raised without movement of the wrist. Weights were put into the funnel in such a way that they could be easily changed." The funnel was lowered into the hand, steadily lifted and lowered, and after 5 sec. lifted and lowered again, but with the weight changed. A weight of 300 gm., including of course the weight of the funnel, was compared with an equal weight and with weights 10, 20 and 30 gms. heavier and lighter. With one subject, V.. differences of 50 gms. were used. Five subjects ments with weights. Each gave 400 judg ch series.

tried

ment

'P.

Sounds. The sound was made by the striking of an ivory ball upon an ebony plate. "The ball was held by an electromagnet and fell at the breaking of the current. The time between the two sounds was 5 sec. The normal height of the fall, 50 cm., was compared with 35, 40, 45, 50, 55, 60 and 65 cm. A signal preceded the sound, and simultaneously with this the adding of the numbers began as they were read."'1 These experiments were tried by two subjects who had had the practice of all the preceding series. Each subject completed four series of 80 experiments, 320 in all.