io HERBERT W0ODROWthe method, however, remains in general the same as that of a distraction method, in spite of the slight change in terminology. The important point is that a detractor may lower the degree of attention without bringing about a state of divided attention.The conditions with which a detractor of attention must comply, in order to be satisfactory for the purpose of measuring' attention, have often been discussed, and I shall emphasize here only two of the most essential. The point of first importance is that the detractor shall act solely upon attention. The detraction effect which it produces (measurable in one of the functions of attention) must be due solely to its effect upon attention. Reaction or any other psychophysical act involves other factors than attention. If the detractor affects any non-attention factor it is unsatisfactory for use in the measurement of attention. The second most important condition is that the detractor shall be of such a nature that its magnitude as a detractor remains constant. A detractor which the subject can avoid or not as he will, to even a slight extent, will vary too much in its magnitude as a detractor to be satisfactory. Again if we find that its effect very rapidly wears off we can be sure that it is unsatisfactory. As a matter of fact, as will be shown later, especially in Chapter V, the effect of a suitable detractor remains approximately constant throughout a large number of successive applications, and we may lay down such a constancy as a test of a satisfactory detractor.The attention that is measured by the detraction method described in the present article, is, as already stated, that involved in a simple reaction. Unfavorable preparatory intervals have been used as the detractor. On this account, my first task was to undertake a careful and extensive investigation of the effect of variation in the preparatory interval upon reaction time. This investigation constitutes Chapter II. The conclusion there reached is that the unfavorable preparatory intervals result in a prolongation of reaction time solely because of their effect as detractors of attention. Unfavorable preparatory intervals further recommend themselves for use as a detractor because they may be varied in unfavoraWeness and so graduated in

The author's problem here is "to explore . . . the constant and variable errors made in reproducing intervals of time . . . to determine the relationship between these errors and the length of the reproduced interval." The intervals used, 13 in number, varied from 0.2" to 30.0", and were empty, bounded by two clicks. Each interval was used alone on any one day and was presented and reproduced 50 times. The author holds that "entirely different results might be expected" if all the intervals were employed on one day, particularly if given in, say, a serial-groups order. The work began always with the shortest interval, proceeding in a daily ascending series. This was followed by a descending series for each S. The S's task was to reproduce each stimulus interval by tapping on a Seashore noiseless reaction key. "From the average of the S's reproductions was subtracted the duration of the stimulus interval . . . to obtain for each duration of the stimulus interval the S's constant error." An analysis of the means and sigmas of these C E's for the 8 S's yielded the following conclusions: (1) There was no universal tendency for long intervals to be under-estimated or for short ones to be over-estimated. (2) Almost every sort of variation in the errors of reproduction occurred among the several S's. (3) There is a daily variation in error for the same S reproducing the same interval. (4) A shift in attitude can reverse the sign (direction) of the constant error. (5) The relative variability was smaller for the shorter intervals, yet "far from . . . constant over the entire range." (6) For intervals of 6 sec. and beyond there is very little change in the relative standard deviations.

No abstract

Initial sitting, any sort of temporal discrimination. Seashore Time and Rhythm tests. 100 trials, filled intervals, standard 850 a. 25 trials, empty intervals. 12 trials, empty intervals. 90 trials (results included in Table II). 30 trials. The same 30 trials then repeated, with subjects looking at the record of their original reports, and the correct report announced immediately after each trial.

No abstract