the connection E-8 he neglected other relations previously learned and gave 8 instead of 10 for D, preceding E. He had for some time been giving 10 correctly for D. This was a case of being so well set for the response 8 at the difficult point that the reaction took place too soon. This sort of behavior is often noticed, and clearly shows overlappings of impulses and conflicts among different attitudes. It has been shown that similar tensions also appear in the learning by animals as low as the white rat. Some blindalleys are frequently eliminated by a piece-meal process. At first the animal runs to the end of the blind-alley; but soon it begins to turn a little short of the end. With successive entrances in different trials the degree of penetration becomes less and less, speaking generally, until, at a late stage, there is hesitancy at the approach of the blind-alley in question and a peculiar and rapid vibration of the head takes place. If, at this stage of inhibition, the impulse to enter breaks through, the animal may run with a blow against the end of the blind-alley and may emerge so much disoriented that it enters several more distant blind-alleys which had, for some time, been successfully eliminated under usual conditions. Thus by a single break the whole system of balanced acts becomes in a measure disturbed.1 We seem to need more detailed analyses of the different 'kinds of learning,' and more attention to the conflict of impulses in the earlier stages of learning and to the manner of their disappearance as the adjustment becomes more and more definitely established. Except to the most superficial observation, 'learning' is not a detached part of psychology that can be divided off into certain discrete types. In all learning we shall probably find the sort of random, trial-anderror adjustment that is so characteristic of the acquirement of new sensory-motor controls, together with inhibitions and balancings of conflicting tendencies to response. The higher forms of learning-adjustments, involving ideational activity, differ from the usual learning by trial and error in that in such forms the conflicts are among systems of acts already 1 See Behavior Monog., 1917, vol. 3, no. 15, 28ff. coördinated within themselves for fairly remote, far-reaching adjustments. At any rate, the present study, which is only at the initial stage, seems to indicate that with finer and finer criteria of errors applied to rational and higher forms of learning, random processes similar in essentials to the usual trial-and-error type will be found. The question of how the maladjusted, excess acts become eliminated seems to be basic to all learning, and fundamental to an adequate understanding of all behavior for which the mechanism is not fully determined by innate structures. Instead of being a process of merely combining a number of elementary unit reactionsystems by association, or of 'conditioning,' to use a term now in favor, learning seems to be fundamentally a smoothing out of conflicts among incompatible impulses aroused by a complexity of external circumstances or stimuli, so that the organism can act in a somewhat unitary or consistent manner toward them. In certain cases these conflicting impulses may, of course, be somewhat local and insignificant, as when conditioned reflexes are established; but it is well known that even such processes do not usually work with great regularity, as if detached from all inhibitions by the general attitudes of the individual. It is becoming clear from recent work in many laboratories 1 that biological research is putting a strong emphasis on the unity of the organism and the interaction of its various part-processes. The results of these investigations are in agreement with those which have been pointed out in the foregoing pages, indicating inner tensions which are not overcome by mere frequency and recency factors in association.2 Factors of inhibition and facilitation seem to play important rôles in effecting the adjustments, however this is accomplished. As to the value of proper distribution of trials in learning we may consider the learning rates in the present experiment of eight subjects-three college men and one woman, and four high school boys-who were severally taken through 1 See, for example, Child, C. M., Physiological foundations of behavior, 1924; and Herrick, C. J., Neurological foundations of animal behavior, 1924. 2 See also Peterson, Jos., Learning when frequency and recency factors are negative, J. Exper. Psychol., 1922, 5, 270-300. nine trials of 156 responses each, in one sitting. Five of these subjects went through ten trials, and two through eleven. In Table III. the results are given in the form of median errors per trial, and of the average deviation from the median. It is obvious that the error-elimination was TABLE III NUMBER OF ERRORS IN SUCCESSIVE SERIES (156 REACTIONS IN EACH SERIES) much more rapid in the case of the eleven daily practices than in that of the long practice all at one sitting, although there are of course too few subjects to make the results more than tentative. On the view of balancing factors in learning by the interaction of different part-processes, it would seem reasonable to suppose that with equal practice the longer period of adjustment would be superior to the shorter time of learning all in a single sitting, because in the former case more time would be allowed for the settlement of the conflicts and for the working out among the several impulses of some more or less consistent or organically complete reaction. THE BRADYSCOPE; AN APPARATUS FOR THE AUTOMATIC PRESENTATION OF VISUAL STIMULI AT A CONSTANT SLOW RATE BY ERWIN A. ESPER University of Illinois In a previous publication1 the writer has described an apparatus for the automatic presentation of visual material exposed serially and at a constant slow rate. This apparatus consisted of a cylindrical drum bearing an endless paper belt on which were pasted the figures to be used. A star-wheel mechanism turned the drum one-fifth of a revolution every 2.7 sec., and also closed a shutter over the exposurewindow while the drum was in motion. In the course of six months and more of constant use two main difficulties appeared. First, the paper belt was very easily torn; and secondly, for every change in the order of presentation of the figures a new paper belt had to be prepared and placed on the drum. In preparing a new series of experiments which require the same type of visual presentation, the writer has devised an apparatus which avoids these difficulties and makes possible quick and easy interchangeability of all the figures of a series. The apparatus consists of a pentagonal spider drum A by means of which an endless chain of metal card-holders E is moved. The spiders are aluminum castings mounted on a steel axle. The flanges on the spiders which carry the cardholders are 14.1 cm. in length and 1.6 cm. in width. An aluminum star-wheel B is mounted on the same axle as the drum; two prongs of this star-wheel ride on the circumference of the internal (driven) piece of the cone clutch C, also of 1A technique for the experimental investigation of associative interference in artificial linguistic material. Language Monog. Linguistic Soc. of Amer., 1925, no. I. |