190 ROBERT M. YERKES AND CHARLES A. COBURN 



boxes. When the exit doors are down, the various parts of the 

 food trough are covered by a horizontally placed sheet of metal 

 which fits closely over them and thus prevents the subjects from 

 obtaining food from the outside of the apparatus. 



The large enclosure is divided into four principal parts: (1) the 

 part which contains the reaction-mechanisms with space for the 

 observer's bench, G, and writing table, H, and a passageway 

 for the subject from the exit doors of the apparatus to the yard, 

 S; (2) second, the reaction space which is labelled B in figure 2, 

 in which the subject responded to the multiple choice situation; 

 (3) and finally, the two yards, S, S, from which the subjects 

 started in the case of each trial and to which they returned on 

 the completion of their reaction. K, L, M, and N, designate slide 

 doors between the several portions of the large enclosure, while 

 J and I represent doors which were used by the experimenter. 



The entire apparatus was constructed in sections, so that at 

 the end of the season it might readily be taken down and stored. 



This brief and very incomplete description will be supple- 

 mented somewhat in the section on experimental procedure. 



PROBLEMS AND GENERAL METHOD 



The four problems enumerated on page 186 were presented to 

 each subject in the order named. For each of these problems, 

 a series of ten settings of the doors was determined upon. These 

 settings differ somewmat from those employed in our study of 

 the crow. It is our intention, so far as possible, to use them 

 with all types of subjects until our observations indicate desirable 

 changes. 



We present below for each of the four problems (1) the num- 

 bers of the settings, (2) the numbers of the doors open, (3) the 

 total. number of doors open in each setting and for the series of 

 ten settings, and (4) the number of the right door. 



It was our plan to give each subject an opportunity to respond 

 to each of the ten settings for a given problem in order and to 

 return then to setting 1 and repeat the series. It was found 

 impossible, however, to give ten trials in succession in our early 

 experiments, and in the case of both problems 1 and 2, as a rule 

 a subject was given five trials in succession. For problems 3 

 and 4 it was found possible to give ten trials in succession. 



* 



