Since figure 17 is drawn to scale, it will be needless to give more than a few of the dimensions of the apparatus.  Each of the boxes was 42 inches long, 18 inches wide, and 72 inches deep, inside measurements.  The alleys D, I, and H were 24 inches, and G 30 inches wide, by 6 feet deep.  The doors of the several boxes were 18 inches wide, by 5 feet high, while those in the alleyways were 24 inches wide by 6 feet high.  The response-compartment E of figure 17 was 14 feet 4 inches, by 8 feet, by 6 feet in depth.  In order that the apparatus might be used with adult human subjects conveniently, if such use should prove desirable, the depth throughout was made 6 feet, and it was therefore possible for the experimenter to walk about erect in it.

The experimental procedure was briefly as follows:  A small quantity of food having been placed in each of the food cups and covered by the metal flanges on the exit doors, the experimenter raised door 11 of figure 17 and then opened door 10 and the door of the cage in which the desired subject was confined.  After the latter, in search of food, had entered the runway D, the experimenter lowered door 11 to keep it in this runway, and immediately proceeded to set the reaction-mechanisms for an experiment (trial).  Let us suppose that the first setting to be tried involved all of the nine boxes.  Each of the entrance doors would therefore be raised.  Let us further suppose that the right door is defined as the middle one of the group.  With the apparatus properly set, the experimenter next raises door 12, thus admitting the animal to the response-compartment E. Any one of the nine boxes may now be entered by it.  But if any except number 5, the middle member of the group, be entered, the entrance door is immediately lowered and both the exit and entrance doors locked in position so that the animal is forced to remain in the box for a stated period, say thirty seconds.  At the expiration of this time the entrance door is raised and the animal allowed to retrace its steps and make another choice.  When the middle box is chosen, the entrance door is lowered and the exit door immediately raised, thus uncovering the food, which the animal eats.  As a rule, by my monkeys and ape the reward was eaten in the alleyway G instead of in the multiple-choice box.  As soon as the food has been eaten, the exit door is lowered by the experimenter, and the animal returns by way of G and H to runway D, where it awaits its next trial.

As rewards, bananas and peanuts were found very satisfactory, and although occasionally other foods were supplied in small quantities, they were on the whole less constantly desired than the former.

Four problems which had previously been presented to other organisms were in precisely the same form presented to the three primates.  These problems may be described, briefly, by definition of the right reaction mechanism, thus:  problem 1, the first mechanism at the subject’s left; problem 2, the second mechanism at the subject’s right (that is, from the end of the series at the subject’s right); problem 3, alternately, the first mechanism at the subject’s left and the first at its right; problem 4, the middle mechanism of the group.