THE ALTERNATION PROBLEM 375 



adapt the alternation to the objective sequence. The curves 

 representing the percentages of successful alternations approx- 

 imate most closely to the learning curves. Next in order of 

 importance is the ability to alternate. The success of the 

 initial choice is the least important factor; this fact is readily 

 comprehensible from two considerations. The number of initial 

 choices constitutes a very small proportion of the total, and the 

 ability to alternate successfully depends but little upon the 

 success of the initial choice except after the problem is prac- 

 tically mastered. 



NATURE OF THE CO-ORDINATION 



Each of the two alternating habits consists of an association 

 between a movement and a certain stimulus. The two stimuli 

 must fulfill at least one requirement; they must be presented in 

 a given temporal order. Four possibilities exist: — 1. The ani- 

 mals may be reacting in a differential manner to the two acts 

 of adjusting the sliding doors, or to two different sensory condi- 

 tions resulting from the adjustment. 2. They may be reacting 

 to two different ways in which they are handled and placed in 

 the starting position. 3. Each movement may be aroused by 

 the cutaneous and kinaesthetic stimuli resulting from the pre- 

 vious act. This hypothesis assumes that the two acts are 

 functionally related to each other in much the same way as are 

 the two leg movements in locomotion. 4. The rats may be 

 reacting to two different motor attitudes maintained during the 

 act of feeding. The arrangement of the apparatus was such 

 that the animals were forced to alternate between two opposite 

 directions of approach to the food. It is possible that the 

 body orientation involved in approach may be continued during 

 the act of feeding, and hence that each run is preceded by a 

 distinctive motor attitude toward the food. 



The first possibility was eliminated by instituting tests in 

 which both sliding doors were left open; in other words the rats 

 were forced to react when the usual acts of adjustment were 

 omitted. Again the doors were adjusted only after the choice 

 of exits was made. Such control conditions did not decrease 

 the percentage of correct responses. 



The second possibility was tested in several ways. 1. The 

 rats were placed in the box as usual with the exception that 



