LABYRINTH HABITS OF THE PIGEON 293 



unable to say definitely. It may be because with training the 

 birds became less susceptible to such distracting stimuli as are 

 incident to novelty of surroundings. It may be that training 

 resulted in the birds acquiring an attitude similar to self-con- 

 fidence in man. 



In figs. 5 and 9 it will be noticed that the error curve in each 

 case descends more rapidly at first than does the time cur\'e. 

 This is just the reverse of the results obtained by Mrs. V. C. 

 Hicks'" for rats. It is in harmony with the view already ex- 

 pressed in this paper as to the complexity of mazes B and C 

 to explain the present relation between the time and the error 

 curve as due to the confusion and consequent slowing down of 

 the pigeon when it is confronted with a problem so near the 

 maximum of dif^culty, As soon, however, as the bird regains 

 "self-confidence," it speeds up and the curve drops rapidly in 

 comparison with errors. From this point on, time and errors 

 run along approximately together. 



//. Second series: Rotation of the maze. It is a familiar fact 

 in work on animal behavior that when a maze, or other problem 

 box with which an animal has been trained, is rotated in refer- 

 ence to the environment, confusion results and a more or less 

 extended period of relearning is necessitated. The work on 

 this problem with birds has been done by Porter" and Watson.'^ 

 Both found that rotation caused confusion. Neither, however, 

 made a detailed analysis of it. 



In the present experiments three different degrees of rota- 

 tion were tested, viz., 90°, 270°, and 360° to the left. Some 

 birds were confused at 90°, but perfect at 270°. For others 

 the reverse was true — or very nearly so, there being a slight 

 confusion at 90°. One bird was confused at both positions. 

 After sixteen days training in these two positions, all the birds 

 were perfect at 360° rotation. 



Fig. 8 summarizes graphically the results of these tests. Table 

 IV presents numerically the results obtained by rotating the 

 maze 90°. It is to be noted that a considerable confusion is 

 indicated by both the average times and the average errors. 



"* Hicks, V. C; The Relative Values of the Different Curves of Learning. Jour. 

 Animal Behavior, 1911, vol. I, p. 145. 



"Op. cit., pp. 256-7. 



12 Watson, J. B.' Behavior of Noddy and Sooty Terns. Carneg. Instit. Publ., 

 1909, no. 103, pp. 254-5. 



