papers ox psychology and education 491 



learning, except for the one change mentioned. The ani- 

 mal, in this second part of the experiment, went directly 

 and immediately to the east side, where the door had 

 been; and, after making eight or nine button-turning 

 movements where the button had been, engaged in ran- 

 dom movements until after three minutes and forty sec- 

 onds it turned the button. Upon the next trial it at 

 once directed its movements toward the button in its 

 new position, and with hardly any superfluous motion. 

 After this the box was placed in other positions — namely, 

 90 and 270 degrees from the original one, and the cat 

 went directly to the door and button, performing the 

 correct act. 



The experiment was repeated with another cat, and the 

 findings set forth above were corroborated in all essen- 

 tial respects. 



The experiment shows that an association can be 

 formed between the button as a stimulus and the appro- 

 priate movement, but that this association does not come 

 about until the box has occupied several different po- 

 sitions. The original orientation, when the box was fac- 

 ing east, was with other, no doubt external, features of 

 the environment rather than with the sense-impression 

 of the interior of the box. 



The results of this experiment accord with Carr 's find- 

 ings relative to the orientation of rats in a maze. 3 The 

 maze was covered with a canvas top about two feet above 

 the glass, flaps extending all the way down on the four 

 sides. There was a peep-hole, and the interior was il- 

 luminated with an electric light. A group of ten rats 

 learned the maze thus covered; then the cover above was 

 turned through ninety degrees, and the rats reintro- 

 duced. Xo disturbance in the habit resulted. After this 

 another group, of seven rats, learned the maze with the 

 same canvas cover, except that one side of the cover re- 

 mained open. Then the cover was closed on this side, 

 and opened on another. Five of the seven rats were 

 disturbed by this new condition. In six trials the total 

 number of errors per rat ranged from nine to fifteen. 



»Jour. of Am. Beh.. 7, 1917. pp. 265f. 



