MAZE STUDIES WITH THE WHITE RAT 269 



between the two sets of values would have been much greater 

 than they were. This difference in complexity can be equated 

 by comparing the initial error record due to rotation with the 

 initial error record in learning. The average error record for 

 the first trial in learning the standard maze was 44, while the 

 average number of errors made in the first rotation test was 10. 

 Rotation in the standard maze produces an initial error distur- 

 bance which is approximately 23 per cent of that in mastering 

 the maze. The initial error record for the sideless maze was 

 10.5, while the corresponding value for rotation was 19.7 The 

 confusion involved in rotation was thus greater than that in 

 learning. Relative to the number of initial errors in learning, 

 rotation in the sideless maze produces a disturbance eight times 

 as great as in the standard maze. 



Certain peculiarities of behavior were apparent in the sideless 

 maze. The rats frequently gravitated to that corner at which 

 the food box had been located before the rotation. Failing to 

 find food, they renewed their explorations of the maze, but 

 came back again and again to this particular corner. One 

 animal finally refused to leave this locality and had to be removed 

 from the maze. After two such unsuccessful trials on successive 

 days, the experimenter guided the animal to the new position 

 of the food box, and thereafter the maze was traversed success- 

 fully on the animal's own initiative. One other rat was un- 

 successful on the third trial. This type of behavior was exhibited 

 in varying degrees during the first three trials by each of the 

 ten rats employed in the test. Such behavior has been observed 

 but rarely in a standard maze and only when certain parts of 

 the maze were flooded by strong daylight illumination. 



5. The cut de sacs in the standard maze were closed by sliding 

 doors. After learning the maze in this condition, the animals 

 were subjected to the usual rotation tests. Obviously all dis- 

 turbance due to rotation must be measured by return errors. 

 Twelve rats were tested and all were affected. The experiment 

 is significant in indicating that the disturbance in maze rotation 

 is not due exclusively to wrong choices at those critical positions 

 from which several paths diverge. Confusion obtains when no 

 choice is possible and when the animals have had no experience 

 with ad de sacs during learning. 



Rotation of Maze and Environment. The maze was learned 



