THE WHITE RAT AND THE MAZE PROBLEM 17 



in the black-white maze and of 17 in the normal maze. It shows 

 the trial at which the zero point was reached and the number 

 making it at each point. It is perhaps more striking than the 

 ordinary error curve. 



If the errors are so few in the beginning, if the co-ordination 

 is acquired so quickly, why is it that there is not a greater differ- 

 ence in the total errors ? This is a question which requires an 

 answer. The answer is that the final accuracy is less. See the 

 curves, Figs. 2, 3 and 5. The normal error curve is regularly at 

 the end of 50 trials below the black-white. What the causes 

 are which produce this greater final variability is a question 

 for our future inquiry. The three main points to be emphasized 

 here are the few beginning errors, the rapid drop to the zero 

 point and the persistence of errors to the very end. 



SURPLUS TIME AND RATE OF ELIMINATION OF TIME 



AND ERRORS 



Professor Carr argues at length for the value of the rate of 

 elimination of surplus time and errors as one of the elemental 

 components of the learning curve which varies independently 

 from the other components. 15 The curves which show elimina- 

 tion in the experiments here reported are seen in Fig. 7. The 

 rate of elimination is very similar in the two mazes. There 

 are, however, some points of contrast. 



The time curve for the normal groups exhibits a fairly steady 

 rate of decline to the 24th trial. Here the final limit is reached, 

 .4%. The time curve for the black- white group reaches its 

 final limit of 1.8% on the 9th trial. It then follows the same 

 course as the curve for the normal maze but with greater varia- 

 bility. The curve for the black- white maze reaches its level 

 much sooner but cannot maintain it with the same constancy; 

 neither can it reach, in the limits of the experiment, the same 

 low level which the normal curve so easily reaches and maintains. 



The error curve of the black-white maze indicates a much 

 quicker rate of elimination than that of the normal maze. There 

 is a steady drop till the 6th trial. From this point on there is 

 only 6% of the errors left to eliminate. It will be remembered 

 that the b lack-white maze had fewer errors to eliminate at the 



15 Hicks, Vinnie, and Carr, H. A. Human Reactions in the Maze. Jour, of 

 Animal Behavior, vol. 2, no. 2, p. 98. 1912. 



