HUMAN REACTIONS IN A MAZE 123 



1:1.9 53. 6. That these ratios express the relative ability of 

 the three, groups with accuracy is a doubtful proposition. The 

 high degree of variability inside of each group must also be 

 considered. The percentage of individual variability for time 

 and error (Table V) is approximately 30%, for each group. 

 It is doubtful if the intelligent capacity of the individual mem- 

 bers of any group varies to any such extent. When corrections 

 are made for the average variability, the above ratios should 

 read 100 ±30:190 ±57 :36c ± 108. Such values can express the 

 relative status of the groups only in a gross and general way. 

 (c) We have expressly refrained from maintaining that no 

 causal relation obtains between intelligent ability and the 

 "number of trials" and the "rate of elimination." If the 

 " surplus values " represent the " amount of effort," the " num- 

 ber of trials " and the " rate of elimination " can be embraced 

 under the caption of the " distribution of effort." Experiments 

 on human memory have demonstrated that the " distribution 

 of effort " is a very 'essential factor in economic learning. Pre- 

 liminary experiments in this laboratory have also demonstrated 

 its importance in such a problem as the maze. To maintain 

 that such features should be totally ignored, in interpretations 

 of intelligent ability, is preposterous. Furthermore, the three 

 factors are not independent variables, but they are interrelated 

 in a very complex manner. Intelligence is related to each 

 factor both directly and indirectly. An insistence upon the 

 intricate complexity of the three factors as to their relations 

 to each other and to intelligence leads to several suggestions. 

 (1) It helps to explain why the " surplus values " are not an 

 accurate index of the degree of intelligence. (2) It suggests 

 that an accurate index must be based upon some formula which 

 shall give due weight to the element of " distribution of effort." 

 (3) Comparative psychology has many problems to solve before 

 making any attempt to rate the intelligent status of groups of 

 animals on the basis of their learning curves. As yet prac- 

 tically nothing is known of the real meaning of a learning curve 

 and the interrelations of its several features. (4) The concep- 

 tion renders explicable several peculiar features of our results. 

 For example, there exists an inverse relation between the initial 

 values and the number of trials. In so far as intelligence operates 



