HUMAN REACTIONS IN A MAZE 103 



of both groups of rats. If but one successful trial had been 

 demanded of the group R II, their results would be 7 ±3. The 

 values for Miss Vincent's group are 12 ±4.6 when the criterion 

 of mastery is the ability to run 8 trials out of 10 without error. 

 While the results indicate that the animals can learn a maze 

 in a less number of trials than human subjects, yet one feels 

 some hesitation in advancing such a conclusion in view of (1) 

 the differences between the mazes, (2) the different criteria of 

 masterv used, and (3) the small number of individuals constitut- 

 ing each group. One is certainly justified in concluding that no 

 correlation exists between the number of trials necessary to 

 learn a problem of this sort and the degrees of rational ability 

 or of the general intelligence of the groups compared. This 

 means that one cannot estimate correctly the relative intelli- 

 gent ability of any two groups of organisms wholly in terms of 

 the relative number of trials utilized in learning a maze. The 

 conclusion does not mean that there is necessarily no causal 

 relation between the two factors, but that if it does exist, it 

 is obscured by more important factors. 



II. GROSS VALUES ELIMINATED 



A pronounced and unambiguous inverse relation obtains 

 between the degrees of intelligence and the amount of surplus 

 values eliminated. In a general way the ratios between the 

 values eliminated are approximately 1 \2 .-3 for the adults, chil- 

 dren, and rats, respectively. Stated in terms of the usual curve 

 of learning, these figures mean that the average height of the 

 rat curve above the base line is three times as great as that of 

 the adults, while the adult curve would possess but one-half 

 the height of the children's curve. 



Tables 1, 2, and 3 give the average surplus time, error, and 

 distance values for the four groups for the successive trials. 

 At the bottom of each column is given the average value elimi- 

 nated per individual in the learning process, and the average 

 individual amount for each trial. Referring to Table II, column 

 I, it is seen that on the average each adult made 58 errors in 

 learning the maze, or on the average 3.4 errors on each trial. 



The surplus values for time and distance are computed by 

 subtracting from the records of each trial the corresponding 

 value for the last trial. If a rat traverses go feet in anv trial, 



