12 THOMAS WILLIAM BROCKBANK 



Every experimenter has observed that in the early trials 

 there is little, if any, definite integration of movement 

 directed by the external senses toward the center; and 

 further, that " errors " may be made in almost every spot 

 in the problem, especially in the first few alleys. That this 

 is true of some rats more so than of others is a matter of 

 individual differences. But that these " errors " have an 

 influence on subsequent behavior is not to be doubted. 

 As the learning progresses it may be observed more clearly 

 than in the first few trials that, as a rule, there seems to 

 be some certain place or places in the maze where the 

 acquirement of perfect integration is more difficult than 

 elsewhere; and as. a consequence a delay arises here in the 

 integration of certain movements which are important links 

 in the chain of responses constituting the habit. In every 

 learning process, such as this in the maze, the series of suc- 

 cessive movements in the habit may be compared to a 

 chain which in the process of making has naturally some 

 relatively weak and strong links. The locus of an imperfect 

 response would indicate a weak link; and as the same 

 imperfect response recurs continuously, or at irregular in- 

 tervals, down through the whole process of learning, it is 

 quite evident that of all the imperfect responses possible 

 to be made in the treading of the maze, this particular 

 response or " error " is dominant. And, therefore, there is 

 to be observed in the process of learning what may be 

 called the " dominant ' error '." By the dominant " error " 

 is to be understood that imperfect integration which is 

 made most frequently as the result of some difficulty in 

 the problem which delays the establishment of perfect 

 integration. The " error " which persists is nearly always 

 the dominant "error "; in any case it is the dominant 

 " error " in the last trials of learning which, as will be 

 shown, has an important bearing on the subsequent trials 

 in redintegration. In such a problem as the round maze, 

 where each turn is a separate problem in itself, there may 

 be expected numerous " errors " more or less dominant 

 and equal in number to the number of separate problems. 



Records 1 and 2 show the daily dominant imperfect 

 response, or " error " records for two individual rats in 

 the learning of the maze. These records present typical 



