FACTORS IN LEARNING BY WHITE RATS 



359 



It will be noted that the summaries of the first three trials 

 of the animals of each group approximate rather closely the 

 results of all seventeen, so far as combined recency and fre- 

 quency relations are concerned; i.e., 38.4% against the expecta- 

 tions based on recency alone and on both recency and frequency, 

 and 48.8% in agreement with the expectations on recency alone 

 and on both recency and frequency. The differences of physical 

 conditions in the three mazes used — slight in the case of the 

 B-mazes, differences only in the relative lengths of the cut de 

 sacs similarly located with respect to the correct path — do not 

 show themselves much in these results. Of course, only a small 

 number of animals were tried on each maze and the present 

 results need corroboration by more extensive studies. It is 

 obvious that there is a gradual increase with successive trials, 

 in all three mazes, in the reactions agreeing with recency expec- 

 tations or with recency and frequency expectations combined; 

 and that there is a corresponding decrease in reactions violating 

 such expectations. Table XV shows this tendency. It would 



TABLE XV 



Showing Gradual Increase with Successive Trials in Reactions 

 Favoring Recency and Frequency Expectations 



seem from this table that the change is most rapid in the easiest 

 maze, IA, as is to be expected. It cannot be too strongly 

 pointed out, as has already been mentioned, that this increasing 

 percentage of reactions agreeing with the expectations based on 

 recency and frequency effects, as learning advances from the 

 first random stages toward the establishment of a regular habit, 

 cannot be safely regarded as evidence that learning is brought 

 about by recency and frequency factors: our evidence seems to 

 justify the contrary conclusion, that this increase in reactions 



