THE ALTERNATION PROBLEM 367 



way. The time devoted to a single run varied somewhat with 

 the animal and the stage of mastery, but it became practically 

 a constant after the first fifty trials. This time was determined 

 for each animal for different stages of mastery. The average 

 time per rat ranged from 21.5 to 25.5 seconds with a group 

 average of 23. Of this time, 6.5 seconds were devoted to the 

 run and 16.5 seconds to feeding and handling between runs. 



ANALYSIS OF THE LEARNING PROCESS 



All members of the group were able to master this simple 

 alternation with a high degree of accuracy. A consistent record 

 of 85% of correct choices for the group was obtained at the 

 end of 600 trials. The number of trials per rat necessary to 

 secure such a degree of proficiency ranged from 168 to 588, 

 with a group average of 412. The number of trials for five of 

 the eight animals closely approximated 450. 



Three graphs representing progress in mastery are given in 

 fig 2. The group curve is represented by the solid line. In 

 its general features it is similar to the usual learning curve. 

 The distribution of choices between the two exits is at first a 

 matter of chance as the initial record is 50% of correct choices. 

 The initial trials are more effective than the later ones though 

 the curve approximates a straight line more closely than does 

 the typical learning curve. There is some indication of the 

 existence of a plateau beginning at the 340th trial. This phe- 

 nomenon is to some extent a group artefact, though four of the 

 eight individual curves give some indication of a plateau in 

 this region. The individual curves exhibit some pronounced 

 differences. Four graphs exhibit a relatively rapid initial ascent 

 followed by a period of slower progress. Only one of these, 

 however, approximates the typical learning curve. The curves 

 for three animals exhibit an approximately straight line ascent; 

 progress is uniform for all stages of mastery. One curve is quite 

 unusual in this respect as it descends rather rapidly for 200 

 trials, then rises abruptly, and this period of ascent is followed 

 by the usual slow progress. This curve is represented by the 

 broken line graph of fig. 2. The dotted line curve represents 

 the case in which the initial trials are relatively the most effec- 

 tive. These two individual curves represent the two extremes 

 between which are to be found all degrees of gradation. 



