342 JOSEPH PETERSON 



The totals are very interesting and seem to throw some light 

 on maze learning. As would be expected on laws of probability, 

 the numbers at the entrance end of the maze are much larger 

 than those at the food end, though the diminution in the size 

 of the numbers as the food end of the maze is approached (the 

 right side of the table) is not entirely regular. This general 

 decrease in the numbers toward the food end of the maze is 

 obviously due in the main to the fact, not usually pointed out 

 in maze learning, that any animal entering the food box does 

 not return. This is not due simply, if at all to the pleasure 

 of eating; it is due to the conditions of the experiment. An 

 animal entering the food box is prevented from returning. There 

 are consequently on our record no returns in the section of the 

 maze leading from cut de sac 10 to the food box. Hence, more- 

 over, only one-fourth of all runs reaching 10 — whether the 

 animal enters or goes beyond it — result in returns from this 

 point. This is the expectation statistically and our results of 

 the four " trials ' of the hypothetical rat only approximate 

 this expectation, — 2 to 4. As we shall see, this is a point of 

 much significance which seems to have been overlooked in 

 studies on maze learning. 



In these results, derived by the laws of chance, the totals for 

 all forward movements in the lettered parts of the maze — the 

 sections making up the true path — amount to 96. For ease 

 of detection these numbers are in bold face in the totals column. 

 The total entrances to blind alleys are 88; and there are in all 

 71 returns over the lettered sections of the maze. These num- 

 bers come close to probability expectations, as will be shown. 

 A forward-direction approach to any blind alley gives a prob- 

 ability of 1/2 that the blind alley will be entered and 1/2 that 

 the correct path will be kept. If the cut de sac is entered the 

 animal will return to the true path, at which point the prob- 

 ability is again 1/2 that the forward direction will be taken 

 and 1/2 that the return direction will be chosen. This makes 

 a total probability of 3/4 (= 1/2 + 1/2 of 1/2) that the animal 

 will keep the general forward orientation whether or not the 

 blind alley is entered, and of 1/4 that it will be turned back at 

 the blind alley (i.e., 1/2 of the blind alley entrances will result 

 in returns). 2 In general, of % approaches to any blind alley in 



2 For simplicity we assume that returns occur only at times of emergence by 

 the animals from cut de sacs, an assumption which does not do great violence to the 

 actual behavior of rats. 



