The backward elimination of errors in mental maze learning.
Joseph Peterson · Journal of Experimental Psychology · 1920
In another paper the writer has shown that an animal through sheer probability will, with a continued series of efforts, reach the food box in a maze; and further, that on the basis of frequency of exercise (if we grant the operation in learning of certain other factors appearing to be indispensable to learning) the errors of entrance to blind alleys should be gradually eliminated in what has been called the backward order, that is, the errors near the food box should first disappear, and the others should be eliminated in order, speaking generally, from this point back toward the entrance place in the maze.In the earlier paper errors were interpreted as entrances to blind alleys; return runs not bringing about such entrances were not counted in the error score.It was shown that the first blind alley is passed, on the basis of probability solely, % of the times that it is encountered in forward runs by an animal, and that the animal is turned back by it toward the entrance 34 of the times.The next blind is likewise passed % of the times that it is encountered, or M of ^ of th e number of times that the first blind is reached.Speaking generally, in n trials the mth blind from the entrance would be passed in forward runs (%)"* of the number of times that the first blind is reached.The tenth blind, for example, would be passed (%) 10 -or a little more than 1/18-of the number of times that the first is encountered.This matter is, of course, vastly complicated by partreturn runs.Disregarding these part-returns for the sake of simplicity, and assuming that each return run will take the 1 Read in part in Section H of the A.