THE ELIMINATION OF ERRORS IN THE MAZE 131 



of other conditioned reflexes as well as can an unconditioned 

 reflex (Bechterew, '13). If this condition obtains in the forma- 

 tion of a maze-habit, the situation involved in the last turn 

 before the food is reached may acquire from the getting of food 

 something (we can not be more definite at present) which makes 

 it capable of serving instead of the food in the fixation of the 

 next preceding activity. In this case the formation of a maze- 

 habit would appear as a series of secondary, tertiary, etc., con- 

 ditioned reflexes starting from the taking of food as the primary 

 reflex and progressing from the food compartment outward to 

 the entrance of the maze. Such an explanation avoids one of 

 the chief difficulties of theories of the fixation of habit; that of 

 accounting for the effects of getting food, or what not, upon an 

 activity which occurred half an hour or more earlier, such as 

 turning in a given direction at the entrance to the maze. 2 



But such an hypothesis must be looked upon with suspicion 

 until the fact of retroactive association is established, and there 

 is little evidence in favor of this at present. With a slightly dif- 

 ferent problem in view Hubbert ('15) has made an analysis of 

 the order of elimination of errors in the maze. The results of 

 this analysis were not very certain owing to confliction in the 

 data given by different groups of animals. One point was quite 

 clear, however; there is no invariable sequence in the elimina- 

 tion of errors, if the records of single animals are considered. 

 In contrast to this, when the averages of very large groups of 

 animals are taken there does seem to be a progressive elimina- 

 tion of errors from the food compartment to the entrance of the 

 maze. This is shown by the following averages based upon all 

 the data given by Hubbert. (See figure 1 for the designation of 

 the alleys.) 



a b c d e f 



30.6 26.4 19.7 19.7 18.7 8.3 trials. 



2 The theories of nervous drainage such as those formulated by Pawlow, Max 

 Meyer, Watson, and others seem to require the immediate succession of the acts 

 associated. In the experiments of Bonn, for example, they must assume that 

 the salivary reflex is excited while the nervous elements of the struggling reflexes 

 are still active and that the efferent fibres of the salivary glands drain off a part of 

 the energy from the cutaneous stimulation, thus opening the synapses from the 

 receptors of the skin to the salivary glands. None of the other theories of the 

 physiology of learning has been formulated in sufficient detail to take into con- 

 sideration the possibility of retroactive association. 



