144 VINNIE C. HICKS 



a new segment, they feel constrained to return to old and familiar 

 landmarks in order to relate the new part with those parts with 

 which they are familiar. 



The second proposition was accepted for several reasons. 

 Blinds differ in complexity and ease of learning, and it seems 

 appropriate to attempt some quantitative evaluation. A rat that 

 is lost and wanders hopelessly around until it returns to the 

 entrance box is surely guilty of a greater error than in the case 

 it returns but a short distance before getting its bearings. Evi- 

 dently the animal knows more about the maze in the latter case. 

 An animal may get lost in a complex blind and wander about 

 hopelessly. Surely this represents a greater error than in the 

 case the animal keeps its bearings and returns immediately 

 to the true path after exploring the blind. Watson has advanced 

 the thesis with a high degree of probability that each runwa}^ 

 with its entrance corner forms a characteristic kinaesthetic unit 

 which is the stimulus for the adaptive behavior to the succeeding 

 runways. The adoption of the runway as the unit of error is a 

 logical outcome of such a conception. This view emphasizes 

 the value of all turns or corners as characteristic landmarks 

 in the process of learning. Consequently an animal is checked 

 with an error whenever it makes the turn necessary to enter 

 an alley irrespective of the distance entered up to the first turn. 

 A year's observation of the behavior of the rat in the maze has 

 strengthened this conclusion in the writer's mind. A stud}' of the 

 graphic representations of the trials reveals the fact that the 

 turns are critical points in learning. The rat usually turns or 

 halts at the corners. When the animal becomes confused, he 

 generally picks up his cue at or near some corner. 



Curves representative of the three sets of data must be equated, 

 or reduced to some common denominator before any comparisons 

 are valid. There is much that is arbitrary in any graphic repre- 

 sentation of the learning process. For example, a time curve 

 can be constructed either "steepled" or "flat" from the same 

 set of data. The ordinate unit must be assigned an arbitrary 

 time value and it is evident that a more steepled curve will 

 result when the ordinate unit represents ten seconds than in 

 the case the ordinate unit is assigned a value of sixty seconds. 

 To avoid any purely arbitrary results, units of time, distance, 

 and error must be equated so that an ordinate unit will repre- 

 sent equivalent values for the three sets of data. 



