LABYRINTH HABITS OF THE PIGEON 



291 



heighth of the curve at the first or second trial. The learning 

 curve, e. g., of a problem the chief points of which an animal 

 has failed to master, would never descend permanently below 

 the height to which reference has just been made. The mere 



T 



na-Ls 





Figure 9— Graphs from the data of table VII, labyrinth C. T, the time curve; 

 E, the error curve. 



fact that an animal has made all of the possible errors is not 

 sufficient to produce a lowering of the curve. If such were 

 the case, it would be difficult to get a graph that would not 

 descend. The animal must recognize an error as an error and 



