THE PERCEPTION OF THINGS. 83 



takes coDsist in using entire wrong words instead of right 

 ones. Only in tlie gravest lesions does he become quite in- 

 articulate. These facts show how subtle is the associative 

 link ; how delicate yet how strong that connection among 

 brain-paths which makes any number of them, once excited 

 together, thereafter tend to vibrate as a systematic whole. A 

 small group of elements, ' this,'' common to two systems, A 

 and B, may touch off A or B according as accident decides 

 the next step (see Fig, 47). If it happen that a single point 

 leading from ' this ' to B is momentarily a little more per- 

 vious than any leading from ' this ' to A, then that little 

 advantage will upset the equilibrium in favor of the entire 

 system B. The currents will sweep first through that point 



Fig. 47. 



and thence into all the paths of B, each increment of ad- 

 vance making A more and more impossible. The thoughts 

 correlated with A and B, in such a case, will have objects 

 different, though similar. The similarity will, however, 

 consist in some very limited feature if the ' this ' be small. 

 Thus the faintest sensations will give rise to tJie perception 

 of definite things if only they resemble those ichich the things 

 are icont to arouse. In fact, a sensation must be strong and 

 distinct in order not to suggest an object and, if it is a non- 

 descript feeling, really to seem one. The aurpe of epilepsy, 

 globes of light, fiery vision, roarings in the ears, the sensa- 

 tions which electric currents give rise to when passed through 

 the head, these are unfigured because they are strong. 

 Weaker feelings of the same sort would probably suggest 

 objects. Many years ago, after reading Maury's book, Le 

 Sommeil et les Reves, I began for the first time to observe 

 those ideas which faintly flit through the mind at all times, 

 words, visions, etc., disconnected with the main stream of 

 thought, but discernible to an attention on the watch for 



