INTERNAL SECRETION IN LEARNING 63 



is conjectural. It can hardly be adrenalin, for this, as Cannon's 

 experiments seem to show, is the endocrinic correlate of excite- 

 ment, which is not conducive to habit formation, but rather to 

 the breaking down of habits. It must be a secretion, which like 

 adrenalin, is discharged directly into the blood (not indirectly 

 through the lymph channels), by which it is carried to the 

 " centers' 7 in which habits are formed, i.e., in which the critical 

 synapes lie: unless indeed the route may be still more direct as 

 seems hardly possible, even from the pituitary body. It may 

 be, however, that the secretion is not formed in a " gland" 

 proper, but in some tissue whose primary function is not 

 secretion. 



The effects of pain, in preventing the fixing of the preceding 

 acts, may not be so specific as are the effects of pleasure. It is 

 possible that adrenalin or some other active principle is the negat- 

 ing agent here, but it is also possible that the effects are produced 

 by the setting up immediately of more powerful reactions which 

 disturb the interconnections left by the preceding algesogenic 

 reaction. By "pain" is here meant the affective content usually 

 (and properly) described by this term, ignoring the unfortunate 

 psychologists' confusion between this and certain specific sen- 

 sations. 



The implications of the theory which admit of experimental 

 verification, or the reverse, are numerous. Those in which I 

 have been most interested in are the following. 



1. Actions performed shortly before the reaction which pro- 

 duces the " isatisf ying" result, and actions immediately follow- 

 ing it, would be fixed, along with the act itself. The normal 

 pause in activity following the " satisfying" reaction (where the 

 reaction itself is not immediately repeated) is probably a useful 

 phenomenon. 



2. If an animal, in solving a simple "problem" makes a short 

 series of reactions, including a number of "wrong" acts and 

 terminating with the correct (satisfying) act the probability of 

 repetition of the "wrong" acts is as great as that of the "right." 

 But after the solving several times, the probability of the" right" 

 act becomes greater than that of any "wrong" act unless a 



