THE ASSOCIATION OF IDEAS 49 



(not necessarily simplified) by the similarity of the reactions, and 

 possibly by the similarity of the stimulations. In learning sig- 

 nificant series, of prose or poetry, still greater complications oc- 

 cur, as the words are not associated as simple terms but as signifi- 

 cant groups: that is, complex serial habits already formed are 

 woven into a new series with attendant inhibitions and facilita- 

 tions. If the several reactions themselves have to be learned as 

 well as the serial connection, as in memorizing series of " nonsense 

 syllables" such as nef, yok, pib, ruz, mog, zal, the learning is more 

 complicated than in learning a series of words already familiar. 11 



The probable changes in the mechanism of serial connections 

 during the learning period, to which we have already referred, 

 must now be considered more concretely. Complete reactions 

 are apparently essential during the learning process; during the 

 learning of serial connections as well as the learning of percep- 

 tions. Action, perception, and thought, in so far as they are 

 learned at all, are learned through action. There is however no 

 evidence that in the final stages of thought-habits and perception- 

 habits, action is essential: nor is it probable that in serially con- 

 nected activities, as in knitting or waltzing, the completion of one 

 action in the chain is the actual stimulus for the next, after the 

 series has been learned to a reasonable degree of efficiency. It 

 may possibly be, of course, that complete reactions always occur, 

 and that each member of the series waits on the completion of the 

 preceding one: but the facts do not at present appear to be such, 

 and such a mechanism, however admirable for purposes of learn- 

 ing, seems needlessly clumsy and inefficient for the finished result. 



It is probable that during the learning process for serial activi- 

 ties and the association of ideas, short circuits are progressively 

 developed, and simplify the reaction chains very much. It is not 

 safe to extend the reaction hypothesis, as applied to learning 

 processes, beyond the learning period; and it is necessary to search 



11 This is an important point which has to an unfortunate extent been over- 

 looked. We can not expect the learning curve for nonsense syllables to follow in 

 all respects the laws of the simpler process of learning significant words, any more 

 than we can expect serial association and paired association to give exactly simi- 

 lar curves although there is even a flagrant example of the latter expectation in 

 the German literature. 



PSTCHOBIOLOGY, VOL. II, NO. 1 



