242 THE INTELLIGENCE OF MAMMALS 
events is too great. But if (A) can be associated with (B) 
over an interval of ten seconds why can it not be directly 
associated with the impulse (C) over an interval but slightly 
greater? Ten seconds after (A) the cat sees (B), eleven 
seconds after she performs (C). That the cat had pre- 
viously associated (B) and (C) does not necessarily play 
any part in the process. The experiment proves only that 
some sort of a neurosis persists from (A) during ten or more 
seconds of time; but it fails to afford any proof of the réle of 
the idea (B) in the process. It shows the persistence of 
impressions, not the association of ideas. Indirectly it 
may support the theory of association. If the neurosis of 
(A) persists and is accompanied by consciousness in the 
form of an idea of (A), ideas of (B) probably occur also, and 
if ideas occur why may they not become associated as well 
as impressions and impulses? Such ideas may be more like 
after-images in ourselves, but no sharp line can be drawn 
between the latter and ideas properly so called. While 
such an experiment as the one described may not prove the 
association of ideas it may serve to make association more 
probable by showing the persistence of impressions. 
The experiments of L. W. Cole on the raccoon yielded 
better evidence of the existence of ideas than the investiga- 
tions of Thorndike, owing perhaps to the greater degree of 
intelligence of the animals employed. Raccoons learned to 
get out of a box with seven fastenings consisting of two 
buttons, two loops, a thumb latch, a treadle and a hook. 
They were first put into boxes with one or two of these 
devices, and when these were learned others were added until 
the above combination was reached, which seemed to be 
about the limit of a raccoon’s learning capacity. The rac- 
coons in attacking the fastenings did not take them up in 
any constant order in successive trials, but they showed a 
