THE INTELLIGENCE OF MAMMALS 243 
good memory for any combination they may have learned. 
Three raccoons which had learned to open the box with 
seven fastenings were put into the box again after an interval 
of 147 days. Only one individual succeeded in undoing all 
of the seven fastenings and escaping. His times in four 
successive trials were 34, 28, 131, and 182 seconds. The 
other two worked most of the fastenings, but generally 
failed to undo a horizontal lock. Davis found that rac- 
coons remembered how to undo the fastenings of a puzzle 
box for over a year. 
One significant feature of the raccoons’ method of attack 
on their problems is that they employ different means of 
accomplishing the same result. They worked fastenings 
with either the right or the left paw or with both together. 
“All of the raccoons turned a button once or twice with 
the nose in early trials, then settled down to working it with 
the paw.” This looks very much as if there were something 
besides the sensori-motor associations assumed by Thorn- 
dike. We cannot say that there is nothing but the asso- 
ciation of the sight of a certain object with a particular 
impulse to movement, if in effecting a certain change an 
entirely different organ is substituted for the one pre- 
viously employed. When an animal moves with its paw a 
fastening it formerly moved with its nose it gives evidence 
of being guided by an idea of what it is setting out to 
accomplish. 
Evidence for the existence of ideas was derived also from 
other sources. It was found that raccoons were able to 
associate being in a box with getting food after they came 
out, so that when they were dropped in, as in the experi- 
ment with the cats, they came after a while to go into the 
box of their own accord. The motor impulse to enter the 
box was not in these cases an element in the association 
