Deborah Smith and I 
performed some experiments 
with tropical fish to test this 
hypothesis. We placed a fish in 
a cylindrical glass tank that 
was surrounded by a drum lined 
with vertical stripes. The top 
was covered except for a small 
hole in the center through 
which the fish could be viewed. 
In one experiment, we rotated 
the tank about its center, thus 
generating a current because of 
friction, while keeping the 
surrounding drum stationary. 
This essentially simulates the 
situation of the fish in the 
river carried downstream by a 
current. We reasoned that the 
fish does not directly react to 
the water current but 
ordinarily responds to it only 
by its visual consequences—–
that is, being carried away from 
visible objects in its field. To 
prove this, we first wrapped the 
tank with white paper so that 
the fish could not see through 
the tank. Under these 
conditions, the fish made no 
effort to swim against the 
current going around the tank. 
When we removed the paper so 
that the fish could see the 
stationary stripes,
however, all fish swam 
vigorously against the current, 
thus remaining in place 
opposite any given set of 
stripes.