the answer to this question is there is a cognitive component to operant learning.
The result of Tolmman's experiments shows that Rats are making less mistakes as it got used to the maze environment. This indicates that rats are actively process information rather than operating on a simple stimulus-response relationship like most people believe.