OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR CHOICE : SOME QUANTITATIVE RELATIONS