The conditional discrimination is a procedure the use of which is widely extended in the EAB (Experimental Analysis of Behavior), especially those known as "Matching to Sample". Although it has been used with a wide variety of species, the behavior of humans with verbal skills in these kinds of tasks may involve other control variables which are different from the scheduled contingencies of four terms. The aim of this work was to verify if conditional discriminations could be acquired, although reinforcement contingencies did not involve the sample. 109 psychology students, who were divided into three conditions, participated in the study. All of them were exposed to two blocks of training (A and B), with one sample and three comparisons, however, the sample did not really function as a conditional stimulus in 75% of the trials in block B. Simultaneity between sample and comparisons, as well as the requirement of a sample observation response, were manipulated resulting in three different conditions. The results showed no differences between acquisition speed in block A and block B in any condition, which suggests that the behavior of the participants was more controlled by the stimuli configuration than by the reinforcement contingencies.