<p>In a matching-to-sample task employed to teach conditional relations among arbitrary stimuli, accurate performance can be achieved by selecting the correct stimulus (S⁺) and/or rejecting the incorrect one (S⁻). This study employed a matching-to-sample task with observing responses requirements (MTS-OR) to investigate the control exerted by the S⁺ and S⁻ during baseline training of arbitrary conditional relations and subsequent tests of derived equivalence relations. Three undergraduate participants were presented with stimuli that were normally covered but could be briefly revealed upon observing responses. All participants learned the baseline relations and successfully formed equivalence classes. Correct choices were predominantly controlled by the selection of S⁺, whereas the rejection of S⁻ was unsystematically observed. Whereas control by S⁺ consistently led to the formation of equivalence classes, control by S⁻ did not necessarily prevent their formation. During testing, indications of select and reject control have also been observed within the newly formed equivalence classes. These findings show that observing responses can effectively track the development of stimulus control during baseline training and suggest that select and reject controls might also act upon emergent relations, although the stimulus control during training does not predict control in tests of emergent relations.</p>