Two pigeons were exposed to a 3-sec delayed matching-to-sample (DMTS) procedure. Following acquisition, reinforcement for sample responding was presented during the delay interval. After baseline recovery, a blackout procedure was instituted during the delay interval. Both procedures reduced the accuracy of responding on the DMTS task. Accuracy was below that seen in baseline sessions. In addition, reductions in sample-specific behaviors were obtained with both procedures. The data support and extend the notion that sample-specific behaviors mediate accurate DMTS performance.