Predictions concerning aggression displacement, derived from Miller's conflict model, were investigated under conditions in which subjects' inhibitions about aggressing against an attacker were manipulated. Half of the high-attacked subjects were placed in a high-inhibition situation, designed so that strong inhibitory tendencies competed with strong aggressive tendencies; the remaining high-attacked subjects were placed in a low-inhibition situation, designed so that weak inhibitory tendencies competed with strong aggressive tendencies. Low-attacked control subjects also received the inhibition treatment. High- and low-attacked subjects were then confronted with one of four target persons, varying in similarity, toward whom they could aggress with electric shock.
As expected, under high inhibition, the target most similar to the attacker received more shocks from high-attacked subjects than did either the attacker or two less similar targets. Contrary to expectation, the attacker did not receive the most shocks under low inhibition; personality evaluations of the attacker suggested that high-attacked subjects in substitute target conditions may have been unintentionally angered further by being denied the opportunity for direct retaliation.