What every skeptic should know about subliminal persuasion

Subliminal Influences on Behavior

Can subliminally presented stimuli influence behavior? Recent investigations suggest that the answer may be yes. For example, Neuberg (1988) has argued that subliminally presented stimuli can influence behavior indirectly, by way of activating concepts that can influence the way individuals interpret the behavior of others. These interpretations, then, can lead individuals to opt for certain behavioral responses. For example, if the concept of hostility were activated subliminally, and caused individuals to "read" hostility into the behavior of others, these individuals might then choose to adopt a hostile course of action themselves. Though such an indirect effect is a far cry from the mindlessly acquiescent behavior conjured by the words "subliminal advertising," it nonetheless would represent an instance of subliminally presented stimuli affecting behavior.

To test this hypothesis, Neuberg confronted participants with a "Prisoner's Dilemma," an exercise in which individuals must choose to either cooperate or compete with another participant (Luce and Raiffa 1957). Before choosing, participants completed questionnaires designed to assess their proclivity toward cooperation versus competition, and were exposed subliminally to either neutral words (e.g., home, water, sound) or competition-related words (e.g., hostile, adversary, cutthroat). Although the primes did not influence the behavior of those with a cooperative orientation, participants predisposed to compete did so to a greater extent when they were exposed to competitive words than when exposed to neutral words (Neuberg 1988).

More recently, Bargh and colleagues have provided even more compelling evidence that subliminal stimuli can influence behavior (Bargh et al. 1996; Bargh 1997). In contrast to Neuberg's notion of an indirect influence on behavior, Bargh suggests that subliminally presented stimuli can influence behavior directly - that the influence is unmediated by conscious thought and results from a direct perception-behavior link that operates not unlike a reflex. How might this hypothesis be put to the test? Previous research has established that exposing white participants to words stereotypically associated with African Americans tends to automatically activate the concept of hostility (Devine 1989). To find out if such exposure might also induce hostile behavior, Bargh and colleagues (1996) asked participants to perform a tedious task on a computer. Unbeknownst to the participants, the computer not only administered the task but also exposed them, subliminally, to photographs of either black or white faces. Then, after many trials, the computer presented them with a bogus error message - "F 11 error: failure saving data" - and informed them that they would have to start the task again from the beginning.

Participants' reactions to this news were videotaped using a hidden camera and were rated by judges (who were unaware of the participants' experimental condition) to determine the amount of hostility they exhibited. Results indicated that those exposed to black faces did indeed respond in a more hostile, frustrated manner than those exposed to white faces.