when participants were thirsty. Specifically, thirsty participants poured significantly more drink from the pitcher and drank more from their cup after happy faces than after angry faces (Study 1). Thirsty participants were also willing to pay about twice as much for the drink after happy than after angry expressions (Study 2). The modulating role of thirst indicates that unconscious emotional reactions acted through basic biopsychological mechanisms that determine reactions to incentives, such as a drink, rather than through cognitive mech- anisms influencing interpretation of the stimulus (Berridge & Win- kielman, 2003; Winkielman et al., 2002).
In summary, the studies just described show that subliminally presented emotional faces can cause affective reactions that alter consumption behavior, without eliciting conscious feelings at the moment the affective reactions are caused. Because the influence of emotional faces on consumption behavior was observed also for those participants who rated their feelings immediately after the subliminal presentation of the faces, these results cannot be explained by failures of memory. Thus, we propose that these results demonstrate unconscious affect in the strong sense of the term—affect that is powerful enough to alter behavior, but that people are simply not aware of, even when attending to their feelings.
Support From Evolution and Neuroscience From the standpoint of evolution and neuroscience, there are good reasons to suppose that at least some forms of emotional reaction can exist independently of subjective correlates. Evolutionarily speaking, the ability to have conscious feelings is probably a late achievement compared with the ability to have behavioral affective reactions to emotional stimuli (LeDoux, 1996). Basic affective reactions are widely shared by animals, including reptiles and fish, and at least in some species may not involve conscious awareness comparable to that in humans. The original function of emotion was to allow the organism to react appropriately to positive or negative events, and conscious feelings might not always have been required.
activates opioid receptors is injected into the nucleus accumbens (a reward-related structure at the base of the front of the brain). Liking reactions to sugar can even be enhanced by injecting a drug that activates other receptors into the brain stem, which is perhaps the most basic component of the brain. Such examples reflect the per- sisting importance of early-evolved neurocircuitry in generating be- havioral emotional reactions in modern mammalian brains (Berridge, 2003; LeDoux, 1996). In short, evidence from affective neuroscience suggests that basic affective reactions are mediated largely by brain structures deep below the cortex, raising the possibility that these reactions might not be intrinsically accessible to conscious awareness.
KEY QUESTIONS FOR FUTURE RESEARCH
As we have argued, there are good theoretical reasons why some emotional reactions might be unconscious, and we suggest that our recent empirical evidence actually provides an example. However, several critical issues need to be addressed by future research.
The studies discussed here focused only on basic liking-disliking, so it is possible that the crucial property of unconscious emotion is simply positive-negative valence, rather than qualitative distinctions associated with categorical emotion (fear, anger, disgust, joy, etc.). However, evidence suggests that subcortical circuitry may be capable of some qualitative differentiation. For example, human neuroimaging studies reveal differential activation of the amygdala in response to consciously presented facial expressions of fear versus anger (Whalen, 1998). If future research shows that subliminally presented expres- sions of fear, anger, disgust, and sadness can create qualitatively different physiological and behavioral reactions, all without conscious experience, then there may indeed exist implicit affective processes deserving the label ‘‘unconscious emotion’’ in its strongest sense. Studies that simultaneously measure psychophysiology, behavior, and self-reports of emotion could be particularly useful to address such issues (Winkielman, Berntson, & Cacioppo, 2001).
The neurocircuitry needed for basic affective responses, such as a ‘‘liking’’ reaction1 to a pleasant sensation or a fear reaction to a threatening stimulus, is largely contained in emotional brain struc- tures that lie below the cortex, such as the nucleus accumbens, amygdala, hypothalamus, and even lower brain stem (Berridge, 2003; LeDoux, 1996). These subcortical structures evolved early and may carry out limited operations that are essentially preconscious, com- pared with the elaborate human cortex at the top of the brain, which is more involved in conscious emotional feelings. Yet even limited subcortical structures on their own are capable of some basic affective reactions. A dramatic demonstration of this point comes from affective neuroscience studies with anencephalic human infants. The brain of such infants is congenitally malformed, possessing only a brain stem, and lacking nearly all structures at the top or front of the brain, in- cluding the entire cortex. Yet sweet tastes of sugar still elicit positive facial expressions of liking from anencephalic infants, whereas bitter tastes elicit negative facial expressions of disgust (Steiner, 1973).
Even in normal brains, the most effective ‘‘brain tweaks’’ so far discovered for enhancing liking and related affective reactions all involve deep brain structures below the cortex. Thus, animal studies have shown that liking for sweetness increases after a drug that
1We use the term ‘‘liking’’ to indicate an unconscious reaction, not a con- scious feeling of pleasure.
The studies discussed here employed basic affective stimuli, such as subliminally presented facial expressions, to influence emotional behavior without eliciting conscious feelings. Future studies might address whether more complex stimuli that derive their positive or negative value from a person’s cultural environment can also influence emotional behavior without eliciting any accompanying feelings. A related question concerns whether stimuli presented above the threshold of awareness can also change emotional behavior and physiology without influencing feelings.
The studies described here suggest that under some conditions emotional reactions are genuinely unconscious. But obviously many emotional states are conscious, even when elicited with subliminal
¨ stimuli (Monahan et al., 2000; Ohman et al., 2000). What determines
when a basic emotional reaction is accompanied by conscious feel- ings? Is it possible for even a strong emotional reaction to be un- conscious? What are the neural mechanisms by which emotion is made conscious? How do behavioral consequences of conscious and unconscious reactions differ?
Finally, a question of practical importance to many emotion re- searchers, as well as clinicians, concerns the meaning of people’s reports of their own emotions. The existence of verifiable but un- conscious emotional reactions does not mean that subjective feelings are merely ‘‘icing on the emotional cake.’’ At least, that is not our view. We believe that self-reports of feelings have a major place in emotion
Volume 13—Number 3