In studies led by Piotr Winkielman, associate professor of psychology at the University of California, San Diego, people altered their consumption behaviors after exposure to subliminal facial expressions.
Hidden smiles persuaded thirsty subjects to pour more and drink more of an unidentified beverage than did neutral expressions. Frowns had the opposite effect.
Study participants who were unconsciously "primed" with happy faces also reported being willing to pay up to triple the price for the mystery drink. And they reported wanting another half cup instead of just a sip or two more.
Remarkably, the test subjects, whose actions had been influenced by these emotional cues, were not aware of their feelings having changed.
"This is the first demonstration that you can influence consequential, real-world behavior without affecting conscious feeling. We can change what you do, without changing how you feel," Winkielman said.
Winkielman, coeditor of the forthcoming book Emotion and Consciousness, believes the findings, presented at the American Psychological Society annual convention in Los Angeles, May 26-29, support the existence of unconscious or "unfelt" emotion.
"Emotional states operating outside conscious awareness can drive behavior. The subjective experience of a feeling is not always necessary to the process," said Winkielman.
"Feelings are often slow," he said. "In a frightening situation, you run first, feel afraid later."
To tease apart emotional reaction and subjective feeling, Winkielman and colleagues devised two different experiments.
In both studies, subjects were first asked to rate how thirsty and hungry they were. Next, they were subliminally exposed to a series of photographs of happy, angry or neutral faces - masked each time by a neutral face. Consciously, the subjects were aware only of seeing the second, neutral image, which they were then asked to classify as male or female. Immediately afterwards, they were asked (in varying order) to interact with the beverage and rate their moods.
Happy and angry expressions were selected as primes, Winkielman said, because it is easy to extract a simple positive and negative interpretation from them: Grins and glowers are flashed at us in approval and reproach since Day 1 and are essentially equivalent to "stop" and "go" signs.
The researchers chose drinking in part because ingesting an unknown substance can have obvious biological consequences and is therefore not a trivial act - even if, as in this case, the drink is made of nothing more than water, sugar and lemon-lime Kool-Aid.
In the first experiment, 39 undergraduates freely helped themselves and drank as much as they wanted. Unknown to them, the amounts poured and consumed were recorded using an electronic scale. Thirsty participants poured and drank more than twice the amount of the beverage after happy primes than after angry primes.
In the second experiment, 29 undergraduates tasted a small, predetermined sample and were then asked to evaluate it after one sip. Those at the high level of thirst reported willingness to pay 38 cents (U.S.) after happy primes and only 10 cents after angry ones. They also expressed desire for an additional half cup instead of one to two sips.
In both studies, thirst proved a necessary precondition for influence. Moderately thirsty participants were only moderately affected. And those not thirsty, not at all. Thirst also correlated positively with ratings of the beverage's deliciousness and thirst-quenching abilities.
"Motivation matters," Winkielman said. "Your motivational state - your level of need - prepares you to process relevant information and gives value to the stimulus. Otherwise, the emotional message falls on deaf ears."
To business-people or politicians tempted to apply these findings to advertising, Winkielman says: It won't work. The effects of subliminal expressions were too short-lived. By the time people arrived at the store or polling booth, all influence would have worn off.
The studies were supported by a grant from the National Science Foundation. Results were published earlier this year in the Personality and Social Psychology Bulletin, with coauthors Kent C. Berridge of the University of Michigan and Julia L. Wilbarger of the University of Wisconsin-Madison.