Scientific inspiration sometimes comes from unlikely sources. Two years ago, Gregory Berns, a neuroeconomist at Emory University in Atlanta, was on the couch with his kids watching American Idol. One of the contestants sang the melancholy hit song "Apologize" by the alternative rock band OneRepublic, and something clicked in Berns's mind.
He'd used the song a few years earlier in a study on the neural mechanisms of peer pressure, in this case, how teenagers' perceptions of a song's popularity influence how they rate the song themselves. At the time, OneRepublic had yet to sign its first record deal. A student in Bern's lab had pulled a clip of "Apologize" from the band's MySpace page to use in the study. When Berns heard the song on American Idol, he wondered whether anything in the brain scan data his team had collected could have predicted it would become a hit. At the time, all 120 songs used in the experiment were by artists who were unsigned and not widely known. "The next day, in the lab, we talked about it."
To find out what had become of the songs, the lab bought a subscription to Nielsen SoundScan, a service that tracks music sales. The database contained sales data for 87 of the 120 songs (not surprisingly, many songs had languished in MySpace obscurity). Berns reexamined the functional magnetic resonance imaging scans his group had collected from 27 adolescents in 2007, looking for regions of the brain where neural activity during a 15-second clip of a song correlated with the subject's likeability ratings. Two regions stood out: the orbitofrontal cortex and the nucleus accumbens. "That was a good check that we were on the right track, because we knew from a ton of other studies that those regions are heavily linked to reward and anticipation," Berns says.
Next, the researchers looked to see whether the activity in either of these two brain regions, averaged across subjects for each song, correlated with the song's sales through May 2010. It did, Berns and co-author Sara Moore report in a paper in press at the Journal of Consumer Psychology. The correlations were statistically significant but modest. Activity in the nucleus accumbens, the best predictor of song sales, accounts for about 10% of the variance in sales, Berns says. "It's not a hit maker," he cautions.
Intriguingly, the brain scan data predicted commercial success better than the subjects' likeability ratings, which did not correlate with sales. "What is new and interesting about this study is that brain signals predict sales in a situation where the ratings of the participants don't," says John-Dylan Haynes of the Bernstein Center for Computational Neuroscience in Berlin. Although several recent studies have shown it's possible to predict consumer choices from brain activity, Haynes says, it hasn't been clear whether brain scans can reveal anything about people's product preferences that couldn't be gained by simply asking them. In this case, at least, it seems they can.
"This is a really cool result," says Brian Knutson, a cognitive neuroscientist at Stanford University in Palo Alto, California. Showing that brain activity in a small group of people can predict the buying behavior of a much larger group of people is a novel and provocative finding, he says. But how does it work, and why would brain activity be better than the subjects' ratings? Knutson suggests that activity in the nucleus accumbens may provide a more pure indication of how much people actually want something, unencumbered by economic and social considerations that might influence their ratings—for example, whether one's credibility as a hard-rocking heavy metal fan would be undermined by a fondness for, well, "Apologize."
There have been many dubious claims about "neuromarketing" strategies for using brain activity to assess consumer sentiment, says Antonio Rangel, a neuroeconomist at the California Institute of Technology in Pasadena. He sees the new study as an exciting proof of principle that in some cases neuroimaging can provide useful information not picked up by traditional methods such as consumer surveys and focus groups. Still, Rangel says, it's a long way from being a viable marketing tool. "I would not invest in a company based on this."