The Science of Viral Videos
What activity in one brain region can tell us about popular online content.
Posted Mar 13, 2014
The last time you sent your friend a link to a video, did you think about why? You probably looked at many stories that day; what made that one worth sharing?
Philosophers have grappled with similar questions for millennia. Aristotle suggested that persuasive ideas shared three traits: They must be credible, elicit emotion, and make sense. More recently, marketers and psychologists have studied why certain videos and articles get passed along on the Internet while others get passed over. Aristotle probably wouldn’t have been surprised to learn that articles that excite a reader and have a positive message are more likely to reach the New York Times website's most-emailed list.
Only in the past few years, however, have we had the chance to get a glimpse of the neural underpinnings that drive us to watch silly cat videos. At a recent talk, Stanford neuroeconomist Brian Knutson mentioned a study that examined whether brain activity can offer clues about how popular something is likely to become. A study by Emory neuroeconomist and PT blogger Gregory Berns suggests it can.
In Berns' study, 32 teenagers completed a functional MRI scan during which they listened to 15-second song clips. Berns also asked them if they liked the song and how familiar it sounded. He then followed the songs three years later to see which became hits.
First, Berns found that subjective liking was associated with greater brain activation in three regions: the ventral striatum, the ventromedial prefrontal cortex (vmPFC), and the cuneus. These regions have been shown in previous studies to encode value in the brain. For example, people have a greater ventral striatum response to the possibility of winning $5 rather than $1. Similarly, they have more activation in the vmPFC when receiving $5 than when receiving $1. So it makes sense that these regions would have more activation when someone reports liking a song.
When Berns looked at sales of music, he found no relationship between copies sold and his subjects' subjective declarations about whether they liked a song. How much the teenagers reported liking the music actually turned out to be a poor predictor of sales.
When Berns looked at brain activity, however, a different pattern emerged. The activity in those 32 teen brains while listening to songs was linked to how many people bought that music over the next three years. The songs that elicited more activity in the ventral striatum racked up more sales.
As exciting as that relationship appears, there are several reasons to interpret the results with caution. For one, Berns didn’t use a robust prediction model or out-of-sample data. Using out-of-sample data would help validate the relation between brain activity and music sales. A recent paper suggests failure to use out-of-sample data leads neuroimaging studies to generate overly optimistic models. It is unclear how well the pattern reported in the study will generalize to other groups.
Further, we don’t know if brain activity represents unique predictive information. It could simply be a biological sign of something we already know predicts popularity, like how much the song elicits emotion. Future studies should address this.
Nonetheless, the results do give us reason for enthusiasm. A number of other recent studies have examined what makes music or videos go viral, but this is the first to explore the brain’s role. The results provide initial evidence that ventral striatum activity from a few people when listening to a song could be a telltale signal of whether other people will buy it.
That idea that woke me in the middle of the night had germinated a few days prior, when Brian Knutson told me about Berns’ study. I wonder what happened in my brain when I first heard about the study and if it would indicate that I’d later want to pass the finding along to a friend—and now to you.
And how likely are you to share this post? If you just had a spike of ventral striatum activity, the chances may be good.
Image credit: ElizaC3
Thanks to Julie Rodriguez for suggestions and to Phil Collins for ideas. Thanks to Greg Berns for sending me a copy of his paper and to Brian Knutson for telling me about it.