A new Stanford University study is suggesting the global popularity, or virality, of a video on the internet can be predicted by looking at how certain areas of a person’s brain are activated while they view the first few seconds of footage. The technique has been dubbed neuroforecasting.
“In many of our lives, every day, there is often a gap between what we actually do and what we intend to do,” explains Brian Knutson, a Stanford neuroscientist who has, for several years, been investigating how brain imaging can help predict what we buy or what becomes popular. “We want to understand how and why people’s choices lead to unintended consequences – like wasting money or even time – and also whether processes that generate individual choice can tell us something about choices made by large groups of people.”
Prior research from Knutson and his team has focused on using brain imaging to predict music sales, crowdfunding support and general money-spending activities. The research ultimately generated a framework called AIM (affect-integration-motivation) using neural measures to predict spending behaviors.
The new study set out to explore whether the AIM framework would be effective in predicting how people choose to allocate their time. To do this, the researchers focused on video engagement and popularity, recruiting 36 individuals to watch videos while being scanned with fMRI.
Participants were shown 32 videos and given the choice to stop the video at some point during the first four to eight seconds. As well as constantly gathering brain activity data from the fMRI scans, the study captured subjective data from all the participants, including whether they “liked” each video, and how each video made them feel.
The previously established AIM framework homed in on four particular brain regions proven to be associated with spending behaviors: the nucleus accumbens (NAcc), the anterior insula (AIns), the medial prefrontal cortex (MPFC), and the posterior cingulate cortex (PCC). It was these areas the new study focused on imaging.
The results revealed specific brain activity during the first four seconds of a video could effectively predict a person’s choice to continue watching the video and predict a person’s feelings towards that video. However, the subjective opinions of people as to what videos they thought would be the most popular did not correlate with the videos that were ultimately the most popular.
Instead, activity in two particular brain regions was the most accurate metric in predicting which videos were the most popular, in terms of YouTube view numbers. Increased activity in the NAcc and decreased activity in the AIns during the first seconds a video could effectively forecast the aggregate view frequency better than any other metric. These two brain regions are related to anticipatory affect, the emotional state we enter momentarily when we are uncertain of the future outcome from a given stimulus.
“If we examine our subjects’ choices to watch the video or even their reported responses to the videos, they don’t tell us about the general response online,” says Knutson. “Only brain activity seems to forecast a video’s popularity on the internet.”
Interestingly, the study suggests we choose how we allocate our time in ways very similar to how we choose to allocate our money. However, the study also raises a whole new host of questions as to what specific kinds of content could be triggering these brain regions in the first few seconds of a video playing and, of course, the literal million-dollar question, how can content creators leverage these findings to make their videos more popular?
“Future research might also systematically deconstruct and label dynamic video content to determine whether specific video features influence aggregate engagement,” the published study concludes, pointing to where the research may go next.
But one of the fundamental ideas to take away from this fascinating study is brain imaging data can sometimes offer insights into future group trends that isn’t reflected in subjective self-reporting or observed behaviors.
“Here, we have a case where there is information contained in subjects’ brain activity that allows us to forecast the behavior of other, unrelated, people – but it’s not necessarily reflected in their self-reports or behavior,” concludes Lester Tong, one of Knutson’s graduate students working on the project. “One of the key takeaways here is that brain activity matters, and can even reveal hidden information.”
The new study was published in the journal PNAS.
Source: Stanford University
