Neuroscientists reveal: When does an article go viral?

It is not only bloggers and online marketing experts who are concerned with the question: Why does one article spread virally like wildfire, but another - very similar - never goes viral? For the first time, neuroscientists at the University of Pennsylvania have now investigated which brain activities lead to an article being read and recommended (going viral) via social media. In fact, their method can even predict the virality of an article. "We all want to read or share content that relates to our own experience and self-image: who we are or who we want to be," says the study's director Dr Emily Falk. "We share with others the information that makes us seem smart, empathetic or otherwise in a positive light and that can improve our relationships with those we share with."

Making virality measurable

These are all very soft, subjective criteria. With this research, the scientists wanted to bring virality into a rational, measurable form. Certain regions in the brain must somehow determine how valuable we assess a piece of information and how "recommendable" it is for us. This value determined by the brain should translate directly into a measure of virality.

Measurement of brain activity

The neuroscientists measured with the help of the functional magnetic resonance imaging (fMRI, see below) Brain activity of 80 test subjects while they read the headlines and abstracts of about 80 articles. Each subject thus read about 80 headlines and abstracts of texts from the health section of the New York Times. They covered topics such as nutrition, fitness or healthy living and were between 21 and 35 words long. Each text was presented on the screen for between 8 and 12 seconds, depending on its length. Afterwards, the subjects were asked what the topic of the text was and whether they would read the whole article and recommend it via Facebook. In the fMRI image analysis, the researchers focused on three regions of the cerebral cortex. These are already known to be associated through previous studies:
  • with self-centred thinking,
  • with determining whether something is relevant to us and our lives
  • and with the idea of what others think.

What do people think when they share?

Of course it is obvious that a reader thinks of himself when he decides to read something. And that he thinks of others when he decides to recommend something. However, the fMRI measurements showed something quite different: whether the subjects decided to read something or whether they decided to recommend something, in both cases they thought of themselves as well as of others. In fact, it was even possible to measure the highest brain activity in both the self-related and the other-related neuronal systems when the subjects were thinking about recommending something to others. Or to put it a bit flippantly, when the subjects thought about recommending something, they thought at least as much about their own benefit. As the scientists found out, it is mainly the latter two brain areas that become active during reading (see below). The brain apparently decides on the basis of the signal strength in these two regions whether a piece of information - in this case an article from the New York Times - is valuable and will be read and shared or not.

Virality can be predicted

As the researchers also showed, it is even possible to predict the probability of an article going viral from measuring brain activity. To do this, the researchers measured the spread of the 80 articles tested via Facebook, Twitter or with emails by all New York Times readers. This resulted in a total of 117,611 shares in the 30 days after publication. In fact, the sharing rate of the articles could be correlated very precisely with the neuronal activity of the test subjects in the regions of the cerebral cortex discussed above. This was despite the fact that the subjects did not match the typical New York Times readers (whose sharing was measured) in age and other demographic factors (students aged 18 to 24). "We can predict how the total number of New York Times readers will behave with a relatively small number of brain scans," said first author Christin Scholz. The brain regions involved seem to be the same in all people. However, what we think about ourselves and about others when we read an article can be different, of course you can't see that in the scan. One person may think the article makes his friend laugh, another recommends it because it helps his friend with a problem or because he simply thinks his friends find it interesting. Nevertheless, the neuronal activities are very similar in all cases and can serve as a common denominator for different types of social and self-referential thinking.

After this study, a reader will then spread an article further:

  • if it casts a positive light on him in any way. I.e. if it makes him appear more intelligent, humorous or empathetic.
  • And if it improves its relationship with the target audience in any way.

How are the results to be classified?

What is new in this work is not the results, i.e. what makes an article viral, that had already been established in the past through simple surveys. What is new is that the virality of an article can actually be predicted by objective brain measurements. This is interesting, but the practical use is limited, because who can use magnetic resonance imaging for their content creation. But the results imply that whether an article goes viral or not is apparently a general phenomenon. "The fact that an article strikes the same chords in very different brains suggests that very similar motivations and similar norms drive these behaviours," the researchers said.  

Background information

Functional magnetic resonance imaging (fMRI)

The fMRI is an imaging procedure that not only shows the organic structures, but also the functional states of the organs. So in the brain, you not only see the individual brain regions, but also their neuronal activity with high resolution. The blood flow is measured, which increases when a brain region is activated.  

The brain regions studied

In this study, the neuroscientists found increased nerve activity in two brain regions in particular:
  • the medial prefrontal cortex (MPFC)
  • the posterior cingulum (PC).
The prefrontal cortex is a part of the frontal lobe of the cerebral cortex. It performs so-called executive functions, such as anticipation of action consequences, planning of future actions, action control, problem solving. The posterior cingulate (Brodmann area 23) is a part of the cingulate gyrus, a brain turn of the cerebral cortex located in the middle of the brain (in the gap between the hemispheres). It is considered the centre of consciousness of the brain, but this is a gross oversimplification because it has many other functions.


Christin Scholz, Elisa C. Baek, Matthew Brook O'Donnell, Hyun Suk Kim, Joseph N. Cappella, and Emily B. Falk: A Neural Model of Valuation and Information Virality. In PNAS. Published online February 27 2017 doi:10.1073/pnas.1615259114 Brain Images Of Sharing Predicts Which Articles Go Viral Science 2.0 Blog, 27.02.2017