New research suggests memorable images are more likely to go viral online

The Brain Bridge study argues that memorability has a greater impact on virality than emotion.

By Sarah Steimer

If you see a photo of a beautiful beach, along with a photo of a watermelon carved to resemble a reptile’s head — which are you more likely to remember? It’s probably the latter, and you wouldn’t be alone. In fact, a new study published in Computers in Human Behavior shows that high memorability may be the reason the watermelon image would go viral.

Wilma Bainbridge, associate professor in the Department of Psychology, and her Brain Bridge Lab focus their research on questions of perceptual inputs and outputs of the human memory system. To further explore what makes certain images more memorable than others, Bainbridge and her coauthor Shikang Peng — a former MA student at the lab and now a PhD student at University of Toronto — looked to social media. They were curious: Do memorable images have different fates on social media than forgettable images?

The researchers focused their study on Reddit, which makes its data publicly available. Specifically, they explored image-based subreddits, judging 1,247 Reddit image posts across three timepoints using ResMem, a specialized AI model trained with human data to predict an image's memorability score. Bainbridge notes that ResMem doesn’t have any cultural measures, offering only an objective measure of memorability. The researchers then compared the memorability scores with the images’ virality, i.e., how many people commented on or “upvoted” — a form of recommending — the posts.

“The first thing we found was that, across all three time points, more memorable images resulted in more comments,” Bainbridge says, adding that the correlation between memorable images and upvotes was less pronounced. “This confirms this idea that more memorable images cause more discussion.”

The researchers then studied the content of the comments for memorable and forgettable image posts. They ran a few different analyses — many using computational linguistic methods — on the post discussions. They found that memorable image comments tend to be more emotionally neutral, while forgettable images tend to have very positive comments. 

“Even if something elicits positive discussion,” Bainbridge notes, “it doesn't necessarily stick in your memory.”

The team also noticed that when people were commenting on memorable images, they talked about more abstract ideas beyond the image. Whereas forgettable images — perhaps a beautiful beach — commonly led to people saying nice things, but rarely resulted in discussion beyond the photo. 

“But for memorable images, we found there was more abstract language,” Bainbridge says. “They might talk about political views or thoughts that the image elicits — or humor, (they might) make a joke. They also talked about more things that were not directly in the image: Maybe this image would make them think about another event that happened to them or something else that they saw. So in a way, the memorable images elicited richer discussion.”

For the study’s practical application, the findings can help steer people toward what type of image to use, based on their goals. For example, an influencer may want to use a memorable image if their goal is to go viral and drive conversations. But if an advertiser wanted to focus on attracting positive comments, they may want to choose a beautiful but forgettable image.

“A lot of the work done on testing what makes something viral has purely focused on emotion,” Bainbridge says. “Here we actually find a totally different factor that impacts virality is memorability, and we even almost find an opposite effect of this emotion work: Some memorable images cause less emotional commenting.”

The study’s findings, she says, further support the idea that memorable images cause more engagement, which then may cause them to become more famous, viral, or long-lasting.