How COVID masks messed with our minds

How the brain processes the emotional expressions of masked faces

In spring 2021, when public schools in Los Angeles finally reopened after a full year of pandemic closures, my kids went back to finish out the last few months of the year. After a year of Zoom school misery, the scene after dropoff looked a little like this.

Although I was elated that my kids could see their friends again in person, the vibes at school were very weird. Recess zones were cordoned off and spaced out, parents could no longer walk freely onto campus to drop off and pick up their kids, and most notably, everyone wore masks.

Mandatory masking continued through the following fall. I taught an undergraduate class that semester, and struggled to read the room while teaching. Were my students bored? Confused? Did they hate my jokes? Were my pop culture references too outdated? (Given that I just referenced a 1992 Simpsons episode, I’m pretty sure that’s a yes). That same semester, my daughter started junior high school. I worried about how she’d make friends or connect with teachers when she couldn’t see anyone’s expressions.

I wasn’t the only one wondering about how masks affected everyday social life. Plenty of other parents and researchers wondered too. One of the most popular paradigms in social neuroscience is to present participants with emotional faces and see what lights up, but what happens when part of the face is covered? A lab meeting conversation about this question sent us all running to the research literature, and we were surprised to find only a small handful of studies had collected data on masks. Even fewer studies looked at kids – although children were wearing masks daily at school. When you’re forming your first friendships and developing brand-new social skills, how might masks affect your understanding of other people?

Luckily, we were starting to collect brain data from kids and their parents, so we could test these questions ourselves.

My lab studies the neurobiological underpinnings of parenthood, and we’ve been following a sample of first-time parents for over a decade. In fall 2021, we were gearing up to pilot a seven-year follow-up wave of data collection, bringing the parents back to the lab along with their kids, who were now first- and second-graders. We had already scanned our fathers’ brains both before and after birth, and this time, we planned to scan dads again, and also scan their kids. As we developed our scan protocol, the question of masks remained on our minds.

Luckily, one of my stellar graduate students, Yael Waizman, was game to develop a project digging into this question. Yael drew on social neuroscience research to create an original scanner task that we called the MASC (Masked Affective and Social Cognition) task. The task shows participants a mix of masked and unmasked faces, using standardized face pictures drawn from publicly available dataset that has been used in many studies. Yael redesigned these faces to add face masks, and created a protocol in which participants saw a mix of both masked and unmasked faces. We were curious about whether masks would affect specific types of emotion, so we included sad, scared, and angry faces. (We wanted to include happy and neutral faces too, but since you need to show each emotion a certain number of times to get reliable data, the task got way too long). We designed the task to have ten gender-balanced trials of every condition, including both child and adult faces. Here’s how all the adult stimuli looked:

Participant saw the faces (sad, scared, and angry) and then rated the emotion on each face.

We ran this task with our own families, and we also partnered with Sara Berzenski, a professor at California State University, Northridge (CSUN), to test a behavioral version of the task with over a hundred CSUN students. In addition to giving participants the MASC task, we also asked them how often they wore masks in everyday life, and assessed their depression and stress levels.

Across all the samples – college students, elementary school kids, and middle-aged dads – we found that people were worse at recognizing facial expressions when the faces were masked. This is not at all surprising, since masks block off at least a third of the information you need to interpret an emotion. The effect of masking was strongest for sad facial expressions. Across all the conditions, participants had the lowest accuracy when trying to label sad masked faces, and had the easiest time identifying fearful faces without masks. The below graph shows accuracy scores by condition - as you can tell, people get way more accurate when you take the mask off a sad face.

When we looked at brain responses to the faces, we found a cluster that lit up when viewing the masked faces but not the unmasked faces. It was in the posterior cingulate gyrus, considered to be part of the “mentalizing network” that helps us think about other people’s minds. We also found clusters in the fusiform gyrus (the part of the brain that recognizes faces) and the insula (a part of the brain that processes our interoceptive sensations, or gut feelings). In all cases, the brain was more active when viewing masked vs. unmasked faces, suggesting that it was exerting more effort to decode the masked emotions (even though it was doing so less accurately).

Sad masked faces were also linked with more activation in the insula, the “gut feelings” region. It’s been theorized that we also use the insula not only for sensing our own feelings, but also as a platform for empathy, using our own bodies to imagine how other people feel too. Interestingly, we found that activation in these insula clusters was correlated with participants’ stress and depression. (We only looked at this in the dads, since they were the only group that provided both brain and mental health data). When fathers of school-aged kids were more distressed, they engaged the insula more when trying to identify sad masked faces. You have to work the brain harder to recognize tricky faces when they’re masked - and you must work even harder when you’re struggling with your own mental health.

We did not find significant overall accuracy or brain differences between the kids, dads, and college students, suggesting that these effects were not shaped by age (so, in other words, reassuringly, elementary-school-age kids were not showing any special deficits in emotion processing). We also did not see any link between everyday mask usage and these results, which surprised us, because we thought that frequent mask wearers might be better at judging emotions.

These results tell us that masks make facial expressions measurably harder to read, and that decoding masked faces costs the brain extra effort in emotion, face processing, mentalizing, and empathy regions. It would be way premature to wonder whether masks ultimately challenge our ability to be empathetic towards each other, but it’s an important question to ask about the mass natural experiment that we conducted in 2020-2022. In particular, empathy for sadness is an important part of social bonding, and may have made it harder for teachers and peers to pick up on students who were in distress during those early days back to school.

There are so many more questions to ask about masking (and other COVID mitigation measures, like social distancing and the rise of remote work). We looked at static faces, but how does masking constrain actual social interaction, like conversational dynamics? How do masks affect children’s acquisition of social skills, and can long-term use of masks contribute to enduring social skills deficits in young kids? We need more research to understand these questions (at a time that the administration is slashing funding for social science and neuroscience research that can answer these questions).

I’ve been trying to steer clear of politics in this post (for once!). It’s regrettable how masking and other COVID adaptations became fractured among partisan political faultlines. Now that the pandemic is (mostly) in the rear-view mirror, it’s easy to say that blue state Democrats embraced pandemic mitigation measures too zealously, and the party’s brand still continues to suffer as a result. We can ascribe malevolent intentions, but my belief is that most blue state leaders were acting cautiously and trying to follow the best evidence available at a time of great uncertainty. Having a divisive president in office for the first nine months of the pandemic, one who undermined his own public health authorities and made wildly inaccurate claims and predictions, also created an atmosphere of negative polarization that increased emotion on both sides. I believe that we kept schools closed way too long in many places (including in my own school district), but we also had underinvestment in education early on; at the same time Congress was passing a giant bailout for the airline industry, they approved education funds that were measly in comparison. We could have thought much bigger, and used the opportunity to invest in updating school infrastructure to improve ventilation and hiring more teachers to reduce class sizes (I made this argument way back in summer 2020)!

If there’s one fatal flaw in how we approached the pandemic, in my opinion, it’s this: In our efforts to protect the physical health of the population, we forgot that mental health and social connection are also essential forms of public health. When we mitigate viral risk by keeping people isolated from each other, we inevitably introduce other forms of risk, because isolation is bad for us too. In fact, even though I’ve been talking about some of the downsides of masks, they were in a fact a very good tool for allowing people to feel comfortable gathering again in groups after lockdowns lifted.

It’s very likely that we haven’t seen our last pandemic (ahem, especially now that we’re defunding and politicizing the CDC), so let’s hope we can engage in more thoughtful and constructive conversations about our mitigation efforts the next time around.

In any case, if you want to read the full paper yourself, we wrote it up in our third Paper in a Day blitz* and published it in the journal Emotion. It looks like it’s embargoed till Nov 11, but feel free to reach out if you want me to send you a copy.

In other news: Last week, I was very excited to have my first piece in a new magazine,

The Argument

, which publishes thought-provoking work about politics and culture. I wrote about a girl being left alone at a roller rink —a story that went viral on social media—and how online outrage culture has convinced us that the physical world is unsafe for kids. The article is paywalled, but I have a couple of gift subscriptions to give away – comment below if you want one. If you’re on the fence about buying a subscription, I think it’s worth it – I ponied up myself after hitting too many paywalls on their pieces. (Hot tip for those overwhelmed by Substack subscriptions: set up a separate email folder just for them. I did this, so my newsletters never clutter up my inbox, and instead offer perfect procrastination fodder whenever I’m ready for a break, aka always).

*This post should really be titled “How my lab cranks out the science, part II,” but no one wants to click on a sequel post. I mean, I couldn’t even get myself to the theater to watch “Alvin & The Chipmunks, The Squeakquel” back in 2009, so why would I want to read a “part two” essay? In last week’s essay, I described our first two Papers-in-a-Day and promised to describe the last three papers and then provide a guide to writing these quickie papers, and as you can see, I achieved less than one-third of this goal. I’ll have to kick the ball down the road and make next week’s post a THREEQUEL.

MUSIC RECOMMENDATION OF THE WEEK

Past recs: Broncho // Alvvays // Capitol Years // The Cairo Gang and Hard Quartet // The Beths // Ballerina Black // James Mercer // Playboy Carti & Car Seat Headrest // Weyes Blood // Matthew Sweet // Fontaines D.C. // Elvis Costello Spanish Model

LILY ALLEN

I try to use my little music corner to shout out lesser-known artists, but I’m going to break this rule and talk about an album that is already dominating the discourse, because I’m obsessed with it too. In case you’ve been living under a rock, that’s Lily Allen’s new West End Girl, which documents the end of her marriage in excruciating detail. I’ve liked Allen ever since Smile came out 20 years ago, but I’ve never been more than a casual fan. Listening to her new album is like reading someone’s most confessional diary entries, full of raw, un-metabolized personal experience. But the songs are also melodic and catchy in a way that makes it hard to stop listening. The back-to-back numbers Tennis and Madeline, in which Allen discovers her husband’s texts to another woman and then reaches out to that woman herself, are particularly devastating.

Listening to the album prompted me to re-listen to my favorite song from Allen’s first album, Alright Still, and it holds up. She’s still picking the wrong guys, two decades later. I hope she eventually finds a nice normal dude who’s into monogamy!