Daniel sent us this one — he's been thinking about how still photographs in news media shape our impressions of people we'll never meet, based on whatever their face happened to be doing in that millisecond the shutter snapped. He mentions Giorgia Meloni as an example — someone where a still photo and a video clip can give you two completely different reads on the same person. And then the bigger question: why do people even have different resting faces? Is it just muscles and biology, or is something deeper going on? And he brings up sloths — how our facial expression gets misread as a smile, which is apparently a real problem in animal welfare. So, resting faces. The gap between what a face is actually doing and what we think it's doing.
There's a name for this, actually — or at least a name for one version of it. Psychologists call it "resting bitch face." The term is informal, but the research is real. There was a study published in 2015 by researchers at Noldus Information Technology and a behavioral scientist named Jason Rogers. They used facial expression analysis software to quantify what people are actually picking up on. And they found the key signal is contempt — a slight unilateral lip tightening or a subtle eye-narrowing that the software registers as low-level contempt, even when the person feels completely neutral.
The face is just...
The face is leaking. And it's not intentional. The person isn't actually feeling contempt. Their resting configuration just happens to match the muscular signature of that emotion. The software picked it up in about 6 percent of the neutral expressions they analyzed, and it was disproportionately attributed to female faces — which is its own whole thing, because the same neutral expression on a male face was less likely to be read that way.
There's a perception layer on top of the muscle layer. It's not just what your face is doing — it's what people are primed to see.
Right, and that's where it gets interesting. The resting face isn't one thing. There are at least three components. One is your baseline facial muscle tone — the actual anatomy. Some people have a stronger depressor anguli oris, the muscle that pulls the corner of the mouth down. Some people have a naturally downturned mouth even when every other muscle is relaxed. That's just structural. Then there's habitual expression patterns — the way your face has learned to settle based on decades of repeated emotional experience. If you've spent years frowning in concentration, that can etch itself in. And then there's the perception side — what the viewer projects onto the face.
It's a three-body problem. The face, the history of the face, and the person looking at the face.
None of them necessarily agree with each other. There's a fascinating 2019 study from the University of Glasgow that found people can reliably judge personality traits from neutral faces — or at least they think they can, and their judgments are consistent enough across raters that something is being transmitted. But whether it's accurate is a completely different question. What's happening is that facial structure correlates with certain personality impressions — a wider face is read as more dominant, a higher brow as more trustworthy — but these are stereotypes, not truths.
We're basically phrenologists with better branding.
That's not entirely unfair. But there's a twist. Some of these structural features do correlate with actual behavior, probably through hormonal pathways. Higher testosterone exposure during development affects facial bone structure — wider jaw, more prominent brow ridge — and also affects behavior. So the face isn't a completely arbitrary signal. But it's noisy. And the problem Daniel is pointing at — the news photo problem — amplifies all the noise and silences all the correction.
Right, because a photograph freezes one sample. You get one millisecond, one angle, one lighting condition, and that becomes the mental file photo for a person you've never met. If they were mid-blink, they look drunk. If they were transitioning between expressions, they look like they're having a completely different emotion than they actually were.
There's a term for this in photography — it's called the "decisive moment," from Cartier-Bresson. But the thing about the decisive moment is that it's often not representative. A photographer can capture a politician looking shifty in a way that, if you watched ten seconds of video, you'd never even notice. The still frame isolates and amplifies.
The Meloni example is perfect for this. She's got a strong facial structure — prominent jaw, intense brow — and in still photos she can look severe, almost combative. But if you watch her speaking in video, the animation changes everything. The severity becomes focus, the intensity becomes engagement.
I've seen this. There's a whole genre of unflattering politician photos, and the Italian press is not exactly gentle with her. But I think what's underappreciated is that this isn't just a media criticism problem — it's a cognitive problem. Our face perception system evolved for real-time, dynamic, three-dimensional interaction. It's designed to integrate motion, to use micro-expressions that flash by in a fraction of a second, to triangulate from multiple angles as someone moves. A photograph strips all of that out and hands your brain a single frozen data point that it was never designed to interpret in isolation.
Your brain just runs with it. It doesn't say "insufficient data." It says "this person looks hostile" and files that as a fact.
Because the alternative — withholding judgment — is cognitively expensive. Snap judgments are cheap. And evolutionarily, they were probably adaptive. If you see a face for two hundred milliseconds and it looks angry, maybe don't wait for more data before you step back.
Which brings us to the sloth thing. Because Daniel's right — our facial expression gets systematically misread, and it's not just a curiosity. It has consequences.
Yeah, this is something I've looked into. Sloths have a facial anatomy that naturally curves upward at the mouth — it looks like a smile. And humans are hardwired to read that upward curve as a signal of contentment or happiness. But in sloths, it's just the resting configuration. It doesn't indicate anything about their emotional state. The problem is that in wildlife tourism, in sanctuaries, in captivity situations, people see a sloth "smiling" and assume the animal is happy and comfortable, when it might actually be stressed, overheating, or in distress.
The smile as a mask.
And it causes real welfare issues. There was a study published in 2022 looking at sloth selfie tourism in Costa Rica and Panama. Tourists would hold sloths, take photos with them, post them online with captions about how happy the sloth looked — and the researchers found that many of these animals were showing physiological signs of stress, elevated heart rate, the works. But the smile made it invisible to the people interacting with them.
"The smile made it invisible" is a terrifying sentence when you think about it. A facial feature that evolved for completely unrelated reasons — probably just the muscle attachment points for chewing or something — becomes this powerful misinformation signal to an entirely different species that's projecting its own emotional vocabulary onto it.
It's not just sloths. Dolphins have a permanent smile-like curve to their mouth. It's just their anatomy. But people look at captive dolphins and think they're happy. Chimpanzees have a "fear grin" that looks like a human smile but means the opposite — it's a submissive signal, often indicating anxiety or fear. We project our facial vocabulary onto animals constantly, and we're wrong constantly.
We're bad at reading human faces, bad at reading animal faces, and the thing we're bad at shapes everything from who we vote for to how we treat captive animals. That's a cheerful thought.
I want to go back to the human side for a second, because there's something I think is genuinely under-discussed. We've talked about the perception error — the viewer misreading a neutral face. But there's also a feedback loop. If you have a resting face that people consistently read as hostile or unapproachable, you get treated differently. People are slightly cooler to you, slightly more guarded. And over time, that can actually shape your personality. You might become more guarded in response.
The face creates the social environment that then reinforces the face.
There's some evidence for this. A 2017 study in the Journal of Personality and Social Psychology looked at something they called "resting face feedback." People whose neutral expressions were rated as less approachable reported more negative social interactions over time. Not because they were actually less approachable — because people approached them less.
It's a self-fulfilling prophecy with cheekbones.
This is where the still photograph problem gets amplified. Because if you're a public figure and the most widely circulated image of you is a still frame where you look severe, that becomes the filter through which everything else you do is interpreted. People watch your speeches already primed to see hostility. The still photo sets the baseline, and the video has to work against it.
Which is harder than it sounds, because once a first impression is formed, it's sticky. There's a whole literature on impression perseverance — people hold onto initial judgments even when they're presented with contradictory evidence.
And the first impression from a photo is formed in something like thirty-nine milliseconds. A study from the University of York in 2006 found that people make trustworthiness judgments that fast from a single glance at a face. And those snap judgments persist even when you give people more time or more information.
Thirty-nine milliseconds. That's less than a tenth of the time it took me to say the words "thirty-nine milliseconds.
It's basically instantaneous. And here's the thing — those judgments aren't random noise. Across raters, they're surprisingly consistent. We agree on who looks trustworthy and who doesn't, even though we're often wrong about whether they actually are trustworthy.
Which means we're not just individually bad at this. We're collectively bad at it in the same way. The error is systematic.
The news media ecosystem is basically a machine for manufacturing and distributing these systematic errors at scale. An editor picks a photo — and editors have their own biases, their own narratives — and that photo gets attached to every story about that person. Google image search reinforces it. Social media reinforces it. The error compounds.
What do we actually know about why faces differ in the first place? You mentioned anatomy, habitual expression, and perception. But what's driving the anatomy differences?
It's a mix of genetics, development, and environment. The genetic component is substantial — twin studies show that facial structure is highly heritable, including the specific configurations that produce different resting expressions. But it's not just bone structure. Muscle insertion points matter. Fat distribution matters. Skin elasticity matters. And all of those change over time.
Your resting face at twenty isn't necessarily your resting face at fifty.
Right, and that's partly aging and partly cumulative expression habits. There's a concept in facial aging called "dynamic wrinkles" — these are wrinkles that form from repeated muscle movements. If you've spent decades furrowing your brow, those lines become permanent even when the muscle is relaxed. Your face literally records your emotional history.
Like tree rings, but for frowning.
I mean, that's basically what it is. And this is where the line between the biological and the psychological gets blurry. Your resting face is partly a fossil record of your emotional life. If you've had a life that involved a lot of worry, a lot of concentration, a lot of anger — that shows up. If you've had a life that involved a lot of smiling, that shows up too, in different muscle patterns.
There's a version of this where resting face is actually a somewhat honest signal — just not of what you're feeling right now. It's a signal of what you've felt a lot of, over a long period.
But I'm cautious about this, because the evidence is mixed. Some studies find correlations between facial appearance and personality traits that are above chance but still quite modest. The signal is there, but it's weak, and the noise is enormous. You'd be wrong more often than you'd be right if you tried to read someone's life history from their resting face.
People do it anyway. That's the whole problem.
People do it constantly and with enormous confidence. And that's really the core of what Daniel is getting at. The resting face is this Rorschach test that we all insist on treating as a photograph.
Speaking of things we treat as photographs — let's talk about the sloth side more. Because I have feelings about this.
I would expect nothing less.
The sloth smile is one of those things where the more you know, the worse it gets. Our facial anatomy evolved for a completely different set of selection pressures than human facial anatomy. The curve of the mouth has nothing to do with emotional signaling and everything to do with the mechanical requirements of the jaw. But humans see it and immediately slot it into their own emotional categories.
There's a real ethical problem here. Wildlife tourism is a massive industry, and sloths are a big draw in Central and South America. People want the photo. They want the sloth selfie. And the smile becomes this justification — "look, he's happy, he's enjoying this." When in reality, sloths are solitary animals that find handling extremely stressful.
The stress isn't always obvious. A stressed sloth might not thrash around. It might just shut down. And the frozen sloth with its permanent smile looks like a contented sloth to a human observer.
There's a 2021 paper that specifically looked at this — the welfare implications of sloth selfie tourism. They found that sloths used for tourist interactions showed elevated cortisol levels, altered feeding patterns, and disrupted sleep cycles. But the tourists, when surveyed afterward, overwhelmingly reported that the sloth seemed "happy" or "content." The smile was doing all the work.
The smile is essentially a welfare liability. It hides the distress signal that might otherwise trigger intervention.
This connects back to a broader problem in animal welfare that's been recognized for decades but is still poorly addressed — anthropomorphism in welfare assessment. We judge animal wellbeing by human facial expressions, human body language, human emotional categories. And it leads us systematically astray.
The dolphin smile. The chimpanzee fear grin. The sloth smile. All these signals that mean "everything's fine" to a human and mean something completely different — or nothing at all — in the animal.
Dolphins are a particularly tragic case. The smile is just the shape of their jaw. It doesn't change. A dolphin could be in severe distress, and the smile is still there. There's been work on actually identifying real dolphin stress indicators — changes in respiration, changes in social behavior, certain vocalizations — but the smile dominates public perception.
Because the smile is what photographs well.
And the photograph is what gets shared. The photograph is what drives tourism. The photograph is what shapes policy.
We've got a loop. The misleading signal generates the photograph, the photograph generates the perception, the perception generates the demand, the demand generates the captive interactions, the captive interactions generate the distress, and the distress remains invisible because the misleading signal is still doing its job.
That's the loop. And breaking it requires public education at a scale that's difficult to achieve, because the smile is so intuitively convincing. You can tell people "sloths aren't actually smiling" and they'll nod and then look at a photo of a sloth and go "but look at that face.
I've been on the receiving end of that. "But you look so peaceful." Yes, that's my face. That's what it does.
There's actually something interesting here about the specific anatomy of the sloth face. The facial muscles of sloths are quite different from most mammals — they have a reduced facial musculature compared to, say, primates. They don't have the same repertoire of facial expressions that we do. The "smile" isn't an expression at all — it's just the default configuration of the lips when the facial muscles are completely relaxed. It's the absence of expression, not the presence of happiness.
The absence of expression, misinterpreted as the presence of happiness. That's basically the whole problem in one sentence.
I think that applies to human resting faces too, in a slightly different way. When someone has a neutral expression that gets read as hostile, it's the absence of expression being misinterpreted as the presence of hostility. The face isn't doing anything — and that "not doing anything" gets filled in by the viewer.
The blank canvas problem. If you give people a blank canvas, they'll paint something on it. They can't help themselves.
There's a classic demonstration of this from the 1940s — Heider and Simmel. They showed people a short animation of geometric shapes moving around — triangles, circles, squares. No faces, no bodies, no expressions. And people immediately constructed elaborate narratives about the shapes' personalities and intentions. "The big triangle is bullying the small circle." The human brain is an attribution machine. It sees agency and emotion everywhere, and when there's insufficient information, it invents it.
A still photograph is the geometric shape problem, but with a human face. Insufficient information, maximum attribution.
The attribution isn't random. It follows predictable patterns. There's research showing that people with certain facial features — a heavier brow, a stronger jaw, deeper-set eyes — are consistently rated as more dominant, more competent, and less warm. These are structural features that have nothing to do with the person's actual personality, but they shape everything from hiring decisions to election outcomes.
There was that study about political candidates, right? People could predict election outcomes at above-chance levels just from brief exposures to candidate photos.
Yes — Alexander Todorov at Princeton has done extensive work on this. In a 2005 study, participants viewed pairs of political candidate photos for one second and judged who looked more competent. Those competence judgments predicted about seventy percent of the actual election outcomes. One second of exposure to a still photograph.
That's not everything, but it's not nothing either. That's a real effect.
What's wild is that the effect held even when people saw the faces for as little as one-tenth of a second. The judgment is essentially instantaneous, and it doesn't get much more accurate with more time.
The still photo isn't just shaping our impression of someone's personality — it's shaping democratic outcomes.
And the counterargument is that maybe the face is signaling something real — that more competent-looking people actually are more competent in ways that voters are picking up on. But the evidence for that is thin. Most of what's being read is just facial structure, which isn't correlated with governing ability.
We're electing cheekbones.
We're electing cheekbones, brow ridges, and jawlines. Among other things. Obviously policy positions and party affiliation matter more in most cases — but in close races, in primaries, in low-information contests, the face effect could be decisive.
The media knows this. Photo editors know this. They're not naive about the power of image selection.
No, and there's a whole visual rhetoric to political photojournalism. The unflattering freeze-frame. The caught-mid-blink shot. The harsh lighting. These are editorial choices, and they're not always neutral.
Which brings us back to Meloni. The Italian press has run some truly unflattering photos of her — harsh shadows, awkward angles, caught mid-expression in ways that make her look aggressive or unstable. And then you watch her speak, and it's a completely different person.
This isn't unique to her — it happens across the political spectrum, across countries. But it's a particularly clear example of the gap between the still image and the person.
What's the fix? If the problem is that still photographs systematically misrepresent people, and we're not going to stop using still photographs, what do we do?
I don't think there's a clean fix at the individual level. You can't train seven billion people to be better at reading neutral faces. But there are things that could help at the institutional level. News organizations could adopt policies about photo selection — avoid the freeze-frame shots that capture someone mid-expression, use photos where the subject is engaged in an activity rather than frozen in a posed moment, provide video alongside still images whenever possible.
Essentially, lower the signal-to-noise ratio by providing more signal.
Give the brain more data to work with. The reason still photos are so misleading is that they're impoverished stimuli — they strip out motion, context, temporal information. Adding some of that back helps.
On the animal welfare side?
Education, regulation, and alternative framing. Some wildlife tourism operators are starting to shift away from the "happy sloth" narrative and toward a more accurate framing that acknowledges the animal's actual state. But it's an uphill battle against the smile.
The smile is very powerful. I should know — I've been accused of looking serene when I'm actually deeply annoyed.
When are you deeply annoyed?
Whenever someone suggests sloths didn't invent pizza.
that's a thing you believe.
It's documented.
It's not documented.
The documents were lost. That's not my fault.
I want to circle back to something we touched on earlier — the feedback loop. Because I think it's actually the most important part of this for anyone who's not a politician or a sloth. If you have a resting face that reads as hostile or unapproachable, you've probably been told about it. People say "smile" or "what's wrong" or "you look angry." And that creates a self-consciousness that can actually make the expression worse.
Because now you're not just relaxing your face — you're tensing it in anticipation of being misread.
The awareness of the problem becomes part of the problem. And there's a gendered dimension to this that's well-documented. Women are told to smile far more often than men are. A neutral expression on a woman's face is more likely to be read as negative. The resting bitch face phenomenon isn't just about facial anatomy — it's about differential social expectations.
The same neutral face, same muscle configuration, different perception depending on whose face it is.
A man with a neutral expression is serious, focused, contemplative. A woman with the same expression is cold, unfriendly, hostile. The face is identical, but the social script is different.
Which means the resting face problem isn't just a biology problem or a psychology problem — it's a social problem. The error isn't just in the muscles or the perception. It's in the expectations we bring to different faces.
Those expectations are culturally specific and historically contingent. What counts as a "pleasant" resting face in one culture might read differently in another. The interpretation layer is not universal — it's learned.
We're back to the three-body problem. The face, the history of the face, and the person looking at the face — and the person looking at the face is bringing their entire cultural training, their gender expectations, their political priors, their editing software.
All of that is compressed into a judgment that takes thirty-nine milliseconds.
It's almost impressive, in a horrifying way. The sheer efficiency of the error.
There's a philosopher, Lisa Feldman Barrett, who's done a lot of work on emotion perception and argues that the whole idea of reading emotions from faces is fundamentally flawed. Her view is that facial expressions don't reliably signal specific emotions — they're more like communicative gestures that vary enormously across contexts and cultures. And the idea that there's a universal facial language of emotion is, in her view, largely a myth.
The whole premise — "this face means this emotion" — might be wrong from the ground up.
That's the strong version of her argument, and it's controversial. But even the weak version — that facial expressions are noisy, context-dependent signals — is enough to make the still-photograph problem really serious. If even dynamic facial expressions are ambiguous, a single frozen frame is basically meaningless.
It doesn't feel meaningless. That's the trap. It feels incredibly information-rich.
Because your brain is filling in all the missing information without telling you it's doing that. The experience of looking at a face feels like perception. But it's actually construction. Your brain is building an interpretation from sparse data, and it's so good at it that you don't notice the gaps.
Like a language model that's very confident and very wrong.
I was not going to make that comparison, but... In some ways, the face perception system is a predictive model that's been trained on a lifetime of facial data, and it's making inferences from partial information. And like any model, it has systematic biases.
No one gave it a safety filter.
Evolution gave it a speed filter, which is sort of the opposite. The system is optimized for speed, not accuracy. In a social species, a fast-but-noisy judgment might be more adaptive than a slow-but-accurate one. If you hesitate too long before deciding whether that face is friendly or hostile, you might not get a chance to use the information.
We're stuck with the fast noisy system, and we've built a media environment that exploits it.
I don't know if "exploits" is always intentional, but it certainly amplifies the noise. The still photograph is the perfect input for the fast noisy system — just enough information to trigger a confident judgment, not enough to correct it.
Video is better, but it's not perfect. You can still be misled by a video clip if it's selected adversarially.
The edited video clip is arguably worse than the still photo, because it feels more complete. You've seen them move, you've heard them speak — it feels like you've got the full picture. But a skilled editor can make anyone look like anything.
The solution isn't "watch more video." It's something more like "maintain awareness that you're operating on incomplete information and hold your judgments lightly.
Which is good advice for basically everything, and almost nobody does it.
Because it's exhausting. Constantly second-guessing your own perceptions is cognitively expensive, and most of the time it doesn't pay off. The fast noisy system is right often enough that overriding it constantly would be a bad strategy.
The problem isn't that the system is noisy — it's that we've built an information environment where the noise gets amplified and the signal gets stripped away, and then we treat the resulting output as ground truth.
We do it to sloths too. That's the part that really gets me. We've taken this completely innocent anatomical feature — the upward curve of a sloth's mouth — and turned it into a justification for stressing out an animal that just wants to be left alone in a tree.
The sloth's smile is the perfect case study because there's zero signal there. It's pure noise. The smile doesn't mean anything at all. And yet humans build entire emotional narratives around it, make tourism decisions based on it, post Instagram captions about it.
"Living his best life." No, he's dissociating. Those are different things.
The tragedy is that the real sloth — the actual animal, with its actual needs and its actual experience — gets completely erased in this process. The sloth becomes a prop in a human emotional narrative that has nothing to do with sloth reality.
The smile as erasure. That's even darker than the smile as mask.
I think it's both. The smile masks the distress, and in masking it, it erases the animal's actual experience from human consideration. The photo gets shared, the likes accumulate, and the sloth goes back to its enclosure stressed and exhausted, and nobody notices because the smile told them everything was fine.
The same dynamic plays out with human faces, just less dramatically. The resting hostile face masks the actual neutrality or warmth of the person. The resting friendly face masks actual hostility. The photo freezes one moment and erases the person's complexity.
There's a good example of the reverse problem — resting friendly face. Some people have a naturally warm, approachable expression even when they're feeling neutral or negative. And they get approached more, confided in more, expected to be more emotionally available. The face creates social demands that may not match the person's actual disposition.
It cuts both ways. You can be trapped by a hostile resting face or trapped by a friendly one. Either way, the face is doing something you didn't sign up for.
Either way, other people are responding to the face, not to you.
Which is a very strange thing to think about. That for most of the people who see your photo in a news article, on social media, wherever — they're not responding to you. They're responding to a frozen millisecond of your facial muscle configuration, filtered through their own perceptual biases, shaped by cultural expectations you may not share.
That frozen millisecond might be the only version of you they ever encounter.
Now: Hilbert's daily fun fact.
Hilbert: In the 1720s, a British naturalist in Belize misattributed a failed muon-catalysed fusion reaction to swamp gas igniting spontaneously, a correction that would not appear in the scientific record for two hundred and ninety-three years.
...right.
Here's my closing thought. We've spent this whole episode talking about how faces mislead, how still photos distort, how the gap between what a face is doing and what we think it's doing is vast and consequential. But I think the deeper point is that this isn't fixable in any complete way. We're never going to stop making snap judgments from faces. We're never going to stop using still photographs. The machinery is too deep and the incentives are too strong. The best we can do is remember that we're doing it — to hold our face-based judgments a little more lightly, to seek out more information when it matters, and to notice when a smile might not be a smile.
Maybe, when you see a sloth in a photo looking happy, remember that what you're seeing is an anatomical accident being misread by a brain that was never designed to read sloth faces. The sloth isn't smiling at you. It's just being a sloth. And that's enough.
Thanks to our producer Hilbert Flumingtop. This has been My Weird Prompts. You can find every episode at myweirdprompts.
If you've got a resting face that's been misread, you're in good company. We'll be back soon.
