Home

Forced Smiles

In an age of augmented reality, old portraits can feel remote and disturbingly unresponsive

On May 11th, a British designer named Olly Gibbs visited Amsterdam’s Rijksmuseum, where some of the canonical works of the Dutch Golden Age are housed. “Went to a Museum armed with FaceApp,” he tweeted, “to brighten up a lot of the sombre looks on the paintings and sculptures. The results …” Appended to the tweet were a series of before-and-after images. Rembrandt’s 1661 self-portrait as the apostle Paul, dressed all in black and with a sword tucked into his cloak to symbolize his execution as an early church martyr; then grinning like an old man who’s just pulled off a card trick, delighted at his own cunning. Ferdinand Bols’ famous portrait of Elisabeth Bas, dated around 1640, showing an old woman seated with her hands sternly folded and the face of a grumpy egg under her starched white cap; then smiling gently like an entirely different kind of grandmother.

FaceApp, which was released earlier this year, uses artificial neural networks to model photorealistic transformations — you can force a photographed face to smile, change its gender, age or rejuvenate it. In most photos, the modifications are unconvincing — the “meet your older self” function usually means meeting a self whose face is dripping off like cake batter — but the smiles on the Rijksmuseum portraits seem to reveal something genuine. When I look at them, I feel as though I’m seeing an actor break character — a 400-year poker face finally cracking. Elisabeth Bas seems relieved to escape her own oppressive dignity. The sad girl with the vampiric hair parting (Clara Veth, painted by her brother Jan) looks like someone just told her women’s lib was on its way. Just kidding, they seem to say, as if the seriousness of high art was always a dupe. The message is to lighten up: A somber face is never the end of the story. “Is the face app AI the best thing that’s ever happened to museum staff on quiet days? The answer is yes,” tweeted the vice chair of the UK’s North East Emerging Museum Professionals Group.

Our faces may say more about when we are than who we are

And yet, I couldn’t help but feel that forcing the past to smile isn’t pure comedy. The smiles made the originals feel withholding, as if they had been deliberately cheating me of a level of interaction I had a right to expect. As the interactivity that governs the digital world influences our relationships with institutions, creators, and art pieces, it can feel as though art that doesn’t respond to us — that doesn’t signal an awareness of our presence — is refusing or rejecting our advances. We pay attention to it; why doesn’t it pay attention to us? The resolute deadness of the past can feel like a form of abandonment.


Realism was the presiding notion of the Dutch Golden Age of painting. The 17th century saw the end of the Eighty Years’ War, after which the newly formed Dutch Republic was among the most prosperous nations in Europe. The emergence of Calvinism as the state religion meant that forms of artistic expression associated with the Catholic church fell out of favor or were actively suppressed — in 1578, Protestant reformers took possession of Amsterdam’s Oude Kerk and destroyed all the adornment and iconography, leaving it with a stark and plain interior. With the rise of a merchant class able to pay for commissions, portraiture became a popular way for artists to make a living. Painters were at pains to walk the delicate line between showing their wealthy clients in a flattering light and inviting criticism of their subjects for the sin of pride.

X-ray analyses of portraits by iconic painters of the era show the immense technical proficiency and flexibility needed to reproduce the realities of 17th century life on canvas. Techniques for painting lace collars ranged from the older method of using a fine brush to trace delicate white lines on top of the dark clothing to Rembrandt’s innovative white backgrounds with black dots and lines to represent the holes in elaborate patterns. Portraits could take years to finish, and clients might pose a dozen times, sitting for hours as the details of their costumes and physiognomy gave themselves up to the painter’s gaze. It is unsurprising that this long, intimate looking did not produce easily legible expressions. In Rembrandt’s self-portrait as the apostle Paul, the quirk of the eyebrows suggests hope, disappointment, and resignation all at once. The bulbous nose is rendered with harsh honesty, and the light glares off the greasy, wrinkled forehead. The grave set of the mouth indicates the painter’s slightly bitter self-knowledge. He looks as if he entertains an educated doubt as to whether humanity can be redeemed.

As Nicholas Jeeves wrote in a 2013 article on the absence of smiles in classical painting, “A smile is like a blush — it is a response, not an expression per se.” Dentistry was not at its best in the 17th century, and good etiquette meant hiding your inner rottenness both figuratively and literally. The social norms of the time reserved open smiles in paintings for drunk, lewd, or otherwise irresponsible characters. During the Renaissance, Western European culture came to value order and clarity, and rigid hierarchies enforced a strict adherence to social norms appropriate to one’s station — musicians and children could smile; civic leaders and wealthy businesspeople should not. “As a tooth-smiler I’m violating every rule of decorum, at least through the 19th century,” says art historian Kathy Galitz in a video on the Met’s website. To the contemporary eye, row after row of unsmiling portraits convey a refusal to respond to the world; people from the past seem unlike ourselves in their self-containment.

Especially with formal portraits, there is a power imbalance between the subject and the viewer — we can be changed by these works without being able to change them. This goes against contemporary intuitions. Over the 20th century, interactivity between the viewer and the artwork became a dominant mode of creation. In the 1920s, Marcel Duchamp’s Rotary Glass Plates consisted of five plates affixed to an axis, and required the viewer to turn a handle to rotate them at speed. The whirling glass produced an optical effect as the afterimage on the viewer’s retina glued the separate pieces into a continuous circle. circle. In Allan Kaprow’s 1964 event Eat, apples dangled from strings in a cave-like space, and visitors could choose to consume them.

As computer technologies and political liberties both advanced, interactive artists became increasingly interested in these twin re-shapings of the social landscape. Hierarchies and divisions of the past began to seem less stable, and artists relinquished sole control over their work. In 1968, London’s Institute of Contemporary Arts mounted the exhibit Cybernetic Serendipity, featuring art in which the viewer and the technology became co-creators — visitors could whistle a few notes and Peter Zinovieff’s studio equipment would embellish it into a song. The vision of reality most people had grown up with, wrote artist and theorist Roy Ascott in 1991, “conjured out of the thinking of the Enlightenment of the 18th century, presented a world of certainty and determinacy in which subject and object, mind and matter, art and science were all quite clearly defined, separated out and neatly categorized. That world is in many ways crumbling; we see now that it was not the world after all.”

The struggle between the original and the altered expression of a painting make new diptychs into the power dynamic between the portrait and the viewer

The new world was less about accepting one’s place in a fixed order than about seeing oneself as an agent with the power to adapt cultural and artistic products, however illusorily. New technologies aided this perception, and not only digital ones: If previously you had attended concerts, now Walkmans made concerts attend you. Movies now came to your home, and car phones and call waiting meant your interlocutors joined you in the interstices of your life. With the rise of mobile digital technology, the individual user is more than ever at the center of the creative imagination, even when it comes to public art. In Montreal, where I live, an ongoing public installation allows members of the public to tweet using certain hashtags and see the Jacques Cartier bridge light up in response. “Montrealers’ energy is continuously expressed through a sparkling on the exterior ‘skin,’” the designers explain.

While modes of art have become more open and responsive, the stoicism admired in an overtly class-based society has fallen out of fashion. The manners of our era prefer these distinctions to remain understated, and public discourse has increasingly come to favor emotional expressivity — since the 1960s, a certain studied informality has become a more reliably endearing public disposition. Contemporary relationships governed by the norms of social media and the demands of an economy that requires lifelong self-marketing, call for sharing as well as the admission of a wide variety of feeling. We also live with the constant feeling that our data precedes us — opting to share feels like a way to get out in front of whatever fragments of our private lives exist hidden in public places, vulnerable to manipulation by strangers. Projecting an appropriate image in today’s world tends to mean signaling openness, and the new decorum consists in responding to one’s public. If political figures of the past were meant to remain unsmiling to show their gravity, today’s political figures must demonstrate likeability. A smile shows a person who likes the world back.


Part of the fascination portraiture holds for us is in this refusal: These paintings challenge us to draw out their meaning. No amount of looking can persuade the subjects to give up more of themselves, but we can feel our way into what inhabiting their expressions might be like. The development of technologies like FaceApp allows us to enter into the inner lives of old portraits in a new way, painting over the sitters’ skin feelings that respond to us and to our time. The FaceApp smiles are all response and no expression. The delight they seem to paint onto Rembrandt’s face is the delight of interactivity, the fun of change. The subjects become carriers for superimposed emotions, and the struggle between the subtle meanings of the original expression and the surface meaning of the altered expression make the new diptychs into a commentary on the fraught power dynamic between the portrait and the viewer.

What does a forced smile say? If realism was the painting’s original mode, its technological grin is also realism of a different kind. Converted into pixels — information granular enough to rearrange — the subject’s face has been hacked. He or she joins us in our contemporary vulnerability to having our information used without our consent, all with a veneer of ease, sociability, and improved prospects. If the rows of unsmiling faces in a portrait gallery seem unrealistic in their gravity, the rows of carefully curated well-being in most Facebook galleries seem equally unreal in their joy. Alone, a face can seem trustworthy, but in the aggregate we become aware of the affective fashions of a particular moment in time. Our faces may say more about when we are than who we are.

Like using fossil DNA to recreate megafauna of the past, FaceApp allows us to reach into the musculature of historical faces and draw out a creature for our own time

Our era’s portraiture overwhelms with the volume of its entries — even the purses of aristocrats of the Golden Age didn’t run to thousands of paintings of themselves. The multiplicity of our images suggests a shifting personality, each moment hungrily captured and disposed of, still without achieving the kind of certainty that marked Elisabeth Bas in her 1640 portrait’s firmly located dress and demeanour. A phone’s worth of selfies may add up to a profoundly anxious group portrait. When responsiveness is imperative, it can be hard to tell whether a face is expressing a feeling of its own, or mirroring a feeling back. When we look at pictures of each other, it’s hard to know whether we are still only seeing projections of ourselves.

If our manipulation of old portraits allows us to critique the realism of somber faces, it’s also a critique of the realism of death. The death of a loved one can provoke anger; the retreat into some unreachable place feels like abandonment, and it hurts to be left behind. Our relationship to the people who lived before us is complicated by our feelings about our own era; in contemporary North America, our high standard of living and relative freedom to control our lives make the people pictured in 17th-century Dutch art seem pitiable. At the same time, it’s hard to imagine living without the dread of climate change and nuclear war that seem likely to wipe us out within an ever-shortening timespan. Reanimating the facial muscles of old portraits using artificial intelligence is one way to wake the dead. Like using fossil DNA to recreate megafauna of the past, FaceApp allows us to reach into the musculature of historical faces and draw out a creature for our own time.

Paying to have oneself committed to canvas was once a bid for immortality. In the 17th century, a commissioned portrait conveyed a sense of solidity and permanence. Today’s images are designed for spontaneity and change, with the expectation that we will be continuously documented and documenting. The deceptive longevity of our snapshots, once uploaded to social networks or other archived sites, ensures that our faces are open to future uses we are unable to imagine. Should we succeed in immortalizing ourselves, we may be subject to the same kind of superimposition, manipulation, and defamiliarization by the future that we now load onto the past. Cheating death means remaining accessible to the unknown societies that will come after us, with the possibility that we will be pressed into service to assuage future needs we can’t anticipate.


Elisabeth Bas’ anachronistic smile takes on a particularly slippery meaning when we realize that the woman in the painting may not be who we think she is. The portrait was for several hundred years attributed to Rembrandt, and thought to show Bas, who ran a successful Amsterdam tavern in the mid-1600s. In 1911, a Rembrandt expert argued that the painting was more consistent with Ferdinand Bol’s style, and in 1992, an art historian wrote that there was no compelling reason to identify the sitter as Bas. Though it seems impossible in our heavily surveilled era, the woman pictured has somehow revealed her face while still concealing her identity. Her response to our digital manipulation becomes taunting. We can make her smile but we can’t make her tell us who she is.

Linda Besner’s most recent book is Feel Happier in Nine Seconds. Her poetry and nonfiction have appeared in the New York Times Magazine, the Boston Review, the Globe & Mail, and Enroute, and aired on CBC Radio. She lives in Montreal.