Joel Lloyd Bellenson places a little ceramic bowl in front of me and lifts its lid. "Before we begin," he says, "you need to clear your nasal palate."
I peer into the bowl. "Coffee beans," explains Bellenson's partner, Dexster Smith. "This is what they use in perfume stores. It's like the reset button."
Dutifully, I reinitialize my nose by sniffing the beans. I'm preparing for a sensory epiphany here, an epochal event in the history of art, smell, and computation. Bellenson and Smith claim they've developed a highly secret process to encode odors as digital data. Just as we can download digitized music and play it through speakers attached to a computer, we should soon be able to acquire online scent data that a little gadget can play back as smells.
Bellenson, who sounds edgy and looks sleep-deprived, has been pitching his new paradigm to me with the manic charisma of an infomercial host. The way he sees it, "scentography" is going to transform the entertainment-media spectrum, all the way from web surfing to Hollywood movies.
"Now we're ready," says Smith, as I set down the bowl of beans. Like Bellenson, he seems tired and nervous. This is, after all, the first-ever demo the scent team has given an outsider. They're not even in beta yet.
We're sitting in Bellenson's immaculately renovated 1920s apartment in Oakland, California, overlooking a grand panorama of downtown and Lake Merritt. But no one's checking out the view. We move closer to an IBMThinkPad that rests on an ornate glass-topped table. Smith rolls the trackpoint, shifting the cursor arrow to a picture of some green grapes. He clicks on them.
Beside the ThinkPad, linked via serial cable, sits a black plastic box 3 inches tall, 2 inches wide, and 5 inches deep, about the size of an electric pencil sharpener. Somewhere inside it a little fan starts whirring, drawing in air at the back and blowing it over tiny vials of oils that are being heated selectively in response to signals from the computer. The air picks up the oily fragrance and wafts it out through a 2-inch vent.
All conversation ceases as I lean forward and inhale. There is a long moment of tense anticipation.
I straighten up and give Bellenson and Smith an apologetic look. "Uh, it smells like cheap perfume."
Bellenson releases a sharp, brittle laugh."Well, yeah. We were generating perfume earlier." He shakes his head. "It'll clear! Try again!"
Smith scrolls to a picture of oranges. He clicks on them.
Once more I breathe deeply—”Orange peel," I report. My hosts relax in their chairs, looking immensely relieved.
But the orange peel is just the beginning, a humble hors d'oeuvre. Marc Canter, sometimes known as the father of multimedia, joins us to serve the multiscented main course.
Canter is a big man with a deep voice and a brooding presence. Back in 1985, when he was founder, president, and chair of MacroMind (now Macromedia), he cowrote and launched the product that became Director and its playback module, Shockwave, still among the most popular tools for CD-ROM, DVD, and web multimedia. He's now CEO of a startup called Broadband Mechanics, developing applications in anticipation of the era of true high-speed Internet access. If Canter has his way, broadband is going to be odor-enabled, and his company will be producing scent demos synchronized with movies, videos, and music.
Canter initiates a sequence of movie clips, and in a window on the screen we're transported to Oz. Dorothy, the Lion, and the Scarecrow are venturing into the forest, and its trees are not merely visible but sniffable. "Ahhhh!" Canter exclaims. "Smell that cedar!" And—it's true! A cedar fragrance emerges from the little black box beside the computer.
Now the Wicked Witch is mixing a poison potion over a crackling fire. "Oh, that wood smoke!" Canter cries with delight. Indeed, the box emits a smoky tang—without the smoke.
From Oz we jump-cut to Donkey Kong, watching the title character roll over a bunch of bananas. Their scent is undeniable. And now a scene in the Orient. "Mmmmm, that incense!" Canter says rapturously, like a chef over a rare truffle. It seems he has been waiting all his life for this sensory experience, advancing us beyond video, beyond audio, to—what should we call it? Odor-on-demand? Streaming naso?
Canter's montage lasts eight minutes, triggering 26 scents, all of them clearly discernible. "That sequence contains so much stimulation," he remarks at the end, "the first time I experienced it, literally my whole body was affected." He clutches his chest, emphasizing the heart-stopping power of aroma. "Luckily," he goes on, "we have a world-class sensory psychologist on staff, who'll be able to work out exactly the median time frame that is appropriate for refreshing the nostrils." He frowns, mentally critiquing the demo. "You know, I don't think the transition from wood smoke to bananas worked very well."
Bellenson shifts impatiently. Canter's aesthetic pretensions are all very well, but this is serious. Scentography promises a vast extension of sensory space, with profound implications. "We've lost touch, as a species, with our sense of smell," Bellenson says, speaking so rapidly that his words run together. "Like, our noses are not on the ground anymore, because we don't have to hunt for food. Scent became an art perpetuated by the big fragrance houses in Europe, and the average person was not empowered. You cannot create a new smell, communicate about it, talk about it. But now we can change that! Our mission is to make scent accessible to everyone! We're giving back to humanity our ability to communicate using scent!"
I examine the little odor synthesizer more closely. This first prototype is just a few days old, with an out-of-the-mold, rough-edged look. Still, someone has found time to paste a brand name on the side. I'm expecting it to say "DigiScents," the name of the company Bellenson and Smith have formed as partners. But the lettering reads "iSmell."
"That's the product name?" I ask. "For real?"
Bellenson emits another nervous laugh. "Yeah. Well, we think so. I mean, it's a powerful, in-your-face name. It's saying, 'Sure, I stink!' And that's its purpose—to create smell!"
Bellenson and Smith didn't set out to be the Prometheans of digitized funk. They met more than 10 years ago as Stanford undergrads, over a chessboard in a coffeehouse. Bellenson, who had grown up in California, was studying biology and international relations; Smith, who came from Florida, was doing industrial engineering.
By 1989 Bellenson had graduated and was running a lab at Stanford's Beckman Center, using early automated equipment to do DNA synthesis. "People would request DNA sequences for use in their experiments," he says. "We began charging fees against people's research grants, and bringing in millions of dollars."
In 1991, sensing an opportunity, he and Smith started Pangea Systems, a company that built genetic databases. Their software was customized to run on Macs and Unix boxes at major pharmaceutical companies. One of the packages, LifeSeq, was commissioned by Incyte to help it examine and compare genes in order to develop new drugs that could target specific diseases. LifeSeq was so far ahead of its competition that Incyte was able to resell copies to other companies for what Bellenson calls "a hefty subscription fee of a few million a year." Bellenson realized that he and his partner should market their own software instead of letting other companies do it.
In February 1997, after raising $10 million to transform Pangea from a consultancy to a product-oriented outfit, they conceived GeneWorld to automate genetic-database searches, then GeneThesaurus, which searches all public genetic databases simultaneously, sorting an incredible mass of data. Most genomics firms now use Pangea software. This fall Pangea announced DoubleTwist.com, a site where biologists will be able to use its search tools. "DoubleTwist is the first of several biotech and Internet plays we have in the pipeline," Smith says, alluding to a web-based "personalized medicine" service.
By 1998, Bellenson and Smith had excused themselves from Pangea's day-to-day operations to focus on strategic planning. (Both remain on Pangea's board.) Meanwhile, they enjoyed the spoils of their success. Bellenson filled his 2,500-square-foot apartment with furniture from Africa, a grand piano, and Bowflex weight-training equipment. Some might have expected the two to take time off. But not them. "I'm a compulsive builder," says Smith. "Creating a new market is where the satisfaction is most intense."
They founded their own venture catalyst company, Libra Digital, to jump-start new businesses, including Marc Canter's Broadband Mechanics. Then, on a brief vacation in Florida, they noticed hordes of women wearing exotic perfumes. "It was an intense sensory experience," Smith recalls with a sly grin. "And since smell is a biological phenomenon, it was in our domain."
When Bellenson returned to California, he started investigating the role genes play in our ability to detect smells. He came across a recent experiment at Columbia University, which built on Harvard neurobiologist Linda Buck's 1991 breakthrough wherein she discovered the family of genes responsible for odor detection. In 1998, researchers in StuartFirestein's lab at Columbia engineered a virus to stimulate a rat's odor receptors, heightening its experience of smell and, theoretically, paving the way for the genetic "transplant" of receptors from one type of animal to another.
The explanation for this proved relatively simple. When odor molecules drift into the nose, each of them binds with a particular protein on the surface of a neuron. There are about 1,000 odor-matching proteins, each with a slightly different configuration, scattered across a human's 10 million odor-detecting neurons. (By comparison, a mouse has about 1 million neurons of this type, while a pig boasts 100 million.) When the shape of an odor molecule matches the shape of a protein, the molecules lock together, triggering the neuron, which sends a signal that the brain recognizes as a smell. DNA is relevant because its instructions—its genes—tell the body how to build the proteins that receive odor molecules and activate the neurons.
Bellenson started looking at promising gene sequences, then determined which proteins they would create. Here, he encountered a problem: No one knew the exact shape of these protein molecules; they were in a category that had never been modeled. Bellenson didn't want to spend the time and money on X-ray diffraction studies to create detailed models, so he searched for other proteins that might be similar—and found them in bacteria. This wouldn't have been close enough for a pharmaceutical application, he notes. But it let him create a smell index. He wrote software algorithms to simulate the binding of odor molecules with proteins, then used trial and error to fine-tune the odors, testing them on an odor output device he'd designed.
Ultimately, just as a computer monitor can display millions of colors by mixing different proportions of red, green, and blue, Bellenson wants to generate billions of odors by blending different proportions of just 100 to 200 "scent primaries." This is the only way to make the iSmell output device versatile and cheap enough for the mass market.
Initially, he wasn't thinking of building an odor synthesizer. He imagined applying his research to make foods and drinks smell and taste better. "Taste is 95 percent aroma," he explains. "Your tongue can only detect sweet, sour, bitter, and salty, plus the sensation you get from MSG. All other tastes are a function of the nose. That's why your sense of taste is degraded when you have nasal congestion. Also, it explains why elderly people don't taste things as well, because the ends of their chromosomes get chewed away."
Gradually, though, he developed a higher ambition—"to integrate scent with all media." DigiScents was incorporated earlier this year.
Because Smith and Bellenson are software people, they don't want to manufacture the odor synthesizer themselves. They plan instead to bulk-license iSmell to hardware companies. Also, anyone who wants to odor-enable a website will have to pay a licensing fee for the smell index, which reveals the correct proportion of iSmell scent primaries for a particular odor. "That's a business model very much like RealPlayer," Dexster Smith says.
The big question, though, is whether consumers will want iSmell as much as they want MP3. The device works; I can attest to that. But there's something funny about putting a computer-controlled smell machine on your desktop. The biggest challenge now facing DigiScents may be to figure out how to overcome this "wacky factor." Solving the marketing problem will determine whether the product becomes a favorite toy for millions or is laughed into oblivion.
Bellenson is well aware that he must convince people to take this seriously. The very notion of digital odor synthesis has been spoofed in several online hoaxes over the years. "Realaroma.com is a joke website," he says, "postulating a fictitious synthesizer and smell markup language. Then there was another website, sniff-and-cough, and an email message announcing WinSmell from Microsoft."
Yet using odors for artistic expression hasn't always been dismissed as a joke; it's just been a while since the concept was explored in any depth. In the late '50s, Americans were treated to two cinematic stinkfests, AromaRama and Smell-O-Vision. Behind the Great Wall, a documentary employing the former, included 72 scent cues, ranging from nightclub smoke to Oriental spice, synced to scents pumped through the theater's ventilation system. Smell-O-Vision was dispensed from under each theater seat.
Two decades later, filmmaker John Waters famously used smell as a self-satirical gimmick in the 1981 Polyester. Waters presented his film in Odorama, distributing numbered scratch-and-sniff cards to audiences. Earlier mass entertainment applications of smell, introduced as a way to compete with the popularity of TV, offered an enhanced experience without asking anything extra of moviegoers; Odorama was more in the tradition of those headache-inducing, cardboard-and-acetate 3-D specs distributed at screenings of '50s horror movies like Creature From the Black Lagoon.
Transcending this kind of gimmick stigma seems intimidating, yet it has been done. At 3-D Imax theaters, audiences have shown they are willing to pay a premium to wear headgear fitted with liquid-crystal lenses synchronized via infrared signals with the movie projector, which runs at twice the normal frame rate. These movie viewers plumb new depths in depth perception—they experience extreme realism. If smart technology can transform 3-D from a crude novelty to a genuine visual enhancement, why shouldn't a sophisticated odor synthesizer follow a similar path?
Certainly smell could be artistically important, because it bypasses our conscious brain and communicates directly with the limbic system. It's an opportunity for filmmakers, advertisers, and web designers to evoke emotions that are literally uncontrollable. That's the power—and perhaps the danger—of communication via smell.
Bellenson foresees informational applications, too. "We can facilitate people experiencing a lot more of the world," he says with intense conviction. "The United States is bland with regard to smell, but the rest of the world is not. There are educational possibilities, because odor is so closely linked with memory. People will learn more about each other and communicate more effectively."
Perhaps more to the point, lucrative web apps are possible. When you shop online, wouldn't you like to sample the bouquet of wines, the aroma of cigars, or the subtle fragrance of flowers before surrendering your credit card number? Surely more companies will want to aromatize their Web presence when they realize there's a device that can produce genuinely pleasing, authentic fragrances.
Likewise, in simulation games, golf software can create not just the click of club against ball, but also the heady aroma of fresh-cut grass. In the Nintendo/PlayStation universe, auto-racing games can reek of gasoline and burning rubber, while Duke Nukem may be enhanced with the tang of underarm sweat or the stench of alien flesh roasted by laser weapons.
Already Smith and Bellenson have commissioned a consultant to create a perfume for tomb raider Lara Croft, and they imagine a unique scent for each fire, water, and earth character in the Pokémon games. "We're in negotiations with [Tomb Raider publisher] Eidos," says Smith. "We're talking to Mattel, too. Imagine a CD-ROM where you pick outfits and activities for your Barbie doll—along with her perfume!" Yeah!
Nor does it end there. Aromatherapy can now go online. Though its healing powers are debatable, Bellenson maintains that some oils are definitely antimicrobial; they may not kill infectious agents in the bloodstream, but they'll certainly disinfect the room. Bellenson also believes that odors can play a significant role in affecting mental states such as depression—and sexual excitement. Pheromones, which seem odorless, Bellenson points out, "can make people feel sexy—or hungry." Thus gourmet-food distributors could induce cravings for their products, while XXX-rated web sites could crank the user's libido.
By early next year, you should be able to seriously sniff around www.digiscents.com. With characteristic zeal, Bellenson & Co. promise online goodies such as free smell samples, ranging from chocolate and roses to "new car"; smell design tools that let you create your very own odors; movie scent tracks to run in sync with your favorite DVDs; and, of course, the iSmell box. (Assuming, that is, they have landed a manufacturer.) DigiScents plans to begin taking beta-user orders by Christmas, and aims to make the gadgets generally available by spring. There's even the prospect, unconfirmed at press time, of odor broadcasting. Smith and Bellenson have had preliminary talks with major US music-television and home-entertainment networks and Asian satellite television broadcasters. Adding odor to the broadcasts wouldn't steal much bandwidth: A typical scent, using the Bellenson-Smith encoding format, can be defined with less than 2 Kbytes of data.
In Bellenson's apartment, Marc Canter has been lying on a postmodern faux-leopard-skin couch with his eyes half closed, listening as Bellenson and Smith outline their grand vision. He rouses himself now, like a lugubrious guru, a veteran of more than half a dozen projects pushing the state of the art. He wishes to make a statement about trends that lie ahead.
"There is a new paradigm for tools," he says. "In the old days, they were shrink-wrapped pieces of software; you sat down and read the manual and used the tool. Nowadays, the tools are free. And what we need are scalable content tools. Look at Hollywood: They take a movie and amortize the cost among multiple forms, from cable TV to toys. On the web, we haven't been able to do that, because it's just a delivery medium. But if all the content can be decoupled"—in other words, if it can exist separately from any particular format—"I can output a low-end website, a medium-res CD-ROM, and a high-end broadband version, all from the same ideas. In the smell world, this means 16-pack cartridges that do only a few smells, or big systems that do thousands."
"We expect to have low-end and high-end iSmell hardware," Smith agrees. "The low end may retail for under $200. The smell cartridges—even at the high end—will probably cost under $50." With moderate use, he guesses, they should last a few months.
"The key, as always, is the installed base," Canter says. "But there's so many different target markets. It'll be easy to get overwhelmed. You'll need a staff of 15 people just to answer the phones. We'll do the usual things—developers' kits, conferences, seminars, T-shirts, hats, all that stuff." The prospect seems to overcome him with ennui, yet he appears convinced it will work.
Currently, about 25 people are refining concepts and hardware for DigiScents, according to vice president of business development John Williams, the third member of DigiScents' four-man executive team. Williams, who has known the founders since their Stanford days, previously directed "bizdev" for Spray, now part of Razorfish, and Red Dot, which was absorbed into AnswerThink Consulting Group. By September, on behalf of DigiScents, he had entered into talks with not only Eidos but also Nintendo, Comcast, InterAct Accessories, and Sega. At press time no deals had closed.
R&D has been rapid: Just eight months have elapsed from concept to prototype. The exterior design of the unit may change slightly to make it resemble the kind of loudspeakers sold with multimedia PC setups—"Reekers, instead of speakers," Bellenson comments. The original iSmell demo unit, built by GeneMachines in San Carlos, California, contains just 36 wells of odor oil. The finished product should have at least 128 primaries, enabling, says Bellenson, 12850 odors. "Not the entire scent space," he adds, "but a substantial portion of it."
Odor output will be boosted automatically for users who live at high altitudes, where vapor dissipates more rapidly. Also, some people may choose to inhibit odors they don't like, just as audiophiles can use a graphic equalizer to fine-tune a stereo system to suit their tastes and perceptions. But I wonder if there's a liability problem here: Some odors could be intolerable to allergy sufferers. "We'll have product disclaimers, naturally," says Bellenson.
"I think aesthetic disclaimers will be more important," adds Canter. "You know, when PageMaker was first released, it created a lot of really ugly pages. I'll be surprised if 10 percent of the first smell output is bearable."
This is, after all, a totally new art form.
"We know when the first visual art was done, in cave paintings," Canter continues. "And the first musical art consisted of tribal people beating drums. Think of all the books written about musical and visual arts since then. Now show me the library on smells."
Avery Gilbert used to run a scent consultancy company named Synesthetics, and was vice president of sensory research at Givaudan-Roure, a division of Hoffmann-LaRoche. He also serves as head of scientific affairs for the Olfactory Research Fund. Gilbert has joined DigiScents to help establish some aesthetic guidelines. But, as Canter implies, odor generation and movie scent tracks won't be controlled purely by experts. Bellenson and Smith ardently believe that we should all gain the same kind of creative power over odor synthesis that we acquired when desktop publishing revolutionized graphic design. Thus, we will brew our own odors, save them as digital files, and email them as attachments—whiffs, not GIFs.
Imagine: Digitized party invitations could carry the aroma of food, wine, or more exotic stimulants. Kiss-off email can smell rancid—or worse. "Make money fast!" chain letters can arrive smelling of new dollar bills. And spam could smell of Spam—or perhaps emit a scent that's so irresistible, you'll let the text linger on your screen just to enjoy it.
On the downside, empowerment always brings an element of risk. Hackers will write their own iSmell software, with unforeseeable results. Virus designers may circulate smell files that cause computers to crank out odious odors for hours at a time.
Meanwhile, if DigiScents is a hit, it will spawn competitors. Already, a Massachusetts company called MicroChips has announced a chip containing many tiny compartments whose liquid, gel, or solid contents can be released individually. An early news item suggested using this system for odor emittance, but Bellenson doesn't think that's viable. "We called the company and asked about it because we thought maybe we could have our software work with their devices," he says. "But the amount of odor they could deliver is minuscule." Originally the MicroChips gadget was intended as an under-the-skin drug-release system, which still seems its most plausible purpose.
Other smell generators have been proposed, and prototypes have been built, but on a less commercial level. Scientists at the Illinois Institute of Technology in Chicago, for example, have a patent pending on a peripheral device, similar to a disk drive, that releases odors from an absorbent pad, but it's not as far along as the iSmell.
Bellenson sounds disappointed that serious competitors don't exist yet. "We want to license our software and our index to run with anybody's synthesizer," he says. "Let them come up with their own primaries, and we'll translate from theirs to ours. We're trying to grow a whole industry, not just our business."
And if it does catch the public imagination, where will it end?
DigiScents has no plans to scale the product for auditoriums. A movie theater would require huge odor synthesizers—and, in the era of multiplexes, elaborate ventilation equipment would be needed to clear the air before the next smell.
But since small-scale smell generation is possible, could it be followed by smell capture? Some of today's digital cameras can record a five-second sound clip with each photo. What would it take for a camera to store a digitized odor sample as well?
"There are a number of paths to a smell camera," Bellenson says. "Gas chromatography works by sucking air through a liquid, then analyzing the liquid. The big fragrance houses already do this. In rain forests, they capture scents from exotic flowers by sucking the air around the flower. Of course, today this process still costs tens of thousands of dollars."
Artificial noses already exist in a rudimentary form. Led by Paul E. Keller, an Environmental Molecular Sciences Lab team at the Pacific Northwest National Laboratory in Richland, Washington, developed a prototype nose in the early '90s that identified eight household chemicals: acetone, correction fluid, contact cement, glass cleaner, isopropyl alcohol, lighter fluid, rubber cement, and vinegar.
More recently, Marc Madou, who runs a microsensors program in the materials research lab at Ohio State University in Columbus, has developed sophisticated gas sensors attached to a neural network.
"Every major university has someone studying chemical olfaction," Bellenson says. A Caltech technology is at the heart of a company called Cyrano Sciences whose smell-sensitive polymers swell to provide a range of odor signatures. "These things are very useful for finding explosives or drugs, or bad batches of food on an assembly line in a warehouse," he adds.
So far, the synthetic nostrils don't know how to select—and capture—the odors we care about and recognize. But if this embryonic technology finds an audience, we could, among other things, create electronic scrapbooks filled with smell clips from our travels, stimulating intense nostalgic experiences. According to Bellenson, "A smell camera could be two to three years away."
In the meantime, while the whole field of smell-enabled computers remains untested, one application seems inevitable. Virtual reality already uses 3-D displays and tactile feedback from data gloves, vests, and chairs. Odor would be a natural extension of this experience.
The concept has long been discussed in academic papers. Woodrow Barfield and Eric Danas examined it in "Comments on the Use of Olfactory Displays for Virtual Environments," published in the winter 1995 issue of Presence. This year, in a paper for the University of Washington, Martin Zybura and Gunnar A. Eskeland discussed the challenges in "Olfaction for Virtual Reality," suggesting that merely wafting odors toward a VR user won't be enough, because "an olfactory interface would have to use head orientation and position tracking to present a localized smell. Further, a separate display for each nostril is required to provide internasal time and intensity differences."
There's at least one business plan out there that defines a commercial odor-enhanced VR environment—and is looking for investors. Digital Tech Frontier promises a Virtual Scentsations kiosk for amusement arcades with "up to three different virtual experiences that stimulate visual, aural, and even olfactory senses by delivering localized fragrances near the user's nose."
This sounds severely limited when compared with iSmell and its billions of primary-scent combinations. Considering the track record for breakneck R&D at DigiScents, and Pangea Systems' history of marketing unique biologically oriented software, DigiScents seems a good ways ahead of any possible competitors. As cyberspace matures into a totally immersive experience, I'm betting it will turn out to be fully odor-enabled.
Just clear your nasal palate with coffee beans before you jack in.
IMAGES: FACE; VIEWPOINT DIGITAL, INC.; SMOKESTACK: ©ROBERT ECKERT/STOCK, BOSTON/PNI; GRAPES: ©KEVIN MORRIS/ALLSTOCK/PNI; FLOWERS: ©GREG VAUGHN/ALLSTOCK/PNI; FRIES: ©MICHAEL NEWMAN/PHOTOEDIT/PNI; BEANS: ©BURKE; TRIOLO; FOODPIX/PNI; CIGARETTES: ©HENRYK T. KAISER/PHOTO NETWORK/PNI; FISH: CORBIS/BARNABAS BOSSHART; ORANGES: ©DAVID RYAN/PHOTO 20-20/PNI
Contributing editor Charles Platt profiled Bill Gross, founder of idealab!, in WIRED 7.09.
