Strolling with my dog became a daily ritual, one that introduced me to a rich array of scents that had previously eluded me amidst the hustle of LA life. It was the first time I had noticed the seasonal variation in smells, and I began collecting an eclectic mix of herbs, flowers, and fruit on every walk. One day it dawned on me: Why not encapsulate the distinctive blend of Southern California sage, citrus, and rosemary into a candle? This idea was not just about preserving the fragrance; it was about holding onto a particular moment in time.

What started as a quest to capture the scent of my neighborhood has become a full-blown obsession. Crafting these scents is both a science and an art, a delicate balance of blending base, middle, and top notes, guided by a fragrance wheel. Despite my arsenal of essential oils, meticulously gathered over the years, my ability to perfectly recreate specific scents remained elusive.

The German philosopher Friedrich Nietzsche once described the sense of smell as “our most refined instrument,” a poignant testament to its ability to help us understand the world around us. Scent wields the unique power to anchor us to a specific moment or emotion, effortlessly transporting us through time to cherished memories and forging deep, inherent connections with people and places. 

Given the elusive nature of scents in my candle-making endeavors, I began exploring how to harness machine learning and advances in sensory science to not only capture but also faithfully recreate these evocative, complex scents.

Since the 1980s, researchers have been developing instruments/hardware called electronic noses (or “e-noses”) to detect discrete scent molecules. While their evolution has been significant, the full potential of e-noses remains largely untapped in commercial products. Researchers in the field believe they are on the brink of a major breakthrough, however, with the development of the Principal Odor Map (POM).

Just as we use RGB and CMYK as foundational color guides, enabling a spectrum of color combinations and hues, the Principal Odor Map offers a similar guide for scents. By defining primary odors, we open the door to recreating any conceivable scent, capturing the multidimensional essences that our noses perceive. Wild, right?

Although the quest for a perfect scent model continues, an AI algorithm known as the graph neural network is well on its way to making it a not-so-distant reality. It’s trained on fragrance-industry datasets of more than 5,000 molecules, with their structures converted into graphs and tagged. So far, it has demonstrated an ability to detect scents with an accuracy akin to human descriptors, marking a significant step toward digitally capturing and replicating the nuanced world of aroma.

The POM has several practical limitations, and further rounds of research will be needed before it can be commercialized.  But it made me wonder what this digitized olfactory future might look like in physical form. What novel interactions might we encounter with this technology, and what design principles are defined within the space? How can we more intentionally integrate scent into our daily lives and accurately recall and reproduce those scent experiences at any time?

To answer some of these questions, I created the concept of the ScentCapture sphere. It is a manifestation of my innermost olfactory desires. Think of it as an abstraction analogous to a camera or a field recorder, but strictly for scent.

ScentCapture can use digitized e-noses to record, organize, tag, and combine smells to create AI-generated unique scents. With a simple click of a button and an interactive touchscreen that fits comfortably in your palm, it provides a portable way to capture smells. The form factor of ScentCapture is designed for convenience. Using the capture button is similar to taking a photo with a camera, while the shape plays into my love of the Poké Ball and its ability to capture and store Pokémon. It’s designed to be light and simple to use. 

After a scent is captured, the device maps it to its closest scent relative—showing the presence of that scent and providing pairing and creative naming suggestions based on the base, middle, and top notes. You can select a range of capture dates to create an amalgamation of scents or encapsulate a singular capture. Once a user has created a custom scent combination, they can submit it to a factory to produce an instant digital sample, similar to Givaudan’s innovative AI scent tool, “Carto.” Once the sample is created, the custom scent can be formulated into a perfume, candle, or essential oil. Looking further into the future, ScentCapture could generate smells on demand and be deployed into diffusers to fill entire spaces.

With this device in hand on one of my walks through Los Angeles, I could effortlessly record the fragrance of jasmine clusters and a ripe guava tree, whose branches drape over my neighbor’s fence. When there’s an elusive aroma I can’t name, my ScentCapture could become my olfactory cartographer, mapping the aromatic journey of my stroll, and identifying the unknown scent as “bay laurel” mixed with earthy undertones of “silty soil.”

During business trips, when I feel waves of homesickness, I could rely on my ScentCapture to transport me back home. I would select the profile I’ve named “Springtime in LA,” and feel those pangs gently fade away.

The possibility of deploying synthetic scents makes me wonder how ScentCapture could improve memory, particularly for individuals with conditions such as dementia. A recent study at the University of California, Irvine, found that when older adults were exposed to a range of scents each night, their memories improved measurably. Another recent study at the University of Oxford found that the olfactory system is highly responsive to training; researchers wrote that the “sense of smell may facilitate transfer of learning to other sensory domains.”

The ability to recreate smells opens doors for our perception of other humans and the natural world. Scent philosopher Annick Le Guérer encapsulates that line of thinking. “Smell is revelatory not just of substances, but also of moods, climates, and even existential states. It is a subtle tool of knowledge that allows for an intuitive and prelinguistic understanding.”

I’m reminded of Jose Chavarria’s explorations into altering human perception of reality through a series of animal-inspired masks. His speculative design project, titled Interface, enables us to experience the way other creatures perceive the world. His bat-inspired echolocation mask simulates the sensation of bouncing sound waves to perceive distance, while his python-inspired infrared-sensing mask enables users to detect infrared emissions, giving them a sense of the nearness of other living creatures.

Drawing inspiration from Chavarria's work and the beauty of nature, we might move toward more organic forms where ScentCapture becomes one of many biomimetic multisensory devices resembling a barnacle.

I imagine this multi-sensory device would be equipped with modular attachments, allowing the user to choose which senses to amplify, fundamentally transforming our perception of reality. With the ability to precisely replicate every sensory experience, we can paint a more comprehensive and vivid picture of our collective human experiences.

Such a device could profoundly enhance our understanding of the self, foster deeper connections among people, and even expand our empathy toward other species. It has the potential not only to replicate experiences but also to bridge between different realms of perception, offering us the opportunity to experience the world through an entirely new lens. This exploration of sensory integration could be a key to unlocking a deeper, shared understanding of our world and each other.