Scientists discovered how to capture sights and sounds, to record and replicate visual and audio information long ago. But mechanically capturing smells has proved trickier. Despite more than a century of evolving, ever more sophisticated audio-visual technologies, scientists still don’t have a satisfying method for recording scents. But researcher David Harel of the Weizmann Institute of Science in Israel believes he has created a theoretical system that could, eventually, help experts more accurately mimic aromas.

Popular Science explains that Harel recently posted a paper to arXiv, an electronic archive, that imagines a theoretical model for producing and assessing artificial scents. Harel explains that part of the challenge when it comes to replicating aromas is judging whether the replicas are accurate: While it’s relatively easy to look at a photograph and judge whether it’s an accurate representation of its subject, scent perception is more difficult to judge. As Popular Science notes, not only do odors depend on our brain’s perception of molecules, but human language is generally limited when it comes to describing aromas.

Harel argues that, in order to accurately replicate smells, a “sniffer” (a device which analyzes a scent) and a “whiffer” (which would create the replica scent based on the sniffer’s analysis) are necessary. Like a camera, the theoretical scent replicator would reproduce scents automatically.

But, in order to judge the accuracy of this theoretical machine, argues Harel, something akin to the Turing Test would be needed. While the Turing Test was designed to determine the existence of artificial intelligence, Harel’s test would judge the success of artificial aromas. In essence, Harel argues, the key to creating replica aromas may lie as much in the tests we design to assess them as in the replicating technologies themselves. Harel writes, “To put it simply, we would like to make sure that if and when promising methods for artificial olfaction are proposed, we will be prepared with carefully worked-out evaluation methods.”

[h/t Popular Science]