We can encode sound and images in radio waves and send them, but presumably there's some physiological reason that we can't easily make a picture or video of a smell. Could we realistically break smells down into "primary scents", in analogy with "primary colors" and encode the information for easy reproduction?

  • $\begingroup$ I guess ultimately you could say the reason is the nature of the sense you're trying to reproduce. While seeing, hearing and touch are all senses based on non-material physical phenomena (sound waves, light rays, forces), the sense of smell serves as a tool to sample gaseous chemicals. This is why it's so intimately connected to taste, which is sampling the chemicals in solid and liquid matter. $\endgroup$
    – Armatus
    Commented Jul 24, 2015 at 11:25

4 Answers 4


There are only three kinds of optical receptors in the eye, but more than 900 kinds of olfactory receptors. Thus you can encode pictures with the three primary colors, but there is no small set of primary scents.

To transmit a smell via "primary scents", you'd have to create an artificial nose that monitors the response of each of the olfactory receptors, and then at the other end release a chemical mixture that has the exact same response pattern. (You won't find a one-to-one mapping of chemical compound to receptor.)

  • 8
    $\begingroup$ I think this is only half the answer. Encoding > 900 different olfactory signals shouldn’t be too big a problem – just create a distinctive signature for each molecule. We have this already in chemistry databases. The problem is actually converting the electrical signal back to a chemical one, since the alternative, plugging directly into the nerves of the nose, would be slightly unpleasant. $\endgroup$ Commented May 3, 2012 at 12:20
  • 2
    $\begingroup$ @KonradRudolph did you see the 2nd paragraph? ;) $\endgroup$ Commented May 3, 2012 at 12:44
  • $\begingroup$ I believe creating mixes of 900 substances could be rather hard. OTOH, electronic neurostimulation - bypassing the receptors and stimulating their respective nerves directly would be more doable (if still difficult). $\endgroup$
    – SF.
    Commented Nov 22, 2014 at 23:46
  • 1
    $\begingroup$ one question: just how specifically can one stimulate those receptors. The fact that any wavelength of light stimulates all three photoreceptors to some degree has the interesting consequence that it's impossible to mimic the full color space with 3 primary colors. I wonder if scents would have a similar limitation. $\endgroup$
    – Shep
    Commented Nov 24, 2014 at 23:22

In expansion to biocs' excellent answer, I would like to highlight some practical limitations of this. Suppose we did manage to create a huge database of exact chemical mixtures which produce all smells recognisable by humans. You would still meet some complications:

  • The output device (analogous to headphones or screen) would either need to be able to produce those substances on demand or have a large enough stack of them.
    • The former would require a whole lab built into one device - even with our miniature technology this would probably end up rather large.
    • The latter would require you to regularly stock up on all of the substances. You could sell these in an all-in-one package (again size becomes a problem) or in smaller "smell sets" (possibly practicable - but it would take a long time until people would understand what this is about)
  • The "smell substances" have to get to your nose somehow. At the same time, you probably don't want them in the rest of the room: While sound boxes are no problem because sound disappears rather quickly after being produced, smell can stick around for longer. The solution would be an apparatus that you insert into your nose. Not very practical, but people might do it.

I can't think of more at the moment but there probably are :)


In my view, the reason we can't transmit smell is that we don't understand it. That is, we don't have a solid understanding of how odor information is coded, so it is hard to imagine how to build a system that could reproduce that information.

This is sometimes discussed in terms of a multi-dimensional "odor space," where each odor could be described by its location along each dimension. The problem is that we don't really know what any of the dimensions are. By contrast, we know that visual space can be described by the intensity of a set of colors in a two-dimensional plane; auditory space is described by frequency and amplitude. It is clear that odors are comprised of chemicals, but we don't yet know where those chemicals fit in odor space. Understanding the odor space is a prerequisite to building a system that could transmit messages from it.

Also, to a great extent, how to engineer a smell transmission system is only partly a biological question. For instance, a CMYK printing system does a fine job at generating images that our RGB photoreceptors understand just fine. A telephone wire can transmit sound without a care about how the cochlea actually works. Transmitted pictures and transmitted sounds work because they activate our photoreceptors and auditory hair cells in the same pattern that the original images and sounds would. We can build systems to do this because we understand the nature of the information sensory systems decode, and we don't need a really deep understanding of the physiology to do that. (Of course, I think figuring out that physiology is great fun, but that's another point.) That is, to build a smell system, it's probably not really necessary to know what every one of the many olfactory receptors does. That's a detail of the way the mammalian olfactory system evolved. The important thing is to understand the nature of odor information. Then it really may be theoretically feasible to create a system to transmit odors.


A smell capsule with primary smells, concentrated non runny oils that can be heated up by a code sent to the capsule Then it will release the code, a built in fan will blow it to you from pc or phone, on a pc it could be renewed, maybe plugged in to a usb, Ive been thinking about this lately an have come to the conclusion that a smell Capsule would only need a collection of smells, if you type out banana there is a code behind the keys for the letter to appear, i think a smell program will be needed so that if some one sends you a coded image of a banana from there smell program the soon as you receive it and click on it the code will activate the same code in your program and you will smell the banana i have no idea how many primary smells there are or even if you can mix them but a collection of most common smells will do the job till the primary and mixing is masted, I believe a pc is missing the smell capability and it will come along just as many other things have, it would be great to smell the ocean while watching ships or even while sailing one on a simulator, a heating of a oil scent will do the job fine, A thin wire sitting in each concentrated thick scented oil and all the different oils in separate compartments with there individual codes inside a complex capsule and a fan to blow out the heated one that the thin wire heated up

  • 1
    $\begingroup$ Can you add some references? $\endgroup$ Commented Jul 22, 2014 at 9:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .