Millions of colors in the visible spectrum can be generated by mixing red, green and blue - the RGB color system. Is there a basic set of smells that, when mixed, can yield all, or nearly all detectable smells ?

  • 6
    $\begingroup$ It would be so cool if this could be used in video games. Like if the player walks into this room, make the smell equal to $x$. But I think we have to wait a long time until this becomes possible haha. $\endgroup$
    – Kevin
    Commented Jan 16, 2017 at 16:37
  • 1
    $\begingroup$ Not an answer, but instead of and RGB style system (with 200 receptors) some of the more experimental systems mix common sents to get a new smell. There are a few "sents" that make up a large selection of artificial smells in our food an fragrances. $\endgroup$
    – coteyr
    Commented Jan 16, 2017 at 20:57
  • 12
    $\begingroup$ Actually, RGB can't reproduce the entire visible spectrum. If you're curious, read up on the color triangle. The mind-blowing thing for me is that how the primary colors were chosen is somewhat arbitrary, but they do correspond to where each type of cone (we have three types) is most responsive. $\endgroup$ Commented Jan 16, 2017 at 23:53
  • 1
    $\begingroup$ It's probably also worth mentioning that there's no consistent theory for what smell actually is - i.e. alternatives like vibration theory which describe things shape theory can't (but vice-versa is also true). Once it's known how it actually works then it can be broken down into its basic components and categorized into the resulting "human perception spectrum" like RGB does for light. $\endgroup$ Commented Jan 18, 2017 at 10:20
  • 2
    $\begingroup$ @Kevin you make me nostalgic. Back in 2001 (I was 10 then) computer magazines were crazy over scent reproduction for movies, web perfume catalogs, web tours (click a door, wait 10 seconds to download, repeat). I remember them noting that audiovisual storytelling will elicit desired reactions (mountains evoke awe) but smells wont (rotten wood may evoke nostalgia instead of the intended disgust). Here is a link from those times: wired.com/1999/11/digiscent $\endgroup$ Commented Jan 19, 2017 at 19:55

3 Answers 3


There are about 100 (Purves, 2001) to 400 (Zozulya et al., 2001) functional olfactory receptors in man. While the total tally of olfactory receptor genes exceeds 1000, more than half of them are inactive pseudogenes. The combined activity of the expressed functional receptors accounts for the number of distinct odors that can be discriminated by the human olfactory system, which is estimated to be about 10,000 (Purves, 2001).

Different receptors are sensitive to subsets of chemicals that define a “tuning curve.” Depending on the particular olfactory receptor molecules they contain, some olfactory receptor neurons exhibit marked selectivity to particular chemical stimuli, whereas others are activated by a number of different odorant molecules. In addition, olfactory receptor neurons can exhibit different thresholds for a particular odorant. How these olfactory responses encode a specific odorant is a complex issue that is unlikely to be explained at the level of the primary neurons (Purves, 2001).

So in a way, the answer to your question is yes, as there are approximately 100 to 400 olfactory receptors. Just like the photoreceptors in the visual system, each sensory neuron in the olfactory epithelium in the nose expresses only a single receptor gene (Kimball). In the visual system for color vision there are just three (red, green and blue cones - RGB) types of sensory neurons, so it's a bit more complicated in olfaction.

- Purves et al, Neuroscience, 2nd ed. Sunderland (MA): Sinauer Associates; 2001
- Zozulya et al., Genome Biol (2001); 2(6): research0018.1–0018.12

- Kimball's Biology Pages

  • 8
    $\begingroup$ It's fascinating. Are there people who can approximately evaluate the 200-parameter RGB smell counterpart? Do dogs have more than 200 or is their smell superiority caused by something else? $\endgroup$ Commented Jan 16, 2017 at 14:15
  • 9
    $\begingroup$ @LubošMotl - Dogs have higher sensitivity, i.e., more receptors in terms of sheer numbers. Their olfactory spectrum is not so different than ours afaik $\endgroup$
    – AliceD
    Commented Jan 16, 2017 at 14:18
  • 2
    $\begingroup$ Thanks! Maybe they're less pretentious, too. I think that if a man weren't feeling painful, he could smell quite some details about many things and other people, too. ;-) $\endgroup$ Commented Jan 16, 2017 at 14:19
  • 4
    $\begingroup$ Also, dogs get a lot more practice, starting from a very young age. Though it turns out humans aren't too bad at it if they really try. Read Surely you're joking Mr Feynman for details. $\endgroup$ Commented Jan 16, 2017 at 21:52
  • 1
    $\begingroup$ @LubošMotl AliceD's second paragraph alludes to the fact that the olfactory responses don't have a great correspondence to the chemical structure... There are certainly variations in odor sensitivity and discrimination between people, but it isn't at all clear that the 200 "smell motifs" have any particular meaning individually. This seems like a difference from RGB where there is a perception of the individual colors, but maybe not: for example, it seems silly from a perceptual standpoint to say that white is "more red" than black, though this is clear in RGB or physical terms. $\endgroup$
    – Bryan Krause
    Commented Jan 17, 2017 at 22:19

There are many, many more parameters than 200! As an example, look at the nomenclature system for olfactory receptors (ORnXm).

  • "OR" is the root name (Olfactory Receptor superfamily)
  • n = an integer representing a family (e.g., 1-56) whose members have greater than 40% sequence identity,
  • X = a single letter (A, B, C, ...) denoting a subfamily (>60% sequence identity),
  • m = an integer representing an individual family member (isoform)

Members belonging to the same subfamily of olfactory receptors (>60% sequence identity) are likely to recognize structurally similar odorant molecules.

So, if we have 56 families, and each family has 26 possible subfamilies that can each detect a specific range of molecules, then you have a 1456-characters "alphabet" to describe all possible smells. Now humans will have more or less of each type, some types will not be there but be present in dogs, etc... Now if you take any actual smell (complex molecules with multiple odorants), it will be represented by any number of specific receptors going on at the same time. So if we set a limit of 100 receptors activated simultaneously per whiff (imaginary, no idea what the real number is), there are 1456^100 possible combinations of receptor activations for any 100-odorants smell.


Glusman G, Bahar A, Sharon D, Pilpel Y, White J, Lancet D (November 2000). "The olfactory receptor gene superfamily: data mining, classification, and nomenclature". Mammalian Genome. 11 (11): 1016–23. doi:10.1007/s003350010196. PMID 11063259.

Malnic B, Godfrey PA, Buck LB (February 2004). "The human olfactory receptor gene family". Proceedings of the National Academy of Sciences of the United States of America. 101 (8): 2584–9. Bibcode:2004PNAS..101.2584M. doi:10.1073/pnas.0307882100. PMC 356993Freely accessible. PMID 14983052.

Glusman G, Yanai I, Rubin I, Lancet D (May 2001). "The complete human olfactory subgenome". Genome Research. 11 (5): 685–702. doi:10.1101/gr.171001. PMID 11337468.

  • 1
    $\begingroup$ This seriously needs a reference or two. $\endgroup$
    – bpedit
    Commented Jan 16, 2017 at 19:03
  • 9
    $\begingroup$ Having the possibility to name such a big number of genes in a family doesn't actually mean, that there are so many genes. Your first reference tells us that there are 330 human olfactory receptors (rather than 1456), so please read these references. $\endgroup$
    – Chris
    Commented Jan 16, 2017 at 19:17
  • 7
    $\begingroup$ This seems to be addressing how many smells there can be, rather than how many axes. Compare: with RGB, there are three axes, but we often encode colors using a level from 0 (black) to 255 (very bright) on each axis, giving 256^3 colors. By comparison, you are saying one way to think of smell is as a 330-axis thing with only levels 0 (not present) and 1 (present), which would give 2^330 possible smells, but then constrained so that at most 20 axes are 1 at a time, which reduces the space to 330 choose 20 (not quite as big as your 330^20). But that doesn't mean there are 330 choose 20 axes! $\endgroup$ Commented Jan 16, 2017 at 21:53
  • 10
    $\begingroup$ Where is your evidence that there are 26 possible subfamilies - from the 26 letters in the alphabet? I'm terribly sorry, but biology doesn't work that way. $\endgroup$
    – MattDMo
    Commented Jan 16, 2017 at 22:45
  • 2
    $\begingroup$ @AliceD The ability of man to discern millions of colors is a bit dependent on circumstance... The studies that report people can determine millions of colors are really talking about contrast sensitivity. The number goes way way down if you would instead present colors sequentially. Even the paint colors in a hardware store can't be reliably distinguished when they aren't held together. $\endgroup$
    – Bryan Krause
    Commented Jan 18, 2017 at 16:13

Yes, it certainly is. But I cannot guarantee that all have been properly determined.

For that, we would have to find the basic unique values (like the Red, Green and Blue in the case of colors).

Currently, we have 10 unique human detectable smells: The wood-like smell, the floral/fragrant one, non-citrus, chemical, sharp(garlic-like), sweet, menthol, peanut-like, lemon and decay/dead-like (those are labeled by researchers as W1 to W10). A reference of the 10.

The your RGB-like-values for smell should be something like that.


You must log in to answer this question.

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