A measure of smell enables the creation of olfactory metamers

ABSTRACT Wavelength is a physical measure of light, and the intricate understanding of its link to perceived colour enables the creation of perceptual entities such as

metamers—non-overlapping spectral compositions that generate identical colour percepts1. By contrast, scientists have been unable to develop a physical measure linked to perceived smell,

even one that merely reflects the extent of perceptual similarity between odorants2. Here, to generate such a measure, we collected perceptual similarity estimates of 49,788 pairwise

odorants from 199 participants who smelled 242 different multicomponent odorants and used these data to refine a predictive model that links odorant structure to odorant perception3. The

resulting measure combines 21 physicochemical features of the odorants into a single number—expressed in radians—that accurately predicts the extent of perceptual similarity between

multicomponent odorant pairs. To assess the usefulness of this measure, we investigated whether we could use it to create olfactory metamers. To this end, we first identified a cut-off in

the measure: pairs of multicomponent odorants that were within 0.05 radians of each other or less were very difficult to discriminate. Using this cut-off, we were able to design olfactory

metamers—pairs of non-overlapping molecular compositions that generated identical odour percepts. The accurate predictions of perceptual similarity, and the ensuing creation of olfactory

metamers, suggest that we have obtained a valid olfactory measure, one that may enable the digitization of smell. Access through your institution Buy or subscribe This is a preview of

subscription content, access via your institution ACCESS OPTIONS Access through your institution Access Nature and 54 other Nature Portfolio journals Get Nature+, our best-value

online-access subscription $32.99 / 30 days cancel any time Learn more Subscribe to this journal Receive 51 print issues and online access $199.00 per year only $3.90 per issue Learn more

Buy this article * Purchase on SpringerLink * Instant access to full article PDF Buy now Prices may be subject to local taxes which are calculated during checkout ADDITIONAL ACCESS OPTIONS:

* Log in * Learn about institutional subscriptions * Read our FAQs * Contact customer support SIMILAR CONTENT BEING VIEWED BY OTHERS PHYSICOCHEMICAL FEATURES PARTIALLY EXPLAIN OLFACTORY

CROSSMODAL CORRESPONDENCES Article Open access 30 June 2023 ODOR DISCRIMINATION IS IMMUNE TO THE EFFECTS OF VERBAL LABELS Article Open access 31 January 2023 A DATASET OF LAYMEN OLFACTORY

PERCEPTION FOR 74 MONO-MOLECULAR ODORS Article Open access 26 February 2025 DATA AVAILABILITY All data generated during this study are included in the Article and its Supplementary

Information. All the odorants used are included in Supplementary Table 1, all behavioural similarity results are included in Supplementary Table 2 and all behavioural discrimination results

are included in Supplementary Table 3. An additional external dataset used can be found in the supplementary material of a previously published study15. CODE AVAILABILITY The custom code

used to process the data collected in this study is available at https://gitlab.com/AharonR/olfaction. REFERENCES * Wandell, B. A. _Foundations of Vision_ (Sinauer Associates, 1995). * Bell,

A. G. Discovery and invention. _Natl Geogr. Mag_. 25, 649–655 (1914). Google Scholar  * Snitz, K. et al. Predicting odor perceptual similarity from odor structure. _PLOS Comput. Biol_. 9,

e1003184 (2013). Article  CAS  Google Scholar  * Khan, R. M. et al. Predicting odor pleasantness from odorant structure: pleasantness as a reflection of the physical world. _J. Neurosci_.

27, 10015–10023 (2007). Article  CAS  Google Scholar  * Zarzo, M. & Stanton, D. T. Understanding the underlying dimensions in perfumers’ odor perception space as a basis for developing

meaningful odor maps. _Atten. Percept. Psychophys_. 71, 225–247 (2009). Article  Google Scholar  * Koulakov, A. A., Kolterman, B. E., Enikolopov, A. G. & Rinberg, D. In search of the

structure of human olfactory space. _Front. Syst. Neurosci_. 5, 65 (2011). Article  Google Scholar  * Keller, A. et al. Predicting human olfactory perception from chemical features of odor

molecules. _Science_ 355, 820–826 (2017). Article  ADS  CAS  Google Scholar  * Weiss, T. et al. Perceptual convergence of multi-component mixtures in olfaction implies an olfactory white.

_Proc. Natl Acad. Sci. USA_ 109, 19959–19964 (2012). Article  ADS  CAS  Google Scholar  * Zhou, Y., Smith, B. H. & Sharpee, T. O. Hyperbolic geometry of the olfactory space. _Sci. Adv_.

4, eaaq1458 (2018). Article  ADS  Google Scholar  * Cain, W. S. Odor intensity: differences in the exponent of the psychophysical function. _Percept. Psychophys_. 6, 349–354 (1969). Article

  Google Scholar  * Olsson, M. J. An integrated model of intensity and quality of odor mixtures. _Ann. NY Acad. Sci_. 855, 837–840 (1998). Article  ADS  CAS  Google Scholar  * Halpern, S.

D., Andrews, T. J. & Purves, D. Interindividual variation in human visual performance. _J. Cogn. Neurosci_. 11, 521–534 (1999). Article  CAS  Google Scholar  * Thiede, T. et al. PEAQ—the

ITU standard for objective measurement of perceived audio quality. _J. Audio Eng. Soc_. 48, 3–29 (2000). Google Scholar  * Yuhong, Y. et al. Auditory attention based mobile audio quality

assessment. In _IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)_ 1389–1393 (IEEE, 2014). * Bushdid, C., Magnasco, M. O., Vosshall, L. B. & Keller, A.

Humans can discriminate more than 1 trillion olfactory stimuli. _Science_ 343, 1370–1372 (2014). Article  ADS  CAS  Google Scholar  * Cain, W. S. Differential sensitivity for smell: “noise”

at the nose. _Science_ 195, 796–798 (1977). Article  ADS  CAS  Google Scholar  * Booth, D. A. & Freeman, R. P. Discriminative feature integration by individuals. _Acta Psychol. (Amst.)_

84, 1–16 (1993). Article  CAS  Google Scholar  * Prins, N. _Psychophysics: A Practical Introduction_ (Academic, 2016). * Ennis, J. M., Ennis, D. M., Yip, D. & O’Mahony, M. Thurstonian

models for variants of the method of tetrads. _Br. J. Math. Stat. Psychol_. 51, 205–215 (1998). Article  Google Scholar  * Ennis, D. M. The power of sensory discrimination methods. _J. Sens.

Stud_. 8, 353–370 (1993). Article  Google Scholar  * Hamwi, V. & Landis, C. Memory for color. _J. Psychol_. 39, 183–194 (1955). Article  Google Scholar  * Rousseau, B. Meyer, A. &

O’Mahony, M. Power and sensitivity of the same‐different test: comparison with triangle and duo‐trio methods. _J. Sens. Stud_. 13, 149–173 (1998). Article  Google Scholar  * Stillman, J. A.

& Irwin, R. J. Advantages of the same‐different method over the triangular method for the measurement of taste discrimination. _J. Sens. Stud_. 10, 261–272 (1995). Article  Google

Scholar  * Laska, M. & Teubner, P. Olfactory discrimination ability of human subjects for ten pairs of enantiomers. _Chem. Senses_ 24, 161–170 (1999). Article  CAS  Google Scholar  *

Sela, L. & Sobel, N. Human olfaction: a constant state of change-blindness. _Exp. Brain Res_. 205, 13–29 (2010). Article  Google Scholar  * Mainland, J. D. et al. The missense of smell:

functional variability in the human odorant receptor repertoire. _Nat. Neurosci_. 17, 114–120 (2014). Article  CAS  Google Scholar  * Brainard, D. H. & Hurlbert, A. C. Colour vision:

understanding #TheDress. _Curr. Biol_. 25, R551–R554 (2015). Article  CAS  Google Scholar  * Jameson, D. & Hurvich, L. M. Theoretical analysis of anomalous trichromatic color vision. _J.

Opt. Soc. Am_. 46, 1075–1089 (1956). Article  ADS  Google Scholar  * Rüfer, F. et al. Age-corrected reference values for the Heidelberg multi-color anomaloscope. _Graefes Arch. Clin. Exp.

Ophthalmol_. 250, 1267–1273 (2012). Article  Google Scholar  * Meister, M. On the dimensionality of odor space. _eLife_ 4, e07865 (2015). Article  Google Scholar  * Gerkin, R. C. &

Castro, J. B. The number of olfactory stimuli that humans can discriminate is still unknown. _eLife_ 4, e08127 (2015). Article  Google Scholar  * Mamlouk, A. M., Chee-Ruiter, C., Hofmann, U.

G. & Bower, J. M. Quantifying olfactory perception: mapping olfactory perception space by using multidimensional scaling and self-organizing maps. _Neurocomputing_ 52–54, 591–597

(2003). Article  Google Scholar  * Fan, M., Qiao, H. & Zhang, B. Intrinsic dimension estimation of manifolds by incising balls. _Pattern Recognit_. 42, 780–787 (2009). Article  Google

Scholar  * Camastra, F. Data dimensionality estimation methods: a survey. _Pattern Recognit_. 36, 2945–2954 (2003). Article  Google Scholar  * Haddad, R. et al. Global features of neural

activity in the olfactory system form a parallel code that predicts olfactory behavior and perception. _J. Neurosci_. 30, 9017–9026 (2010). Article  CAS  Google Scholar  * Kleiner, M. et al.

What's new in psychtoolbox-3. _Perception_ 36, 1–16 (2007). * Pelli, D. G. The VideoToolbox software for visual psychophysics: transforming numbers into movies. _Spat. Vis_. 10,

437–442 (1997). Article  CAS  Google Scholar  * Brainard, D. H. The psychophysics toolbox. _Spat. Vis_. 10, 433–436 (1997). Article  CAS  Google Scholar  * Dragon: software for the

calculation of molecular descriptors v.6.0 (Talete srl, 2011). * Macmillan, N. A. & Creelman, C. D. _Detection Theory: A User’s Guide_ (Psychology Press, 2004). * Rousseau, B. &

Ennis, D. M. A Thurstonian model for the dual pair (4IAX) discrimination method. _Percept. Psychophys_. 63, 1083–1090 (2001). Article  CAS  Google Scholar  * Kaplan, H. L., Macmillan, N. A.

& Creelman, C. D. Tables of d′ for variable-standard discrimination paradigms. _Behav. Res. Meth. Instrum_. 10, 796–813 (1978). Article  Google Scholar  Download references

ACKNOWLEDGEMENTS This work was primarily supported by the Horizon 2020 FET Open project NanoSmell (662629). Additional support from grant 1599/14 from the Israel Science Foundation, by a

grant from Unilever, and by the Rob and Cheryl McEwen Fund for Brain Research. AUTHOR INFORMATION AUTHORS AND AFFILIATIONS * Department of Neurobiology, Weizmann Institute of Science,

Rehovot, Israel Aharon Ravia, Kobi Snitz, Danielle Honigstein, Maya Finkel, Rotem Zirler, Ofer Perl, Lavi Secundo & Noam Sobel * DreamAir LLC, New York, NY, USA Christophe Laudamiel *

Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot, Israel David Harel Authors * Aharon Ravia View author publications You can also search for

this author inPubMed Google Scholar * Kobi Snitz View author publications You can also search for this author inPubMed Google Scholar * Danielle Honigstein View author publications You can

also search for this author inPubMed Google Scholar * Maya Finkel View author publications You can also search for this author inPubMed Google Scholar * Rotem Zirler View author publications

You can also search for this author inPubMed Google Scholar * Ofer Perl View author publications You can also search for this author inPubMed Google Scholar * Lavi Secundo View author

publications You can also search for this author inPubMed Google Scholar * Christophe Laudamiel View author publications You can also search for this author inPubMed Google Scholar * David

Harel View author publications You can also search for this author inPubMed Google Scholar * Noam Sobel View author publications You can also search for this author inPubMed Google Scholar

CONTRIBUTIONS A.R., K.S., L.S., D. Harel and N.S. developed the concepts. A.R. and N.S. designed experiments. A.R., R.Z. and M.F. ran experiments. A.R., K.S., O.P. and N.S. analysed data.

C.L. developed scent formulas. A.R., D. Honigstein, K.S., O.P. and N.S. constructed the web-tool. A.R., O.P., D. Harel and N.S. wrote the paper. CORRESPONDING AUTHORS Correspondence to

Aharon Ravia or Noam Sobel. ETHICS DECLARATIONS COMPETING INTERESTS The Office of Technology Licensing at the Weizmann Institute of Science is filing for patents on the algorithms developed

in this study. A small portion of this work was supported by a research grant from Unilever, a company with interests in the fragrance industry. Unilever had no input or impact on the design

of experiments, or on analysis and presentation of the results. C.L. is the owner of DreamAir LLC, a company with interests in the fragrance industry. DreamAir had no input or impact on the

analysis and presentation of the results. ADDITIONAL INFORMATION PEER REVIEW INFORMATION _Nature_ thanks Tatyana Sharpee and the other, anonymous, reviewer(s) for their contribution to the

peer review of this work. PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. EXTENDED DATA FIGURES AND

TABLES EXTENDED DATA FIG. 1 THE ODORANTS USED PROJECTED INTO PERCEPTUAL SPACE. A, As in the main text, the 148 molecules used across experiments overlaid on 4,046 molecules within the first

and second principal components of the 21-descriptor physicochemical space. B, The same molecules within the first and second principal components of perceptual space. Perceptual space data

for 470 molecules as background (data from previously published studies4,7), containing 115 of the 148 molecules that we used. C, Histograms showing the experiment odorant distribution on

each principal component (PC) in the range of PC1–PC6. The principal components were computed as in A, on the 21-descriptor physicochemical space. There is a large decline in the explained

variance from the third principal component onward. D, Histograms showing the distances between all odorant pairs, per experiment. The distances are summed (black line) for the overall

distribution. Although monomolecules were not used as a stimulus for discrimination, this is to show that there was no bias in their selection, because for each experiment the distances of

the pairs spanned a range of distances. EXTENDED DATA FIG. 2 EXPERIMENTAL FLOWCHART. Ordered depiction of the tasks across the seven reported experiments. EXTENDED DATA FIG. 3 FACTORING AND

PREDICTING ODORANT INTENSITY. A, B, Factoring odorant intensity. A, In experiment 1, the overall MC-odorant intensity could have been used to determine similarity, _n_ = 23 participants for

intensity ratings and 22 participants for similarity ratings. Correlation coefficient _r_ = −0.61, _P_ < 6 × 10−11, _n_ = 95 (_r_ = −0.57, _P_ < 6 × 10−11, _n_ = 91, for comparisons

excluding identical pairs). To check whether intensity similarity and angle-distance similarity account for overlapping information, we built a linear model considering the two factors. We

found that this two-factor model could account for larger variability than each of the models alone (adjusted _R_2 = 0.37 versus adjusted _R_2 = 0.32 for intensity difference and adjusted

_R_2 = 0.16 for angle distance). Both factors were significant in this model (both _P_ < 0.005). In other words, although intensity differences could explain variance in the results,

angle distance was a significant factor as well, and could explain independent variance. B, The same analysis for experiment 2. Here, MC-odorant intensity was weakly, albeit significantly

correlated with MC-odorant similarity (_n_ = 30 participants for intensity ratings and 29 participants for similarity ratings, correlation coefficient _r_ = −0.22, _P_ = 0.03, _n_ = 95) and

this correlation was entirely explained by comparing odorants to themselves, and once these comparisons were removed, the correlation was lost altogether (_r_ = 0.04, _P_ = 0.68, _n_ = 91

for comparisons excluding identical pairs). Thus, experiment 2 largely negated this overall concern. C–I, Predicting odorant intensity. C, Estimated performance of predicted intensity model

as correlation between actual and predicted intensity on _k_-fold test-set (Supplementary Methods). Expected variance estimated using cross-validation (_k_ varied according to the number of

molecules (_n_) used in each concentration; _k_ = 8, 10, 10 and 5, and _n_ = 134, 422, 346 and 58 for concentrations of 10−1, 10−3, 10−5 and 10−7, respectively). In the violin plot large

points are averages of _k_-folds, vertical lines are quartiles 2–3. All four models have correlations significantly larger than zero, with peak at the 10−3 concentration (average _r_ = 

0.67). D–I, We used the 10−3 concentration data (Supplementary Methods) to devise a predictive model for intensity ratings, this time excluding molecules used in experiments 1 and 2 to avoid

overfitting. D, G, Intensity predictions generated by this model for monomolecule intensities in experiments 1 (D) and 2 (G). The _x_ axis is actual intensity (averages of _n_ = 23

participants, 2 repetitions each for experiment 1; and _n_ = 29 participants, 3 repetitions each for experiment 2) and the _y_ axis is predicted intensity. We show correlations in black and

in red to be compatible with other panels, although no zero intensity odours were included. D, Correlation coefficient _r_ = 0.36, _P_ < 0.02, _n_ = 44 monomolecules. G, Correlation

coefficient _r_ = 0.68, _P_ < 7 × 10−7, _n_ = 43 monomolecules. E, H, Angle distance estimation using the intensity factor. The intensity factor was calculated based on predicted

intensity (D, G) as in Fig. 1e; these predicted factors were then used to model MC-odorants. Finally, angle distances between pairs of MC-odorants were calculated according to predicted

intensity compared to those obtained by rated intensity (as used in the main text). E, Correlation coefficient _r_ = 0.53, _P_ < 3 × 10−8, _n_ = 95 (_r_ = 0.29, _P_ < 6 × 10−3, _n_ = 

91 for comparisons excluding identical pairs). H, Correlation coefficient _r_ = 0.73, _P_ < 2 × 10−17, _n_ = 95 (_r_ = 0.56, _P_ < 7 × 10−9, _n_ = 91 for comparisons excluding

identical pairs). F, I, Prediction of measured similarity from angle distances calculated using predicted intensity (similar to Figs. 1f, 2c). In the scatter plot, each dot is a pairwise

comparison of MC-odorants; the _y_ axis shows their actual similarity as rated by participants (for experiment 1, _n_ = 22, 2 repetitions; for experiment 2, _n_ = 29, 2 repetitions) and the

_x_ axis shows their angle distance according to predicted intensity. Red regression lines include comparisons of identical MC-odorants (zero angle distance), black regression lines are with

those comparisons removed. F, Correlation coefficient _r_ = −0.50, _P_ < 3 × 10−7, _n_ = 95 (_r_ = −0.29, _P_ < 6 × 10−3, _n_ = 91 for comparisons excluding identical pairs). I,

Correlation coefficient _r_ = 0.74, _P_ < 9 × 10−19, _n_ = 95 (_r_ = 0.54, _P_ < 5 × 10−8, _n_ = 91 for comparisons excluding identical pairs). F, I, Correlations between previous and

current results were not significantly different. F, Experiment 1, difference between result using rated and predicted monomolecule intensities (_r_ = −0.41 and _r_ = −0.29, respectively)

was not significantly different (_Z_ = 0.91, _P_ = 0.36, two-tailed, _n_ = 91 comparisons). I, Experiment 2, same procedure, difference between _r_ = −0.69 and _r_ = −0.54 was not

significantly different (_Z_ = −1.62, _P_ = 0.011, two-tailed, _n_ = 91 comparisons). We summarize that this is a promising direction for the future, but beyond the scope of this manuscript.

EXTENDED DATA FIG. 4 VARIABILITY IN PREDICTIONS OF PERCEPTUAL SIMILARITY FROM STRUCTURE IN OLFACTION AND AUDITION. A, Recreation of Fig. 2c, which shows our underlying results, with the

point of maximal variance highlighted with a blue ellipse. B, Data extracted from figure 22 from a previously published study13, which shows the state-of-the-art predictions from around ad

2000 of sound similarity from sound structure (overlaying points may be missing, as these data were extracted from the graph). Correlation coefficient _r_ = −0.80, _P_ < 2 × 10−103, _n_ =

 462. C, Data extracted from figure 3 of a previously published study14, which shows the state-of-the-art predictions from around ad 2014 of sound similarity from sound structure. Note that

we formatted the data to compare the datapoints to our data by putting the data into the same graph colour and structure and by reversing the axes. Correlation coefficient _r_ = −0.84, _P_ 

< 3 × 10−26, _n_ = 96. D, Comparison of points of maximal variance across datasets (blue, olfaction; red and green, audition). In audition technology, the major standard is PEAQ—the ITU

standard for objective measurement of perceived audio quality. PEAQ defines the subjective difference grade, which is the equivalent of our ‘perceived similarity’, and the objective

difference grade (ODG), which is the equivalent of our ‘angle distance’. The field is tasked with developing different objective difference grades, which can be made of various combined

measures such as frequency, timbre, power, and so on. We observe that the overall correlation in audition is not very different from olfaction, and that the variability at a given physical

distance is perhaps even greater in audition compared with olfaction. EXTENDED DATA FIG. 5 FROM ANGLE DISTANCE TO PERCEIVED SIMILARITY. A, C, Scatter plots om which each dot is a pairwise

comparison of two odorants; the _y_ axis shows their actual similarity as rated by participants and the _x_ axis shows their distance according to the model. A, Data from the experiment

containing rose, violet, asafoetida and 11 additional MC-odorants. All comparisons containing rose are shown in red, all comparisons containing are shown violet in violet and all comparisons

containing asafoetida are shown in mustard (_n_ = 29 participants, 2 repetitions each). Correlation coefficient _r_ = −0.55, _P_ < 3 × 10−5, _n_ = 52 (_r_ = −0.31, _P_ < 0.03, _n_ = 

48 for comparisons excluding identical pairs). B, Rated similarity versus angle distance between rose, violet and asafoetida comparisons in this experiment. The rated similarity data (dark

blue) are the average of _n_ = 29 participants, mean of 2 repetitions. Data are mean ± s.e.m. Blue circles are individual ratings of similarity. C, Data from experiments 1 and 2 used for

model building, taken from Figs. 1f, 2c. Correlation coefficient _r_ = −0.66, _P_ < 3 × 10−25, _n_ = 190 (_r_ = −0.55, _P_ < 2 × 10−15, _n_ = 182 for comparisons excluding identical

pairs). D, End result of predicted versus actual similarity of rose, violet and asafoetida, rated similarity (dark blue) is as in B. Data for predicted similarity (light blue) presented as

mean prediction using the linear regression model described in C (red line); the error bars show the confidence intervals (_P_ = 0.05) for this model prediction. See Supplementary Methods

for transformation from angle distance to predicted similarity. EXTENDED DATA FIG. 6 VARIABILITY IN INDIVIDUAL PERFORMANCE. A, Performance displayed by individual participant rather than by

odorant comparison, sorted by performance. The _z_ axis and colour both code participant performance accuracy. White, 41.8% accuracy or _d_′ = 1; red, _d_′ < 1; blue, _d_′ > 1. B,

Performance displayed by individual participants rather than by odorant comparison, sorted by performance. Colour codes are shown for the participant _d_′ as estimated in Fig. 3c. white,

_d_′ = 1; red, _d_′ < 1; blue, _d_′ > 1. EXTENDED DATA FIG. 7 TESTING OF SIGNIFICANCE BY SHUFFLING. We randomly shuffled performance outcome in the previously published dataset15, and

in experiments 4–6. For each MC-odorant pair, we assigned performance (means of the participants) randomly 10,000 times, and then computed the correlation between angle distance and

‘shuffled’ performance. A, A copy of Fig. 3b. B, A set of 100 traces (randomly picked for visualization purposes) of a moving average of shuffled data, similar to the black line in A. Red

dashed line in A and B is performance of d′ = 1 (41.8% correct) C–F, Histogram of correlations between angle distance and shuffled performance. Red line is the correlation of the observed

data. C, The previously published data15. The correlation of observed data (_r_ = 0.50, _n_ = 310 comparisons) outperforms the correlation of shuffled data (_P_ < 10−4, _n_ = 10,000

repetitions). D–F, Angle distance is shown on a log scale. D, Experiment 4, the correlation of observed data (_r_ = 0.51, _n_ = 50 comparisons) outperforms the correlation of shuffled data

(_P_ < 10−4, _n_ = 10,000 repetitions). E, Experiment 5, the correlation of observed data (_r_ = 0.42, _n_ = 50 comparisons) is significantly stronger than the correlation of shuffled

data (_P_ = 0.0009, _n_ = 10,000 repetitions). F, Experiment 6, the correlation of observed data (_r_ = 0.53, _n_ = 40 comparisons) is significantly stronger than the correlation of shuffled

data (_P_ = 0.0013, _n_ = 10,000 repetitions). G–I, Same as D–F, only here angle distance was analysed using a linear rather than logarithmic scale. G, Experiment 4, the correlation of

observed data (_r_ = 0.61, _n_ = 50 comparisons) outperforms the correlation of shuffled data (_P_ < 10−4, _n_ = 10,000 repetitions). H, Experiment 5, the correlation of observed data

(_r_ = 0.43, _n_ = 50 comparisons) is significantly stronger than the correlation of shuffled data (_P_ = 0.0015, _n_ = 10,000 repetitions). I, Experiment 6, the correlation of observed data

(_r_ = 0.45, _n_ = 40 comparisons) outperforms the correlation of shuffled data (_P_ < 10−4, _n_ = 10,000 repetitions). J–L, Here we verify the validity of the choice of performance

threshold, namely _d_′ = 1, in our data. For this verification, we calculate the null distribution for _d_′ for the discrimination tasks in experiments 4–6. To generate a meaningful

distribution, we carefully choose the shuffling in this analysis. For our data, we shuffled the correct responses for each participant in each session, and assigned the responses to

different MC-odorant pairs. For each participant, we used a different label assignment; this way we disentangle the difficulty of the task, and produce a statistic on the frequency at which

one would expect each _d_′ by chance. The histograms of performance in the different experiments are shown in the case in which the data of the participants have been shuffled participants.

the red areas show the bottom and top 5%; the grey line is _d_′ = 1. J, Experiment 4. K, Experiment 5. L, Experiment 6. EXTENDED DATA FIG. 8 PERCEPTUAL INDEPENDENCE OF METAMERS. We wondered

whether metamers are simply instances of ‘olfactory white’. This would imply that the difference between (not within) the 3 metamer pairs would be under 0.05 radians. To address this

question, we measured the distances between the 3 metamer pairs, which are as follows: pair 1 and pair 2, 0.11 radians; pair 1 and pair 3, 0.13 radians; pair 2 and pair 3, 0.07 radians. In

other words, each metamer is a distinct odour. Moreover, we next compared the metamers to ‘olfactory white’. We selected the ‘best’ white from a previously published study8 and measured its

distance from each of the metamers. The obtained minimal distances were 0.25, 0.24 and 0.24, all of which are much higher than 0.05 radians. One may note that the white in the previous

study8 may not have been ‘true White’, as indeed that study did not have the underlying computational framework developed here. Moreover, that study was restricted to about 30 components. To

address this, we generated 1,000 virtual versions of white odours, by combining different sets of 100 components. We observe that all mean distances between the metamers and these whites

are above 0.1 radians, and that the minimal distance of any pair to any white is larger than 0.05 radians. A–C, Histograms show distances between current metamer pairs to the 1,000 different

white odours that we generated. Distance between one odour (of the metamer pair) to the whites is shown in blue, and distance between the other odour (of the metamer pair) to the whites is

shown in red. Circular points show distances of each odour in the pair to the three previously described white odours8. Each panel shows one of the three metamer pairs reported in this

paper. SUPPLEMENTARY INFORMATION SUPPLEMENTARY INFORMATION Supplementary Information document containing Supplementary Discussion and Supplementary Methods. REPORTING SUMMARY SUPPLEMENTARY

TABLE Supplementary Table 1: containing all manuscript odorants and their intensities. SUPPLEMENTARY TABLE Supplementary Table 2: containing all manuscript similarity ratings. SUPPLEMENTARY

TABLE Supplementary Table 3: containing all manuscript discrimination results. RIGHTS AND PERMISSIONS Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Ravia, A., Snitz, K.,

Honigstein, D. _et al._ A measure of smell enables the creation of olfactory metamers. _Nature_ 588, 118–123 (2020). https://doi.org/10.1038/s41586-020-2891-7 Download citation * Received:

20 December 2018 * Accepted: 19 August 2020 * Published: 11 November 2020 * Issue Date: 03 December 2020 * DOI: https://doi.org/10.1038/s41586-020-2891-7 SHARE THIS ARTICLE Anyone you share

the following link with will be able to read this content: Get shareable link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer

Nature SharedIt content-sharing initiative