Unsolved Mystery The Human Sense of Smell: Are We Better Than We Think? Gordon M. Shepherd

“... a complete, comprehensive understanding of by reassessing several overlooked other mammals—is directly correlated odor ... may not seem a profound enough problem factors: the structure of the nasal cavity, with the evolutionary decline in the to dominate all the life sciences, but it contains, retronasal smell, olfactory brain areas, human sense of smell. piece by piece, all the mysteries.” — Lewis Thomas and language. In these arenas, humans may have advantages which outweigh Behavioral Studies ne of the oldest beliefs about their lower numbers of receptors. It Although these conclusions seem human perception is that appears that in the olfactory system, incontrovertible, they are challenged Owe have a poor sense of olfactory receptor genes do not map by some recent behavioral studies. smell. Not only is this a general belief directly onto behavior; rather, behavior One type of study shows that much of among the public, but it appears to is the outcome of multiple factors. If the olfactory system can be removed have a scientifi c basis. Recent genetic human smell perception is better than with no effect on smell perception. studies show a decline in the number we thought, it may have played a more The olfactory receptor genes map of functional olfactory receptor genes important role in human evolution topographically onto the fi rst relay through primate evolution to humans. than is usually acknowledged. station, a sheet of modules called Human evolution was characterized glomeruli in the olfactory bulb. Up to by the gradual ascendance of vision Gene Studies 80% of the glomerular layer in the rat and reduction of smell, evidenced From rodents through the primate can be removed without signifi cant in the anthropological record by the series to humans there is a progressive effect on olfactory detection and progressive diminution of the snout as reduction in the proportion of discrimination (Bisulco and Slotnick the eyes moved to the middle of the functional olfactory receptor genes 2003). If the remaining 20% of the face to subserve depth vision (Jones et (Rouquier et al. 2000; Gilad et al. glomeruli—and the olfactory receptor al. 1992). Concurrently, the use of an 2004). Mice have approximately 1,300 arboreal habitat and the adoption of olfactory receptor genes, of which some Copyright: © 2004 Gordon M. Shepherd. This is an an erect posture moved the nose away 1,100 are functional (Young et al. 2002; open-access article distributed under the terms of from the ground, with its rich varieties Zhang and Firestein 2002), whereas the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduc- of odors. humans have only some 350 functional tion in any medium, provided the original work is However, some recent behavioral genes of approximately 1,000 (Glusman properly cited. studies suggest that primates, including et al. 2001; Zozulya et al. 2001). Gordon M. Shepherd is in the Department of Neu- humans, have relatively good senses The conclusion seems obvious: the robiology at the Yale University School of Medicine, of smell. Resolution of this paradox low number of functional olfactory New Haven, Connecticut, United States. E-mail: may come from a larger perspective receptor genes in humans compared [email protected] on the biology of smell. Here we begin with rodents—and presumably most DOI: 10.1371/journal.pbio.0020146

PLoS Biology | http://biology.plosjournals.org May 2004 | Volume 2 | Issue 5 | Page 0572 genes they represent—can subserve the to conscious perception of ordinary By this hypothesis, during human functions of 1,100 genes, it implies that smells. Pheromones, and the rich world evolution the snout could be reduced 350 genes in the human are more than of unconscious effects of odors and in dimensions and complexity without enough to smell as well as a mouse. pheromones, are beyond the present compromising the ultimate amounts Another type of study has tested scope (cf. Jacob et al. 2004), though of odorized air reaching the olfactory smell perception in primates, and they undoubtedly will add to the epithelium. The reduced snout allowed has shown that, despite their reduced general conclusions. the eyes to come forward and lie closer olfactory receptor gene repertoire, together to promote more effective primates, including humans, have The Filtering Apparatus stereoscopic vision. Thus, vision could surprisingly good senses of smell (Laska of the Nasal Cavity become more dominant in humans et al. 2000). Comparing the data on A marked difference between the without sacrifi cing unduly the sense smell detection thresholds shows that noses of primates and other mammals is of smell. Tests of this hypothesis are humans not only perform as well or that in nearly all nonprimate mammals, needed, including calculations of better than other primates, they also the nasal cavities contain at the front air fl ows and odor losses through perform as well or better than other a much-convoluted fi ltering apparatus the fi ltering apparatus in mammals mammals. When tested for thresholds (formed by the ethmo- and maxillo- with extensive fi ltering apparatuses to the odors of a series of straight-chain turbinals) covered with respiratory compared with the simpler nasal (aliphatic) aldehydes, dogs do better membrane. This fi ltering apparatus is a cavities of primates. on the short chain compounds, but biological air conditioner (Negus 1958) humans perform as well or slightly with three key functions: cleaning, Humans Receive Richer better than dogs on the longer chain warming, and humidifying the inspired Retronasal Smells compounds, and humans perform air. An important function of the Being carried in with inhaled air signifi cantly better than rats (Laska fi ltering apparatus is presumably to (the orthonasal route) is not the et al. 2000). Similar results have been protect the nasal cavity from infections. only way for odor molecules to reach obtained with other types of odors. In many mammals, air drawn into the the olfactory receptor cells. Odor A third type of study demonstrating nose is often highly contaminated with molecules also reach the olfactory human olfactory abilities shows that bacteria from fecal material, decaying receptor cells via the retronasal in tests of odor detection, humans animal and plant material, and noxious route, from the back of the oral cavity outperform the most sensitive fumes from the environment, all of through the nasopharynx into the measuring instruments such as the gas which attack the olfactory epithelium. back of the nasal cavity. Although the chromatograph. Rodents are susceptible to chronic orthonasal route is the one usually These results indicate that humans rhinitis, which causes substantial loss used to test for smell perception, the are not poor smellers (a condition of functioning olfactory receptor cells retronasal route is the main source of technically called microsmats), but (Hinds et al. 1984). the smells we perceive from foods and rather are relatively good, perhaps This fi ltering, however, might liquids within our mouths. These are even excellent, smellers (macrosmats) have negative consequences for odor the smells that primarily determine the (Laska et al. 2000). This may come detection. Warming and humidifi cation hedonic (i.e., pleasurable or aversive) as a surprise to many people, though presumably enhance the odor- qualities of foods, and that, combined not to those who make their living stimulating capacity of the inhaled with taste and somatosensation, form by their noses, such as oenologists, air, but cleaning would remove odor the complex sensation of fl avor. It is perfumers, and food scientists. Anyone molecules by absorbing them into likely, for several reasons, that this is an who has taken part in a wine tasting, or the lining of the epithelium, an effect important route for smell in humans. observed professional testing of food which could be large depending on First, with the adoption of fl avors or perfumes, knows that the the size of the fi ltering apparatus. If so, bipedalism, humans became human sense of smell has extraordinary mammals with large snouts might have increasingly wide ranging, with capacities for discrimination. a large inventory of olfactory receptors concomitant diversifi cation of diet and at least in part to offset the loss of odor retronasal smells. Second, the advent The Mystery molecules absorbed by the fi ltering of fi re, perhaps as early as 2 million Here, then, is the mystery: how apparatus. years ago (Wrangham and Conklin- can one reconcile a relatively high How do these considerations relate Brittain 2003), made the human diet sensitivity to smell with a relatively to humans? The evolution of humans more odorous and tasty. From this time low number of olfactory receptors in involved lifting the nose away from also one can begin to speak of human the nose? To answer this question, the noxious ground environment as cuisines of prepared foods, with all I think we need to look beyond they adopted a bipedal posture (Aiello their diversity of smells. Wrangham and the olfactory receptor genes and and Dean 1990). This would have Conklin-Brittain (2003) support the consider olfaction in its full behavioral reduced the need for the fi ltering view that prepared cuisines based on context. This requires considering apparatus and with it the losses of cooked foods are one of the defi ning several overlooked aspects of the absorbed odor molecules. The large characteristics of humans. Third, added olfactory system: the nasal cavity, the numbers of olfactory receptors and to the cooked cuisines were fermented oropharyngeal cavity, the olfactory receptor cells would have come under foods and liquids, with their own strong brain, and the role of language. In this reduced adaptive pressure and could fl avors. These developments occurred article I focus on behaviors related accordingly be reduced in proportion. among the early hunter-gatherer

PLoS Biology | http://biology.plosjournals.org May 2004 | Volume 2 | Issue 5 | Page 0573 human cultures and continued through The reduced repertoire of olfactory et al. 2003). It has been hypothesized the last ice age. With the transition to receptor genes in the human is thus that these odor images provide the agricultural and urban cultures 10,000 offset by the expanded repertoire of basis for discrimination between odors, years ago, human cuisines changed by higher brain mechanisms. Rather than analogous to the way that retinal the advent of animal domestication, being restricted to a tiny part of the images are the basis for discrimination plant cultivation, use of spices, and brain, olfactory processing of complex of visual pattern stimuli. The complex of complex procedures, such as those smells, such as those produced by patterns constituting odor images may for producing cheeses and wines, all human cuisines, draws on the enlarged be considered as analogous to the of which produced foodstuffs that processing capacity of the human complex patterns constituting visual especially stimulate the smell receptors brain. images of faces. And just as we are very in the nose through the retronasal good at recognizing a human face, yet route and contribute to complex Language Is Necessary have diffi culty describing it in words, fl avors. for Human Smell we have a hard time describing and These considerations suggest the In the enlarged processing capacity verbally comparing odor images. hypothesis that the retronasal route for perceiving and discriminating Because of this diffi culty, describing for smells has delivered a richer odors, language plays a critical role. a smell or a taste in words is very repertoire of smells in humans than This seems paradoxical, for we have demanding. A professional wine in nonhuman primates and other great diffi culty describing a smell tasting, for example, requires many mammals (see Figure 1). Research on in words. Insight into this diffi culty steps: analysing both orthonasal and retronasal olfaction is being actively comes from the fi nding that different retronasal perception, comparing the pursued (reviewed in Deibler and smells are represented in the olfactory two in memory with each other and Delwiche 2004). Studies are needed of bulb by different patterns of olfactory with all other wines to be compared, the evolutionary pressures on this route glomerular activity. These patterns identifying the constituent properties in addition to the pressures on the function as virtual “odor images” (Xu separate from the hedonic qualities, evolution of the snout. Humans Smell with Bigger and Better Brains Comparisons of the decreasing size of the olfactory system relative to expansion of the visual, auditory, and somatosensory systems usually focus on the olfactory bulb and lateral olfactory tract, which are relatively small. However, what matters more are the central olfactory brain regions that process the olfactory input as the basis for smell perception. These regions are more extensive in humans than is usually realized. The dedicated olfactory regions include the olfactory cortex, the olfactory tubercle, the entorhinal cortex, parts of the amygdala, parts of the hypothalamus, the mediodorsal thalamus, the medial and lateral orbitofrontal cortex, and parts of the insula (Neville and Haberly 2004). These regions are involved in immediate processing of odor input and probably subserve the specifi c tasks of smell detection and simple smell discrimination. For more complex DOI: 10.1371/journal.pbio.0020146.g001 tasks, memory becomes important in comparing smells, thus involving Figure 1. Hypothetical “Odor Wheel” Representing and Comparing the Odor Worlds of Mouse and the temporal and frontal lobes Human The inner part represents the different categories of odors for the mouse; the relative (e.g., Buchanan et al. 2003) and the importance of each category for mouse smell-dependent behavior is indicated by the specifi cally human higher association area of each wedge. The outer part represents the same categories for the human; the areas. It may be hypothesized that these importance of each category for human smell-dependent behavior compared with the regions enable humans to bring far mouse is indicated by the area of each wedge. Note the greater importance of food odors for the human, refl ecting the factors discussed in the text. Note also the retention more cognitive power to bear on odor of some sensitivity in humans to social odors and other odors prominent in rodents, discrimination than is possible in the though in many cases to still undetermined degrees. Based on numerous sources and rodent and other mammals. the hypotheses discussed in the text.

PLoS Biology | http://biology.plosjournals.org May 2004 | Volume 2 | Issue 5 | Page 0574 and fi nding the words to describe of all the relevant mechanisms that are and a Multiuniversity Research Initiative the process as it unfolds, leading to involved. grant from the Department of Defense. the fi nal formulation to characterize It may be hoped that the hypotheses the quality of the wine and identify it and mechanisms discussed here can References as distinct from all others. It may be help to address and resolve the mystery Aiello L, Dean C (1990) An introduction to human evolutionary anatomy. New York: characterized as hard cognitive work of the apparent noncorrelation of Academic Press. 608 p. that only a human, among all the olfactory receptor gene numbers with Bisulco S, Slotnick B (2003) Olfactory animals with olfactory organs, can do. It smell acuity, and in doing so stimulate discrimination of short chain fatty acids in rats with large bilateral lesions of the olfactory may be argued that this is what humans a major reassessment of human smell bulbs. Chem Senses 28: 361–370. are adapted to do (Wrangham and perception. Such an effort cuts across Buchanan TW, Tranel D, Adolphs R (2003) A specifi c role for the human amygdala in Conklin-Brittain 2003). many academic disciplines. Molecular olfactory memory. Learn Mem 10: 319–325. This cognitive work is largely biologists need to continue their Deibler KD, Delwiche J, editors (2004) Handbook independent of the numbers of efforts to characterize the olfactory of fl avor characterization: Sensory analysis, chemistry, and physiology. New York: Marcel peripheral receptor cells and their genomes of humans and nonhuman Dekker.493 p. genes. A good analogy is with language. mammals more closely, to compare Gilad Y, Wiebe V, Przeworski M, Lancet D, There are some 17,000–20,000 auditory how different organisms sample odor Pääbo S (2004) Loss of olfactory receptor genes coincides with the acquisition of full nerve fi bers in the rat and cat and space. Physiologists need to devise trichromatic vision in primates. PLoS Biol. 2: e5 some 25,000–30,000 in the human high-throughput systems to test these doi://10.1371/journal.pbio.0020005. Glusman G, Yanai I, Rubin I, Lancet D (2001) (cf. Hall and Masengill 1997). This odor spaces. Behavioral neuroscientists The complete human olfactory subgenome. modest increase in the input from the need to develop increasingly accurate Genome Res 11: 685–702. peripheral auditory receptors provides tests of olfactory function that enable Hall RD, Massengill JL (1997) The number of primary auditory afferents in the rat. Hear Res little basis for the development of comparisons across different species. 103: 75–84. human speech and language, which Psychologists need to explore even Hinds JW, Hinds PL, McNelly NA (1984) An had much more to do with the increase more vigorously the subtle ways that autoradiographic study of the mouse olfactory epithelium: Evidence for long-lived receptors. in the central brain mechanisms smells can infl uence human behavior. Anat Rec 210: 375–383. that elaborate the input. It may be Anthropologists and paleontologists Jacob S, Spencer NA, Bullivant SB, Sellergren SA, Mennella JA, et al. (2004) Effects of hypothesized that a similar conclusion need to study the olfactory parts of breastfeeding chemosignals on the human applies to human olfaction. the cranium and face from this new menstrual cycle. Hum Reprod 19: 422–429. perspective, to reassess the role that Jones S, Martin R, Pilbeam D (1992) The Implications for Systems Biology Cambridge encyclopedia of human evolution. both orthonasal and retronasal smell New York: Cambridge University Press. 520 p. A general result from these may have played in primate and human Laska M, Seibt A, Weber A (2000) “Microsmatic” considerations is that there appears not evolution. primates revisited: Olfactory sensitivity in the squirrel monkey. Chem Senses. 25: 47–53. to be a one-to-one relation between The factors reviewed here suggest Negus V (1958) The comparative anatomy and the number of olfactory receptor genes that the sense of smell is more physiology of the nose and paranasal sinuses. and the detection and discrimination London: E & S Livingstone. 402 p. important in humans than is generally Neville KR, Haberly LB (2004) Olfactory cortex. of odors. This implies that we are realized, which in turn suggests that In: Shepherd GM, editor. The synaptic dealing with a fundamental problem it may have played a bigger role in organization of the brain, 5th ed. New York: Oxford, pp. 415–454. in relating genes to systems behavior: a the evolution of human diet, habitat, Rouquier S, Blancher A, Giorgi D (2000) The given set of genes may not map directly and social behavior than has been olfactory receptor gene repertoire in primates onto a given behavior. In this respect appreciated. All of these considerations and mouse: Evidence for reduction of the functional fraction in primates. Proc Natl Acad the mystery being addressed here is should stimulate a greater interest in Sci U S A. 97: 2870–2874. a caution for the new era of “systems this neglected sense. Wrangham R, Conklin-Brittain N (2003) Cooking biology” and against any belief that as a biological trait. Comp Biochem Physiol A Mol Integr Physiol 136: 35–46. behavior can be related directly to Acknowledgments Xu F, Liu N, Kida I, Rothman DL, Hyder F, et genomes, proteomes, or any other type I am grateful to C. Greer, L. Bartoshuk, and al. (2003) Odor maps of aldehydes and esters D. Waddle-Stock for valuable discussions. revealed by functional MRI in the glomerular of “-ome.” We are reminded instead layer of the mouse olfactory bulb. Proc Natl that the functional ecology of the body The work of my laboratory has been Acad Sci USA 100: 11029–11034. is dependent on many factors. supported by the National Institute for Young JM, Friedman C, Williams EM, Ross JA, Deafness and other Communicative Tonnes-Priddy L, et al. (2002) Different Disorders, the Human Brain Project evolutionary processes shaped the mouse and Conclusions human olfactory receptor gene families. Hum (National Institute for Deafness and Much about the sense of smell seems Mol Genet 11: 535–546. other Communicative Disorders, National Zhang X, Firestein S (2002) The olfactory receptor enigmatic and confl icting. This is partly Institute for Neurological Diseases and gene superfamily of the mouse. Nat Neurosci because of the inherent diffi culties in Stroke, National Institute for Mental Health, 5: 124–133. presenting smell stimuli, and partly Zozulya S, Echeverri F, Nguyen T (2001) The National Institute for Aging, and National human olfactory receptor repertoire. Genome because there is not yet a recognition Institute for Alcoholism and Drug Abuse), Biol 2: RESEARCH0018.

PLoS Biology | http://biology.plosjournals.org May 2004 | Volume 2 | Issue 5 | Page 0575