Smell identification and the role of labels

ABSTRACT

There has recently been a reevaluation of our sense of smell, which is now considered a very sensitive and discriminating sense modality by scientists and philosophers. However, the consensus in the literature is that humans, and certainly Western subjects, are very poor at identifying smells: they produce the “veridical label” for an odor in just 30–50% of cases and there is wide inter-subjective variation in their responses. This suggests that we rarely know what we smell. Is this the right conclusion to draw from the evidence? This paper takes a closer look at the empirical evidence on the smell naming performance of Western subjects and argues that a comparative model of olfactory language and categorization is more effective at explaining the evidence than a model on which each smell kind is supposed to correspond to one label. One result of applying a comparative model is that we are not quite as poor at naming smells as the commonly cited data would suggest. Another result is a better understanding of the kinds of knowledge we may gain by smelling and how these relate to the linguistic resources, experiences, and practices of different speakers and communities.

KEYWORDS:

• Smell
• olfaction
• identification
• labels
• olfactory language

Acknowledgements

For their helpful comments and questions on previous versions of this paper, I am grateful to audiences at the universities of Milan, Bergamo, and Tübingen, and to members of the Philosophy of Mind Twitter work-in-progress group. Thanks also to Hong Yu Wong and Simon Wimmer for discussion and encouragement, and to two reviewers for this journal for their constructive suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. The estimated magnitude of discriminable odourants is debated, but Young and colleagues note that the majority of the 166 billion molecules with 17 or less atoms that have been studied has a smell that can be distinguished from that of all others; additionally, mixtures of chemicals smell different from the sum of their components (Young et al., 2014).

2. Thai speakers also consistently use a range of smell descriptors (Wnuk et al., 2020), though some are generic terms for pleasant and unpleasant smells, similar to the English “fragrant” and “stinky”, and some are compounds formed with those generic terms.

3. This is a guarded exception, with some researchers pointing out that these subjects’ smell naming performance is still significantly inferior to the color naming performance of Western subjects (Olofsson & Pierzchajlo, 2021).

4. The earlier studies report that information about the odor and its source is available to subjects who cannot name the odor, while the later studies report that this information is very limited. The evidence is difficult to conclusively assess also because responses are not always scored for accuracy (Jönsson & Olsson, 2012, p. 126).

5. Holley also says that “odor naming turns out to be odor-source naming”. The thesis I defend, following Martina (2022), is instead that we name, and talk about, odors using words for odor sources.

6. Engen (1987) connects the collocation model to the idea that episodic memory plays a distinctive role in olfactory cognition, claiming that “what is stored about odors is not likely to involve semantic categories … Rather, odor memory involves perceptually unitary episodes” (1987, p. 501). The comparative model proposed here is neutral on this issue: categorization and label choice are affected by personal past experiences, but this does not per se show that identification is based on episodic memory specifically.

7. Dubois and Rouby worry that the “veridical label” approach unjustifiably presupposes that there are names that really refer to odors. I do not think this is quite the problem: as we have seen, the words we do have, even if source-based, can be used to characterize smells, albeit comparatively.

8. Besides, given that the stimulus was an artificial odourant, we cannot exclude that the scented cleaner was even a better example than lemon for the kind of smell presented.

9. As a reviewer pointed out, research on perceptual categorization indicates that experts tend to categorize entities in a domain at a different level of abstraction than non-experts (e.g., bird experts would categorize a particular bird as “sparrow”, non-experts as “bird”, cf. Johnson & Mervis, 1997). This raises interesting questions about potential differences between visual and olfactory categorization, and about the relation between labels and categories in the two domains.

10. Interestingly, in the last study, 84% of subjects who had named licorice odor as “anise”, chose the label “anise” in a subsequent cued identification task even when the label “licorice” was available.

11. This description is actually not so idiosyncratic: the authors of another study decided to include a French label referring to adhesive paste or glue as a second correct label for bitter almond smell (Sulmont et al., 2002).

12. It is controversial whether these different responses amount to changes in olfactory perception itself (e.g., Martina, 2021); this issue does not affect the current discussion.

13. I discuss some of this evidence further in Section 3.3.

14. Cued identification with very few alternative labels, the authors observe, is not the best way to investigate identification because subjects may proceed by elimination of alternative labels.

15. Sometimes, subjects identify an aspect or note in a qualitatively complex smell, say the apricot note in a coffee; they may convey that the current smell given off by coffee is similar to the smell to the typical smell of apricots just in some respects, without conveying that it is so similar overall that the current smell may be coming from apricots, rather than coffee. For a discussion of multifaceted smells, see Martina (2021).

16. On the different communicative uses of comparative language, see Martina (2022, Sec. 3).

17. For further discussion of these explanations, see e.g., Majid (2021), Olofsson and Gottfried (2015), Jönsson and Olsson (2012).

18. Majid (2021) provides a summary of relevant studies.

19. Some of this data comes from research developing effective tests to diagnose impairments and disorders of olfactory sensitivity, such as anosmia and hyposmia, where cued identification is often taken as a proxy for olfactory discrimination capacities.

20. Performance on an even simpler identification task suggests that these associations can be made quickly: when subjects are first presented with a label (e.g., “wood”), then with the odor, and only need to press a button to indicate whether the odor corresponds to the label, they can respond within 200 to 5000 milliseconds after stimulus delivery, with close to “ceiling-level” accuracy (Olofsson et al., 2012).

Additional information

Funding

This work was supported by a Thyssen Stiftung Postdoc Fellowship; and the PRIN Project ‘The Mark of the Mental’ (Grant 2017P9E9NF) at the University of Turin.

Notes on contributors

Giulia Martina

Giulia Martina is a postdoctoral researcher in the philosophy of perception and mind. She received her PhD at the University of Warwick in 2020 and has worked as a postdoc at the universities of Salzburg, Turin, and Tübingen. Her research focuses on olfaction, perceptual variation, perceptual language, and the metaphysics of perceivable entities.