Lexical olfaction recruits olfactory orbitofrontal cortex in metaphorical and literal contexts
Introduction
Recent theories of how lexical and semantic information is represented in the brain often point to embodied accounts (Pulvermüller, 2013). For example, comprehension of action words such as ‘lick’, ‘pick’ and ‘kick’ has been shown to activate portions of the motor and premotor cortices which are typically associated with movements of the tongue, arm and leg, respectively (Hauk, Johnsrude & Pulvermüller, 2004). Hence, bodily movements are partially simulated in response to words that refer to them. Empirical work in support of embodied accounts has mainly focused on concrete concepts, which are associated with one of the five senses (“arm”, “cake”, “bell”). However, little is known about how abstract concepts (e.g., “thought”, “indifference”) are represented. One way to address this question is to embed the same concrete word in different sentential contexts, with the result that the crucial word is used figuratively and therefore acquires a more abstract meaning. Figurative expressions include metaphors, which are thought of mapping abstract conceptual domains onto concrete ones, therefore facilitating comprehension (Gibbs et al., 2004, Lakoff and Johnson, 1980). For example, in the expression “She had a rough day”, the abstract concept of “bad” day is mapped onto the concrete concept of “rough” surface. It has been shown that, to understand such sentence, linguistic representations are activated, along with a simulation of the concrete domain of texture (Lacey, Stilla, & Sathian, 2012; for other domains, see Citron and Goldberg, 2014, Desai et al., 2011). This should be particularly true for novel metaphors (e.g., “My husband is an elephant.”) compared to more conventional, “dead” metaphors (e.g., “This is a hard task.”; Keysar et al., 2000, Lai and Curran, 2013). In idioms, whose meaning cannot be derived from the meanings of their constituting words (Cacciari, 2014, Gibbs et al., 1989), this effect of sensorimotor activity has been shown to be reduced or even absent (Raposo et al., 2009, Schuil et al., 2013; see also Cacciari et al., 2011).
Metaphors establish an analogical relationship between two concepts: a source, concrete concept and a target, abstract concept (Yang, Bradley, Huq, Wu & Krawczyk, 2013), which can be activated in parallel. In fact, using conventional textural metaphors such as “She had a rough day”, Lacey et al. (2012) showed activation of the parietal operculum, a somatosensory area that is texture-selective. This finding indicates that metaphors indeed activate areas associated with the source domain, namely perception of texture. Similar results were obtained by Citron and Goldberg (2014), who employed German gustatory metaphors like “Sie bekam ein süßes Kompliment.” (i.e., she received a sweet compliment) and compared these to their literal counterparts (“Sie bekam ein nettes Kompliment” – she received a nice compliment). Their main findings are an activation increase in a left inferior frontal cluster including gustatory cortices for metaphors compared to literal sentences as well as involvement of the left hippocampus and anterior parahippocampal gyrus, including the amygdala. These results provide further evidence for the grounding of metaphors in sensorimotor representations by generalizing to the taste domain. Furthermore, the authors concluded that the metaphorical formulations are more emotionally engaging than their literal renderings at an implicit processing level. In fact, the two types of stimuli had been matched for explicit emotional valence and arousal ratings, but still activated emotion-related brain areas (i.e., amygdala and hippocampus) during reading comprehension. While reverse inference could be a limitation of this interpretation, the latter finding is supported by a meta-analysis of neuroimaging studies (Bohrn, Altmann, & Jacobs, 2012), by physiological responses to translations of English metaphors (Rojo, Ramos, & Valenzuela, 2014), and has been replicated and generalized to other metaphors not restricted to taste (Citron, Güsten, Michaelis, & Goldberg, 2016). Nevertheless, this novel finding was not originally predicted by the authors. Importantly, the difference between metaphorical and literal expressions was not restricted to their metaphoricity but included the presence versus absence of a taste word. Therefore, the stronger amygdala response could be due to the use of taste words per se. Although a control condition showed no amygdala activation in response to taste words in isolation (“sweet” meant literally) compared to their counterparts (“nice”), this null finding could be due to less statistical power. A similar limitation was mentioned by Lacey et al. (2012) in their textural metaphor study.
One sensory system that has been mostly neglected regarding the neural representation of lexical meaning is olfaction. This sense offers very interesting characteristics that can further elucidate the nature of neural representations. Olfaction, as gustation, is a chemical sense. However olfaction, in contrast to gustation, is very difficult to verbalize (Wilson and Stevenson, 2006, Yeshurun and Sobel, 2010). While humans are able to distinguish among an incredible number of odors (McGann, 2017), there seems to be a lack of abstract olfactory categories in most languages, including German and English; however, other languages such as Jahai verbalize specific odor features in the same way in which German verbalizes different colors, shapes or textures (Majid & Burenhult, 2014; see also Wnuk & Majid, 2014). Concerning cognitive processes, verbalization of olfactory impressions seems critical as olfaction strongly relies on verbal translation for efficient working memory maintenance and episodic memory encoding (Jehl et al., 1997, Rabin and Cain, 1984). This particular link of olfaction to language makes it compelling to investigate.
Odors are processed as configurations, e.g., peanut butter, with no access to their chemical components, and this makes it very difficult to name odors (Howard & Gottfried, 2014). In the case of vision, instead, we have access to both configurations, e.g., a happy face, as well as to their individual features, i.e., nose, mouth, eyes, therefore allowing mapping precision between visual features and lexico-semantic representations. Hence, bottom-up connections between the odor and the language systems are weak, whereas top-down connections, i.e., activation of the olfactory neural network upon presentation of odor-related words, are much more robust (Olofsson & Gottfried, 2015).
Overall, the cortical response to odors remains poorly understood (Price, 1991, Weiss and Sobel, 2012, Wilson and Rennaker, 2010). From the piriform cortex, i.e., the main primary olfactory area, olfactory information is projected to the orbitofrontal cortex (OFC), which receives direct afferent connections from the piriform cortex as well as indirect connections from the dorso-medial nucleus of the thalamus; the latter constitutes the olfactory thalamo-cortical pathway (Wilson & Rennaker, 2010). Since the OFC receives direct afferent inputs also from the amygdala and the entorhinal cortex, thus from most primary olfactory areas, it is considered to be the secondary olfactory area (Gottfried & Zald, 2005). In fact, lesions to the OFC in humans have been reported to induce impairments in discrimination and identification of olfactory stimuli (Jones-Gotman and Zatorre, 1988, Zatorre and Jones-Gotman, 1991). We will use the terms ‘OFC’ (as identified in the functional localizer) and ‘secondary olfactory cortex’ interchangeably.
The present work aims to investigate the neural representation of lexical olfaction in literal and figurative language. The term lexical olfaction refers to the perceptual olfactory vocabulary. To prevent the above-mentioned lack of dissociation between metaphoricity and sensory source domain, the present study implements three experimental conditions: olfactory words were used to build conventional olfactory metaphors, their literal paraphrases, which did not contain any olfactory expression, and literal olfactory sentences. Furthermore, this work aims to examine whether olfactory metaphors are more emotionally engaging than literal olfactory sentences and non-olfactory literal paraphrases. A functional localizer of odor processing was run to examine the neural activation in more detail. The following predictions were generated: (1) the processing of lexical olfaction activates olfactory regions (despite the poor verbalization abilities), (2) reading olfactory metaphors shows a processing difference to reading literal olfactory expressions in olfactory regions, and (3) metaphoricity enhances activation in emotion-related areas.
Section snippets
Participants
Eighteen native German speakers (Mage = 25.83 years, SD = 5.04, age range = 19–39 years, 6 women) recruited by advertisement participated in the present study. Six additional participants were eliminated from the analysis due to technical equipment failure (five participants) and a pronounced functional and anatomical cerebral deviation (one participant). One participant reported to have learned speaking German not until kindergarten. All participants were right-handed by self-report,
Olfactory contrast for the functional localizer
The contrast of processing odorants (positive, negative and neutral) versus sniffing ambient air (across all conditions) revealed an olfactory network comprising five, left-lateralized regions (displayed in Fig. 2): the piriform cortex (primary olfactory cortex), the amygdala and hippocampus, the inferior OFC, the medial OFC and the inferior frontal gyrus (IFG). See Table 3.
Reading network
At the whole-brain level, the contrast of lexical conditions versus visual baseline (olfactory metaphors & literal
Discussion
The present work investigates the neural representation of lexical olfaction. Furthermore, it examines whether emotion-related areas are involved in the processing of olfactory metaphors relative to literal sentences. Olfactory metaphors, their literal paraphrases and literal olfactory sentences were silently read for comprehension while we recorded the hemodynamic responses of the participants using functional MRI.
There are three main findings regarding our predictions: (1) the processing of
Conclusion
Lexical olfaction recruits olfactory orbitofrontal cortex in metaphorical and literal sentential contexts. This result indicates that modality-specific areas of the human brain contribute to the neural representation of lexical meaning. Considering the weak connection between olfaction and language, it is particularly interesting to witness this sensory contribution in lexical olfaction. The absence of primary olfactory activation is probably due to olfactory hedonicity being central to
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Author contribution
JP, AKB, FC and LK designed the study. JP, AKB, PS, CM and LK collected data. JP, AKB, PS and LK analyzed data. BS and LK contributed with scientific support, supervision, and coordination. JP, AKB, PS, CM, FC, BS and LK wrote and reviewed the manuscript.
Conflict of interest
None.
References (66)
- et al.
Psychophysiological correlates of affects in human olfaction
Neurophysiologie Clinique
(2002) - et al.
An extension of olfactometry methods: An expandable, fully automated, mobile, MRI-compatible olfactometer
Journal of Neuroscience Methods
(2016) - et al.
Looking at the brains behind figurative language: A quantitative meta-analysis of neuroimaging studies on metaphor, idiom, and irony processing
Neuropsychologia
(2012) - et al.
Literal, fictive and metaphorical motion sentences preserve the motion component of the verb: A TMS study
Brain and Language
(2011) - et al.
Conventional metaphors in longer passages evoke affective brain response
NeuroImage
(2016) - et al.
Metaphor is grounded in embodied experience
Journal of Pragmatics
(2004) - et al.
How to kick the bucket and not decompose: Analyzability and idiom processing
Journal of Memory and Language
(1989) - et al.
Reading cinnamon activates olfactory brain regions
Neuroimage
(2006) - et al.
On the scent of human olfactory orbitofrontal cortex: Meta-analysis and comparison to non-human primates
Brain Research Reviews
(2005) - et al.
Somatotopic representation of action words in human motor and premotor cortex
Neuron
(2004)
The influence of olfactory-induced negative emotion on verbal working memory: Individual differences in neurobehavioral findings
Brain Research
Configural and elemental coding of natural odor mixture components in the human brain
Neuron
Olfactory identification deficits in patients with focal cerebral excision
Neuropsychologia
Bilateral brain processes for comprehending natural language
Trends in Cognitive Sciences
The two-level theory of verb meaning: An approach to integrating the semantics of action with the mirror neuron system
Brain and language
Conventional language: How metaphorical is it?
Journal of Memory and Language
Metaphorically feeling: Comprehending textural metaphors activates somatosensory cortex
Brain and Language
Odors are expressible in language, as long as you speak the right language
Cognition
Anatomic localization and quantitative analysis of gradient refocused echo-planar fMRI susceptibility artifacts
Neuroimage
The muted sense: Neurocognitive limitations of olfactory language
Trends in cognitive Sciences
A review and synthesis of the first 20 years of PET and fMRI studies of heard speech, spoken language and reading
Neuroimage
Semantic embodiment, disembodiment or misembodiment? In search of meaning in modules and neuron circuits
Brain and Language
Modulation of motor and premotor cortices by actions, action words and action sentences
Neuropsychologia
The emotional impact of translation: A heart rate study
Journal of Pragmatics
Functional neuroanatomy of different olfactory judgments
Neuroimage
fMRI of emotional responses to odors: Influence of hedonic valence and judgment, handedness, and gender
Neuroimage
Deriving meaning: Distinct neural mechanisms for metaphoric, literal, and non-meaningful sentences
Brain and Language
Meta-analyzing left hemisphere language areas: Phonology, semantics, and sentence processing
Neuroimage
Revisiting the limits of language: The odor lexicon of Maniq
Cognition
Contextual effects on conceptual blending in metaphors: An event-related potential study
Journal of Neurolinguistics
Functional neuroimaging of the olfactory system in humans
International Journal of Psychophysiology
Olfactory predictive codes and stimulus templates in piriform cortex
Neuron
Reading salt activates gustatory brain regions: fMRI evidence for semantic grounding in a novel sensory modality
Cerebral Cortex
Cited by (25)
Losing the sense of smell does not disrupt processing of odor words
2022, Brain and LanguageCitation Excerpt :The existing findings suggest the connection between language and olfaction is symmetric, with equally weak connections (for English speakers) between language and olfactory areas in comprehension as production (Majid, 2021; Speed & Majid, 2018). Instead, the evidence suggests odor language may involve high-level representations, such as hedonic information (Pomp et al., 2018; Speed & Majid, 2018). This contrasts with other behavioral paradigms that suggest single words can activate sensory odor information.
Making sense of sensory language: Acquisition of sensory knowledge by individuals with congenital sensory impairments
2022, NeuropsychologiaCitation Excerpt :Cross-sensory expressions, or synesthetic metaphors (cf., Day, 1996; Winter, 2018), are another linguistic strategy, wherein words typically associated with one sense describe another (e.g., “loud color”, “bright sound”). Intriguingly, cross-sensory expressions trigger neural activations associated with the source sense (e.g. vision for “loud color”; Lacey, Stilla, & Sathian, 2012; Citron and Goldberg, 2014; Pomp et al., 2018). This suggests that cross-sensory expressions facilitate connections between target and source perceptual domains.
Conventional metaphors elicit greater real-time engagement than literal paraphrases or concrete sentences
2021, Journal of Memory and LanguageCitation Excerpt :Moreover, the results demonstrate that the increased engagement is not due to greater imageability nor the inclusion of concrete words in metaphors. This is important because the only prior study to investigate imageability or concreteness as a possible source of greater engagement had not found significantly greater engagement even in metaphorical sentences (Pomp et al., 2018). Further work is required, but we suspect that lack of increased amygdala activity in that study was due to the fact that the sentences included words related to smell, and olfaction may independently evoke amygdala activity.
Human Olfaction at the Intersection of Language, Culture, and Biology
2021, Trends in Cognitive SciencesA Quantitative Meta-analysis of Neuroimaging Studies of Pragmatic Language Comprehension: In Search of a Universal Neural Substrate
2018, NeuroscienceCitation Excerpt :The problem with using those individual studies to draw definite conclusions about the neural basis of pragmatic language comprehension is their variation in terms of natural language and stimulus modality, what we will designate as the “pragmatic content configuration”. These studies have been conducted in a diversity of natural languages: Italian (Lauro et al., 2008; Bambini et al., 2011; Bosco et al., 2017), French (Spotorno et al., 2012; Obert et al., 2014, 2016), and Spanish (Licea-Haquet et al., 2016; Zamora-Ursulo, 2018) from the family of Romance languages; English (Lee and Dapretto, 2006; Stringaris et al., 2006; Wang et al., 2006; Stringaris et al., 2007; Chen et al., 2008; Boulenger et al., 2009; Hillert and Buračas, 2009; Schmidt and Seger, 2009; Yang et al., 2009; Desai et al., 2011; Diaz et al., 2011; Diaz and Hogstrom, 2011; Lacey et al., 2012; Obert et al., 2014; Egorova et al., 2016; van Ackeren et al., 2016, Rapp et al., 2004, 2010; Kircher et al., 2007; Forgács et al., 2012; Citron and Goldberg, 2014; Citron et al., 2016; Pomp et al., 2018), and Dutch (Zempleni et al., 2007; van Ackeren et al., 2012; Bašnáková et al., 2014; Bašnáková et al., 2015; Samur et al., 2015) from Germanic languages; and others, such as Japanese (Uchiyama et al., 2006; Shibata et al., 2007; Wakusawa et al., 2007; Shibata et al., 2010, 2011, 2012; Uchiyama et al., 2012; Akimoto et al., 2014), Hebrew (Mashal et al., 2009; Mashal et al., 2013), Mandarin Chinese (Ahrens et al., 2007; Yang et al., 2016; Feng et al., 2017), and Hungarian (Schnell et al., 2016). The stimulus modality is also heterogeneous and includes text, audio, image, videos, and a combination of these.