A persistent belief holds that humans can imagine visual content but not odors. While visual imagery is regarded as recreating a perceptual representation, it is unknown whether olfactory mental imagery shares a perceptual format. Visual imagery studies have demonstrated this perceptual formatting using distance and shape similarity judgments, whereas olfactory studies often use single-odor vividness ratings, complicating the establishment of perceptual formatting for odors. Using odor pair similarity scores from two experiments (odor-based: 8,880 ratings from 37 participants, including 20 women; label-based: 129,472 ratings from 2,023 participants, including 1,164 women), we observed a strong correlation (r =.71) between odor-based and label-based odor pairs. The correlation was unaffected by gender and age and was present in a wide range of self-perceived olfactory functions. Pleasantness similarity was the main determinant of overall similarity for both odor-based (r=−.63) and label-based (r=−.45) odor pairs. We then used a large language model to derive semantic similarity scores for the labels of all odor pairs. Semantic similarity only mediated a small part of the observed correlation, further supporting our conclusions that odor imagery shares a perceptual formatting with vision, that odor percepts may be elicited from verbal labels alone, and that odor pair pleasantness may be a dominant and accessible feature in this regard.

Whereas human cognition develops through perceptually driven interactions with the environment, language models (LMs) are “disembodied learners” which might limit their usefulness as model systems. We evaluate the ability of LMs to recover sensory information from natural language, addressing a significant gap in cognitive science research literature. Our investigation is carried out through the sense of smell — olfaction — because it is severely underrepresented in natural language and thus poses a unique challenge for linguistic and cognitive modeling. By systematically evaluating the ability of three generations of LMs, including static word embedding models (Word2Vec, FastText), encoder-based models (BERT), and the decoder-based large LMs (LLMs; GPT-4o, Llama 3.1 among others), under nearly 200 training configurations, we investigate their proficiency in acquiring information to approximate human odor perception from textual data. As benchmarks for the performance of the LMs, we use three diverse experimental odor datasets including odor similarity ratings, imagined similarities of odor pairings from word labels, and odor-to-label ratings. The results reveal the possibility for LMs to accurately represent olfactory information, and describe the conditions under which this possibility is realized. Static, simpler models perform best in capturing odor-perceptual similarities under certain training configurations, while GPT-4o excels in simulating olfactory-semantic relationships, as suggested by its superior performance on datasets where the collected odor similarities are derived from word-based assessments. Our findings show that natural language encodes latent information regarding human olfactory information that is retrievable through text-based LMs to varying degrees. Our research shows promise for LMs to be useful tools in investigating the long debated relation between symbolic representations and perceptual experience in cognitive science.

We spend every day using our senses to interact with the world. Though we use language as a way to understand the sensory world, language might have different roles for different senses. Freely identifying odors in naming tasks is more difficult than with senses like vision, making olfaction an interesting place to study the intersection between language and the senses. While free olfactory identification is poor, word cues strongly increase our ability to identify odors. This has led some to conclude that olfaction is more dependent on supporting information from other senses, and that odors are encoded in a coarse way, so it is particularly dependent on language and sensory cues to function capably. This has further led to debate regarding whether language can activate olfactory-related representations in the brain, or whether odor and language systems are disconnected. The general aim of this thesis was to investigate whether and how word cues can affect olfactory processes and representations.

Study I investigated whether the olfactory system is more reliant on object-based predictive verbal cues than the visual system. Using two behavioural experiments and one neuroimaging experiment, Study I found that reaction time is most delayed when words incorrectly cue an (unexpected) odor target. Study I further demonstrated that the primary and secondary olfactory cortices are more activated by odors that were unexpected, something similar to predictive coding in the visual system, but also that visual and cognitive activations were observed by unexpected odors. Study II investigated whether word cues could activate olfactory representations similar to actual odor perception. Using an in-person olfactory psychophysics experiment, an online experiment, and embeddings from a custom trained Large Language Model, Study II found that odor similarity scores estimated from odor names were very strongly correlated with their odor-based counterparts, and that this correlation was only mediated to a small degree by the semantic similarity of the word pairs, suggesting word-based representations of odors closely resemble odor representations. Study III assessed whether people can make accurate judgements about odor similarities based on viewing words and when effective odor imagery was prevented by nose blocking. Study III found that people can very accurately determine how similar two odors are using only word cues, and that people can seemingly do this when their  nose is blocked.

Overall, this research supports the idea that there are unique mechanisms whereby word cues guide olfactory processes, and that odor representations can be accessed by perceiving odor names, without the presence of odors or elicited vivid mental images of odors. Even though odors are often hard to name or describe verbally, olfaction may thus be strongly shaped by language cues.

Vi använder varje dag våra sinnen för att interagera med världen. Även om vi använder språk som ett sätt att förstå den sensoriska världen, kan språk ha olika roller för olika sinnen. Att fritt identifiera lukter i namngivningsuppgifter är svårare än med sinnen som synen, vilket gör luktsinnet till ett intressant område att studera samspelet mellan språk och sinnen. Medan fri luktidentifiering är dålig, ökar tillgången till alternativ starkt vår förmåga att identifiera lukter. Detta har lett vissa till slutsatsen att luktsinnet är mer beroende av stödjande information från andra sinnen, och att lukter kodas på ett grovt sätt, vilket gör det särskilt beroende av språkliga och andra sensoriska signaler för att fungera väl. Detta har ytterligare lett till debatt om huruvida lukternas namn kan aktivera luktrelaterade representationer i hjärnan, eller om lukt- och språksystem är bortkopplade. Det övergripande syftet med denna avhandling var att undersöka om och hur lukters namn kan påverka luktprocesser och representationer.

Studie I undersökte om luktsystemet är mer beroende av objektbaserade prediktiva verbala signaler än det visuella systemet. Med hjälp av två beteendeexperiment och ett hjärnavbildningsexperiment fann Studie I att reaktionstiden är särskilt fördröjd när ett ord felaktigt anger en lukt (som därför blir oväntad). Studie I visade vidare att den primära och sekundära luktbarken aktiveras mer av oväntade lukter, vilket liknar prediktiv kodning i det visuella systemet, men också att visuella och kognitiva hjärnområden aktiverades av oväntade lukter. Studie II undersökte om ord kunde aktivera luktrepresentationer som liknar faktiska luktupplevelser. Med hjälp av ett luktpsykofysiskt laboratorieexperiment, ett online-experiment och ordinbäddningar från en specialtränad stor språkmodell fann Studie II att luktlikheter som uppskattades endast med hjälp av luktnamn var mycket starkt korrelerade med deras luktbaserade motsvarigheter, och att denna korrelation endast förklarades i liten grad av den semantiska likheten mellan ordparen som språkmodellen genererade, vilket tyder på att ordbaserade representationer av lukter liknar luktrepresentationer. Studie III bedömde om människor kan göra korrekta bedömningar om luktlikheter baserat enbart utifrån luktnamn och när den mentala luktföreställningsförmågan förhindras genom att näsan blockeras. Studie III fann att människor mycket väl kan avgöra hur lika två lukter är, enbart med hjälp av lukternas namn, och att människor till synes kan göra detta utan att föreställa sig lukter.

Sammantaget stöder denna forskning idén att det finns mekanismer som är unika för luktsinnet genom vilka ordsignaler påverkar luktprocesser, och att luktrepresentationer kan aktiveras av lukters namn, utan närvaron av lukter eller framkallandet av livfulla mentala bilder av lukter. Även om lukter ofta är svåra att namnge eller beskriva verbalt, kan luktsinnet således starkt formas av språkliga signaler.

Creating and evaluating predictions are considered important features in sensory perception. Little is known about processing differences between the senses and their cortical substrates. Here, we tested the hypothesis that olfaction, the sense of smell, would be highly dependent on (nonolfactory) object-predictive cues and involve distinct cortical processing features. We developed a novel paradigm to compare prediction error processing across senses. Participants listened to spoken word cues (e.g., “lilac”) and determined whether target stimuli (odors or pictures) matched the word cue or not. In two behavioral experiments (total n = 113; 72 female), the disparity between congruent and incongruent response times was exaggerated for olfactory relative to visual targets, indicating a greater dependency on predictive verbal cues to process olfactory targets. A preregistered fMRI study (n = 30; 19 female) revealed the anterior cingulate cortex (a region central for error detection) being more activated by incongruent olfactory targets, indicating a role for olfactory predictive error processing. Additionally, both the primary olfactory and visual cortices were significantly activated for incongruent olfactory targets, suggesting olfactory prediction errors are dependent on cross-sensory processing resources, whereas visual prediction errors are not. We propose that olfaction is characterized by a strong dependency on predictive (nonolfactory) cues and that odors are evaluated in the context of such predictions by a designated transmodal cortical network. Our results indicate differences in how predictive cues are used by different senses in rapid decision-making.

Although our understanding of human olfactory perception has increased dramatically, it remains less well understood how olfaction interacts with cognitive processes. This chapter reviews the claim that olfaction predominantly supports ‘novelty detection’, a framework that leaves little room for cognitive representations and processes. It is argued that this framework has several possible limitations. Instead, it is proposed that human olfaction is best understood in terms of its significant reliance on top-down processes; a fundamental role of olfaction is to evaluate predictions that are constructed based on the context, and these predictions may be explicit and cognitive in nature. Olfaction is, according to this view, not limited to passively responding to environmental changes, but actively interacts with cognitive systems to shape goal-directed behaviours. It is proposed that cue-target matching tasks provide useful and flexible experimental designs to study these effects.

Prior laboratory research has suggested that humans may become more prosocial in stressful or threatening situations, but it is unknown whether the link between prosociality and defense generalizes to real-life. Here, we examined the association between defensive responses to a real-world threat (the COVID-19 pandemic) and everyday altruism. Four independent samples of 150 (N = 600) US residents were recruited online at 4 different timepoints, and self-report measures of perceived COVID-19 threat, defensive emotions (e.g., stress and anxiety), and everyday altruism were collected. Our operationalization of defensive emotions was inspired by the threat imminence framework, an ecological model of how humans and animals respond to varying levels of threat. We found that perceived COVID-19 threat was associated with higher levels of everyday altruism (assessed by the Self-report Altruism scale). Importantly, there was a robust association between experiencing acute anxiety and high physiological arousal during the pandemic (responses typically characteristic of higher perceived threat imminence), and propensity to engage in everyday altruism. Non-significant or negative associations were found with less acute defensive responses like stress. These findings support a real-life relation between defensive and altruistic motivation in humans, which may be modulated by perceived threat imminence.

Specific phobia can be treated successfully with exposure therapy. Although exposure therapy has strong effects on self-reported ratings and behavioral avoidance, effects on measures derived from electroencephalography (EEG) are scant and unclear. To fill this gap, spider-phobic individuals received either in-vivo or virtual reality exposure treatment. Patients were tested twice (one week before and after treatment), and control subjects once. In each session, EEG was recorded to spider pictures as well as other positive, negative, and neutral pictures. During EEG recording, participants performed a simple detection task while task-irrelevant pictures were shown in the background. The task was used to reduce potential confounding effects from shifts of attention. After the task, subjects were shown the pictures again and rated each in terms of their emotional reaction (arousal and pleasantness). The results showed that before treatment, patients rated spiders as more negative than did control subjects. Patients also showed elevated early posterior negativity (EPN) and late positive potential (LPP) to spiders. After treatment, the negative emotional ratings of spiders were substantially reduced. Critically, Bayesian analyses suggested that EPN and LPP were unaffected by treatment and that the treatment groups did not differ in their responses (EPN, LPP, and ratings). These findings suggest that the effects of in vivo and virtual reality exposure therapy are similar and that the initial stages of motivated attention (EPN and LPP) are unaffected by treatment.

Odor identification is a common assessment of olfaction, and it is affected in a large number of diseases. Identification abilities decline with age, but little is known about whether there are perceptual odor features that can be used to predict identification. Here, we analyzed data from a large, population-based sample of 2,479 adults, aged 60 years or above, from the Swedish National study on Aging and Care in Kungsholmen. Participants performed both free and cued odor identification tests. In a separate experiment, we assessed perceived pleasantness, familiarity, intensity, and edibility of all odors in the first sample, and examined how odor identification performance is associated with these variables. The analysis showed that high-intensity odors are easier to identify than low-intensity odors overall, but also that they are more susceptible to the negative repercussions of old age. This result indicates that sensory decline is a major aspect of age-dependent odor identification impairment, and suggests a framework where identification likelihood is proportional to the perceived intensity of the odor. Additional analyses further showed that high-performing individuals can discriminate target odors from distractors along the pleasantness and edibility dimensions and that unpleasant and inedible odors show smaller age-related differences in identification. Altogether, these results may guide further development and optimization of brief and efficient odor identification tests as well as influence the design of odorous products targeted toward older consumers. 

The sense of smell, olfaction, is currently in the spotlight. Although often a neglected sense, smell loss and smell distortions due to coronavirus disease 2019 (COVID-19) has increased public awareness of olfaction and its role in health, well-being, and nutrition. In this unusual context, Majid provides a timely, well written, and thought-provoking review of olfaction and its cognitive aspects. Majid focuses on how smells are expressed in languages around the world, yet the review covers a vast literature that includes also experimental, cognitive, and biological fields. The thrust of the argument is directed against a perceived dogma; that ‘there is no language of smell and humans are bad at naming odors’. Majid argues that this dogma is refuted by cross-cultural research and concludes that ‘rather than focusing on constrained experimental tasks’, olfactory language should be studied in terms of ‘how people across the globe use, manipulate and talk about odors in their day-to-day contexts’. Later, we show how empirical and theoretical considerations lead us to partly different conclusions. Empirically, we believe that odor naming remains poorer than naming in other senses and that careful laboratory experiments will remain critical for understanding olfactory-based language and cognition. Toward this goal, we emphasize the distinction between multisensory and unisensory olfaction, which is not highlighted in Majid's review but which we believe is of theoretical importance.