RSA Lecture, March 3 1999

Smell: Can we use it to manipulate behaviour?

Dr G Neil Martin
Senior Lecturer in Neuropsychology
Cognition and Brain Sciences Research Centre
Middlesex University
Tel: 0181 362 6292
Fax: 0181 362 5343
Copyright ‘ G N Martin

To Kipling, they were surer than sights of sounds to make your heart-strings crack. To Proust, they were the catalyst that provoked a flood of childhood memories. The ability of odours to generate emotion and to change behaviour has a long literary pedigree. But is there any scientific merit in these observations? Can odour influence emotion, thought, sexual behaviour- even health? This lecture presents some of the more interesting recent findings in olfactory psychology- the study of the effect of odour on behaviour. It will address questions such as: Do pheromones exist? Do they make you attractive to the opposite sex? Can an ambient odour improve cognitive performance? Can it improve vigilance? Do odours make you significantly happier? Do they change the brain’s behaviour? Are any reliable patterns of brain activity associated with perceiving different types of odour? Can odours improve your mental health? Does aromatherapy really work? Can smelling help you lose weight? Can we use odour to sell? A few tentative and a few conclusive answers to these questions will be discussed.

Lecture Outline

• The human sense of smell: Some fragrant facts
• What is the use of smell?
• Do human pheromones exist?
• The effects of smell on thinking and feeling
• Marketing and the smell of baked bread: half-baked or hot stuff?
• The Middlesex experiments I: Chocolate aroma and the brain
• The Middlesex experiments II: The cognitive consequences of smelling a relaxing odour
• The Middlesex experiments III: How pleasant and unpleasant smells affect performance on easy and difficult cognitive tasks

The human sense of smell: Some fragrant facts
Imagine that you were asked to live without one of your senses. Which one would you elect to relinquish? Vision? Probably not. Audition? Again, probably not. Most people, if they had to face this -admittedly, hypothetical- dilemma would chose to ditch the olfactory sense, the sense of smell.
For this reason, amongst others, the sense of smell has suffered paranoiacally from a Cinderella complex. It may be the beauty of the senses but it hardly ever has the opportunity to show this beauty at the sensory Ball. The visual and auditory ugly sisters have usurped it as the sense which we most prize and respect. In one sense, this is not surprising. Ever since we became bipedal and raised our noses from the ground, vision came to dominate our ‘view’ of the world. Much of the information we receive to enable us to live enjoyably and effectively is visual or auditory. This shift in the sensory hierarchy is reflected in the development of the brain. The oldest part of the brain contains those structures thought to mediate the work of the olfactory system (and is called the rhinencephalon or ‘smell brain’). Primitive brains were those allowing touch, smell and taste but little else.
This considered, however, is the sense of smell that peripheral to our lives? We spend small fortunes on perfumes; we delight in the aroma of food; we are sexually enticed by the odour of our partner. Whereas there are only four or five basic taste qualities, there are a myriad odours (there may be at least 400, 000 different, identifiable odours). We are not particularly adept at naming these different smells, however. We often experience the tip-of-the-nose phenomenon: we think we know what the smell is but we find it difficult to retrieve the correct label. Sherlock Holmes believed that any good detective should have knowledge of at least 75 perfumes; a random sample of detectives would probably reveal knowledge of fewer than five.
Our literary heritage also places great importance on the sense of smell. Shakespeare, Kipling, Proust, Joyce, Flann O’Brien and others have regarded smell as an emotional stimulus capable of changing thought, provoking memories and altering mood. So, the question that begs itself is: is smell important to modern day human living? More specifically, can smell be used to manipulate behaviour -thought, feeling, perception, action- in the same way that sights and sounds do?
This lecture seeks to find an answer to this question by focusing on development in modern psychology and neuroscience. It will presents some of the more interesting recent findings in olfactory psychology- the study of the effect of odour on behaviour. It will address questions such as: Do pheromones exist? Do they make you attractive to the opposite sex? Can an ambient odour improve your thinking and decision-making? Can we use odour to sell? Does aromatherapy really work? Do odours change the human brain’s behaviour? Are any reliable patterns of brain activity associated with smelling different types of odour? Can odour improve vigilance? First, we take a look at some of the traditional uses of smell.

What is the use of smell?
Traditionally, smell has been considered important to two activities: avoiding potentially harmful stimuli and contributing to food flavour. We are used to using malodour as warning signals: infected food or gas leaks are easily detected by sensing a ‘bad’ odour in the environment (gas is actually odourless; mercaptans added to it make it malodorous). There are exceptions. Some pungent cheeses are regarded by some as an olfactory treat. Smell is also the most important contributor to food flavour. When we say that a food ‘tastes’ good what we really mean is that it ‘smells’ good. There are only a few basic taste qualities; there is a vast number of olfactory qualities, and receptors located inside the mouth and at the back of the throat sense olfactory molecules released by food (the sense of smell is a chemosense, that is, it is stimulated by chemical stimuli- unlike vision or sound). These two functions are the undisputed uses of smell. More controversial is the idea that smells can do more than just warn, or enhance flavour. Most controversially, some investigators have suggested that odour can influence sexual behaviour in humans. This is the issue we look at next.

Do human pheromones exist?
Some species in the animal kingdom have a terrifically efficient sense of smell. Dogs have a remarkable ability to detect and discriminate between odours, hence their use in drug-sniffing operations. The odour of vaginal copulins can arouse male monkeys. Some odours affect the oestral cycle and sexual development of rodents. Some ‘scents’ are thought to produce stereotypical responses in a receiving organism without the scent being overtly detected. Such scents are called pheromones, chemicals secreted by the body which induce a stereotypical behavioural or physiological response. A well-known example of a pheromone is androstenone (its full, chemical name is 5alpha-16-androst-16-en-3-one), a steroid developed in the testes of pigs which has a musk-like odour and is secreted in the saliva of male pigs during mating.
The unusual feature of androstenone is that it can induce a sow to adopt the mating position when it is sprayed on her. Farmers and vets can even buy the chemical in aerosol form, Boarmate, so that the sow can be prepared for mating. (In fact, truffle hunters use sows to detect the delicacy because it contains androstenone). Pigs are not the only species to secrete androstenone. It is also present in men’s sweat glands and in the urine of men and women, although at stronger concentrations in men (Brooksbank et al, 1974).
The behaviour induced by androstenone in sows, together with the presence of the chemical in humans, has led some researchers to believe that a human pheromone effect exists (Cohn, 1994). Much of this work has focused on examining whether certain of the body’s chemicals alter the length of the menstrual cycle and whether they increase sexual attractiveness in a person of the opposite sex.
The evidence
One of the earliest studies of the pheromone effect in humans was conducted by Martha McClintock in the 1970s (McClintock, 1971). She found that 17-22 year old women students who lived and slept in the same halls of residence reported menstrual synchrony. That is, their menstrual cycles began on or about the same time. The effect was unrelated to food intake, lifestyle pattern or stress. The result is difficult to explain because no mechanism that we know of can account for the finding. McClintock suggested that the mechanism might be pheromonal or mediated by an awareness of another’s menstrual cycle. To explain the result, replications would be needed. If the effect was pheromonal then a controlled experiment in which menstrual cycles were deliberately manipulated would show this. This what Russell et al (1980) did.
They applied the sweaty secretions of a woman who had a history of 28 day cycles and experience of ‘driving’ (that is influencing) other women’s cycles on the upper lips of five women, three times a week for four months. Six individuals wore odourless alcohol (the control group). The mean difference in the cycle onset in the experimental group was three to nine days before the experiment; three to four days during driving. Controls’ figures were 8 days and 9.2 days’ difference pre-testing and during testing, respectively. A significant difference, therefore, was found between the experimental group’s cycle onset and that of controls. However, there were some important limitations to the study. The experiment was not single- or double-blind which means that the experimenters knew which participant was in each condition and each participant knew the purpose of the experiment. The woman who provided the samples was also one of the experimenters!
In a similar experiment, Stern and McClintock (1998) examined whether such pheromones could affect the length of ovulation. The experimenters took the odourless compounds from the armpits of women in the late (follicular) or ovulatory stages of their menstrual cycle and applied them to the top lips of 29 healthy young women aged between 20 and 35 years old. The length of the recipients’ menstrual cycle was then measured. The experimenters found that women reported shorter cycles when receiving follicular compounds and longer cycles when receiving ovulatory compounds. There was a significant difference between the baseline cycle length and the cycle length reported when carriers ‘wore’ the compounds. 68% of women responded to the compounds of both kinds. Does this suggest that ‘pheromones’ can modulate ovulation? Or are the effects limited to healthy young women? Would the same effect occur in women on various types of oral contraceptive? And are there alternative explanations, other than pheromonal ones, for the results?
Questions such of these crop up quite often in human pheromone research. They become even more pertinent when examining the results of studies investigating the effect of ‘pheromones’ on sexual attractiveness. One of the earliest reports suggested men and women exposed to androstenol (a form of androstenone) rated photographs of women as sexier and more attractive than did a control group (Kirk-Smith et al, 1978). In this experiment, photographs of objects such as buildings were also rated. Oddly, these buildings were also described as ‘less sensitive’ in the presence of androstenol. This result -and others like it- suggests that the androstenol did not have sex-specific effects. More to the point, there was no control odour used in the experiment- participants could have made such ratings to any odour. Results since have not been particularly encouraging.
The majority of experiments, for example, has tended to expose subjects to volatile chemicals in contexts which are not generally appropriate. For example, participants have rated imaginary verbal descriptions of people, pictures and slides of buildings, people and animals or have indicated the number of sexual partners they have recently had. They have worn masks impregnated with an odour, a necklace impregnated with the odour, worn the odour on the top lip, sat on a chair impregnated with the odour, or used a doctored changing room cubicle. None of these conditions realistically reflects the behaviour and contexts normally observed during physical attraction.
In the most ecologically valid of these experiments, Black and Biron (1982) required participants to interact with a confederate of the opposite sex who wore either androstenone or a control odour. The participant was later asked to rate the confederate for attractiveness. The experimenters found no effect of these chemicals on the rated attractiveness of the confederate.
Given the importance of vision to human beings and the apparent irrelevance of olfaction to most important social behaviour, it should not be surprising that odours do not produce the same reliable and strong sex-related effects in humans as they do in other animals. Vision is our dominant sense. There may also be another reason why pheromones ‘don’t work’ in humans. There is evidence to suggest that the organ in the brain responsible for detecting and acting on these pheromones (this is called the vomeronasal organ) may be absent in humans (Moran et al, 1995). If this organ is important for sensing pheromones then humans will have difficulty sensing pheromones. Meanwhile, however, the evidence suggests that if you want to attract a member of the opposite sex, a bottle of good perfume or cologne would be a better option than would exposing your armpits, or investing in a can of Boarmate.

The effects of smell on thinking and feeling
Recent years have witnessed an increasing interest in the potential ability of ambient odour to influence behaviour (Knasko, 1996; Martin, 1996). From the use of scents to enhance the pleasantness of stores to the more deliberate marketing ploys of using odour to manipulate consumer behaviour, anecdotal and experimental evidence (Gulas & Bloch, 1995; Spangenberg, Crawley & Henderson, 1996) indicates that ambient odour has the ability to alter behaviour, although the precise nature of this alteration depends on the behaviour observed, type of odour used and the delivery system used to present the odour.
Experimental evidence suggests that when olfactory stimulation is effective, this effectiveness depends on a complex interaction of odorant, personality characteristics and experimental manipulation. Memory for odour is markedly resistant to time, easily accessed and tends to be characterised by a high degree of emotion, clarity and vividness (Laird 1935; Engen & Ross, 1973; Hertz and Cupchik 1992). In one experiment (Schab, 1991), subjects learned and retrieved antonyms either (i) in the presence of chocolate odour, (ii) in an unscented environment or (iii) by following instructions to imagine and think about the odour. Recall was superior in the chocolate condition smell, compared that in the unscented conditions. Details of the time of presentation, the duration of the presentation and the concentration of the odour were not provided, however, and neither were sample sizes for each of the experimental conditions given. In another experiment, recall was better when the same odour is present at encoding and retrieval. Herz (1997b) recently observed that exposure to odour at encoding and retrieval of information resulted in superior recall when compared to a no-odour control condition. Odours appear to be comparatively no better or worse in enhancing autobiographical retrieval than stimuli in other sensory modalities, however (Rubin, Groth & Goldsmith, 1984).
Experiments examining the direct effect of odour on cognitive task performance have produced mixed results. Although negative effects have been reported (Ludvigson & Rottman, 1989; Lewis et al, 1970) and some tasks, such as risk-taking, altruism and optimism tasks, certain memory recall exercises and creativity tasks, appear to be immune to the effects of olfactory stimulation (Ludvigson & Rottman, 1989; Ehrlichman & Bastone 1991; Knasko, 1992), improvements in task performance have been noted following exposure to odour during photograph recognition, anagram formation and word construction.
In an experiment by Baron (1990), subjects were asked to complete a clerical coding task in the presence of one of two pleasant odours (or no odour) and to estimate how successful they would be at completing this task. Subjects were also required to engage in negotiation with a confederate. In this scenario, the subject was allocated one million dollars for distribution among parts of a company but was also required to make cuts. Estimations of how well they would perform in this task were measured.
While there was no significant effect of odour on the number of clerical coding procedures completed, a significant effect was seen on some aspects of the negotiation task. Subjects undertaking the negotiation task in a room sprayed with a pleasant-smelling air freshener reported setting significantly higher negotiation goals and rated their estimated self-efficacy much higher than subjects in an unscented room. These subjects also indicated that they would attempt to attain a higher share of available funds. In practice, subjects in the scented environment gave away slightly more monetary concessions and reported weaker tendencies to resolve future conflict through avoidance and competition than did those in the non-scented room. They did, however, rate themselves as more positive, pleasant and good, a finding which may be related to evidence suggesting that elevated positive affect increases confidence in one’s ability or judgement. These subjects also had high expectations of themselves and of their performance. Similar effects on mood have also been obtained with the odour of chamomile (Roberts and Williams, 1992). This odour was reported to produce significant reductions in self-reported negative mood when compared with an inert placebo in female subjects asked to visualise negative scenes. Subjects also tended to think that there had been more positive than negative images presented during the chamomile condition.

Marketing and the smell of bread: half-baked or hot stuff?
Some time ago, a well-known British supermarket used to pump the artificial aroma of baked bread to its entrance to entice customers. The reasoning behind this was that the smell provided a positive, attractive ‘entry-point’ into the store and that customers would run straight to the bakery or feel more positively disposed towards the store. With developments in modern supermarket retailing, artificial smells are a thing of the past; now, the odour of freshly baked bread suffuses the store.
The use of smell as a marketing tool is not new. Manufacturers have long-since enhanced their products by claiming spring- or alpine-freshness. Carsalesmen have been known to spray leather-scent on the seats of their cars and one American bank scented its cheque books with rose oil. Smell is an important factor in sensory evaluation of products and commerce invests a lot of time and money to achieve the right scent for its products. However, a more interesting question is: does ambient odour make consumers buy more, stay in the store for longer or rate the store environment as positive?
Two well-known companies have used smell as a marketing tool. Woolworth’s have during two Christmas periods, introduced the aroma of Christmas dinner or mulled wine into 20 of its stores. On Valentine’s Day, 1997, Superdrug, based on research at Middlesex University, introduced the odour of chocolate into its flagship Strand store. This marketing tool is part of a new trend in commerce called ‘atmospherics’ which manipulates aspects of the atmosphere (music, smell or appearance) in order to manipulate consumer behaviour.
One experimental study diffused either lavender, ginger, spearmint or orange odour into a mocked-up store as business studies students evaluated its environment and products (Spangenberg, Crowley & Henderson, 1996). The evaluation of the store and the environment was more positive in the scented conditions: they were rated as more favourable, positive, liked and modern. The students expressed greater intent to visit the store in the scented conditions and regarded the merchandise as more up-to-date, varied and of higher quality. According to the authors, “the presence of an inoffensive scent in a store is an inexpensive and effective way to enhance consumer reactions to the store and its merchandise.”
Along similar lines, Bone & Jantrania (1992) found that lemon-scented cleaner was rated more positively than was coconut-scented cleaner. Conversely, coconut-scented sunscreen lotion was rated more positively than was lemon-scented lotion. If a disinfectant was scented with lemon or coconut, it was regarded as disinfecting better. Both odours were regarded as pleasant.
What this finding illustrates is that any use of odour to sell merchandise must ensure that the smell is compatible with the product. This may seem, on first inspection, to be stating the obvious, but it is not. It suggests that simply adding a pleasant smell to a product will not increase the perception of it as effective, liked, nice, etc. The scent has to be associated in some way with the product (coconut is not usually associated with cleaner nor lemon with suntan lotion). Any pleasant scent will make a store seem nice; specific scents paired with appropriately specific products will lead to more positive, merchandise-related consumer decisions.
The Middlesex experiments I: Chocolate and the brain
At Middlesex University, we are engaged in a programme of research into the behavioural consequences of exposure to odour. We have been interested in whether different types of odour affect performance on different types of tasks in different ways. We have also been interested in exploring why different odours affect performance differently- are these odours more or less distracting, more or less relaxing or alterting or more or less pleasant? Are any of these characteristics inter-related?
In 1998, we published the world’s first systematic study of the effects of food aroma on the activity human brain (Martin, 1998). The aim of this study (in essence, two experiments), was to explore the influence of synthetic and real olfactory food stimuli on the human Central Nervous System.
In the first experiment, individuals were exposed to a variety of pleasant and unpleasant synthetic odours such as chocolate, spearmint, almond, strawberry , vegetable, garlic and onion and cumin while EEG (the brain’s electrical activity or ‘brainwave’) was recorded from 19 electrodes placed on the scalp. The brain’s electrical activity can be divided into various types depending on the size of the brainwave and how often it occurs per second. From largest (and least frequently occurring) to smallest (and most frequently occurring), these are: delta, theta, alpha and beta. Each is associated with various functions and states. Alpha, for example, is the adult resting EEG, seen when individuals are relaxed with eyes closed. Delta and theta are seen in deep sleep and theta may also be produced by being successfully engaged in an intellectual task that requires attention.
Our results showed that the brain responded to synthetic odours in a very specific way. Theta activity was reduced during exposure to the odour of chocolate compared with almond and cumin. Smelling spearmint was also associated with less theta when compared with a no-odour control condition. To see whether these results were specific to synthetic odours (or to specific odours), we set up a second experiment in which individuals were exposed to the aroma of real baked beans, chocolate, coffee and rotting meat while their brain activity was recorded. This time, the effect of chocolate odour was even more pronounced. It was associated with less theta activity than was any other odour. Why?
One reason may be that theta activity occurs more often when individuals engage deeply in a cognitive or perceptual task than when engaging less deeply. Other EEG experiments have shown that successfully remembered words in a memory task are associated with greater theta activity at the time when successfully remembered words are learned. There is also evidence to show that the greater the attention paid to a task, the greater the theta activity seen. When we analysed the psychometric data to see which odours were most pleasant, relaxing, alerting and so on, spearmint and chocolate were rated not only the most pleasant of the odours but also the most relaxing. If relaxation (or lack of attention) decreases theta activity, the odour of chocolate may have affected the brain in this way. This finding may suggest that the odour of chocolate may be used to relax or to cause distraction. The relationship between the relaxing nature of odour and its effect on behaviour was the subject of our next experiment.
The Middlesex experiments II: The cognitive consequences of smelling a relaxing odour
My research student, Alison Gould, and I set up an experiment in which we would observe the effects of exposure to ambient relaxing and altering odour on human visual vigilance. An earlier experiment by Warm et al’s (1991) had required subjects to detect when two parallel lines were a distance of 10mm from a dot in a series of lines that were either 10mm or 12mm from a dot, in one of three conditions. In one condition, subjects received 30 second whiffs of peppermint odour (via a face mask) every 10 minutes; in the second, subjects received muguet for the same duration (and using the same method of delivery); in the third condition, subjects received whiffs of clean air. In the odour conditions, the percentage of correct detections was significantly greater than that found in the clean-air condition, although the post-hoc tests and means were not provided for confirmation. The authors suggest that two hypotheses arise from these findings. The first is that any pleasant odour will be sufficient to enhance visual vigilance; the second is that any odour (pleasant or unpleasant), provided it is rated as relaxing or alerting, will enhance vigilance, suggesting that odours of diametrically opposite psychometric hue (alerting vs calming) might produce the same effect.
In our experiment, we exposed subjects to the odours of bergamot or peppermint (or no odour) while they engaged in a visual vigilance task. This involved observing a computer monitor on which appeared a series of asterisks. Occasionally, one of these asterisks (the target) would flash for 150ms. When this occurred, the subject was required to press the space bar of a keyboard as quickly as they could. The task lasted 15 minutes. We found that those subjects in the bergamot condition detected fewer targets within 1.25 seconds of the stimuli appearing than did those in the peppermint or no-odour conditions. When we broke down task performance into early and late components, those in the bergamot condition detected fewer targets in the later part of the experiment than the early part when compared with subjects in the peppermint or no odour conditions. This finding would seem to indicate that exposure to a pleasant relaxing stimulus is associated with a reduction in vigilance and attention, which is compatible with the finding of the EEG experiments.

The Middlesex experiments III:
How pleasant and unpleasant smells affect performance on easy and difficult cognitive tasks
In our next experiment, we wanted to examine whether pleasant and unpleasant odours had the same or different effects on cognitive tasks varying in complexity. Evidence regarding whether some tasks are easier to perform in the presence of a pleasant odour than are others is unclear. There is a great deal of evidence to suggest that the presence of a malodour is detrimental to the performance of a complex task.
After extensive pilot testing, we struck upon two odours that were regarded by most people as pleasant or pleasant: a well-known supermarket air freshener and sour milk. In the experiment, we asked subjects to complete easy or difficult versions of three standard cognitive tasks in the presence of one of these odours (or no odour). The tasks tested visuospatial ability, vocabulary and verbal fluency. The spatial task required individuals to indicate which of six patterns completed an incomplete figure (Raven’s Progressive Matrices). The vocabulary task required subjects to define six commonplace and six rare and abstract words (the Mill Hill Vocabulary Test). Finally, the verbal fluency task required subjects to write down as many words beginning with a specific letter under timed conditions (Thurstone’s Verbal Fluency Task).
We found a significant and unusual interaction between odour and task complexity. Those subjects exposed to sour milk scored higher on the easy tasks than did subjects in the control (no odour) condition; furthermore, those exposed to pleasant odour scored higher on difficult tasks than did those exposed to no odour. When we examined individuals’reactions to the pleasantness of the environment in which they were tested, we found that the perception of the odour in the room as ‘unpleasant’ was associated with improved vocabulary and impaired spatial task performance. When subjects perceived the smell in the testing room as pleasant, this was associated with an improvement in spatial task performance.