An investigation into the effect of androstanone on the emission of dominant responses

Nik Makepeace, 1999 (unpublished)

Social facilitation is the robust effect of increased/decreased ability to perform easy/hard tasks respectively in the mere presence of conspecifics. This study examines the claim that semiochemicals such as androstanone can have behavioural effects on human subjects, by using its presence to substitute for conspecifics in a social facilitation experiment with 80 subjects. Analysis showed the predicted significant social facilitation effect occurred with an audience, p < 0.005, but androstanone was found to elicit only a non-significant trend towards social facilitation, p = 0.08. Limitations of the experimental design are discussed along with suggestions for more useful studies of this phenomenon.




It is well known that when there are members of the same species present, the behaviour of individual animals is different to those times when the animal is alone, and there are many different ways in which this manifests itself. In some animals, the presence of another will lead to conflict, such as the territorial disputes of robins. In other animals, the presence of others is required for their proper functioning, for example the honey bee. The effect of others differs widely between species, as a function of what significance another animal of the same species might have. However, there is one effect of the presence of others that exists across many species, wherein the performance of animals engaged in a task (such as eating, running, etc.) is exaggerated. Examples of this effect include runners who run faster in a filled stadium than on a deserted track, and professional musicians who produce their best performances in a band or orchestra. This effect was first examined by Triplett (1898), who thought it was an effect of competition, such that people were 'energised' by competitors. Allport (1920) widened the concept, and coined the term 'social facilitation' which is how this effect is described today. Allport proposed that the performance increase could be attributable not to competition, but due to the mere presence of conspecific co-actors or even entirely passive observers. However, for several decades the understanding of the effect was confused. Even though it was rapidly established that there was a performance increase with a co-acting or passive audience in many species performing many tasks (see, for example, Chen, 1937; Travis, 1925, Harlow, 1932; all in Zajonc, 1965) there were findings from the same experiments that showed that in some situations, the performance wasn't improved but degraded. Husband (1931) had subjects learn a finger maze. He found that in the presence of others the subjects took 12% longer to learn the maze and made 20% more errors. Travis (1928) repeated Allport's (1920) experiment, and found a reversal of Allport's results. Allport used a word-association task with ordinary people, and found an improvement in performance with an audience. Travis used the same experiment, trying to improve the performance of stutterers, but found a decrement in performance with an audience.

The contradictory results were not explained and the study of social facilitation died out. It was not until 1965 that this problem was taken on successfully. Zajonc (1965) published his subtle and astute drive theory, which has informed the study ever since. This explanation for the inconsistent results is that the observed subjects perform better or worse not according to their species, environment, the passivity of conspecific, or even the attentiveness of the audience, but according to their pre-existing ability to perform the task. The theory is that the presence of others leads to generalised drive, which leads to an increase in the predominant behaviour associated with any particular task, and a decrease in the subordinate response (Zajonc and Sales, 1966). In the case of hard, or unfamiliar, tasks the subject's dominant response will be to perform badly and in the presence of others this response will be facilitated, so the subject will perform worse still. Likewise, with familiar or easy tasks subjects will perform better in the presence of others than they would individually.

This explanation of the contradictory results is effective and cogent (see Zajonc, 1965) but since then there have been many further developments of Zajonc's theory. One aspect of social facilitation research which has been investigated particularly thoroughly is the effect of the mere presence of an audience. It does not seem intuitively plausible that where there is no kind of social interaction, the mere presence of another person would have any social influence; it is certainly hard to imagine that one would, for example, make more mistakes learning a maze if there was an inattentive audience, or that a performance of a piece of music might be better if performed in the presence of a deaf person reading a book. It is this sort of thinking that inspired the argument that mere presence wasn't alone sufficient, and that what was required was an attentive, evaluative audience, or at the very least, a potentially attentive and evaluative audience. Cottrell (1972) proposed exactly this. We learn as a child that social rewards and punishments are strongly linked to others' evaluations of us. As adults, this remains true, and in social contexts we are usually aware of what others think of us. Thus, the evaluation apprehension model, in which fear of evaluation leads to arousal and thus emission of dominant responses, was suggested and tested. In support, Cottrell et al. (1968) came across no social facilitation effect when the audience used was blindfolded (and therefore unable to evaluate) or merely present while apparently waiting to take part in another experiment. Several further experiments also support this model (see Klinger, 1969; Paulus and Murdoch, 1971) but, as Markus (1978) pointed out in a criticism of evaluation apprehension studies, none of the studies ever measured a state in which the subject was, and felt, truly alone; when subjects take part in a laboratory psychology experiment they are often aware that they are either being watched or will be evaluated later. She attempted to circumvent this issue by getting the subjects to perform the task when the subjects thought the experiment was yet to begin, and secretly measured them through a one-way mirror. Markus seems, though, to have not heeded her own advice; as Schmitt et al. (1986) pointed out, 'most undergraduates know about one-way mirrors.' Furthermore, the merely present audience was capable of evaluating at any time; despite facing away from the subject, there was nothing to stop the observer from turning round and evaluating at any stage.

Aware of these shortcomings in existing research on mere presence, Schmitt et al. performed the ultimate in mere presence studies. They used an audience that was neither attentive, nor capable of becoming attentive; the confederate 'observer' was thoroughly unobservant, due to his being blindfolded, wearing headphones and with his back turned, apparently in preparation for an experiment on sensory deprivation. The subjects had answer several questions posed by a computer, supposedly prior to the experiment actually commencing. In one condition, the subject was left entirely alone in the office to answer the questions, and in another, the incommunicado confederate was present. Only two of the questions were relevant to the study: the first asked for the subject's name, and the second for a code name for the subject, which was the same name backwards, with the letters interspersed with ascending numbers. The computer recorded the amount of time taken to answer these two questions. Typing the subject's name was used as a familiar task, and the code name was a difficult task. Schmitt also took the opportunity to compare the definitive mere presence condition with a classic evaluation condition: in another condition the experimenter peered over the subject's shoulder. He found that the ability of the audience to evaluate was entirely unnecessary; the subjects in the mere presence condition exhibited a strong social facilitation effect.

Schmitt's conclusion that mere presence is enough to produce social facilitation is well supported by Bernard Guerin (1985) who conducted a review of some 287 social facilitation studies. After having rejected 274 as not having strict enough mere presence conditions he found evidence for the effect of social facilitation due to mere presence. Bond and Titus (1983) in a meta-analysis of 241 social facilitation studies also found that effects are "surprisingly unrelated to performers' evaluation apprehension," which suggests that social facilitation through mere presence can be accepted as a robust, well-established effect.

The effect on the performer's task performance must be related, therefore, to the detection of the observer by the performer, and not to the perceived ability of the observer to detect the performer. In Schmitt et al.'s experiment, the subjects were unaware that, when they thought they were alone, their performance was actually being monitored by the computer, so no social facilitation response was recorded in that condition, while even when they knew the observer had no chance of making any observation at all, subjects performed as social facilitation predicts. One must assume that most social facilitation experiments, the subjects primarily are aware of their audience because they can see them. Indeed, it is reasonable to say that in most situations one is primarily aware of the presence of another through visual detection. Social facilitation does not, however, rely on the detection of another person using the eyes: Hogg and Vaughan (1993) describe a study in which social facilitation was effected with sound rather than sight. Cohen and Davis (1973) found an effect even when no audience was present just by telling the subjects they were being observed. If it is the case that social facilitation can be induced by the sight or sound of another person, is it the case that it can arise from the smell of an audience? In the Cohen and Davis study, just the implication of the presence of a person was required, and it could be said that the smell of another person strongly implies their proximal presence.


Olfaction as a field is relatively young, and has only recently challenged the long-held view that the sense of smell is rather unimportant in humans (Vroon, 1997). Studies into chemical communication between humans are consequently limited, but recently there have been claims of great advances made in this field. Chemical communication among insects is well established, with many studies having established insects' ability to use specific signalling chemicals, or phermones, to communicate with other insects of the same species. These semiochemicals have been known as pheromones since 1959 (Karlson and Luscher, 1959), and are used by animals to introduce a change in behaviour of the recipient of the chemical (known as signalling pheromones), and also to effect a change in the biology (typically hormone levels) of the recipient (Shorey, 1976)

In insects the influence of pheromones on behaviour seems occasionally to be startling in its extent. Kohl and Francoeur (1995) give us the example of the mating scent of a queen bee: when this scent was collected and sent up in a balloon, male bees 'piled on, desperately trying to mate.' Reactions such as these suggest that insect pheromone responses are not mediated by any other factors, such as the appearance of the pheromone emitter or its behaviour.

Although the phenomenon is most startling in insect behaviour, it is present in animals with more developed (and less predictable) behaviour than insects. Mammals also use pheromones in a variety of ways, although research seems to have highlighted sexual behaviour as the most common sphere of pheromonal influence (Taylor, 1997; Albone and Natynczuk, 1992). Nowadays, though, many researchers are keen to stress the the differences in effect between chemical communication in insects and in mammals. In mammals the effects are seldom as direct as the insect pheromone reflexes, and as such there is a tendency to talk of mammalian semiochemical responses rather than pheromone effects.

Nevertheless, some mammalian semiochemical communication is highly reflexive. Many farmers, for example, use artificially produced versions of the semiochemical androstenone in order to assist pig breeding. Boar saliva contains androstenone which, when detected by sows ready to breed, causes them to arch their backs and assume the mating position. By spraying the noses of sows with the artificial semiochemical androstanone, farmers can more easily artificially inseminate the sows in heat and avoid wasting expensive semen on sows who won't conceive (Melrose, Reed and Patterson, 1971, in Kohl and Francoeur, 1995). Mice, too, have been shown to use semiochemicals in a sexual context (Vroon, 1997). A chemical present in the urine of a male mouse can lead to female mice aborting their fetuses if they are not those of the male in question. Moreover, male mice can control nearby females' fertility through exposure to male urine, and respond to semiochemicals in females' urine with mating calls.

It may well be that these kinds of automatic response are responsible for the lack of investigation into human responses to potential human semiochemicals. It is not even universally accepted that humans secrete functional semiochemicals. In mammals, non-odorous semiochemical precursors are secreted by the apocrine glands, whereupon they are processed by symbiotic bacteria into semiochemicals (Van Toller et al., 1983). In humans too, the apocrine glands are the source of secretions of chemicals which strongly take after the semiochemical precursors of other mammals (Kohl and Francoeur, 1995). These secretions, too, are converted by surface dwelling bacteria into the 16-androstenes (Labows and Preti, 1992), a family of odorous chemicals, more prevalent in males, which includes the molecule, androstenone, that causes the strong mating reaction in female pigs.

However, the debate as to the functionality of human semiochemicals is not so much to do with what is secreted, but whether these chemicals are detected. At present it seems as though there is more research to suggest that they are detected (see, for example McClintock, 1971; Weller and Weller, 1993; McClintock and Stern, 1998; Grammar and Jütte, 1997) than there is to say how this happens. The research situation is further complicated by the fact that the leading proponents of the most likely method of semiochemical detection (Moran et al., 1995) are partners in a commercial research organisation that has strong financial links to a company that markets a perfume, the effect of which (and its consequent market value) hinges on the method of the detection they propose (Taylor, 1997).

McClintock's recent work is perhaps the most convincing to date. It concerns the existence of chemical communication between women of their position in the menstrual cycle, and the consequent synchronisation of cycles. In the original study, she eliminated common environmental and dietary factors and concluded that the communication had to happen some other way. Preti et al. (1986, in Weller and Weller, 1993) supports the view that this communication is an example of primer-semiochemical effects; they found a significant effect of samples taken from the armpit of women at one part of their cycle on the length of the cycle of women these samples were given to. With no other contact than these samples, the length of the cycles of the recipient was altered. Since McClintock (1971) there has been a great deal of research in this field, but much of it has been criticised for poor methodology (see Wilson's (1987) critique of such studies), and much of it not well enough controlled to exclude non-semiochemical reasons for the phenomenon. Weller and Weller (1997) conclude that 'results show unequivocally the existence of menstrual synchrony,' but are not able to pinpoint semiochemical communication as the cause. In an attempt to do just this, McClintock and Stern (1998) eliminated all non-chemical contact between the women, and also ensured the communication went only one way. This was achieved by swabbing their female subjects under the nose with axillary secretions from women at specific stages in their cycles. They found that if the donors had not yet ovulated, the recipients' cycles would shorten, and vice versa. By extending these findings they established that if these trends continued over a longer period than that over which the experiment took place, the recipients' cycles would eventually synchronise with the donors.

This is very strong evidence of chemical communication between humans, and goes a long way to substantiating many years of speculation on the subject, but there is far more to the field than menstrual synchrony. As a student at the University of Vienna, Astrid Jütte has been examining the effect of synthetic copulins, volatile fatty acids secreted in the vagina and strong candidates for female semiochemicals, on men's ratings of women's attractivity based on their photographs and voices. She found that while there was no interaction between rated attractivity and the point in the menstrual cycle that the copulins were synthesised to resemble, there was a strong physiological response. The level of testosterone in the subjects' saliva rose by 50% when they were smelling copulin that mimicked the ovulatory part of the cycle, as compared with no increase for the pre- or post-menstrual copulins (Grammer and Jütte, 1997). This is strong evidence for the communication of women's fertility to men by scent, as 'heightened testosterone levels seem to increase selective attention for erotic stimuli.' (Grammer and Jütte, 1997)

In males, too, there is evidence of physiological changes resulting from semiochemical exposure. Also studying behavioural effects of semiochemicals at Vienna, Atzmüller (unpublished) reported a significant decrease in testosterone levels in men exposed to androstenone for 15 minutes, when compared with men exposed to a control scent of lemon oil. At the same time, a significant negative correlation was reported between testosterone levels and co-operative behaviour, i.e. males with high testosterone levels co-operated significantly less than those with lower levels of testosterone. This has implications for the role of androstenone in social behaviour; males who produce high levels of androstenone may well be inducing greater levels of co-operation amongst other males by virtue of the testosterone-reducing effect of androstenone exposure.

Despite all this positive research the effect of semiochemical exposure on human physiology and behaviour, the problem of how the semiochemicals may be detected still remains. In mammals that communicate using semiochemicals, an organ known as the vomeronasal organ (VNO) has been found to be a necessary part of the system that detects semiochemicals, the Accessory Olfactory System, or AOS (Beltramino and Talesnik, 1983, in Moran et al., 1995). Despite being identified in humans almost 300 years ago, there has to date been no widely accepted proof in humans of its functional existence. The highest profile champion of its relevance to human biopsychology is Pherin., a commercial research enterprise the co-founder of which also owns a company which specialises in the production of a range of perfumes based on the effects of wearing the hypothetical human pheromones. Many of these key researchers in the field of the human VNO are also partners in this enterprise, so an understandably cynical response to claims of a fully functional VNO (Moran, et al., 1995) is to wait for impartial research to emerge that supports their claims. Many experts in the field are doing exactly this, wary of any organisation which makes a claim of scientific importance that also happens to be justification for the sale of a product it manufactures. In a summary of the research conducted by Pherin, Taylor (1997) quotes Michael Meredith, a mammalian VNO specialist as saying he would be more comfortable accepting the Pherin research 'if the results were corroborated by another source whose fortunes do not rise and fall on the result'

Furthermore, there are certain further issues with the AOS in humans: whether or not the VNO is functional or not, there still must be a link to the brain, and this has still eluded researchers (Kohl and Francoeur, 1995). It could be that the VNO has survived as a mere vestige of the functional organ pre-human ancestors may have had, while the vomeronasal nerve has been lost altogether. Meredith claims that 'no one has found anatomical proof of the nervous connection between the VNO and the brain', and Moran himself has been unable to find the vomeronasal nerve (Taylor, 1997)

Despite the uncertainty over the existence of the VNO as functional part of modern humanity's anatomy, the research is still there that shows that there is some degree of chemical communication occurring between humans. Whether it is via the VNO or some other means, the research by Jütte, McClintock and Stern, Preti et al., Atzmüller and others strongly implies that, after bacterial processing, chemicals secreted by humans can have behavioural and physiological effects on other humans.


The question this study attempts to answer is: can social facilitation be accounted for in terms of chemical communication alone? By examining subjects' task performance with an audience, without an audience and with and without a semiochemical such as androstanone present, it should be possible to establish whether the presence of the chemical is a sufficient condition for the performance of easy tasks to be facilitated and for that of hard tasks to be impaired. The assumption is that the presence of a human semiochemical will inform task-performing subjects of the nearby presence of another human, so the prediction is that when performing tasks with no audience but a semiochemical present, a social facilitation effect will occur.

A test of semiochemical effects on social facilitation would examine both the extent of the effects of semiochemicals, and reduce still further the conditions for mere presence. Schmitt (1986), Guerin (1986) and Bond and Titus (1983) all found that social facilitation can occur with nothing more than a person present. Cohen and Davis (1973) showed that the person does not even need to be there, just a suggestion of an audience. This study attempts to show that instead of suggestion, we can cause this effect with the scent of a person not even consciously detected.

This study will use androstanone, the synthesised version of androstenone, a member of the 16-androstene group as the semiochemical in question. The artificially made androstanone is functionally identical to the natural chemical androstenone. There is a subtle chemical difference between the two molecules, but it lies at the inactive end of the molecule (Van Toller et al., 1983), so it can be confidently inferred that any effect found with androstanone in the laboratory would occur with the naturally occuring androstenone.

Unlike many other studies involving semiochemicals (see for example Van Toller et al., 1983, McClintock and Stern, 1998) this experiment will use an ambient presence of the semiochemical, rather than have it applied to or held near the top lip of the subjects. To counteract the dilution in air that it will undergo, a very strong solution of 0.8mg androstanone per millilitre of solvent will be used.

In order to ensure a long enough exposure to the chemical, the study will have to gather data over a longer period of time than Schmitt's very brief trials. Similar typing tasks will, however, be used, but will be presented as a specific tasks to be completed. This way, although subject to Markus' (1978) criticism that subjects will know they are being tested and thus a true 'alone' condition may not be achieved, the base 'alone' category will at least be consistent throughout the study. Any effect of androstanone will therefore be on top of any effect caused by subjects' by the experimental environment.


H0-1 Subjects will not type significantly more characters in the easy task than in the hard task.
HA-1 Subjects will type significantly more characters in the easy task than in the hard task
H0-2 Subjects tested in the presence of an observer will not type significantly more characters in the easy task and will not type significantly fewer characters in the hard task than subjects who are not tested in the presence of an observer
HA-2 Subjects tested in the presence of an observer will type significantly more characters in the easy task and significantly fewer characters in the hard task than subjects who are not tested in the presence of an observer
H0-3 Subjects tested in the presence of an androstanone will not type significantly more characters in the easy task and not type significantly fewer characters in the hard task than subjects who are not tested in the presence of androstanone
HA-3 Subjects tested in the presence of androstanone will type significantly more characters in the easy task and significantly fewer characters in the hard task than subjects who are not tested in the presence of an androstanone.

Turn to the Method...