Discussion
The results validated the choice of tasks used in that the disparity between them in terms of their ease of completion was easily enough to make the hard task significantly harder than the easy task. This significant effect of task difficulty was necessary for the test of social faclitation. As predicted by decades of research, this also proved to be significant, with the interaction between task difficulty and audience condition being highly significant. The interaction between the presence of androstanone and task difficulty did not prove significant in these tests, but did show a strong trend in the correct direction, i.e. subjects in the androstanone treatments typed more quickly when performing the familiar and easy task of typing their names, and more slowly when having to perform the mentally tortuous, unfamiliar task devised by Schmitt et al. (1986).
One possible reason for the lack of a significant interaction could be the choice of tasks used. While they were found to be significantly different in terms of ease of completion, the easy task elicited a wide range of responses, from 488 to 2045 characters typed in the time available, and a standard deviation of almost 329 characters. This exposes enormous differences between subjects in terms of the ability to type their names: from around 16 to 66 words per minute. In other words, the best typist was more than four times better than the worst. The experimental design did assume that undergraduates would be highly familiar with the process of typing their own names, and so most would be able to do so at a high speed. The large variance in the scores in this task suggests that many undergraduates are expert typists and many are little more than beginners. The expert typists did not appear to be any better at completing the hard tasks than the less adept typists (no correlation was found during data exploration) so despite the significant effect of task difficulty on task performance, the hard task used seems to have been a better hard task than the easy one was easy.
That there was a trend (albeit non-significant) for the easy task to be performed more quickly in androstanone's presence, combined with the significant effect the semiochemical had in degrading performance in the hard task supports the suggestions that humans can and do use the detection of semiochemicals as part of the fundamental apparatus of everyday sensation. However, further research needs to be done in order to establish firmly the semiochemical's effect in this context, due to the overall non-significant outcome of the analysis. Besides the widely varying keyboard skills of the subjects used, there are several other ways the experiment could be improved. The power analysis showed that using 250 subjects a significant effect could be shown if the difference between the androstanone and non-androstanone means remained the same, but this is an unrealistic number of trials. Each trial in this design takes around 15 minutes, so a study of the size recommended by the power analysis would take 75 hours in trials alone. A way of reducing the variance shown seems to be a much more feasible solution, and it would also establish more convincingly whether or not the audience and sex interaction with task difficulty had any relevance.
The setting could be made much more naturalistic, like Schmitt et al.'s waiting room: the androstanone used could be made to be an ambient odour throughout the whole room, rather than in a very small chamber. The subject could be made to wait for a set period in the presence of this ambient odour, and in this way the exposure could be well controlled while at the same time not alerting the subjects to the presence of any olfactory stimulus. If the subject was thus exposed prior to rather than during the experiment, Schmitt's less time consuming tests could be used. One problem with a study of this kind is the high molecular weight of the molecule androstanone. Its very high weight (around 272) and longevity means it is very difficult to clear from any closed environment, which means in any study using ambient androstanone it is not feasible to randomly assign subjects to androstanone or no androstanone treatments, with potential order effects as the consequence. Furthermore, any experiment of this kind will result in a room permanently awash with androstanone, making it very difficult to conduct controlled studies in that environment again.
There is also the problem that afflicts much university-based psychology research, namely the experimental environment and the subject pool. All trials for this experiment took place, as stated, in the University's anosmic chamber, which is situated within the (well labelled) Chemoreception and Olfaction laboratory, so the typical subject will be a psychology student who probably knows a bit about social facilitation (and may even spot they are being tested with social facilitation in mind) and may well be looking out for something related to odour. Despite the assurances otherwise, it is difficult to believe that all these informed, intelligent people were truly naive subjects. This may account for some lack of social facilitation effect in the semiochemical treatments; if the effect was genuine, but subtle, then it may have been masked by the effect of the unusual environment and subject anticipation.
While the likelihood that the difference in means between the two levels of the androstanone treatment was due to chance is too high to be statistically significant on this occasion, it was low enough to warrant another, more precise investigation. While there are plenty of studies linking the putative human semiochemicals with sexual behaviour, there is a dearth of well designed, well-controlled research in the field of semiochemical effects on social psychological phenomena. If these effects (such as Atzmüller, 1998 unpublished, and this present study) were shown to be reliable then it would impel new thinking and possibly new research into any interpersonal phenomena involving close contact.
However, further research into this area should also take care not to ignore biologial links between this kind of social phenomenon and the biological stimuli. McClintock and Stern (1998) are certain that donor women's secretions led to the recipients' menstrual cycles changing. A more useful study would be to examine the hormonal changes in the recipeints that occurred when exposed to the axillary secretions, and see if artifically induced hormonal changes of these kinds could lead to the same changes in cycle length. In Van Toller et al. (1983) it was shown that a drop in skin conductance often accompanied unconscious exposure to androstanone. As the link between skin conductance and autonomic arousal is well established, it might well be more profitable to examine the effects of androstanone on autonomic arousal and arousal on social facilitation (or other social psychological phenomenon) separately. Similarly, given Atzmüller's findings that androstenone exposure results in significantly lower saliva testosterone levels in men, and Jütte's (Grammer and Jütte, 1997) findings that exposure to copulin resulted in higher saliva testosterone levels in men, the effects of the levels of this hormone on behaviour could be tied with more research on different semiochemicals' effects on hormone levels to produce a clearer, more theoretically structured account of semiochemical effects on behaviour. Such an account would be of far greater scope and worth to the discipline as a whole than a large number of isolated and narrow examination into the specific effects of specific semiochemicals in specific circumstances.
What is needed is a co-ordinated study of the immediate biological effects of semiochemical reception, together with the psychological effects of those biological responses. Furthermore, this study needs to be aimed at arriving at a general academic consensus concerning semiochemical effects in humans. While people have been selling elixirs and aphrodisiacs for many centuries, often claiming totally unsubstantiated properties for their concoctions, the company Pherin is publishing its research findings, in an attempt to be more credible. This means the effects it is examining and the procedures used are open to informed and impartial analysis and controlled testing. If the VNO was found to be active, as they suggest, then research into its properties could be conducted which would examine the organ in its entirety, rather than just those properties that might help sell perfumes and potions.