Tobacco Documents Online

Rijksuniversiteit Utrecht 4 Psychoiogisch baborafioriwrn -V P-W...z*_y Vartcenmarkd' 2 3511 BZ Utrecht Tetetoon 030 - 328711 1C-' l*.1. 101 w' G"b N, c:;7,0 E.P. Koster, Psychological Laboratory, Utrecht University, The Netherlands PitfaSlis in the sensory analysis of smoking, behaviour. . Introduction Sensory analysis, the systematic study of humani reactions to the properties of foods, beverages, pleasure providing ma~terials (including, cigarettes) and the environment, is still a young,science with many pitfalls ('Os.ter 1198'1, 1983). In ord;er to avoid these a thorough analysis~of the consumption behaviour involved is necessary. Some of'these pitfalls are general and do apply, to all products tested, others are specific for the type of product tested and are related to the specific consumer behaviour involved in using that product. In the case of cigarette smoking, only recently all' the intricacies involved in the natural smoking behaviour are becoming clear and with them, a number of misuses of sensory testing, of cigarettes can be cl'arif ied. The problems involved can, be classified into the following,categories. 1. The variation in, smoking behaviour. 2. The selection of the right panel members.. 3. The change in smokiing behav~ur under unnatural circumstances._ Each of them will be treated here separatel'y, but since thev are ~ closely'interconnected there will be some overlap between, the topics. For the sake of ciaritv, we will take the risk of' some redundancy in suchicases. Crl

2' 2. The variationlin smoking, behaviour. 2.1. General Strange as it may seem to the person who has not given attenti'oni to the matter, smoking behaviour is in many ways a much more complex behaviour than eating or drinking,. It involves a number of manipulations which each in itself may be carried out in a number of different ways. These manipulations are lighting the cigarette, taking, the first puff, taking a number of successive puffs, holding, the cigarette in the hand,,bringing the cigarette (and often the hand),in contact with the lips and extinguishing the cigarette at a certain point with a certain pressure. People differ on all of'tfiese aspects to different degrees but everybody has his own peculiar way in performing them. Smoking, is also a social habit. Some people smoke a lot more when in the company of'others than when alone, others smoke heavily duri'ng, work, but only occasionally, in their free time, etc. Thus, there is a tremendous variation in smoking behaviour which is also strongly related to the particular type of'ci'garette to which the person is used. Thus, smokemof heavy eigarettes are used to obtain a great satisfaction from a relatively easy dYaw on the cigarette. To them smoking a light cigarette which has only li,ttle,impact from the same amount of effort, is a very disturbing,experience. Usually, they try to compensate for this by all sorts of'unnatural manipula~tions such as drawing very hard and'at the same time putting more finger pressure on the cigarette or changing,the insertion depth of the cigarette in the mouth. On the other hand, smokers of light cigarettes, when smoking heavy cigarettes tend to be abhorrediby the,strong,tas!te and if so, often start reducing,their puff volume or smoke the cigarette wilth pointed lips in order to permit the intake of false air througla openiingsbetween the lips. 2L2_ Differences in lighting the ciSarette. ~ N`^- Since smoking is a form,o,f addition, people tend to light cigarettes. CA,w- n,2'(amorng other things such as social behaviour) in accordance to I;k

- 3 - their bodily needs. It is likely that when the nicotine 1'evei'in the body decreases the urge to light another cigarette increases. This is much more evident in heavy smokers than in, light smokers. In, fact, the frequency of cigarette 1lghting under normal conditions varies tremendously from person to person. Forcing,pebple, who at that moment do not feel the,need to smoke, may drastically change thei~r behaviour. Thi's will problaby become most apparent in, the depth of inhalation andlthe strength of the draw on the first li'ghti'ng,puff, where in the case of'a high need for nicotine stimulation, immediate satisflaction is demanded. In the case of a low need only a light puf'f will be taken. When people accustomed to heavy cigarettes are forced to smoke light cigarettes this effect may still be enhancedl. Also, the number of cigarettes smoked in the test period and!the intervals between cigarettes allowed, willi account for Iarge! differences in the inhalation on the first puff between, subjects and: even withinithe same subject. In a well formed experiment ample time should be given and people.should be allowed! to smoke according to their needs. This requirement is seldom met in laboratory experiments. If it cannot be met, care must be taken not to overload'l the subjects and the experimenter should use only habitual smokers of the same type: of cigarettes for comparison purposes. 2.3. The smokinQ pattern of subsequent puffs. People vary considerably in the rate at which they take puffs from one cigarette. This is also depend~ent upon factors in the cigarette itself, such as: ease of draw, burning characteristics of the cigarette, the properti'es of' the,filter used and tightness of packing,of the tobacco rod. -Thus, some cigaret2es'burtr much quicker between draws than others andl as a result liess puffs will normal!ly be taken from, them or the,f,requency of puffing is speeded up because the subject will want to "use" the cigarette. Iif' in the laboratory one forces people to take puffs on,the basis of' a fixed schedule and one limits the number of' puffs, this willi lead certainly to a form,of srandardisati'ond but it will limit the validityof the results as an indication ot normal smokino -p,attcrnti. Also. Iorcink~ people to .roic,~ in iin

_ 4 _ unnatural way willi lead again to different forms of compensatory behaviour and thus increase,already existing differences between, subjects. It is a very awkward experience to wait for a sign before one can take ai new puff or to be f'orcedlto take a next puff before one really wants it. 2'.4. Differences in hol!ding, the cigarette in the hand. Holding a cigarette in the hand is a very characteristic gesture, which is performed differentliy by people. Non-habitual smokers can usually be immediately spotted by the: uneasy way in which they hold their cigarettes. Habitual smokers on the other hand, have very fixed patterns in dealing with cigarettes. For the study of cigarette smoking these differences are important because they may influence the results considerably, expecially in the case of ventilated filter cigarettes where the fingers may partially block the filter pores. Some characteristics of the.cigarette, such a& the length of the cigarette and of'the filter and'the compressability of the filter and the tobacco rod also may influence the behaviour. Since the gestures inmvol'ved' in smoking are almost as characteristic for a person as his fingerprints, interfering with these habits by forcing people to smoke with a cigarette holder or other instruments may completely change their smoking behaviour. In studies intended to predict the intake of tar and nicotine during normal human smoking, the use of such instruments should be avoided. 2'..5. Bringing the cigarette (and the hand) in contact with the lips. Lip stimulation is an, integral part of the smoking, pleasure. The lips are among the most sensiltive areas of the huma~n body as far as the sense of' touch is concerned. They play a verv large role in, the contact with the mother in, early infancy and they, are recognized erogenous zones in most' people of the world. Furthermore, the lips provide the first contact with materials: from, the external world which are to be ingested. Manv people stimulate their lips regularly when not smoking. Whille smoking, the lip stiimulation by the cigarette is accompanied by touching,

- S - the lips with the fingers., At the same time there is a strong tendency to avoid touching foreign subjects with the lips, especially when they are cold or are supposed to be unhyg3.enic. Nobody would consid'er to use somebody else's cigarette holder or pipe, for instance. Onithe other hand many people bilte their own pencil or pen and touch it with their lips. People vary widely in the way they stimulate their lips during normal smoking. Differences in the type of' filter paper and thee heat transmission properties.of the filter material will influence aSmost certainly the way in which people bring the cigarettes in contact with their lips. Here again external objects such as cigarette holders and other instruments will change the smoking behaviour to such an extent, that comparison of'the data obtained with such devices with data from normal smoking,, is invalid., 21.66 Insertion depth of the cigarette. The part of'the cigarette taken into the mouth varies also greatly among smokers. On the average the insertion length is about 10-11 mm, but some smokers actually put 20 mm.in their mouth. Here again, no valid prediction can be made about the insertion, depth of the cigarette when different cigarettes.as the ones a smoker is used to:, are used. Length of the filter and length of the total cigarette are probably among the most important factors determining changes'in the insertion depth. In laboratory studies the insertion depth; i's seldom a controlled variable and if it is, this leads to unnatural smoking behavi~ouir. The devices employed for measuring the ventilation air and the smokers stream separately,, dol not leave enough room on the filter end to permit normal: smoking without touching the exsernal object itself. This leads to very abnormal smoking behaviour. 2'.7. The diecilsion to stop smoking, a cigarette. People not only may differ considerably in the, rate of taking puffs, but also in the absolute number of ' puffs they take before deciding to extinKV ish the cigarette. Mere, the intricate interplay betveen the different nicotine needis, of people and the nlicotine delivery of the ci,izarette smoke, plavs of course an important 202839'753+6

- 6~ - role, but other factors are important as well. Thus, the amount one wants to smoke of &cigatette is dependent on ehanges in the taste of it over puffs. The taste of all cigarettes changes over puffs, but in some cigaretes this:effect is stronger than in others and people may form strong,habits on~the basis of their use of a particular brandi. Another reason.fo:r differences between peop:le in the point at which they extinguish a,cigarette is clearly economical. There will probably be a nice:negative correlation between the socio-economic.status of the~smoker and the length of the extinguished butt.. The price of cigarettes must also have an influence.. Strangely enough, we have the impression.that:there are also important differences in,the.extingwishing behaviour itself, which, may also have their reflections onidifferences in the smoking behaviour. Some people extinguish their cigarettes in a.very gentle way, but others go about it with great gusto and, agressiveness. For the latter group it seems that the extinguishing behaviour has in itself a pleasure component. It is often combined with reaching the decision to go on with another task (back to work, gping,out of the room). This illustrates once more that normal smoking behavi&r is closely interwoven with the everyday behaviour of normal life. Experiments in,the laboratory are very unnatural in this respectt also.. Letting,people smoke~to a certain:mark on the cigarette not onlv disturbs this normal integration of smoking behaviur in everyday life, but alsb,forces ttie subject to constantly focus his attention on the smoking behaviour i~tself. This leads to changes in the normal behaviour pattern,similar to those in the walking patterns of people who are asked tocarefullv consider their ownmotion patt'erns in walking. 3. Select'ion of panel members for smokinQ tests., 3.1 General Since people varv so~ widelv in their smokinig, behaviour aind' [huss smoking behaviour and this variation is partlv related to the

- 7 - experiences they have with~their own brands of cigarette, great care must be: taken in selecting panel membe!rs for smoking tests:. In many pubiications on cigarette'smoking the description of the subjects employed is ve'ry'unsatisfactory. Without such information the results!are usually invalid. In a recent study on the effectss of different kinds: of venti~latedi filters, it could' be demonstrated that there are drastic differences between smokers of different brands smoking the same cigarettes (Ko'ster 1985). The main pitfalls encountered with respect to selection of panel members are: a. Not specifying the sex, age:. smoking habits and socio-economic status of the panel memb!e!rs. b. The use of smokers of heavy cigarettes to judge light cigaretes and vice versa. c. The use of informed and non-independent subjects. d. The use of insufficient numbers of subjects. 3.2. Not specifying the sex, age:, smoking habits an& socio-economicc status of the panel memb'ers. Men and women, are equal in, many ways, but it is: very unlikely that smoking behaviour is among them. There are still large differences in the consumption patterns of inen and women. The same holdss for differences in age. Smoking,habits, such as usual brand and frequency of smoking are also important factors to be accounted for. Finally some! indication of the,socio-economic status of the subject is important in view of some of the questions discussed in the previous section. In many countries i~n Europe people have stopped smoking for economic reasons. This means that the smoking habits of those who continue may also have been affected (smokers of self roll cigarettes for instance)'. Iin all sensory work such details about the composition of the panel should be mentioned. Without,them the relevance of the data can not be properly judged. Especia!lly in the case where differen~t groups are used to compare the effects of different cigarettes, it is neces,sarv to show that the different groups do not dif:er in theiir composition, resairdirnV these aspects.

8 3.3'. Using smokers of heavy cigarettes to judge light ones and vicee versa. As has been pointed, out in the previous sections of this paper,, presenting cigarettes with, very different character to people. accustomed to a particular type of cigarette, may lead to all sorts of abnormal and compensatory smoking,behaviour. In the selection of panel members for a given experiment great care should be taken.to avoid this pitfall and in all cases details about the usual brand of'the panel members shouldibe specified. 3.4. The use of informed andlnon-independent subj'ects. In many instances in the literature, people have been used as panelists who are employees of one oflthe tobacco companies. Usually these people are accustomed to the cigarettes from: their own companies. This means that they are not representative of the total population. Also, these people often have more knowledge about'cigarettes, properties of fil'ters etc. than the ordinary layman. This may make it possible for them to recognize the cigarettes from their own company an& those of competitors. Since most people are very loyal to the group to which they belong, it i& very d'ifficult for them not to let their judgements be influenced by suc'h-knowled'ge. In studies in which different brands and cigarettes of different companies are compared, even under the strictest double blin& conditions, such non-independent panel members should'never be used. In general, it is'better to use groups of completely independent andluninformed subjects: panel members. 3'.5. The use of insufficient members of subjects. Since people differ so widely in their smoking behavi'our, the numbers of'people needed to do good comparative studies is large. This especially is true if so-called "between suibjecrs designs" are used. in which the results of different groups, each judging, a different cigarette, are used. Even when "within subject designs" are emploved, in which eachi subject judSes all different cigarettes, a number of at least twentv well chosen persons is necessarv

to guarantee the relevance of the experiment. In "between subjects design" studies, the numbers of paneli'sts to judge each cigarette shouldibe well over 50 tolprovide sufficient eonfidence in the obtained results. A1'so, the groups should'be well matched according to the criteria mentioned under 3.2. Smoking under unnatural condi'tions.. As was pointed out above!, smoking,i's also ai social habit. In many laboratory experiments people are seated separately in boxes and asked toldevote all their attention to the smoked cigarette. This is~a very unnatural situation, which is only then beneficial when one wants to obtain purely analytical datalfrom the subjects. For instance, in making the:same group of persons judge the strength and ease of draw of the.same set of cigarettes. In~such a case: good experimental precautions should be taken, such as the systematic variaition of the order of the presented cigarettes over the panel members, but if such measures are taken,,the results are validd as a comparison withinithe restricted'set up of'the laboratory experiment. If, however, one wants to draw conclusions about normal smoking,behavioar, the:validity of this type of laboratory experiments is limited. Certainly, the fact that al subjects have testedlall the cigarettes under the same conditions is very important, but the fact that the normal social context was completely changedd and!tha~t speciaL attention was asked for the smoking activity itself may have influenced1the normal! smoking behaviour. In everyday life smoking is an activity which accompanies other activities, but is very seldom in the forefront of one's attention. The very fact, that in sensory e,xperilments explicit attention to the smoking activity is demanded, may disturb the results. This is,a dilemma in all sensory research. Elsewhere (Koster 19'8'1') we have suggested to use direct observation of normal behaviour rather than explicit questions to the subjects as a means to circumuent this problem. Thus, in a study oni beverages, we were able to register the frequency of drinking and the intervals between gl!asses for each member of a, group of 96 peoplk in a normal bar situation. It could be demonstrated,that the an5wers to questions asked aib~.~ut preferences for driniks and Iah,rr:3corv tests done with t,he s:nme

- 10 - group of people, were not correlated with the actual drinking behaviour of the people. In the case of' smoking, it might be well worth the effort to develop similar methods for cases in, which a truthful picture of'the normal smoking,behavi'our and the preference for certain cigarettes is demanded. 5. Conclusions and'discussi;on Does this mean that all laboratory experiments should be stopped? In no way. As indicated earlier, many analytical questions (strength, ease of dhraw, etc.) can be answered wi'th, well designed laboratory experiments, carried out wilth properly selected, and matched grou!ps of independent people. This is true, especially when designs can be used in which each individual tests all the samples to be compared, andiprovided that people can smoke in the way they like. This means that they are not forced to use cigarette holders or other instruments that prevent the normal~ contact with the cigarette. As soon, as such diemands. are made, or when the normal individual smoking,pattern is broken by asking people to adhere to rigid time schemes for lighting the cigarettes or taking their puffs or to rigid prescriptions about the number of puffs or the amount of the cigarette.to be smoked, the results wiLi no longer reflect the normal smoking behaviour. As was.also pointed out, such deviations of' the normal behaviour may have different effects on the results obtained with different types of ' cigarettes. On the basis of such experiments no conclusions can be drawn about what would happen in the situation in which normall smoking had been permitted. The clearest demonstration of this are the experiments with the so-called Puff' Parameter Analyzer, an instrument desigrned to measure the ventilation air and the smoke stream; separately under "normall human smoking" conditions. The instrument is ilIustrated in figure 1. The idea of the apparatus is the following. The air used for venti~latio~, of the cigarette which is suckedlin through the pores in the filter is measured by placing r.he cigarette in a holder with rubber membranes on,both sides of the pores. The air taken in through this holder is the ventilation air. Bw f ittin}; the holder vithi the burning end of the cigarette in another ve,,sel, the srream of air through

the tobacco rod can be measured. A1'though in itself, the idea, to measure the two streams separately is scienti'fically very soundl, it is absolutely impossible to mimic normali smoking,with this device. Normal contact wilth the cigarette, one:of the most importantt factors in smoking, has become impossible. The instrument iis aa heavy glass instrument which prevents normal finger contact with the cigarette. Also, since the pores in the filter are at 131mmi from the mouth end a!nd •one of ' the rubber membranes has to be fitted' between mouth,and'po,res only 1'1-12'mm are left for insertion of the cigarette. This is not sufficient to alilow for normal smoking, because people will not want to touch the foreign object with thei!r lips. Although the normali insertionidepthiis 1'0-11 mm, the lips curve outwards above andibelow the place where they make contact withithe cigarette and thus at least 151mm should be allowed'to have the lips free from the instrument at normal average insertion depth. Even then, normal lip contact would'not take place, because people have difficulty handling the voluminous and heavy instrument withithe same precision as.a cigarette andi will therefore by very reluctant to put it to their mouth. Instead, they will bring their lips outward and will take only the very end of the filter between the softer iinner si'des~of the lip, which will fall partly over the roundlend of the filter. In this way the stream of air through the filter is blocked to a certain extent. This will be even more so, because the weight of'the instrument forces the person to put more lip pressure on the end'of the filter than normally. All these effects lead to a completely abnormal smoking situation, which has different effects on the results obtained with the instrument for different types of filters. For one type of filter, the so-called KActron filter, which allows the ventiliation, air into the mouth trough ventilation channels which are situated at the outside boundary of the filter, the results are changed dramatically. The reason is that the softer parts of the lips who are normally not in contact withithe end of the filter, are now closing off the ventillation holes, thus preventing the normal ventiliat,ion to take place. The results of' other types ot filters are less affected' bv the umnatwral wav of smoM1cim4 forced upon

the su!bject, because they'do not feed the ventilation air in the mouth through channels, but through the inner part of' the filter. This illustrates perfectly that creating unnatural smoking conditions may have differential effects on the results obtained for different types of cigarettes. Results obtained in this way have no relevance for normal smoking,behaviour.

References Koster, E.P. 1981. Sensory evaluation in the natural environment. . In: Proceedings of Flavour '81 (the third Weurman Symposium4 Munchen (P. Schreier, ed.)1, Walter de Gruyter, Berlini, p. 93-101. Koster, E.P. 1981. Time and frequency analysis: a new approach, to the measurement of some less-well knowniaspects of food preferences? In: Criteria of Food Acceptance {Ji. Solms and, R.L. Hallls eds.), Forster Verlag,AG/Forster Publishing Ltdi., Zurich, p. 240-2'51. Koster, E.P. 1983. Uses and!misuses of sensory analysis - psychologist'' point of view. In: J. Adda andl H'. Richards (eds.). International Symposium on,Food Flavors, Technique et Documentation (Lavoisier), Paris, pp. 129-137. Koster, E.P. 1985. Sensory analysis: aims, objections, constraints, possibilities. Proceedings of the EBC Symposium on @uality Assurance, Zoeterwoude, 12-1'4 novenber, 1984 (in press). E.P. Koster, 1985. A psychophysical study on the effectiveness of cigarette filters. CONFIDENTIAL unpublished preprint.

Vf'n 1'4 A+J'.,. c -fic Ali,:, fti.*&f4 Vo 4r Lcv r, .!