Tobacco Documents Online

R&D-B057-95 SENSORY CHARACTERIZATION OF BODY-MOUTHFUL N. AYYA, R. A. DE WlJK September 5, 1995 RESEARCH & DEVELOPMENT THIS REPORT MUST NOT BE COPIED OR SHOWN TO UNAUTHORIZED PERSONS. 373.00030;~

:B&Vv" BROWN & WILLIAMSON TOBACCO CORPORATION RESEARCH & DEVELOPMENT RESTRICTED ' R&D-B057-95 SENSORY CHARACTERIZATION OF BODY-MOUTHFUL N. AYYA, R. A. DE WIJK September 5, 1995 DISTRIBUTION 1. E. Kausch (BATG) 2. W. Hass (BATG) 3. G.A. Read (BATCo) 4. R.R. Baker (BATCo) 5. L.T. Caruso (Souza Cruz) 6. I. Volpon (Souza Cruz) 7. S. Massey (ITC) 8. P. Dunn (ITC) 9. T.F. Riehl (B&W) 10. R.H. Honeycutt (B&W) 11. N. Ayya (B&W) 12. R.A. de Wijk (B&W) 13. Souza Cruz Library 14. BTC Library 15. BATG Library 16. B&W Library 17. B&W Library 18. ITC Library Issued by: R. H. Honeycutt Copy Number: /'-JJ Project Number: 468 371000303

Page 2 R&D-B057-95 RESTRICTED BROWN & WILLIAMSON TOBACCO CORPORATION RESEARCH & DEVELOPMENT R&D-B057-95 SENSORY CHARACTERIZATION OF BODY-~MOUTHFUL N. AYYA, R. A. DE WIJK September 5, 1995 SUMMARY The purpose of this study was to generate a set of descriptors which describe Body-Mouthful sensations for nonmentholated cigarettes. These descriptors could then be used to provide additional insight for product development. A set of six descriptors was generated in nine group sessions in which ten R&D expert smokers smoked and rated ultras, lights, and full- flavored cigarettes. The generated set consisted of the following descriptors: Amount of Tobacco Taste, Amount of Smoke, Texture, Stinqin~. Creaminess and Flow Over Tonque. Subsequently, the smokers were trained in the use of the descriptors, and finally they entered the test part in which thirteen brands were rated three times. The set of descriptors distinguished successfully between brands; even between brands within the same tar category. Descriptor ratings increased between the first and the last puffs. In general, Light cigarettes received lower Body-Mouthful ratings than Full Flavored cigarettes of the same brand, except for VICEROY where Full Flavored and Lights received similar ratings. Body- Mouthful ratings for VICEROY and Marlboro were well matched. Basic FF and Lights received relatively high Body-Mouthful ratings. Within tar categories, Body-Mouthful ratings were associated with tar/puff, nicotine/puff, and CO/puff, which corresponds well with earlier findings. The correlations with nicotine/puff and CO/puff are most likely due to covariance with tar/puff. Body- Mouthful was also associated with chloride levels in the tobacco. This might reflect an association with blend composition. Future studies are proposed to investigate systematically the effects of blend composition on Body-Mouthful sensations. 371.000304

Page 3 R&D-B057-95 Backqround Body-Mouthful sensations elicited by cigarette smoke have been the subject of a number of investigations in the past decade. Summarizing, the results were: •. Body and Mouthful ratings are somewhat higher for moist tobacco than for dry tobacco. The effect was only found for nonmenthol cigarettes (St.Charles, unpublished results). • Ammonia seems to enhance body and mouthful. Interestingly, at higher levels of ammonia, mouthful becomes mouth coating. This suggests that mouthful and mouth coating are part of the same sensory continuum (St. Charles, 1986). • Band cast reconstituted tobacco, or RCB, seems to reduce harshness/irritation and increase body and mouthful. Since RCB contains ammonia, this finding is consisting with St. Charles' findings. Furthermore, RCB is a component of Marlboro FF, which may explain the relatively high body and mouthful ratings elicited by Marlboro KS FF (Roth, 1984). • The hydrophobic compound n-hexadecane seems to enhance body and mouthful, without affecting impact or irritation. The hydrophylic compound glycerol had no effect on either of these variables (Bevan & Shepperd, 1992). • Reducing the density of the tobacco in the tobacco rod seems to increase body and mouthful sensations (Hook, 1991). Perceived mouthful of smoke seems to be more dependent on pressure drop than the actual amount of smoke (Powe, 1991). ° Mouth dryness may be important for body and mouthful. The role of filter ventilation on mouth dryness has been investigated and seems to be unimportant (Shepperd, 1992). • Other, more anecdotal findings indicate that body and mouthful sensations increase with the levels of CPCL and with the basis weight of the cigarette paper. Waxes like triglycerates also seem to enhance body and mouthful sensations. Additives like cocoa, butterfat and irritation reducers also seem to act as enhancers to body and mouthful. 371000305

P~ge 4 " R&D-B057-95 All these investigations concern variables that affect Body-Mouthful sensations, which implies that Body-Mouthful sensations exist and that they are important. However, the number of studies on the existence and importance of Body-Mouthful sensations is surprisingly limited and are summarized below. Blindfolded subjects who puff but do not inhale cigarettes can differentiate between lit and unlit cigarettes, which suggests that body and mouthful sensations are genuine (Dixon and Whitehead, 1992). Body and mouthful may play a role in self-controlling puffing behavior (Whitehead, 1994). Smoke from cigarettes with high body and mouthful was inhaled in smaller volumes and over shorter periods of time than smoke from cigarettes with low body and mouthful. In addition, the smoker seems to be able to adjust smoking behavior almost instantaneously while puffing. Interestingly, none of the investigations deals with the question of what Body-Mouthful sensations are, i.e., Body-Mouthful sensations are not defined. A sample of descriptors for body and mouthful sensations used by R&D personnel illustrates the lack of consensus: "Feel", "Warmth", "Coating", "Lack of Airiness", "Chewy", "Smoothness", "Balance of Irritation, Taste, Impact, and Draw", and "Balance of Volume and Texture". The goal of this study is to generate an unambiguous descriptive profile of Body-Mouthful sensations which can be used in other studies on Body-Mouthful. For this purpose, group sessions were held in which expert smokers generated and discussed descriptors for Body-Mouthful sensations while smoking cigarettes with a wide range of tar deliveries. Group sessions were followed by individual training and test sessions. 371000306

' -- ~age 5 R&D-B057-95 Methods and Materials Thirteen Ultra, Light, and Full Flavored nonmentholated KS brands were selected with the help of Baran Chakraborty. These brands are thought to elicit a wide range of Body- Mouthful sensations. The brands are given in Table 1. Nine of the thirteen brands were presented in all parts of the study, while the remaining four brands were presented only in the test part. Table 1: Brands used in this study ii~e:~8:~iiiiiiil;ii!B~:6:~ili:.ii~i i; ::~;;:;c;;~;~; ~;;;~;:::: ~CZ~I~6~ii~i~i~iiiiii!iiiii~ii~!~i!.iiB~!~i!!!ii~iiiiiiiii~i~!i!!i~ii!i~ii~i~i! .... ........ ," ,'", .....:" '" ':"';':':'"" " ', .............................. :":'T" "'"':':'::':'T':'?;¢':''':': 1. Ultra Now 8. Full Flavored 2. Ultra CARLTON 9. Full Flavored 3. Ultra Merit Uitima 10. Full Flavored 4. Light 5. Light 6. Light 7. Licjht Marlboro VICEROY Winston Marlboro 11. Full Flavored State Express 555* VICEROY 12. Full Flavored GPC* GPC 13. Full Flavored Basic* Basic* * Brands used only in the test part of this study. Ten subjects, recruited among the R&D personnel, participated in the descriptor generation part of the study. Nine of the subjects completed the stiJdy. All subjects except one are members of expert panels. The study consisted of three parts: descriptor generation, training and testinq. The procedures for each of these parts are described below. The results of the descriptor generation and the training will be given in this section. The testing results will be given in the Results section. Descriptor generation: Nine group sessions were held at a rate of three sessions per day over a period of three days. The sessions were used to generate, discuss and select descriptors for Body-Mouthful sensations following the smoking of cigarettes. ..aT .oooSO"/

• Page 6 " R&D-B057-95 The group of subjects smoked three cigarettes per session, one Ultra, one Light, and one Full Flavored presented in a random order, and rated the intensity using a rating scale with two anchors placed on a black board. Subjects smoked nine brands once across three sessions. On the second and third days feedback was given on how the ratings compared with the corresponding ratings on the previous days to improve the consistency in the ratings of the group. The group of subjects agreed upon a set of 6 descriptors for an adequate description of Body-Mouthful sensations. These descriptors are given in Table 2, together with the selected anchors. Table 2: Descriptors used in this study Amount of Tobacco Taste Little A Lot Creamy Thin Thick Texture Smooth Stinging No Sensation Amount of Smoke Air Flow over Tongue Light Rough Irdtating Enough Heavy The set of descriptors covers sensations of taste (Amount of Tobacco Taste), irritation (Stinging), tactility (Texture and Flow over Tongue), density (Creaminess), and volume (Amount of Smoke). "Enough" was selected as high end anchor for Amount of Smoke because subjects never take in more than "enough" smoke. Amount of Tobacco Taste and Amount of Smoke were typically assessed during exhalations. Texture was assessed during inhalations, whereas Flow over Tongue was assessed during puffing. Creaminess and Stinging were assessed during inhalations and exhalations. The degree of (dis)similarity between the descriptors was assessed in a Multi Dimensional Scaling exercise in which the similarity for each pair of descriptors was rated twice on a 9.2 cm line scale ranging from "descriptors with the same meaning" to "descriptors with very different meanings". The matrix of averaged similarity ratings served as input for a Multi Dimensional Scaling procedure. This procedure is described in more detail elsewhere (PGS-B003-95). The results, shown in Figure 1, showed distances between the descriptors indicating that none of the them have similar meanings. Distances are smaller between Creaminess, Texture, Amount of Smoke, and Tongue Flow, and larger between Tobacco Taste and Stinging. _ 371L000308

I~age 7 R&D-B057-95 Figure 1" Multi Dimensional Scalinq Plot For 2 6 Body-Mouthful Descriptors O4 c- O E £3 -1 Stin@ng Ore#my Texture Amount of Smol + + Tongue Flow + Tobacco Taste + -2 ' -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 Dimension 1 e Trainin,q: All subjects, except subject 7, individually rated the nine brands twice over a period of four days in their offices. Subject 7 rated the brands only once. Ratings were made on a 13 cm line scale with anchors (see Table 2) placed at the extremes. The subjects were not explicitly instructed on when to make their ratings. The presentation order of the brands was randomized per subject. Individual test-retest correlations, a measure for the consistency with which subjects rate the same brands on different presentations, averaged 0.60 (range from 0.23 to 0.90). Test-retest correlations per descriptor ranged from 0.44 for Amount of Tobacco Taste to 0.72 for Amount of Smoke. Subject-Rest correlations, a measure of the degree of correspondence between ratings of individual subjects and the group, averaged 0.80 (range from 0.62 to 0.90). It was decided that the subjects did not need additional training because virtually all correlation coefficients were relatively high and significant. Results of the test sessions would later indicate that subjects' performances would continue to improve after the training sessions. 371000309

P~ge 8 ~ R&D-B057-95 Testinq: Subjects rated all thirteen brands three times over a period of six days. Tile subjects were instructed to make two rounds of ratings: one after the first 2 puffs and one at the after the last puff of the ciqarette. Subjects met twice a day for approximately 20 minutes in the smoke panel room where they smoked and rated three or four brands. Each subject received the brands in a different randomized order. Results Individual Test-Retest correlations in the test part averaged 0.70 (ranging from 0.51 to 0.95), up from 0.60 in the training part. Subject-Rest correlations averaged 0.89 (ranging from 0.74 to 0.97), up from 0.80 in the training part (see Appendix A). Test-Retest correlations per brand showed no effect of previous presentations of brands in the training part ("trained brands") on response consistencies for those brands in the test part, i.e., Test-retest correlations for trained and untrained brands were essentially the same (r=0.40 and 0.41, respectively). Results of the sensory ratings will be discussed for all brands combined and per tar category. Results across tar cateqories: Ratings, averaged across stimulus replicates and subjects, are shown per descriptor in Figures 2 and Appendix B. Analysis of Variances (ANOVAs), performed on ratings per replicate averaged across subjects, showed significant differences in ratings between brands and between descriptors (see Table 3). In addition, ratings increased significantly between the second and last puff. This increase varied with descriptor (see Figure 3) but not with brand. The large differences in ratings between brands are not unexpected because the set of brands covered a large range of tar deliveries. Consequently, ANOVAs were also performed on brands per tar category. 374_.00034--0

Page 9 R&D-B057-95 Figures 2: Brand Ratinqs for 6 Body-Mouthful Descriptors d~ 8.0 8 ~_ 6.0 4.0 2.0 12.0 ii i = i = i i = i i ; i " J- I 10.0 "T" O.O BRAND 12.O ,,,,,,,,,,,,, 10.0 8.0 == 6.0 O3 4..0 2.0 0.0 BRAND • Last Puffs D First Puffs • Last Puffs I-I Rrst Puffs 8.0 IT 6.0 2- 4.0 12.0 , = , , , , , , , , , , , 10.0 t O ii o.o, BRAND 12Q , , , , , , , , , , , , , 10.0 BRAND • Last Puffs I-I First Puffs • Last Puffs r-i First Puffs 12.0 , , , , , ; , , , , , , , 10.0 BRAND E O3 == < • Last Puffs [] Rrst Puffs 10.0 ! ] I i 6,0 4.0 J rl J • Last Puffs O.O I rl Rrst Puffs 3?1000311