Tobacco Documents Online

V. 1-l. Winston & Sons. Inc., Puhlis ~in I: SNrn, sSV. n•aQNqUMb ul•. CODING PROCESEES IN HUHAN MEMORY [/~MM Anhw I/. M./wn W [LwnMwNn AN ixn ~M Nyn[ .M. F NWn/ wwm/Ma 1n ni n~ W I.M N. LM.waxl N•E.M~w N FlRhMM'. x.M w~eMx u/ l.un.n inma.. Lv.~k.Mp.aryY.+ .n~lh. ~x•N .1 M6.1/.Y Y/xuuw. M. ~Yealwl M W F^ ~x L~e..wM. LANGUAGE COMPREHEM5ION AND THE ACOVIeIT1ON OF RNOYflETNE [Fnlb IN. AL../..Ma.M [. fr..x ~x[MtwnN.nhMx~~l/ M1.Uw ATTACHMENT AND DEPENDENCY [[ryM.Y IwM L .u Ywlpw..l.~fM WNMIIIY x~.1M14 MYw M IM MNW.I~~pw M N .mehmnl N Mn'Jx.ry. E.A N~M.M.{mnV.lilMl W~w111ry _ ewyw~lx.• ,N AIMUT AYtfMIMtl X wM M w NRWMFU Mi[ 11NeHI~n Ne.W w w.nx,.,axl.x..wnm.hwn..e. ]11 N/x Ller CONTEMPORARY A/IROACHE$ 10 CONDITIONING AND LEARNING EN~M.Y [. A hkU.Nmw4G. 4xY LvnW.n , htl m~InR ~IM1 t~x~µMmw .1 Wn iM+~ IM Wl MMI W .+n • 6M.1 W WrMry-.N.w.M FMynY ~.nVM~Nm ~~ w M tlx..ryn/Nn N pYk PIUNNI qrln/ Yww M~pW N In•x owni: i ' N:Ne"..w':IO~AwPw NvYa M~M~Pn W~al. ewp.. MMxw x I.wM~E• eNJIHn•h1.wuMnM.d.My .MIMyn.4Yh~.Nr. ' CONTEMMHARY N9UEE IN COGNITIVE RYCNULOOYI THE LOYOLA SYMIOSIUM [MtNF11.MIl1Mx M/iv.N.~~w p•e.n,qlx. xINM.wxx.nuh Ir.n ~aln. Wy M+.6.IV MwM ~M~\ry m iwOU..yJ laelMnry ne IwM Mw Iww~ N•M/.rv..v~ MY 1. rwrA MMXyn~. tM ~• wrWN wy InMw.Yw~ MwwWw. .w~~~Lwnw.ala/•4Ii M/NMIN w.x PMMn M J.MY/Xw Y,11 Ynnxen\f ~Wk A E..µwMwwtA/w.1a.M, VwrYx. NI/.

SMOKING BEHAVIOR: MOTIVES AND (NCENTtVES. ;i roa EDtrED BY WILLIAM L. DUNN, JR. V. H. WINSTON & SONS 1973 Waazhington, D.C. DISTBlBD7ED BY THE HAISTED PfiESS DIVISION OF JOHN WILEY & SONS New York Toronto London Sydney 2060489393

V n PREFACE Early in his career some 50 years ago, Clark L Hull, foremost among American psychologists, described his ongoing research efforts as a search for "... a clue to the chum which tobacco has for those accustomed to its use." Hull never found that elue. When I first became associated with the cigarette industry in 1961,1 came across a contractual survey in which one of the questions asked of the smoker respondents Was °Why do you smoke?° Of those whose reply went beyond the cliche, "It's a habit," about half of them said "It stimulates me:" The other half said "It rclaxes ma." How to reconcile the temackabte polarity of these responses has intrigued but etaped me over the years. - In January, 1972, a representative group of life, behavioral and social scientists eomened on St. Martin Island of the Lesser Mtillea to reflect upon human cigarette smoking behavior. It was hoped that such a conference would red'uect the scientifio community's interest to the fundamental motivational question which has gone unanswered since Clark Hull posed it a half century ago, and which has not been given the priority it deserves during the past decade. It was further hoped that the conference would correct for a dearth of Interdiscipline cross talk among those conducting research on smoking. The phamucologists were bent upon understanding the modifying effect of nicotine and other smoke constituents upon biochemical processes. The psychologists were aaekfng to fit their observadons into theoretieal models of learning and personality, looking chie0y at the differences between smokers and nonsmokers. The sodologists were documenting the refraetoriness of the habit in face of massive dawasion campaigns. The anthropologists were still recording the bnuhfiralike Ro&aphie diffusion of cigarette smoking and its integration into the cultural petterns of «chnology-temote peoples. Common to all of these efforts was the RMra<tee of the data to the problem of smoker motivation, yet little was happening f~mew.yoftaeae>ehange. C 03G0S 3t~:~ .~ 2060489394

xii PREFACE This, then. is the question that was put to the St. Martin conierees: "R9vt are the mottvattonai mechanisms sustaining clgzrette smokmg behanor?" Thetr replies make up this volume. These published proceedings do not oLthemseives rorrect for the aioremenuoned cross talk def~ctenctes among the disciplines. Nevertheless the contributors have been well rubbed against each other in the course of this exerci3e. Only with time will we know if cross-terrilizauon occurred. 7,_onference was sponsored by The Council for Tobacco Research-USA., Inc, December. 1972 SMOKING BEHAVIOR: MOTIVES AND INCENTIVES E'01G09 2060489395

EXPERIMENTAL METHODS AND CONCEPTUAL MODELS AS APPLIED TO THE STUDY OF MOTIVATION IN CIGARETTE SMOKING' William L. Dunn, Jr. Phlfip Morrls Research Center INTRODUCTION Clark L. Ilull (1924) explained hls work as a search for "a clue to the charm whlch tobacco hee for those accustomed to Ils use [p.53j:" Today, almost a half century later, while smokers around the world are smoktng cigarettes al the rate of three trillion (3 X I O") annually, we still seem a long way from a generally accepled explanation for that charm. In view of Ihls state of affairs, and after ten years of closo-in acquaintance with the problem and Its literature, I propose to take a critical look et tlm methods and models that Investigetors have been using In search of that clue. The first part of this chapter Is a discussion of methods. The second part la a discussion of models. ON METHODS Only three experimentel dcsigns are to be found among all the studies of the motivation of the cigaretta smoker. In fact, the great preponderance of the work has made use of but one design. This is the comparative analysis of smokers and nonsmokers. The direct Interrogation approach has been frequently used in surveys, but has not often been reported out In the literature. The third design, the classical experimental approach, has been used hardly at all. I shall discuss each of these three methods In turn. The Smoker/Nonsmokar Design If we discount the great volume of posG19fi5 activity directed toward the development and evaluation of smoking control techniques, the preponderance of 'The vkws expressed herdn are thnse of the author and ne In no wq to be conddered as represenllna the vlewt of 77w CouneO for Tobaceo Research ar of tldgp Mutrh, Inc. I 93

94 W.L.DUNN,JR. the research on the psychology of unoking has been aimed at differentiating smokers from nonsmokers on measures other then their dgaretle-relaud behavior. The standard procedure Ites been to sample among ulf-selected smokers end nonsmokers from in otherwise common population and to obtain "measures" on same personality or behavioral trait or traits, and thereafter to detcrmine whether the d)f7erence between the mean measures for the two groups was too greet to ascribe to chance. A greaJ number of differences between groups of smokers and groups of nonsmokers have been IdentlBed through the application of the smoker/nommoker difference design. I have listed tiese differences In Table I, following the Matsrazzo and Saslow (1960) pmcedent of summarizing the reported differences. The references accompanying the Irallsare not Inclusive; they are provided as access references. For example, Smith (1970) Is cited for a number of the penonality variables because of his systematic summary of all of the empirical work on the relationship between each of these traits end smoking. Some of the traits or characteristics which appearln Table I have been Ineluded despite the fact that the evidence for their differentiating power is conflicting. A good example is anxiety level. There have been repeated observaltons of differences on this measure such that although the difference has not been reproducible under all circumstances, one is led to anticipate some ultimate Identification of in Isolalable varlable that oan account for the Instances In which differences have been observed. Similarly, not sll of the studies reporting on morphological traits have yielded results supporting the exlstence of the listed traits. As In the casa of anxiety, the data are conflicting but provocative. Even a casual inspection of Table I will make evident the redundancy and overlap among the traits. The compilation represents Ihe work of many Investigatars using d(verro conceptual and descriptive frames of reference. Were Ihe rommunality Idenlifiable and the "pum" traits measurable, our list would be shortened (ndeed. This, then, is a summary of our knowledge of the differences between sntokets and nonsmokers In the psychological and related spheres. A question ar(ses at this point as to the direct applicability of this knowledge to understanding the nmtivatlon of the smoker. It )s disturbing In find liull (1924), long before the proliferation of studies using the smoker/nonemoker difference design, issuing a warning which haagone unheeded: We aeeordlnaly find Inaesll9ators rcsortln6 to statisncai analysls usually or acclden- telly avallable data beuing on /he generd mentel efndency of habitual smokers and cumpadng these with similar data from nonsmokers. The chief Interest in lhere InvutlaaUons has been tim Innuenze of smokin8 on scholarship In the aecondary sahools. Accordingly echoal marks have genen0y been used es the criterion. A very large number of such studies bave been reported .... Since Meylan's investigation was reported with mom aero and inn8ht than the olhen and Nvslntes both the weeknesmz and possibllltles of Ihis method of approach. It aMne MB be examined-... Dr. Mayr.n Invasnaakd 115 smoken and 108 nonsmokers in Columbhl Unherslly.... Over a period of two years the aeerage mark of the nonsmokers wae 69% while that or Ibe smokerf was 62%. The same tendency wu thown by the fact that Ihem were only 4% of failures anlonS the nonsmokers as a8ainsl 10%amon8 the smokers. It became evident, however, that other factors besidee smoking were probably eontributine to produce METHODSANDMDDELSAPPLIEDTOMOTIVATIDNINCIGARETTEBMOKIN6 99 TABLE I Individual Traits and Group G7mracterislics by Which a Group ofSmokers Can Be Distinguished from a Group of Nonsmokers hrsonellty ftaRs More Independenl (PRaum,1965) Creater antisocid lendencks (SmI1h,1970) More aclhe, enerRde (Schubert, 1959; Stralls, 1965) Iilgher mean extroversion rttln5 (Smith, 19v0) IlnppyZO-lucky (5mlth, 1969) Ifiglrtr mean meamrn of'braBly"(Smhh, 1970) Poorer mental health (Smith, 1970) Less rIaid, less orderly, mote Impulsive (Smllh, 1970) Caaler mllance on "external" than nn "Intemal° mnlmis (Smilh, 1970) Mare ehance-0tknled (StuHS A Sechmst,1963) Mote emotional (Smith, t967) Len eateeabk (gmilh, 1969) "7ype A^peHonalily (mnrc timeeoneelou; competltNq etc.) (Rounmm, pdedman, Jenkins, Strauq Wurm, A Kosi(chek, 1966) Lees "strenSth or chancler" (Smlth, 1969) Hfaher anxiety lewl (Walket, Nlcolay, Kluczny, A Reidel, 1969;SteM, t968;Thomu,1968) Life-Style Chamcresistkr Mote bulnenarlented In uccuPalion (Snllzer, 1964) Poorer academic performanee (Veldman A Bown, 1969; Pumtoy, 1967: Sai(wr, MaMahon, A Welsh, 1962) Mure urcn of alcohul (111geint, KklsberS, & Mctzner, 1967; Lillenfeld, 1959) ReHgious serWee altendatce kss frequenl (Canell A JGuR 1967; Slnlb A Seehrest, 1963) Pruporlionalrly higher ftequence or mmrleaes and job chanaee (LOknRld, 1959) Higher hxeldenca or pdor hospltalizatlona (Llllenfeld, 1959) Higher Ineldence of smokln6 among parents (Salber A: AbeBn, 1967) Maro active parlicipalion In sports (LBlenfcid,1959) Mare auln acefdents (Iann1A nack, 195g) Morc users of coffee and les (LBknfcld,1959) Mcrphoroefrat helrr Cmuer body welght (Seltxer, 1963) Greater cwight (Seltzcr, 1963; Beer, 1966) Thlnner (Hiyglns A Klelsberg. 1967) Telfer, mlatlve to eube root of weiaht (Damon, 1961) thlnner skin falds (dcepnnd subecepulsr) IHiagins R Kklsber6 1967) Dtuweraph/c ChanefeNstkr Mure men (U.S. Public Health Seviw; 1970) Proportionately morc 25-to 45-year-oMs (U.S. Public Health Servln, 1970) Lower mean saleeeonomlt cBsr (Salber d MacMshnn, 1961) Propornonately fewer caneae men (Higgins, el d., 1967; LWenfald, 1959) More urban resldenn (Hladns,et el., 1967)

96 W.L,OtINN,JR. lheee difhrcnam. lnw:sligatlon dtowed, fot example, that Ihe 66 fraternity men /nvolved In the Imestl8allan avcreaed anly 59,1% as against 69.9% for the non-fraternity men, whlle the nonfraternily men made up tha gmat bulk of Ihe nonsmokers.This raised the question s to whenmr Inlernity lire might not be rhe nuset or at bast setectl.e fnctor, mtlter Iban tobacco? It wes elsu found that lhe athletes or Ilw group averaged only 63.2% while Ihe nonalhktes averaged 68.376, and that the athletes were much mora apt to be both emnken and fraternity men. This raixd Ibe question u to how much or the dlfferenn round between the smaken and the nansmoken wu in reality due to engrossment In athletics on the put of the smokersi After tabuleling these complex deta in various ways, Maylen finally concluded nut wldk bed tcholarsidp Is distinctly associated with smoking, It Is also distinctly auode/ed with athlelim and rnlemilles and Ihot ir ts rmpassibM to tell how much, If any, of Ihe bad schoNrship aseel.ted with lobacco war rea7ly caused by It. The effect of smaking on scholarship Is thus lefl undetermined though Meytan deserved much credit for clearly recognhing the exlremely complex nature and uncertain meaning of sucil date. IHull, 1924, pp. 15-171. The study of Meyian (1910), cited by Hull (1924), was performed 62 years ago. The pitfalls are stiB real. Fodor, Class, and Weiner (1969) collected extensive psychological and physiological data from 200 smoker and 200 nonsmoker healthy young males, They reported higher triglycerlde levels (stalistically signl0cant) and faster blood-clotting time for smokers. A rccond study, Ihat of Rosenman, Fricdman, Jenkins, Slreus, Wurm, and Kos(tchek (1966), made similar measures over severel yeanon more than 3,000 males between 39 and 60 yean of age whom they grouped according to behavioral types (Type A, greatet sense of time urgency, etc.;Type B, more complacent). Type A's within the 39•49 year•old group hed higher lriglyceride levels and shorter bloodclotting time. There was a greater ineidence of smoking In the Type A group. But the observed differences In triglycedde levels and blood-clotting time were independenl uf smoking habits. Could, then, the Fodor•Class-Welner findings, interpreted by those authors to be "certain changes In blood chemistry, with detrimental long-range effects [p, 1251" among young lltokers, be attributable to a preponderance of Behavior Type A subjects among their smokers? Does tlre Behavior Type A syndrome cause smoking7 Does smoking cause Type A behavior? Does faster blood-clotling time cause smoking7 Or do all of these plrenomena-Type A behavlor, faster bloodtlotting time, elevated triglycedde level, and smoking-have some corrunon antecedent? The 1910 case cited by Hull (1924) and the casa involving the Iwo recent studies which I have cited are botil Illustrations of a very fundamental methodological problem. The Invesligator of smoking behavior who sets out to determine whether differences exist on some personality or behavioral dimension between smokers and nonsmokers is compelled to allow each of his subjects to self-eclect his group assignment by becoming, or not becoming, a smoker. This fact automatically Lrommils the Investigetlon to an ex post facto design, whether prospective or retrospective. All or the differentlating trells listed In Table I were Identified by means of ex post facto investlgalions. Kerlinger (1967) de(tnee ex post facto research es "that research In which the Independent variabte or variables have already occurred and In which the researcher starts with the observation of a dependent variebie or varlables. He then sludies the independent variables In retrospect for their possible relations to, and effects on, the dependent variable or vadables". (p.360j The deficiency In this design rests In the fact that the Invesligatar is unable to control for the effects of extraneous METHODS AND MODELS APPLIED TO MOTIVATION IN CIOARETTE SMOKING 97 variables. As a consequence, the researcher who, however reluctantly, accepts tire ex post facto design must also (but usually does not) accept the limitalions Imposed upon the Interpretatlon of hts results. And the limitations are simply lhese: Variations in the dependent variable y(smoking vs. nonsmoking) which are observed to be associated wilh variations In the Independenl variablex (personality traits, etc.) cannot he Interpreted to be attributable to the variations In x. Nor can Inferences be made as to the permutative relationships (x theny, y then x), unless it Is known that the one predates the other In the clrronologicil history of tlee subject. Even then, no other than a permulalive relationship can be Inferred. In Ihis context It is Interesling to note that the pre-1924 tendency among invesugaton of smoking was to assume a y then x relationship and attribute differences In x(xwdesnic performance) to differences In y(smoking vs. nonsmoking). Mosl post-1950 Investigators have tended conversely to assume an x then y relationship and attribute dffferences in y (smoking vs. nonsmoking) to differences In z(personality traits). In neither case are the assumptions about permutation nor the attributions of contingency jus110ed by tlre date. Apart from these methodological limitations, how fruitful hes the application of this experlmental design been? One index of fruilfulness would be the degree to which our power of predlcliug smoking, or even our power to retrospectively classify smokers vs. nonsmokers, ls enhanced, Smith (1967) writes: "Examinalion of current literature concerning personality differences of snrokers and nonsmokers indicates that accuracy of classi0cationa typically ranges from 50% (chance accuracy) to about 60%(.50 SD separd(ons) [p. 3101." Hitting 6 out of 10 rather than al the chance expectation of 5 out of 10 Is not great Improvement In our power of prediction. Smith's quoted statement referred to data in the literature treated lhrough unirariale analytic melhods. lie expanded his own data collection to Include peer ratings and applied nsore powerful mullivarlale analylic methods. Under these conditions hl=medlan percent correct classi0cation Index In a study of some 1,400 college and nursing students was 68% in classifying smokers vs. nonsmokers, When Smith (1969) later npplied Ihesa multivariate, peenraling procedures to a preadult group (N a 562), cross•valIda- tion analyses yielded 66% accuracy of classlfscation of snwkersvss nonsmokers. In hit group of junior high school females, fhe anatysis (without cran-validation) yielded a 79% accuracy of classification. Such findings are encouraging, suggesting lhat benenih tlse surface complexity there may well be variables awailing measurement with even greater discriminatory power. It Is also axiomatic that the smoker/nonsmoker design ahell continue to be a potential tource of leads as to the molivatlonal determinants In smoking. In Ilte final analysis, however, tlse hypotheses generated from such studies must be corroborated or rejected through controlled experimentation. Sanford (1967), Veldman and flown (1969), and Hunt and Malerazzo (1970) have called for 'Accuracy or chuinntion Is the percentage of "hits' In classifying subjects into caleaodes with anly predictor nriabk(s) known (in ILis asa, personality °seores'). ' Whera N- 100, e 60% classification accuracy Is sfgnificant at Ihe.651evel. And, of eourse, as N increaws, a correspondingly smaller devNtlon from SO%(chmceaccurary) h required foe tlgniflcena.

98 W.L.OUNN,Jn. the application of more sophisticated approaches to the question of maintenance motlvatlon In tmoking. The Direct interrogatlon Approach Same Invesligators have accepted the premise that one ean find out why a man tsnokes by asking him. Vnrious techniques have been used: the direct Interview (face•to-fece, telephone), and the Indlrect interview (completion of a questionnaire supplled by personal contact, by mall, or through the public media). Schedules for eliciting responses range from loosely structured formats with openended questions to mullipaged questionnaires with multiplechoice Items and rating scales. The Information so collected has been treated by techniques ranging from qualitative clinical interpretation to complex mulOvarlate statistical analysis. But whatever form the contact, or the schedule, or the Item, or the data analysis may take, the datum unit remains a deliberated-upon, voluntary, retrospectlve statement about himself by a respondent. The question is raised as to how much credence can be given to the Introspective reports of naive smokers7 Factors sueh as the need for social approval of opinions and actions, the need to justify a preference commitment, order of presentation effects, brand Imagery effects, halo effects, and the yea-saying tendency are collectively more determinative of a report of a smoke-lnduced sensory experience than Is the sensory experience Itself. If such is the equivocal nature of the neive respondent'e report of conscious sensory experience associated with smoking, how veridical can a respondent's report be of those far less verbalizable affective experiences associated with smoking? One of the more sophisticated examples of this appro,ah Is the study reported by Ikard, Green, and florn (1969), In which a 23-item questionnaire was adminlstered to 2,094 smokers. The items In the questionnaire were statements of subjective conditions or effeclive states associated with smoking-viz. "I smoke cigarettes to stimulate me, to perk myself up"; "When 1 reel blue, or want to take my mind off cares and worries, I smoke cigarettea"; "When I feel uisamed or embarrassed about something, I light up a cigarette:"The respondent Indicated how Ihe behavior In each statement was typical of his own by selecting from ansong five alternatives ranging from "Always" to "Never". The results were subjected to a multiple factor analysis which Identifsed six fectors underlying Ihe response patterns to the 23 Items. Tisese six factors were labeled "Habitual," "Addictive," "Reduction of Negative Affect," "Pleasurable Relaxation;' "Stimulation," and "Sensorimotor MmsipulnllonP The close parallel of these labels to the types of smokers proposed In the Tomkins (1966) model Is not surprising; the ilems of the questionnaire were drafted with the four Toinklns types In mind. It Is Inleresting to note that Tomkins had no sensorlmotor manipulation type In his model. Horn, apparently of the opinion that a sensodmotor manipulation dimension also existed, made an ad hoc addition of four sensorimotor manipulation items to the scale which did, of course, "uncover" such a factor. The llsree factors, "Stimulation," "Pleasurable Rclaxation;" and "Scnsorlmotor Manipulation," were then proposed to be Ibrce subclasd0cations of Tomkina'To increau posilive affect" type. Although we can ill afford not to collect Introspective reports from smoking respondents, there Is some Justllicatlon for the contention that the construction of METHOOSANOMOOELSAPPLIEOTOMOTIYATIONINCIOAAETTESMOKINO 99 theory solely upon the self-reports of naive respondents as to why they tmoke Is an overly optimistic enterprise. Not even with a computer nn one make a silk pursa from a sawi ear. An Alternative Approach In tlse smoker/nonsmoker dlfference approach we ask the question, "What's different about the smokerT" Not only does this force us Into an ex post facto design, but It Is also a question tangential to the problem of motivatlon. In the direct approach we esk the questlon, "Why does one continue to smokeT' 17ds is the right question, hut the Investigator Is dependent upon the verbalizatlons of the smnker, a dale transmission system known to carry a high level of background noise and extraneous signals which obscure and distort the essential message. We need a design which permits us both Io ask the right question and to obtain relevant data-I propose the controlled experiment design. This, of course, Is the design of cholce always, but In researching human behavior we often cannot use It. Society would not tolerate our taking a hundred youlht and randomly selecting fifty to become ,moken. Hence, tlie smoking habit must be studied In situ. But do we need to study the smoking habit? Perhaps the emphasis upon the recurrence of Ihe behavioral pattern ie misguided. If it is en undentanding of motivational mechanisnn that we seek, It would seem more pertinent were we to focus upon the discrete smoking act and the attendant phenomene. Investigators of smoking at the biochemicel and physiological levels have routinely used the controlled experiment. Typically, measurements are made on the chosen y variable (acetylchollne Icvel, BEG patterns), the independent variable x is Introduced (injection of nico(lne, Inhalatlon of whole emake), and y is again measured. Note that in the model physiological study, smoking (or Its equivalent) is x, not y es In the psychologist's ex post facto design. Herein lies a profound difference between the typical physiological and the typical psychologlcal study of smoking; the former has been predictingfrom smoking, the latter to emoking. Since the basic question in the study of the motivational aspects of smoking Is, "Why do people smoke?" It follows that we should be asking at the psychological level what hu been asked all along at the physlological level: "What are the immediate effects of mtoking7" Some significant reports of Ihis kind ofinvesllgatlon have already appeared in the Bterature: Morrison and Anniinge (1967) report reduction In free activity level in rats. Increases in lever-prening activity In rats and cets are reported by Armitage, Hall, and Morrison (1968); and Silverman (1969) concluded that the principal effect upon rela In his experiment was to reduce aggression. In Chapter I I of this volume, Hutchinson reports a reduction In aggressive responsivity In squirrel monkeys. Bovet, Bovet-Nitti, and Oliverlo (1966) describe facgitalion in learning effects arpong rats under their conditions. Inhibition ofthe extinctlon of en avoldant response was observed in rats by Ddscall and Battig(1970). Oattig (1970) inlerprets Ihe Iicbb-WOilams test performance of his rats as evidence off bene0clal effect of nicotine on highcr Integrative functions. Rssman. Steinberg, and Colod (1968) report evidence from a study of electrically shocked mice suggestive of a facilitation of memory consolldation. Rarlow, Oliverlo, Satta, and Tlsompson (1970) report

100 W.L.DUNN.JR. possible enhanced memory-consolidating processes. EEG activation In animals associated with nicotine administration has been reported by Armilage and Hell (1968), Domino (1967), Schaeppl (1968), and elhen. B. B. Brown (1968) and Ulett and Itil (1969) have observed EEC activation In humans as a function of smoke Inhalation. In other behavioral modalities among humans inhaling smoke: Warwlckand Eysenck (1963) report changes In Ihe critical flicker fusion Ilueshold; Clark and Rand (1968) and Domino and von Baumgarten (1969) report pateliar reflex reductions. The list is not inclusive, its purpose being merely to point up the Increasing Interest In and successful application of the pretreatment and posttreetment design to the study of the immediate behavioral and psychologlcal effects of smoking. In assessing the trend exemplified by the above studies, I am Impressed with the Investigators' apparent lack of concern over the motivational mievance of the reported changes induced by nicotine and smoke. Larson. Haag, and Silvette (1961) have pointed ralher obliquely to the problem In remarking that there are so many peripheral effects of nicotine that the wntral effecis tend to be obscured. Their frame of reference was clearly neurological rather than motivallonal, as was Domino's (1967) when he made a similar observation: "This agent [nicotine] has so many pharmacological actions that they easily confound, reinforce, and obscure Its direct central actlons.... nicotine acts at a variety of sltes In the peripheral nervous system. These multiple actions must be considered In sny study of nicotine's effects upon the central nervous system. These writers are unquestion- ably correct In despairing over the confounding plurality of effects Induced by nicotine; but I submit that the task is not so much one of separating out central from peripheral effects, but rathcr one of differentiating the motlvatlonally relevant from Ilse molivationally, irrelevant. On Synergism lnvestigators of Ihe bchavioral effects of nicotine and smoke are often stmck by the markedly wide Intraindividual and intedndlv,dual variability In response patterns. Amlilage and Hall (1968) heve observed wide variation suggestive of dichotomous response patterns in the behavioral effects, In the electroencephalographie effeets, and In acetylclmllne release from the padetal cortex of cats. Silverman (1969), In studying the social behavior of rats, observes, "There seem to be Individual differences between rats to which the effects of nicotine are additive [p. 5061:' B. B. Brown (1967) mreported nicotine effects on the behavior and on the EEG patterns IdentlEable with specific types of cats classlEed gccording to behavior. Bovet (1965) reported the necessity of using split litter techniques in order to reduce variability. Bovet, Bovet-Nilti, and Oliverlu (1967) alm reported opposite effects of nicotine associated with environmental faptors and the strains of the animal genetically classified as "bright" and "dull." Domino (1967) reported differences In the effect of nicotine upon avoidance responses between rats classified es slow and fast In lhelr Jump response times. In that same report Domino noted that the observed BEGactivating aetions of nicotine, or its behavioral wakeup effect In cats with chronically implanted brain electrodes appeared only when the nicotlne was administered during a state of mild central nervous system METHODS AND MODELS APPLIED TO MOTIVATION IN CIOARETTE SMDKINO 101 depression. Pradhan and Bowling (1970) report differential effects from inJected nicotine on self-ttimulallon response patterns between high• and low•response-rate relt. Wide variab011y In mcasurement lras also Iong been reported among invesliguors using human subjects In smoking studles. McArthur, Wddron, and Dickinson (1958) attempted to extract mean(ng from the great mass of highly variable data obtained on smoking among Harvard students: "In summary tlten, we may hypotheslae Ihat slarling to smoke Is largely brought about by one's soclal environment but that reactions to smoking, once it has started, seem to depend In good part on Ihe personal needs thnt the newly established habit is able to gratify [p. 2721:" Seltzer (1962) paraphrased these observations In blogenelic terms: "Rather than a superficial habit overlaid Indiscriminately upon various persons, smoking appears to be a response to a wide variety of pcrmnality and behavioral characteristics which have their origin, In part, In the biologic and genetic makeup of the Individual [p.43[:' In view of all this variability, It is likely Ihat we shall ultimately Bnd that the critical mechanisms involved In smoking require lhe synergistic presence of some other factor, such as anxiety or possibly some constitutionally determined reactivity. The observation that only about iulf of (hon who try smuking cigarettes go on to take up the habit certainly suggests some kind of interaction process. Smoking apparently is not sufficiently pleasurable among the disinterested half of the tryers for the aversiveness of smoking to be overridden. Smoke is smoke- whether its Inhalation is pleasurable Is dependent upon what the Individual brings to the situation. In a sense, this statement brings us full circle, for In proposing a synergism model, I am in fact conceding that smokers are probably different from nonsmokers. Does !t not then follow that we should continue to look for smoker/nonsmoker dlfferencesl Perhaps the apparent conflict in the position can be clarified schematically. The Brst-row, entry In Table 2 is the simplistic smoker/nommoker difference paradigm. In actual practice the sequence Isreversed from that shown, which Is the aforementioned liability of the ex post facto method; one "measures"y by asking,"Do you smoke?" and then measures x (e.g., TABLE 2 A Schemmic Representation of Three Research Designs Type Model Independent rcrlrble Dependent Variable Typlcd Psychological Typted Ttdl orsrate of arpnirm Smakelnhdallnn t- -~ 8moker/nnnsmoker on er Efrc l n anl.m Phyelolopetl (Nicotine Inkcllon) Trait or slale p c g Proposed ~-~ Dlrferendd effect Syrxrdstic Smoke Inhtlaeon •~ upen nrgrnhm

102 W.L,DUNN,Jn, penonallty test acores). In the proposed synergistic experlmenlal model, one manipulates or systematlally selects for levels of xt (a psychological or physiological trait or state), Introduces xs (smoke inhalation), and observes for a differential response level y. The significant relationship in such a study would not be between smoking and the measured effect, nor between the trait or state and the measured effecq but between the lraitJsmuking Interactlon and the measured effect. ON MODELS Having touched upon some of Ihe methodological Issues and pltfalh confronting the Invesligator of smaker motivation, I would like to turn now to a nosological exerdse. The first part is an an analysis of the temporal phases of the smoking habiL The second part categorizes the motlves which have been proposed by various investigators. Finally, I will perform a cross tabulation of phase and motive. Phates of the Habit Jarvlk (1970) has proposed that the smnking habit be viewed as consisting of three phases: the initiation phase, Including the smoking of those first cigarettes whereby the novice learns of the effects of smuking upon his person; the maintenance phase, wherein the behavior hasbecome repelltive and pattemed;and the terminatlon phase, wherein other motive forces counter to those of the maintenance phase become dominant and determine behavior. I have found it useful to insert a phase between Jarvik's Inltlalion and maintenance phases. This I will call the transitlonsl phase. In this manner we can clearly distinguish between two classes of events-those lending up to trial, and those associated with trial. Preceding trial, awareness Is largely exlroceptlve; the indlvidual's attention Is turned upon the Interpeeonal slluatjon, either real or imeginary. Awareness at the time of lrial is largely Introceptive as the smoker appraises what if happening to him as a consequence of [lre novel experience. t am suggesting that we simply limit JarNk's initiatlon phase to pretrial events, where clearly the motivational factors are psychosocial In nature, and that we subsume events assoclated with smoking under the transittonal phase. It may well be that a straightforward algebraic summing of the positive and negative experiences occurring In the suggested transltional phase will separate those who will become smokers from those who wlll remain nonsmokers. Nearly everyone hes at one time experleneed the shock of the first Inhalatlon, often accompanied by gagging and nauseousness. On the other hand, nearly evaryone who has continued smoking an reali the pleasantness af the BpJrt- headedness which followed inhelatlon, a sensory experlence that could be made to recur following a bdef time lapse by smoking another cigarette. Whether or not the total experience Is judged by the smoker as positive or negative might In time prove to be associated with biochemical, endoerinologial, or neurological variables which are either constitutlonally Bxed (Fisher, 1959) or shaped by experience prior to Intlial experimentation with tobacco smoke. There is another justification for the transitional phase. The Initial subjective effect, which presumably has a physlologial bnis, becomes progressively less I METHODS ANO MODELS APPLIED TO MOTI VATION IN CIQANEtTE SMOKING 103 discernible with continued smoking until most habituated smokers are no longer aware of It. Thae habituated smokers who do continue to experience it do so only on accasion and onty after pedods of extended abstinence. Sina the effect Is not likely to be sought aRer by the beginning smoker who has had no occesion to experience It prior to smoking, and since it Is no longer aupralimind on the regular smoking regimen for the habituated emoker, the effect and any others that may be peculiar to early trifl =hould be treated Independently of both initiulon and maintenance events. The possibility remains, however, that we are dealing with a physiotoglal response at its maxinmm In early trials, diminishing In intenslty over time, but continuing on at some subthreshold awareness level of sufBcient Intensity to alter the affective state of the smoker. The Motiva Classes So much for the temporal phases of smoking. Consider now the ategodes of motives imputed to the smoker. The rationale of the categories resides in the locus of tlre events which heve been proposed as having reward value. There are four rather obvious classes. Physiolepleal. A number of physiological changes occur as transient reactions to the Inhalation of smoke. It Is not within the scope of this paper to consider the metabolic and biochemlcal changes which have been reported, nor the myriad of reactions at all physiological levels In antmals and In enlmal tissue upon administration of small to massive duses of ntrotine. Table 3 lists severs] systemic changes which have been observed In humans upon smoke inhalation. They are transient, short-lived changes. The fact that a change does accur upon smoking In sufltrlent to Inctude It in the llst, but as I have stated earlier, the mere occurrence of a given change Is not prima fade evidence of Its motivational significance. Any physialogial or pharmacological reward explanation for smoking must Identlfy the motivationally significant change, either from among those In Table 3 or as identlBed insubsequentresearch. For a critlal assessment of the subject, see Jarvik (1970); and for a comprehensive review of the litetature on thin aspect of smoking, see Larson, Heag, and Silvette (1961) and Iarson and Silvette (1968). For our purposes it 1s best to avoid the many controversies attendant upon this area, e.g., dosage levels, animal vs. human, nicotine vs. whole-smoke, direct vs. Indirect actlon.lVe shall simply press TABLE 3 TransientPhyslologlcalResponsesto Smoke Inhalallan 1. Elevated heut nle 2 pJeveted eomnary flow 3. kle.ated blood sugar level 1. Lowered cutaneous lemperature In the extmnines 5. Increased blood now In skdet.l musculature 6. A«actiee releue of adrtndln 7, Alterations In etaetdeal potential patterns of tha btain g. Inhiblaon of Patdlv re0ex