Psychopharmacology (1991) 104:485~490 0033315891001506
Psychopharmacology © Springer-Verlag 1991
Electroencephalographic effects of cigarette smoking* Walter S. Pritchard
Biobehavioral Division, Biochemical/BiobehavioralR&D, Bowman Gray Technical Center, 611-12, R.J. Reynolds Tobacco Company, Winston-Salem, NC 27102, USA Received October 16, 1990 / Final version November 9, 1990
The effects of cigarette smoking on the electroencephalogram (EEG) of smokers were examined in a study involving both task and no-task conditions. Nonsmoking subjects were employed as controls. In light inhaling smokers, (depth of inhalation inferred from preto post-smoking changes in tidal breath carbon monoxide), smoking was found to attenuate EEG activity in the delta, theta, and alpha frequency bands, as well as facilitate behavioral performance. For theta, the attenuation was lateralized over the right frontal cerebral hemisphere. In deep inhaling smokers, smoking produced a symmetrical central midline increase in beta2 magnitude, an EEG effect that in the benzodiazepine literature is associated with anxiety relief. Abstract.
K e y w o r d s : S m o k i n g - N i c o t i n e - E E G - Carbon monoxide - Cerebral lateralization - Performance
An enduring question regarding human cigarette smoking is the basis of the so-called "nicotine paradox". Although the peripheral effects of smoking appear to be stimulatory (e.g., increased heart rate, especially for the initial cigarette of the day; Benowitz 1987) and many smokers say that smoking increases their mental alertness, other smokers report that smoking helps them to function in the face of environmental stress by having a calming effect on their mood (Gilbert 1979). The majority of electroencephalographic (EEG) studies of smoking have not been helpful in delineating the basis of this "paradoxical" calming effect. Subjects in these studies have been tested under conditions atypical of normal smoking situations and not designed to provoke mental effort and/or emotional arousal (e.g., sitting/reclining with eyes closed or staring at a fixation point). Further, * A preliminary version of these data was presented during the symposium "Psychophysiologyof Nicotine", held at the Thirtieth Annual Convention of the Society for Psychophysiological Research, Boston, October, 1990
these studies have varied widely along a multitude of dimensions such as experimental design [whether or not non-smoker controls were tested; whether a "sham smoking" (subject puffs on unlit cigarette) control was employed], the nature of the EEG recordings (montages; monopolar verses bipolar derivations), type of cigarettes smoked, smoking schedule (ad lib verses controlled), degree of prior smoking deprivation, etc. Despite these differences, the results of "no task" EEG/smoking studies have been fairly consistent in indicating that smoking suppresses activity in the lower frequency delta (1-4 Hz), theta (4.5-7.5 Hz), and alpha (8-12 Hz) EEG bands, tends to raise the dominant alpha frequency (frequency within the alpha band at which EEG activity is maximal), and increases activity in the higher frequency betal (12.5-17.5 Hz) and beta2 (18-30 Hz) bands (Ulett and Itil 1960; Philips 1971; Knott and Venables 1977; Griffith and Crossman 1983; Herning et al. 1983; Cinciripini 1986; Golding 1988; Knott 1988). This pattern of change (a switch from lower to higher frequency activity) is indicative of cortical activation, and is not surprising given such monotonous, unengaging testing conditions. Similar results have recently been reported for subjects performing an easy vigilance task (Gilbert 1987, 1988). Two studies have been reported in which the EEG effects of smoking were measured while subjects were exposed to unpleasant stimuli. Golding and Mangan (1982) tested non-deprived smokers under conditions of both mild "sensory isolation" (resting in bed with open eyes) and "stress" (periodic exposure to bursts of 106 dB white noise). Recordings were made from a bipolar montage consisting of a midline frontal (Fz) placement and a lateralized occipital placement (01 for right handers and 02 for left handers). As expected, in the sensory isolation condition smoking significantly suppressed alpha activity relative to a pre-smoking baseline period. In the stress condition, however, smoking increased alpha activity. In the second study, Gilbert et al. (1989) had subjects watch a brief safety film containing three graphically portrayed scenes of serious industrial accidents.
486
EEG was recorded from left and right parietal sites. The results following smoking (high nicotine cigarette condition) indicated that, relative to control segments of the movie, viewing of the accident scenes was associated with increased alpha activity over the right hemisphere and decreased alpha activity over the left hemisphere. The results of these two studies indicate that part of the basis of the nicotine paradox may be a deactivation of the cortex produced by smoking when a smoker is being aversively stimulated. The findings of Gilbert et al. further indicate such effects may be lateralized to the right hemisphere, which in most individuals is neurologically specialized for the processing of emotion (see Gilbert et al. for discussion and references). However, the ecological validity of these studies was still rather low in that subjects had no task and the stimuli employed were not representative of everyday life. The present study sought to examine EEG effects of smoking within a more realistic context of relatively much less aversive "cognitive stress". This was done by giving subjects tasks to perform that produced levels of mental workload representative of those encountered in day-to-day living.
Procedure. Four different testing conditions were employed. These conditions were: (1) eyes closed, no task; (2) eyes open, no task; (3) eyes open, count forward in a soft voice as rapidly as possible by 3s from a starting number varying randomly between 0 and 3; (4) same as 3, except subjects counted backwards by Ys from a starting number varying randomly between 450 and 550. Each condition lasted 30 s. The four different conditions grouped together comprised one "block". Each subject experienced four blocks. Within each of these blocks (of the four different conditions), the order of the conditions was randomly counterbalanced across blocks and subjects to the extent possible. During the two counting conditions, subjects held a small microphone close to their lips and were tape recorded. To help motivate subjects, it was emphasized to each subject that her/his performance would be compared to that of other subjects. In the eyes-open conditions, subjects were instructed to stare at a fixation point on the wall and not blink. Two leading commercial brands of filtered, 85 mm non-menthol cigarettes having FTC yields of 1.1 mg nicotine and 16.8 mg "tar" were smoked by the subjects who were smokers between blocks 2 and 3. Half the smoking subjects were regular smokers of one of the brands, and half were regular smokers of the other brand, and each subject smoked her/his own brand. Smoking was ad lib and usually took around 5 min; otherwise, all inter-block rest periods were 3 min. A sample of tidal breath was taken at the end of each rest period for CO analysis. Data analysis. For each 30-s data epoch, as many artifact-free 2-s
Materials and methods Subjects. Thirty-two subjects participated in exchange for monetary compensation. Half the subjects were non-smokers (who had never been smokers) and half were regular smokers (eight females and eight males each group). The smoking subjects regularly consumed > 1 pack/day of cigarettes yielding 1.1 mg nicotine and 16.8 mg "tar" by the US Federal Trade Commission (FTC) method (Federal Register 1967). Subjects were tested in the morning (for the smokers, following overnight smoking deprivation). Upon arrival, subjects gave informed consent following an explanation of the nature and possible consequences of the study. They then gave a tidal breath sample that was analyzed for the level of carbon monoxide (CO) using an Ecolyzer Model 2000. Only persons with readings less than 16 parts per million (ppm) were allowed to participate. Based on the literature, it was estimated that around 95% of persons having readings in excess of 16 would have failed to comply with the instructions not to smoke after 11 P.M. the previous evening. It is possible, however, that non-complying subjects could have had readings lower than 16, and this fact should be borne in mind in evaluating the results of the present investigation. In actuality, three smokers had readings in excess of the cut-off (19, 20, and 24 ppm) and were dismissed from the study and replaced with other subjects. Data from one (male) smoker and one (female) non-smoker were later dropped due to technical reasons related to the EEG (see below); the final sample of 15 smokers had an average initial CO reading of 8.33 [range 5-15, Standard Error of the Mean (SEM) = 0.68]. The final sample of 15 smokers had an average age of 27.40 [range 18-37, SEM = 1.54], while the final sample of 15 non-smokers had an average age of 27.80 [range 19-34, S E M = 1.28], a difference that was not significant [t(28)=0.21]. Smokers completed Spielberger's Smoking Motivation Questionnaire (SMQ; Spielberger 1986). EEG recording. Tin EEG recording electrodes were attached to the 19 standard loci of the international 10-20 system. A reference electrode was placed on the nose and a ground electrode on the forehead. High- and low-pass filter settings of 0.3 and 30 Hz (12 dB/octave rolloff) were employed with a sampling rate of 128 Hz. All electrode impedances were < 3 kohm. Subjects were tested in a soundproof, electrically shielded room, while seated in a comfortable chair.
EEG segments as possible were submitted to a Fast Fourier Transform (FFT) in conjunction with a Hanning window, yielding a magnitude spectrum in microvolts from 1 to 28 Hz. Artifacts screened for by visual inspection included head movements (which showed up as large deflections in all channels), eye blinks (which showed up as large spikes in channels Fpl and Fp2), horizontal eye movements (which showed up as prominent deflections in channels F7 and F8; cf Torello 1989, Fig. 2, middle left panel), and electromyographic (EMG) activity (which showed up as high frequency spiking primarily in channels T3 and T4; cf Torello 1989, Fig. 2, middle right panel). Artifact rejection decisions were made blindly with respect to subject and condition. As alluded to earlier, the data of one smoker and one non-smoker were judged to have excessive artifacts and were not submitted to the FFT. For the non-counting conditions, average magnitude in the delta, theta, alpha, beta1, and beta2 bands were computed. The beta2 band was defined as 18-28 Hz; the other bands were defined as in the introduction. For most subjects, the EEG in the counting conditions had low frequency vocalization artifacts extending partway into the theta band. For the counting conditions, therefore, average magnitudes were computed for alpha, beta1, and beta2 only. Dominant alpha frequency was computed for all conditions. For the counting data, number of counts/30 s and number of counting errors/30 s were computed.
Segregation of smoking subjects into light- and deep-inhaling groups. Smoking subjects were divided into two groups based on the size of the increase in tidal breath CO following smoking relative to the pre-smoking level ("DCO" for "delta CO"). Previous non-EEG smoking studies had found significant differences along this dimension (Nil et al. 1986, 1987; Michel et al. 1987; Zacny et al. 1987; Nil and Battig 1989); thus, the decision to segregate subjects on the basis of DCO was not post hoc. The dividing point of 10 ppm was determined post hoc on the basis of inspection of the distribution of DCO values across the 15 smoking subjects. This distribution appeared bimodal, with all subjects having DCO values that were either greater than or less than 10. This split yielded a group fo six "low D C O " smokers and nine "high D C O " smokers. The low and high DCO groups did not differ in terms of initial CO level: 8.17 ppm for the low DCO group versus 8.56 ppm for the high DCO group [t(13)= 0.19]. They also did not differ significantly in terms of age: 25 years for the low DCO group versus 29 years for the high DCO group It(13) = 1.31], or in terms of sex: two females/four males
487 for the low DCO group versus six females/three males for the high DCO group [chi square(l)= 1.61]. Statistics. Data were analyzed by means of analysis of variance
(ANOVA). For the EEG data, the ANOVAs were 19 (EEG channels or "loci") x 4 (blocks) x 4 (conditions; for the delta and theta bands, 2 conditions) x 3 (groups; non-smokers, low and high DCO smokers). For the counting data, the ANOVA was 4 (bocks) x 2 (easy/difficult task)x 3 (groups). All P-values reported, including those used in post hoc testing, are Greenhouse-Geisser (G-G) corrected.
Results EEG
As expected, non-smokers showed no significant changes in any EEG measure as a function of blocks (or blocks interacting with loci). For low DCO smokers, delta, theta, and alpha magnitude were all attenuated globally across electrode loci by smoking, with the high DCO smokers being unaffected: for the delta block x group interaction, F(6,81)=6.23, P
Psychopharmacology © Springer-Verlag 1991
Electroencephalographic effects of cigarette smoking* Walter S. Pritchard
Biobehavioral Division, Biochemical/BiobehavioralR&D, Bowman Gray Technical Center, 611-12, R.J. Reynolds Tobacco Company, Winston-Salem, NC 27102, USA Received October 16, 1990 / Final version November 9, 1990
The effects of cigarette smoking on the electroencephalogram (EEG) of smokers were examined in a study involving both task and no-task conditions. Nonsmoking subjects were employed as controls. In light inhaling smokers, (depth of inhalation inferred from preto post-smoking changes in tidal breath carbon monoxide), smoking was found to attenuate EEG activity in the delta, theta, and alpha frequency bands, as well as facilitate behavioral performance. For theta, the attenuation was lateralized over the right frontal cerebral hemisphere. In deep inhaling smokers, smoking produced a symmetrical central midline increase in beta2 magnitude, an EEG effect that in the benzodiazepine literature is associated with anxiety relief. Abstract.
K e y w o r d s : S m o k i n g - N i c o t i n e - E E G - Carbon monoxide - Cerebral lateralization - Performance
An enduring question regarding human cigarette smoking is the basis of the so-called "nicotine paradox". Although the peripheral effects of smoking appear to be stimulatory (e.g., increased heart rate, especially for the initial cigarette of the day; Benowitz 1987) and many smokers say that smoking increases their mental alertness, other smokers report that smoking helps them to function in the face of environmental stress by having a calming effect on their mood (Gilbert 1979). The majority of electroencephalographic (EEG) studies of smoking have not been helpful in delineating the basis of this "paradoxical" calming effect. Subjects in these studies have been tested under conditions atypical of normal smoking situations and not designed to provoke mental effort and/or emotional arousal (e.g., sitting/reclining with eyes closed or staring at a fixation point). Further, * A preliminary version of these data was presented during the symposium "Psychophysiologyof Nicotine", held at the Thirtieth Annual Convention of the Society for Psychophysiological Research, Boston, October, 1990
these studies have varied widely along a multitude of dimensions such as experimental design [whether or not non-smoker controls were tested; whether a "sham smoking" (subject puffs on unlit cigarette) control was employed], the nature of the EEG recordings (montages; monopolar verses bipolar derivations), type of cigarettes smoked, smoking schedule (ad lib verses controlled), degree of prior smoking deprivation, etc. Despite these differences, the results of "no task" EEG/smoking studies have been fairly consistent in indicating that smoking suppresses activity in the lower frequency delta (1-4 Hz), theta (4.5-7.5 Hz), and alpha (8-12 Hz) EEG bands, tends to raise the dominant alpha frequency (frequency within the alpha band at which EEG activity is maximal), and increases activity in the higher frequency betal (12.5-17.5 Hz) and beta2 (18-30 Hz) bands (Ulett and Itil 1960; Philips 1971; Knott and Venables 1977; Griffith and Crossman 1983; Herning et al. 1983; Cinciripini 1986; Golding 1988; Knott 1988). This pattern of change (a switch from lower to higher frequency activity) is indicative of cortical activation, and is not surprising given such monotonous, unengaging testing conditions. Similar results have recently been reported for subjects performing an easy vigilance task (Gilbert 1987, 1988). Two studies have been reported in which the EEG effects of smoking were measured while subjects were exposed to unpleasant stimuli. Golding and Mangan (1982) tested non-deprived smokers under conditions of both mild "sensory isolation" (resting in bed with open eyes) and "stress" (periodic exposure to bursts of 106 dB white noise). Recordings were made from a bipolar montage consisting of a midline frontal (Fz) placement and a lateralized occipital placement (01 for right handers and 02 for left handers). As expected, in the sensory isolation condition smoking significantly suppressed alpha activity relative to a pre-smoking baseline period. In the stress condition, however, smoking increased alpha activity. In the second study, Gilbert et al. (1989) had subjects watch a brief safety film containing three graphically portrayed scenes of serious industrial accidents.
486
EEG was recorded from left and right parietal sites. The results following smoking (high nicotine cigarette condition) indicated that, relative to control segments of the movie, viewing of the accident scenes was associated with increased alpha activity over the right hemisphere and decreased alpha activity over the left hemisphere. The results of these two studies indicate that part of the basis of the nicotine paradox may be a deactivation of the cortex produced by smoking when a smoker is being aversively stimulated. The findings of Gilbert et al. further indicate such effects may be lateralized to the right hemisphere, which in most individuals is neurologically specialized for the processing of emotion (see Gilbert et al. for discussion and references). However, the ecological validity of these studies was still rather low in that subjects had no task and the stimuli employed were not representative of everyday life. The present study sought to examine EEG effects of smoking within a more realistic context of relatively much less aversive "cognitive stress". This was done by giving subjects tasks to perform that produced levels of mental workload representative of those encountered in day-to-day living.
Procedure. Four different testing conditions were employed. These conditions were: (1) eyes closed, no task; (2) eyes open, no task; (3) eyes open, count forward in a soft voice as rapidly as possible by 3s from a starting number varying randomly between 0 and 3; (4) same as 3, except subjects counted backwards by Ys from a starting number varying randomly between 450 and 550. Each condition lasted 30 s. The four different conditions grouped together comprised one "block". Each subject experienced four blocks. Within each of these blocks (of the four different conditions), the order of the conditions was randomly counterbalanced across blocks and subjects to the extent possible. During the two counting conditions, subjects held a small microphone close to their lips and were tape recorded. To help motivate subjects, it was emphasized to each subject that her/his performance would be compared to that of other subjects. In the eyes-open conditions, subjects were instructed to stare at a fixation point on the wall and not blink. Two leading commercial brands of filtered, 85 mm non-menthol cigarettes having FTC yields of 1.1 mg nicotine and 16.8 mg "tar" were smoked by the subjects who were smokers between blocks 2 and 3. Half the smoking subjects were regular smokers of one of the brands, and half were regular smokers of the other brand, and each subject smoked her/his own brand. Smoking was ad lib and usually took around 5 min; otherwise, all inter-block rest periods were 3 min. A sample of tidal breath was taken at the end of each rest period for CO analysis. Data analysis. For each 30-s data epoch, as many artifact-free 2-s
Materials and methods Subjects. Thirty-two subjects participated in exchange for monetary compensation. Half the subjects were non-smokers (who had never been smokers) and half were regular smokers (eight females and eight males each group). The smoking subjects regularly consumed > 1 pack/day of cigarettes yielding 1.1 mg nicotine and 16.8 mg "tar" by the US Federal Trade Commission (FTC) method (Federal Register 1967). Subjects were tested in the morning (for the smokers, following overnight smoking deprivation). Upon arrival, subjects gave informed consent following an explanation of the nature and possible consequences of the study. They then gave a tidal breath sample that was analyzed for the level of carbon monoxide (CO) using an Ecolyzer Model 2000. Only persons with readings less than 16 parts per million (ppm) were allowed to participate. Based on the literature, it was estimated that around 95% of persons having readings in excess of 16 would have failed to comply with the instructions not to smoke after 11 P.M. the previous evening. It is possible, however, that non-complying subjects could have had readings lower than 16, and this fact should be borne in mind in evaluating the results of the present investigation. In actuality, three smokers had readings in excess of the cut-off (19, 20, and 24 ppm) and were dismissed from the study and replaced with other subjects. Data from one (male) smoker and one (female) non-smoker were later dropped due to technical reasons related to the EEG (see below); the final sample of 15 smokers had an average initial CO reading of 8.33 [range 5-15, Standard Error of the Mean (SEM) = 0.68]. The final sample of 15 smokers had an average age of 27.40 [range 18-37, SEM = 1.54], while the final sample of 15 non-smokers had an average age of 27.80 [range 19-34, S E M = 1.28], a difference that was not significant [t(28)=0.21]. Smokers completed Spielberger's Smoking Motivation Questionnaire (SMQ; Spielberger 1986). EEG recording. Tin EEG recording electrodes were attached to the 19 standard loci of the international 10-20 system. A reference electrode was placed on the nose and a ground electrode on the forehead. High- and low-pass filter settings of 0.3 and 30 Hz (12 dB/octave rolloff) were employed with a sampling rate of 128 Hz. All electrode impedances were < 3 kohm. Subjects were tested in a soundproof, electrically shielded room, while seated in a comfortable chair.
EEG segments as possible were submitted to a Fast Fourier Transform (FFT) in conjunction with a Hanning window, yielding a magnitude spectrum in microvolts from 1 to 28 Hz. Artifacts screened for by visual inspection included head movements (which showed up as large deflections in all channels), eye blinks (which showed up as large spikes in channels Fpl and Fp2), horizontal eye movements (which showed up as prominent deflections in channels F7 and F8; cf Torello 1989, Fig. 2, middle left panel), and electromyographic (EMG) activity (which showed up as high frequency spiking primarily in channels T3 and T4; cf Torello 1989, Fig. 2, middle right panel). Artifact rejection decisions were made blindly with respect to subject and condition. As alluded to earlier, the data of one smoker and one non-smoker were judged to have excessive artifacts and were not submitted to the FFT. For the non-counting conditions, average magnitude in the delta, theta, alpha, beta1, and beta2 bands were computed. The beta2 band was defined as 18-28 Hz; the other bands were defined as in the introduction. For most subjects, the EEG in the counting conditions had low frequency vocalization artifacts extending partway into the theta band. For the counting conditions, therefore, average magnitudes were computed for alpha, beta1, and beta2 only. Dominant alpha frequency was computed for all conditions. For the counting data, number of counts/30 s and number of counting errors/30 s were computed.
Segregation of smoking subjects into light- and deep-inhaling groups. Smoking subjects were divided into two groups based on the size of the increase in tidal breath CO following smoking relative to the pre-smoking level ("DCO" for "delta CO"). Previous non-EEG smoking studies had found significant differences along this dimension (Nil et al. 1986, 1987; Michel et al. 1987; Zacny et al. 1987; Nil and Battig 1989); thus, the decision to segregate subjects on the basis of DCO was not post hoc. The dividing point of 10 ppm was determined post hoc on the basis of inspection of the distribution of DCO values across the 15 smoking subjects. This distribution appeared bimodal, with all subjects having DCO values that were either greater than or less than 10. This split yielded a group fo six "low D C O " smokers and nine "high D C O " smokers. The low and high DCO groups did not differ in terms of initial CO level: 8.17 ppm for the low DCO group versus 8.56 ppm for the high DCO group [t(13)= 0.19]. They also did not differ significantly in terms of age: 25 years for the low DCO group versus 29 years for the high DCO group It(13) = 1.31], or in terms of sex: two females/four males
487 for the low DCO group versus six females/three males for the high DCO group [chi square(l)= 1.61]. Statistics. Data were analyzed by means of analysis of variance
(ANOVA). For the EEG data, the ANOVAs were 19 (EEG channels or "loci") x 4 (blocks) x 4 (conditions; for the delta and theta bands, 2 conditions) x 3 (groups; non-smokers, low and high DCO smokers). For the counting data, the ANOVA was 4 (bocks) x 2 (easy/difficult task)x 3 (groups). All P-values reported, including those used in post hoc testing, are Greenhouse-Geisser (G-G) corrected.
Results EEG
As expected, non-smokers showed no significant changes in any EEG measure as a function of blocks (or blocks interacting with loci). For low DCO smokers, delta, theta, and alpha magnitude were all attenuated globally across electrode loci by smoking, with the high DCO smokers being unaffected: for the delta block x group interaction, F(6,81)=6.23, P
