Marijuana’s Acute Effects on Cognitive Bias for Affective and Marijuana Cues

Abstract

Marijuana produces acute increases in positive subjective effects and decreased reactivity to negative affective stimuli, though may also acutely induce anxiety. Implicit attentional and evaluative processes may explicate marijuana’s ability to acutely increase positive and negative emotions. This within-subjects study examined whether smoked marijuana with 2.7–3.0 % delta-9-tetrahydrocannabinol (THC), relative to placebo, acutely changed attentional processing of rewarding and negative affective stimuli as well as marijuana-specific stimuli. On two separate days, regular marijuana users (N=89) smoked placebo or active THC cigarette and completed subjective ratings of mood, intoxication, urge to smoke marijuana, and two experimental tasks: Pleasantness Rating (response latency and perceived pleasantness of affective and marijuana-related stimuli) and Emotional Stroop (attentional bias to affective stimuli). On the Pleasantness Rating task, active marijuana increased response latency to negatively-valenced and marijuana-related (vs. neutral) visual stimuli, beyond a general slowing of response. Active marijuana also increased pleasantness ratings of marijuana images, although to a lesser extent than placebo due to reduced marijuana urge after smoking. Overall, active marijuana did not acutely change processing of positive emotional stimuli. There was no evidence of attentional bias to affective word stimuli on the Emotional Stroop task with the exception of attentional bias to positive word stimuli in the subgroup of marijuana users with cannabis dependence. Marijuana may increase allocation of attentional resources towards marijuana-specific and negatively-valenced visual stimuli without altering processing of positively-valenced stimuli. Marijuana-specific cues may be more attractive with higher levels of marijuana craving and less wanted with low craving levels.

Introduction

The primary psychoactive ingredient in cannabis, delta-9-tetrahydrocannabinol (THC), has been shown to both acutely increase positive subjective effects and decrease amygdala reactivity to negative affective stimuli in humans (Phan et al., 2008; Somaini et al., 2012). However, THC can also acutely increase negative feelings of anxiety and panic (D’Souza et al., 2008; McDonald et al., 2003), in part due to acute increases in physiologic arousal (Wachtel et al., 2002). Salient drug expectancies and experience with marijuana contribute to its acute impact on positive and negative affect post marijuana-smoking (Metrik et al., 2011). Nonetheless, putative mechanisms whereby marijuana impacts emotional processing are generally not well understood.

Marijuana may acutely modulate cognitive processing of motivationally salient stimuli including both drug-related cues and non-drug reinforcers. Positively-valenced stimuli eliciting approach reaction as well as negatively-valenced stimuli signaling initiation of avoidant behavior (Powell et al., 2002; Salamone, 1994) may mobilize additional cognitive resources to process such motivationally salient cues. The dual-process model of addictive behavior postulates that controlled cognitive processes related to conscious deliberation and more automatic implicit cognitive processes that may be outside of one’s awareness are involved in appraisal of stimuli in terms of their motivational and emotional significance (Wiers & Stacy, 2006). Consideration of these cognitive processes may further explicate findings suggesting that marijuana acutely increases both positive and negative emotions. Attentional bias for motivationally salient cues is one such implicit cognitive process and may be a key behavioral marker for drug misuse (Cousijn et al., 2013; Stacy & Wiers, 2010) as it has been related to escalation of drug-related problems and relapse (Carpenter et al., 2006; Waters et al., 2003; Waters et al., 2012). This placebo-controlled study examined whether marijuana with (2.7–3.0% THC) acutely changed attentional and evaluative bias to positive and negative affective stimuli as well as to marijuana-specific stimuli.

The Incentive Sensitization theory of addiction (Robinson & Berridge, 2001) postulates that, through the “incentive sensitization” process, individuals learn to associate drugs like marijuana and marijuana-related stimuli with pleasure, which in turn increases their incentive motivational significance. Stimuli with high incentive salience become attractive and “wanted,” eliciting approach behavior via implicit processes, resulting in greater attentional bias. Substance use and misuse are characterized by biases in the attentional processing of substance-related stimuli, and such stimuli develop the ability to capture the user’s attention (Field & Cox, 2008). Attentional bias can be measured by a modified Emotional Stroop task (Williams et al., 1996), in which participants are asked to indicate quickly and accurately the “ink color” of words (e.g., red, green, blue) presented on a computer screen by pressing the corresponding color response key. Participants are typically slower to color-name words of emotional relevance, with the degree of interference indexing the potency of the attentional bias of stimuli (Williams et al., 1996).

Individuals with heavy or problematic patterns of substance use and dependence, including cannabis users (Field, 2005; Field et al., 2006), display automatic cognitive biases to drug-related cues (Powell et al., 2002; Waters et al., 2003). Compared to non-dependent cannabis users, cannabis dependent individuals exhibited significant attentional bias for marijuana-related words as measured by a modified Marijuana Stroop task (Cousjin et al., 2013; Field, 2005). Furthermore, as compared to cannabis users with low levels of craving for marijuana, those with high craving have shown greater attentional bias for marijuana-related words (Field et al., 2004a). Regular cannabis users exhibit biases by maintaining their gaze on cannabis cues longer than on matched control pictures and rate cannabis pictures as more pleasant (Field et al., 2006).

Furthermore, attentional bias to substance cues has been shown to be sensitive to drug state. Under nicotine deprivation conditions, smokers report increased perceived pleasantness of smoking-related pictures and display increased attentional bias for tobacco smoking-related cues as compared to non-deprivation conditions (i.e., following acute tobacco cigarette administration; Field, Mogg, & Bradley, 2004); the degree of attentional bias has been shown to predict latency to first cigarette of the morning (Waters & Feyerabend, 2000). In addition, acute alcohol administration increases both cognitive bias for tobacco smoking cues and the perceived pleasantness of smoking cues among smokers (Field, Mogg, & Bradley, 2005), indicating that other substances may have the propensity to interfere with cognitive and attentional biases. Attentional bias and pleasantness ratings for marijuana cues has also been assessed following completion of marijuana craving induction paradigms (Eastwood, Bradley, Mogg, Tyler, & Field, 2010; Nickerson et al., 2011). Finally, marijuana with low levels of cannabidiol (CBD) and high levels of THC was shown to increase attentional bias to cannabis stimuli on a dot-probe task in users who were intoxicated with their own chosen cannabis (Morgan et al., 2010). In this study where participants smoked own cannabis in their homes, marijuana with high CBD low THC did not acutely affect attentional bias. Although findings from the latter study are informative, no previous laboratory study has examined experimentally whether marijuana stimuli can grab attention in intoxicated users relative to users not under the acute influence of marijuana.

It is also unknown whether marijuana acutely changes cognitive bias to emotional (positively or negatively valenced) stimuli. After smoking marijuana (versus placebo), marijuana users may display bias in emotional reactivity by selective attention and/or evaluation of affective versus neutral cues, although this has not been previously examined. It is also unknown whether marijuana has greater acute impact on positively or negatively valenced cues. In general, positive subjective effects are most relevant in the initiation and progression to regular drug use, and negative reinforcement, or reduced negative affective state, becomes increasingly salient with regular use and in the development of substance use disorders (Baker et al., 2004; Robinson & Berridge 2003).

In addition to continued marijuana use to avoid unpleasant withdrawal symptoms (Budney et al., 2003), regular heavy users may use marijuana to acutely reduce situational negative affect (e.g., in conjunction with stressful events). This may be why marijuana use is highly comorbid with affective disorders (e.g., Posttraumatic Stress Disorder) characterized by high levels of negative affect and arousal leading to heightened drug motivation (Bonn-Miller et al., 2007). In fact, non-drug stimuli associated with negative mood can implicitly activate processing of drug information (Curtin et al., 2006). Thus, it is possible that heavy marijuana users may take more time to process negatively-valenced affective cues and display greater implicit attentional bias to such cues.

Given these considerations, the purpose of this placebo-controlled study was to evaluate, in experienced regular marijuana users, processing of positively-valenced, negatively-valenced, and marijuana-related visual stimuli on the Pleasantness Rating Task and attentional interference by affective word stimuli on the Emotional Stroop tasks after smoking marijuana. We hypothesized that active marijuana (2.7–3.0% THC¹), relative to placebo, would increase cognitive processing and increase valence of ratings (i.e., evaluative bias) of positive, aversive, and marijuana-specific visual stimuli (versus neutral stimuli) on the Pleasantness Rating Task. We also hypothesized that active marijuana, relative to placebo, would increase attentional bias (latency of response) to positive and aversive word stimuli (versus neutral stimuli) on the Emotional Stroop Task. Because of some evidence of individual differences in attentional processing of marijuana (e.g., cannabis dependence diagnostic status; Cousijn et al., 2013; Field, 2005) and emotional stimuli (e.g., sex differences; Bradley, Codispoti, Sabatinelli, & Lang, 2001; Kring & Gordon, 1998; Lang, Greenwald, Bradley, & Hamm, 1993), we examined sex and cannabis dependence diagnosis as potential moderators of responses on the two experimental tasks in additional set of exploratory analyses. These hypotheses were tested using a within-subjects design in which 89 regular marijuana users completed two marijuana administration sessions. Participants smoked a marijuana cigarette with 2.7–3.0% THC and a placebo marijuana cigarette on separate days in counterbalanced order and then completed subjective measures of intoxication, state affect, urge to smoke marijuana and the two experimental tasks.

Methods

Design and Randomization

The study involved a 2 X 2 within-subjects design crossing drug administration (2.7–3.0% or 0% THC) with cues (positive/negative vs. neutral; and marijuana vs. neutral). Participants completed a screening session and two experimental double-blind sessions in which they smoked either an active marijuana or marijuana placebo cigarette, with the order of administration counterbalanced across participants. We experimentally controlled for expectancy effects (i.e., the effect of believing that one is smoking marijuana rather than a placebo cigarette), which may influence outcomes (Metrik et al., 2012). Using established procedures (Haney et al., 2008), we informed participants via the Study Informed Consent that they would receive a marijuana cigarette of a different strength (“Dose A” and “Dose B”) varying in concentration of delta-9-THC in the low to moderate potency range from session to session.

Participants

The study was approved by the Institutional Review Board of Brown University. Marijuana smokers recruited from the community met the following inclusion criteria: native English speakers, 18 to 44 years of age, non-Hispanic Caucasian (due to additional genetic aims of the study), self-reported marijuana use at least two days a week in the past month and at least weekly in the past 6 months, and ability to abstain from marijuana for 24 hours without withdrawal (to avoid confounding effects). Exclusion criteria were: intent to quit or receive treatment for cannabis abuse; pregnancy; nursing; positive urine toxicology screen for drugs other than cannabis; current DSM-IV Axis I affective disorder or panic disorder, psychotic symptoms or suicidal state assessed by the Structured Clinical Interview for DSM-IV Non-Patient Edition (First et al., 2002); contraindicated medical issues by physical exam or BMI > 30; and 20+ tobacco cigarettes a day (Metrik et al., 2012).

Potential participants were screened by telephone and were first asked to complete a genetic screening visit that included DNA collection to address additional genetic aims not presented in this report. They were invited to return for a baseline screening interview, at which they signed informed consent and completed a physical exam. Of the 115 potential participants completing baseline screening, 12 were deemed ineligible, 10 withdrew prior to completing the first marijuana administration session, three withdrew prior to completing the second session, and one participant was inadvertently administered an active marijuana dose on both of the experimental days without the placebo control. Results are based on the remaining 89 participants, all of whom completed both of the marijuana administration sessions, with the order of administration of two doses counterbalanced across participants.

Procedure

The two experimental sessions were, on average, 11.8 (SD = 10.8) days apart. Participants were told to abstain from marijuana and tobacco smoking for 15 hours, alcohol for 24 hours, and caffeine for 1 hour before both sessions. An alveolar carbon-monoxide (CO) of ≤ 6 ppm was used to confirm no recent smoking (Cooper & Haney 2009; Metrik et al., 2012) with a Bedfont Scientific Smokelyzer®. Tobacco smokers were given an opportunity to smoke a tobacco cigarette following the CO test to prevent nicotine withdrawal at all sessions. Zero breath alcohol concentration was verified with an Alco-Sensor IV (Intoximeters, Inc., St Louis, MO., USA). Positive THC urine screens were obtained from 80% of participants at the beginning of the marijuana administration session and from 83% at the beginning of the placebo administration sessions. Research assistants and participants were blind to the marijuana content of the cigarettes.

Sessions occurred in a 75 sq ft ventilated smoking room with a one-way mirror and intercom. At baseline, participants completed interview and self-report assessments including demographics, diagnostic interview, and marijuana use questions. At the experimental sessions, participants completed measures of subjective effects and underwent heart rate monitoring before and after drug administration. After the smoking, two experimental tasks were completed in a counter-balanced order. The first task always started 17 minutes from the start of smoking. The second task started at approximately 30 minutes from the start of smoking, immediately following completion of the first task, heart rate reading, and self-report assessment of marijuana urge and state affect. Participants remained in the laboratory for 4 hours after smoking, passed a field sobriety test, and were transported home by taxi. All participants were paid $275 upon completion.

Marijuana Administration

Marijuana cigarettes (active THC: 2.7%-3.0% and placebo made of marijuana from which THC had been removed) were provided by the National Institute on Drug Abuse, rolled at both ends, humidified, and smoked according to the standardized paced puffing procedure (Foltin et al., 1987; Metrik et al., 2012) until the ash reached a mark 10 mm from the end. This THC dose has significantly affected subjective measures and behavioral tests (e.g., Lukas et al., 1995; McDonald et al., 2003; Metrik et al., 2012) including subjective measures of affect (Metrik et al., 2011). Participants took a mean of 9.03 (SD = 1.27) puffs on a cigarette and completed smoking in a mean of 8.09 minutes (SD = 1.10).

Baseline Measures

DSM-IV Axis I diagnoses were determined with the Structured Clinical Interview for DSM-IV Non-Patient Edition (SCID; First et al., 2002). The Time-Line Follow-Back (TLFB; Dennis et al., 2004) assessed past 60-day number of marijuana, alcohol, and tobacco cigarette use days using a calendar assisted structured interview.

Marijuana Withdrawal Checklist Diary (MWC) assesses the presence and degree of perceived marijuana withdrawal symptoms (Budney et al., 1999; Budney et al., 2003). Participants rated each symptom possibly experienced due to the overnight abstinence from marijuana based on a 4-point scale where 0 = none, 1 = mild, 2 = moderate, and 3 = severe. Ten of the symptoms comprise a withdrawal discomfort score (Budney et al., 1999). The average marijuana withdrawal discomfort score on the MWC at each smoking session was 2.4 (SD = 2.9) and 2.8 (SD = 2.3), not clinically meaningful.

Dependent Measures

Heart Rate (beats per minute) was recorded prior to smoking at baseline and 15, 30, 45, and 60 minutes post smoking task commencement via a blood pressure cuff attached to the non-dominant arm (Datascope Accutorr Plus NIBP).

Subjective Drug effects. Subjective effects of marijuana were assessed with the Addiction Research Center Inventory-Marijuana scale (ARCI-M; Chait et al., 1985; Martin et al., 1971) and with the 0–100 visual analogue scale (VAS) Marijuana Rating Form (Haney et al., 2005). The VAS included 5 cigarette rating items of liking, desire to smoke again, feeling high, good drug effect, and bad drug effect. Participants also rated the potency of the cigarette smoked relative to their usual marijuana cigarettes on a 5-point Likert scale (from 0 “No effect at all” to 4 “a very strong effect”) and THC content in terms of % concentration (scored: 0= none 0%, 1= low dose <2%, 2= moderate dose 2–3%, 3= high dose 3–4%; Metrik et al., 2009). Subjective drug effect measures were completed twice following marijuana administration: at T1, immediately after the smoking, and at T2, after the completion of both computer tasks (within 45 minutes after the start of smoking). ARCI-M was also completed prior to smoking.

Urge to smoke marijuana was assessed with the 10-item Marijuana Craving Questionnaire (MCQ; Budney et al., 2003). MCQ items are rated on a 1=“strongly disagree” to 7=“strongly agree scale,” with higher scores indicating greater urge, and summed to yield a total craving score.

Positive and Negative Affect Schedule (PANAS) is a 20-item measure of state affect (Watson et al., 1988). Participants rated the degree to which they currently felt affective adjectives (e.g., positive affect subscale: interested, proud, excited; negative affect subscale: irritable, afraid, distressed) on a 5-point Likert scale (from 1 “very slightly, or not at all” to 5 “extremely”). Participants also completed five items using the same scale from the vigor-activity subscale of the Profile of Mood States questionnaire (POMS; McNair et al., 1971). Measures of marijuana craving and affective state were completed before and three times after marijuana administration: at T1, immediately after the smoking, and following the completion of each of the two computer tasks.

Measures of Cognitive Bias

Pleasantness Rating Task assesses perceived attractiveness (valence) of reward, aversive, and marijuana-related (Field et al., 2006) visual stimuli that have been previously validated (Gilbert & Rabinovich, 1999; 2003; 2005). The affective pictures were taken from the standardized International Affective Picture System (Lang et al., 2008). Images included 18 positive, 18 negative, 18 neutral, and 18 marijuana-related (Field et al., 2006) pictures presented in random order for three seconds (maximum time). A fixation cross was presented in the center of the computer screen for 500ms, followed by an image stimulus. Cannabis-related images depicted a cannabis use scene (e.g., preparation of cannabis ‘joints,’ a model smoking cannabis). Pictures were closely matched on perceptual characteristics such as brightness and complexity. There were initially five practice trials, in which practice stimuli were presented, before 72 critical trials, in which each affective, marijuana, and neutral picture was presented once. Trials were presented in a new random order for each participant. Participants rated pleasantness and unpleasantness on a 7-point scale (−3 to +3) displayed on the screen until a rating was completed (3 seconds maximum). Perceived attractiveness (mean valence rating) was recorded. Evaluative bias (average response latency (ms) between stimulus presentation and participant response) was measured. Data for one participant were lost due to technical problems.

Emotional Stroop Task assesses automatic attentional bias to affective relative to neutral word stimuli (MacLeod & Mathews 1991; McKenna & Sharma 1995). Participants were asked to press a designated key as quickly as possible in response to the color of a word (red, green, blue, and yellow) from one of three categories (neutral, positive, negative). Affective stimuli (10 words per category) were based on Affective norms for English words (ANEW; Bradley & Lang 1999) and matched for length and usage frequency (Balota et al., 2007; Kahan & Hely 2008). Each word was presented (3 seconds maximum) once in each of four colors, in category blocks, with block order counterbalanced across participants. A practice sequence of 20 letter strings (e.g., HHHH) was included. Mean reaction time of response and accuracy (i.e., latency of color-congruent response) for each word category was recorded.

Data Analysis Plan

Repeated measures analyses examining subjective intoxication effects, subjective drug rating, and changes in heart rate were conducted using generalized estimating equations (GEE). Experimental condition was dummy-coded with placebo as the reference group. A variable carrying the linear effect of time and a condition by time interaction were also included. All models, with the exception of VAS subjective effect ratings, included the pre-smoking value of the respective dependent variable as a time-varying covariate.

Separate GEE analyses were used to examine 1) response latency (ms) to positive, negative, and marijuana-related visual stimuli on the Pleasantness Rating Task and 2) attentional interference by appetitive and aversive word stimuli on the Emotional Stroop Task. Additional GEE analysis tested perceived attractiveness (i.e., mean valence ratings) of the three types of visual cues (positive, negative, and marijuana vs. neutral) following marijuana administration on the Pleasantness Rating Task. All of the GEE models included dummy-codes contrasting positive, negative, and marijuana cues with neutral cues and the dummy-coded experimental condition (marijuana vs. placebo). Interaction terms between experimental condition and dummy-coded visual (or word) cue comparisons were entered on the last step in the model. We included session number in analyses (e.g., first or second) in case of order effects. However, session number was not associated with outcomes and thus was dropped from analyses. Additional moderation analyses mirrored the GEE models described above but also included dummy-coded sex or cannabis dependence diagnosis as predictors in separate models. First, main effects of sex (or dependence diagnosis), drug condition, and cue type were examined. These were followed by the test of all 3-way interactions (e.g., sex by drug condition by each cue type) with all of the two-way interactions included in the models (drug condition by each cue type, drug condition by sex, cue type by sex). We predicted a three-way interaction between sex (or cannabis dependence diagnosis), drug condition, and cue type on response latency (and on the pleasantness task ratings). All analyses were conducted using SPSS 19.0 for Windows.

Results

Table 1 presents descriptive statistics for the sample.

Table 1.

Sample demographics and marijuana use characteristics (n = 89)

Variable	M	SD
Age	21.4	4.5
Years of education	13.8	1.7
Percent marijuana use days	72.5	21.9
Times used marijuana on average day	2.0	1.2
Percent alcohol use days	29.5	19.6
Alcohol drinks per drinking day	4.2	2.4
Percent heavy drinking days	13.2	13.5
Percent smoking tobacco days (for n = 42 tobacco smokers)	58.0	41.3
Tobacco cigarettes per smoking day	4.1	3.8
	n	%
Gender
Male	59	66
Female	30	34
Marital Status
Single/Never Married	79	89
Other	10	11
Age of initiation of marijuana use
14 or younger	33	37
15–16	30	34
17 or older	26	29
Age of initiation of regular marijuana use
14 or younger	9	10
15–16	24	27
17 or older	56	63
Marijuana ounces used per week
1/16th or less	30	34
1/8th	26	29
1/4th –1/2	26	29
More than ½	7	8
DSM-IV Diagnosis
Lifetime cannabis dependence	18	20
Past year cannabis dependence	12	14
Lifetime cannabis abuse	46	52
Past year cannabis abuse	27	30

Heart Rate

Results of GEE analyses show that active marijuana, compared to placebo, significantly increased heart rate, B = 69.01, 95% CI (60.54, 77.49), p < .001. A significant drug by time interaction, B = −7.07, 95% CI (−8.56, −5.57), p < .001, showed that active marijuana’s effect on heart rate was greater immediately post-smoking than at the end of the post-smoking period.

Subjective Effects Ratings

Significant main effect of drug was seen for ARCI-M ratings indicating that receiving active marijuana vs. placebo was associated with greater subjective intoxication (B = 4.0, 95% CI (2.54, 5.47), p < .001. As shown in Table 2, active marijuana, compared to placebo, also significantly increased VAS ratings of feeling high, drug liking, good drug effect (all p’s < .001), desire to smoke it again (p < .05), and ratings of potency and THC content. Urge to smoke marijuana on the MCQ decreased following active marijuana vs. placebo administration, B = .43, 95% CI (.79, .07), p < .05. There were no main effects of drug on ratings of affect on the PANAS or the POMS vigor subscale (p > .05). There was a significant negative effect of time for ARCI-M and ratings of feeling high, liking, desire to smoke again (p’s < .05), negative affect (p < .05) and positive affect on the PANAS (p < .01), reflecting the fact that these subjective ratings were lower towards the end of the testing period. There were no significant drug by time interaction effects.

Table 2.

Means and Standard Deviations for the Subjective Effects by Drug Condition Post-Smoking

	Placebo				Marijuana
	Time 1		Time 2		Time 1		Time 2
Measure	M	SD	M	SD	M	SD	M	SD
1. Drug Liking	44.66	26.33	38.94	26.80	65.54	23.42	62.09	25.12
2. Desire to Smoke Again	44.94	31.21	39.74	30.67	63.57	28.25	62.31	28.15
3. Feeling High	29.34	22.43	26.53	22.04	66.12	23.64	64.62	25.72
4. Good Drug Effect	37.78	25.88	35.76	25.95	64.61	23.10	62.25	25.37
5. Bad Drug Effect	11.66	17.08	12.64	17.66	15.64	18.42	17.74	17.30
6. Potency	1.21	0.73	1.20	0.77	2.53	0.76	2.54	0.81
7. THC Content	0.91	0.49	0.90	0.50	1.96	0.66	1.96	0.72

Note. THC = delta-9-tetrahydrocannabinol; Time 1 = immediately after the smoking; Time 2 = after the completion of both of the computer tasks within 45 minutes after the start of smoking. Mean subjective effect ratings (measures 1–5) scored on a 0–100 mm visual analogue scale (VAS). Potency = potency of the cigarette smoked relative to usual marijuana cigarette scored on a 5-point Likert scale (from 0 “No effect at all” to 4 “a very strong effect”); THC content = percent concentration scored 0= none 0%, 1= low dose <2%, 2= moderate dose 2–3%, 3= high dose 3–4%. Relative to placebo, active marijuana significantly increased all subjective effect ratings with the exception of “bad drug effect”.

Pleasantness Rating Task: Mean Reaction Time

Figure 1 and Table 3 present mean reaction time (ms) participants used to evaluate neutral, positive, negative, and marijuana images after smoking active marijuana or placebo cigarette. Response latency to negative and marijuana images variables were square-root transformed to correct positive skewness but raw means (SD) are shown for ease of interpretation. Relative difference in response times (RT) between all sets of affective/marijuana and neutral pictures decreased after smoking marijuana compared to placebo.

Fig. 1.

Mean reaction time (ms) to evaluate neutral, positive, negative, and marijuana images on the Pleasantness Rating Task following administration of active marijuana and placebo. Error bars represent standard error of the mean (SEM). The top panel shows the full scale ranging from 0 to 1200 ms. The bottom panel illustrates close-up of the top range of the scale for visual demonstration of group differences.

Table 3.

Means and Standard Deviations for Experimental Task Variables by Drug Condition

	Placebo		Marijuana		Drug X Cue Interaction
	M	(SD)	M	(SD)
Mean reaction time (ms) on the Pleasantness Rating Task
Neutral images	777.4	218.0	936.0	386.1	___
Positive images	834.2*	273.9	943.9	397.0	ns
Negative images	911.2***	365.8	1008.7**	417.1	p < .001
Marijuana images	892.9***	301.2	1022.9**	508.8	p < .001
Mean pleasantness ratings on the Pleasantness Rating Task
Neutral images	−0.1	0.5	−0.1	0.5	___
Positive images	1.8***	0.6	1.8***	0.6	ns
Negative images	−2.3***	0.6	−2.2***	0.6	ns
Marijuana images	1.1***	0.9	0.9***	1.0	p = .001
Mean reaction time (ms) on the Emotional Stroop Task
Neutral words	637.5	145.7	638.7	166.3	___
Positive words	645.1	151.4	650.1	167.5	ns
Negative words	642.6	151.2	656.9*	168.4	ns

Note. Affective/Marijuana cue vs. neutral cue comparisons:

^*p < .05;

^**p < .01;

^***p < .001. Mean valence ratings on the Pleasantness Rating Task scored on a seven-point Likert scale (−3 to +3)

Results of GEE analysis showed significant drug X negative images interaction, B = −157.84, 95% CI (−235.82, −79.87), p < .001, and significant drug X marijuana images interaction, B = −158.22, 95% CI (−235.79, −80.66), p < .001. The differences between RT to negative and neutral cues were significant in both drug conditions, t(88) = 3.94, p < .001 and 2.53, p = .01. Relative to neutral images, reaction time to negative images increased in the placebo (mean increase = 133.7 (319.9) and to a somewhat lesser degree in the marijuana condition (mean increase = 72.7 (268.0)). As expected, marijuana acutely slowed RT across stimuli (i.e., as evidenced by a difference in RT to neutral cues between M = 777.4, SD = 217.9 in the placebo vs. M = 936.0, SD = 386.1 in the marijuana condition). This increase in RT to neutral stimuli after smoking marijuana (i.e., general slowness of response) does not explain an additional significant increase in RT to negative cues in the marijuana condition. In fact, the greatest RT in the marijuana condition was in response to negative images (M = 1008.7, SD = 417.1). In support of our hypothesis, taking into account the general slowness of response after smoking marijuana, the significant increase in RT to negative vs neutral stimuli in the marijuana condition confirmed the hypothesis that marijuana acutely increases cognitive bias towards negative affective stimuli.

Similarly for the marijuana vs. neutral images, post-hoc tests indicated the differences in RT were significant in both drug conditions, t(88) = 5.55, p < .001 and 2.96, p < .01. Relative to neutral images, reaction time to cannabis images increased in the placebo (mean increase = 115.5 (196.2) and to a somewhat lesser degree in the marijuana administration condition (mean increase = 87.0 (274.0)). Accounting for the general slowness of response after smoking marijuana (i.e., as reflected by a change in RT to neutral stimuli from placebo to marijuana conditions), the greatest RT in the marijuana condition was also to cannabis images (M = 1022.9, SD = 508.8). The increased RT to cannabis vs neutral stimuli in the marijuana condition confirmed the hypothesis that marijuana acutely increases cognitive bias towards cannabis-related stimuli.

Drug X positive images interaction was at trend level, B = −45.49, 95% CI (−98.40, 7.43), p = .09. Paired samples t-tests indicated that, relative to neutral images, reaction time to positive images increased in the placebo (mean increase = 56.84 (223.9), t (88) = 2.40, p = .02) but not in the marijuana condition (t (88) = .29, p = .77). These RT data did not confirm the hypothesis that marijuana acutely increases cognitive bias towards positive affective stimuli.

Main effect of drug condition was significant (p < .001) so that participants were slower in responding to images after acute active marijuana dose vs placebo. Main effect of picture type was significant for negative and for the marijuana images (p < .001) that participants took longer to evaluate but not for positive vs neutral images.

Pleasantness Rating Task: Valence Ratings

GEE analysis with valence ratings on the Pleasantness Rating Task showed a significant drug X marijuana images interaction, B = −.25, 95% CI (−.39, −.11), p = .001, but interactions between drug and affective images were not significant (p’s > .21). Paired samples t-tests indicated that participants rated cannabis pictures as more pleasant than the control pictures in both conditions, t(87) = 10.5 and 12.54, p < .001. Relative to neutral images, marijuana images were rated as more pleasant in the placebo (mean increase = 1.2 (.90)) and, to a slightly lesser degree, in the marijuana condition (mean increase = .94 (.84)), possibly due to reduced marijuana urge after smoking. Main effects of picture type: positive vs. neutral images, B = 1.88, 95% CI (1.75, 2.01), p < .001, and negative vs. neutral images, B = −2.14, 95% CI (−2.29, −1.98), p < .001, were significant. As expected, participants rated positive pictures as significantly more pleasant than neutral images and rated negative images as significantly more unpleasant than neutral images (see Figure 2 and Table 3). Main effect of drug was not significant (p = .67). Contrary to our hypothesis, marijuana did not acutely influence affective picture ratings.

Fig. 2.

Mean valence ratings on a seven-point Likert scale (−3 to +3) of neutral, positive, negative, and marijuana images on the Pleasantness Rating Task following administration of active marijuana and placebo. Error bars represent standard error of the mean (SEM)

Pleasantness Rating Task: Moderation by Sex

Additional analyses by sex were conducted to explore whether men and women differed in their evaluations (response latency and valence ratings) to neutral, positive, negative, and marijuana images on the Pleasantness Rating Task after smoking active marijuana or placebo cigarette. Main effect of sex was not significant in both outcome models (p. = 78 for response latency and at the trend level p = .07 for valence ratings). None of the three-way interactions with sex were significant; all models indicated similar main effects and two-way interactions between drug condition and cue type in both sexes as reported above for the whole sample. The only significant two-way interaction with sex was sex by negative images interaction, B = −.38, 95% CI (−.68, −.09), p = .01 indicating that female participants rated negative images as significantly more negative (M = −2.49, SD = .49) than male participants (M = −2.08, SD = .60), p < .001.

Pleasantness Rating Task: Moderation by Cannabis Dependence (CD) Diagnosis

In the analyses with the valence ratings, main effect of CD was at the trend level, B = .12, 95% CI (−.01, .25), p = .08. None of the three-way interactions with CD were significant; all models indicated similar main effects and two-way interactions between drug condition and cue type as reported above for the whole sample. The only significant two-way interaction with CD was CD by marijuana images interaction, B = .52, 95% CI (.11, .93), p = .01, indicating that marijuana images were rated higher by participants with the current CD diagnosis (M = 1.38, SD = .78) relative to those without the diagnosis (M = .93, SD = .97), p < .05. There were no significant effects of cannabis dependence diagnosis in models with response latency as the dependent variable.

Emotional Stroop Task: Affective Cues

Results of GEE analysis of the Emotional Stroop Task data showed significant main effect of negative vs. neutral words, B = 11.62, 95% CI (.60, 22.64), p <.05, and main effect of positive vs. neutral words at the trend level, B = 9.44, 95% CI (−1.34, 20.22), p =.09, with increased RT to affective stimuli. Participants were significantly slower to color-name the negatively-valenced words, M = 649.7 (SD = 15.5), than the control words, M = 638.1 (SD = 15.1) but smoking conditions did not significantly differ in color-naming RT. Interactions with drug effect were also not significant, p’s > .29: relative to neutral words, increase in reaction time to negative words was not significant in the placebo condition and was only at the trend level in the marijuana condition (mean increase = 18.2 (87.5), t (88) = 1.96, p = .05), (see Table 3). Analyses with the mean response latency of correct color-congruent responses were consistent with these findings. Contrary to our hypothesis, active marijuana administration did not result in attentional bias to affective word stimuli.

Emotional Stroop Task: Moderation by Sex

Main effect of gender was at the trend level, p = .08. All interaction effects with gender and negative words were not significant. Three-way interaction between positive words by drug condition by sex was at the trend level, B = −42.58, 95% CI (−87.56, 2.40), p = .06, and two-way interaction between positive words and sex was at the traditional significance level, B = 21.55, 95% CI (−.09, 43.18), p = .05, with other results unchanged from those of the whole sample. Follow-up independent samples t-tests indicated that women (M = 691.73, SD = 163.92) were slower to color-name positive words relative to men (M = 625.12, SD = 152.64), p < .01.

Emotional Stroop: Moderation by Cannabis Dependence (CD) Diagnosis

Main effect of CD was not significant, p = .14. All interaction effects with CD and negative words were not significant. Three-way interaction between positive words by drug condition by CD was significant, B = 50.91, 95% CI (2.48, 99.34), p < .05, with other results unchanged from those of the whole sample. Follow-up analyses indicated that the interaction between drug condition and positive words differed significantly between those with the CD diagnosis (B = 47.87, 95% CI (6.10, 89.64), p < .05) and those without (p > .80); indicating that, relative to placebo, marijuana acutely increased time to color-name positively-valenced words only among those with the CD diagnosis.

Analyses of Association between Marijuana Craving and Attentional/Cognitive Processing Variables

Correlations among the experimental tasks variables and urge to smoke marijuana on the MCQ were examined within each drug condition. On the Pleasantness Rating Task in the placebo condition, MCQ urge correlated significantly with the valence ratings of the marijuana images (r = .35, p = .001) but not with the ratings of neutral, positive, or negative images. In the marijuana condition, MCQ urge correlated significantly with the valence ratings of the neutral (r = .21, p < .05), marijuana (r = .40, p < .001), and negative images (r = .35, p = .001) but not with the positive images. There were no significant associations between marijuana craving and response latency variables on this task. On the Emotional Stroop Task in the placebo condition, MCQ urge correlated significantly with reaction time to neutral (r = .21, p < .05) and negatively valenced (r = .24, p < .05) words but not with the positively-valenced words. In the marijuana condition, MCQ urge correlated significantly with the negatively-valenced words (r = .25, p < .05) but not with the neutral or positively-valenced words.

Discussion

This is the first placebo-controlled evaluation of marijuana’s acute effect on cognitive interference by appetitive, aversive, and marijuana-related visual and word stimuli using the Pleasantness Rating Task and the Emotional Stroop tasks in regular marijuana users. In support of our hypothesis, there was evidence of marijuana acutely altering affective processing of negatively-valenced and marijuana-related visual stimuli. Overall, findings suggest that marijuana does not acutely change processing of positive emotional stimuli beyond a general slowing of response. The only exception to this was some evidence of attentional bias to positive word stimuli on the Emotional Stroop task following marijuana administration in the subgroup of marijuana users who met DSM-IV criteria for cannabis dependence.

Relative to placebo, active marijuana significantly increased response latency to negative as compared to neutral images on the Pleasantness Rating Task. Cognitive bias to negative images due to active marijuana was evident beyond the general slowness of the response on marijuana and beyond the general increased attention customarily seen to aversive neutral pictures. Contrary to our findings for negative visual stimuli, active marijuana did not acutely change attentional processing of positive emotional stimuli. On placebo, participants spent more time in processing positive versus neutral stimuli, but on active marijuana, pleasant images did not differ from controls in terms of response time to these stimuli. These data suggest that pleasant experiences may become more positively reinforcing after marijuana without a specific change in attentional cognitive processing.

Consistent with the Incentive Sensitization conceptualization of marijuana “wanting” (Robinson & Berridge 2001), implicit motivation as measured by attentional bias to visual marijuana-related cues was observed after active marijuana administration. As compared to placebo, active marijuana significantly increased response latency to images with marijuana content relative to neutral content. Just as was the case with the negatively-valenced stimuli, cognitive bias to cannabis images was evident beyond the general slowing of response after smoking marijuana. Furthermore, there was additional evidence in support of biased cognitive processing of marijuana content on the Pleasantness Rating task. Participants rated cannabis pictures as more pleasant than controls in both of the drug conditions but significantly lower after having smoked marijuana. This finding signaling “reduced wanting” is not surprising, as it coincides with reported reduction in marijuana urge after the smoking. In previous attentional bias work, cannabis users have shown increased attentional bias to cannabis-related cues, particularly during withdrawal or following abstinence (Field et al., 2006). Our marijuana-administration study builds on previous non-acute findings with regular cannabis users with high marijuana craving demonstrating greater attentional bias for marijuana-related words relative to users with low levels of craving (Field et al., 2004a). Having smoked active marijuana, participants are arguably satiated such that marijuana cues, although very salient, are no longer as appealing as they are under states of increased drug motivation. Our findings on the Pleasantness Rating Task showing higher ratings of marijuana images by marijuana users with current cannabis dependence relative to those without this diagnosis are also consistent with previous non-marijuana administration studies that demonstrated greater attentional bias for cannabis words by users with the CD diagnosis as compared to non-dependent users (Cousijn et al., 2013; Field et al., 2005). Cousjin and coauthors also demonstrated greater attentional bias to cannabis words in heavy marijuana users relative to the matched controls.

Contrary to our hypothesis, active marijuana, relative to placebo, did not generate attentional bias to affective word stimuli on the Emotional Stroop Task in the whole sample. However, in the subsample of marijuana users meeting DSM-IV criteria for current cannabis dependence, marijuana acutely increased time to color-name positive words. Regardless of the drug condition, participants were significantly slower to color-name the negatively-valenced words than the control words. Findings across both cognitive tasks used in this study consistently suggest that regular marijuana users allocate greater attentional resources to negatively-valenced than positively-valenced cues. These data are consistent with findings in non-marijuana users on negative emotional stimuli having larger effects on interference with color naming than positive emotional stimuli. Negative emotion-laden words have been found to have larger interference effects than positive emotion-laden words on the Emotional Stroop task (McKenna & Sharma, 1995)

Furthermore, our findings on sex differences in evaluations of negative visual stimuli on the Pleasantness Rating task and of positive word stimuli on the Emotional Stroop task are consistent with previous research that demonstrated that, relative to men, women displayed heightened physiological responses to emotional stimuli (Bradley et al., 2001) and evaluated aversive pictures as more unpleasant and arousing (Lang et al., 1993). Women also displayed different brain activation patterns from men when viewing unpleasant compared to pleasant pictures on an fMRI task (Lang et al., 1998). Recent evidence from neuroimaging studies of chronic and acute cannabis users suggests that affective processing is acutely affected by marijuana use and may be altered in individuals who smoke marijuana. Chronic heavy marijuana smokers in a non-acute fMRI study demonstrated a relative decrease in amygdalar activity during a face processing task (Gruber et al., 2009). The authors suggest that marijuana users have difficulty regulating or controlling emotions and may not process affective stimuli in the same way as non-users. Two fMRI studies examined the acute effect of oral THC to social signals of threat on a similar emotional face processing task. Phan and colleagues (2008) found reduced amygdala reactivity to social signals of threat in the absence of self-reported changes in anxiety ratings, interpreting this as an anxiolytic effect of low THC dose (7.5 mg). Fusar-Poli and colleagues (2010) found increased self-reported anxiety and psychotic symptoms following a 10 mg oral THC dose as well as increased skin conductance during the processing of intensely fearful faces. Both studies involved infrequent marijuana users with limited lifetime exposure to the drug with somewhat more experienced users in the Phan et al., (2008) study reporting less anxiety after THC, and almost drug-naïve users in the Fusar-Poli et al., (2010) study reporting more anxiety after THC. In this regard, relative to naïve users, frequent marijuana users are known to report weaker subjective effects (Chait & Perry 1992; Kirk & de Wit 1999; D’Souza et al., 2008; Metrik et al., 2009).

In concert with data from our study with regular marijuana users, these findings suggest that functional neuroimaging (e.g., Phan et al., 2008) or cognitive processing tasks may uncover marijuana-induced changes in processing of affective stimuli that are produced automatically, without conscious thought, even in the absence of self-reported changes in state affect. Important associations between negative emotions and marijuana may be revealed with more covert measures of attention such as response latency, but may be more difficult to detect with explicit self-reported ratings of state affect (hence, our non-significant findings of active marijuana’s effect on mood post-smoking). Alternatively, PANAS may have not been sensitive enough to detect more subtle mood changes in this context. In contrast, self-report measures of subjective drug effects (e.g., the VAS ratings) may not call for the same level of introspection. Furthermore, affective cognitive processing may indeed be altered in chronic users, particularly when under the influence of marijuana. Given reduced amygdala reactivity among chronic marijuana users (Gruber et al., 2009) and acutely (Phan et al., 2008), participants in our study may have attended to negative visual stimuli longer on active marijuana, relative to placebo, as they may have been less affected by negative images after marijuana smoking.

Limitations

Different measures (e.g., visual probe task, gaze tracking tasks) may be more sensitive to identifying attentional bias that is not evident when using response time tasks (Field et al., 2004b; Field et al., 2006; Mogg et al., 2003). This study used only one dose of THC. It is possible that larger THC doses would have produced greater effects on cognitive processing of affective stimuli and would have been more comparable to increasing potency of recreationally-available cannabis (Mehmedic et al., 2010; Vindenes, Strand, Kristoffersen, Boix, & Mørland, 2013). Results may not generalize to less experienced users with less tolerance (Kirk & de Wit 1999) or to individuals naïve to marijuana’s effects. Furthermore, as the sample in the current study was exclusively Caucasian, results may not generalize to racially diverse samples of marijuana users.

Conclusions

Fundamental associations between negative emotions and marijuana including negatively-valenced visual stimuli are in line with the extant clinical literature on comorbidity between cannabis use disorders and affective disorders (Agosti et al., 2002; Lynskey et al., 2002; Stinson et al., 2006; Zvolensky et al., 2008). A proliferation of scientific evidence also supports the major role of the endocannabinoid system in regulation of mood and anxiety states (Witkin et al., 2005). Findings from our study demonstrate increased allocation of cognitive processing resources to negative visual affective stimuli and to marijuana-related visual stimuli after marijuana smoking. Future research should consider examining individual differences (e.g., executive memory, baseline mood symptomatology, and variability in cannabinoid-related genetics) that may modulate attentional bias following active marijuana or placebo. Although the variability in participants’ expectancy effects was at least partially controlled in this study with explicit instructions about the active dose, future balanced-placebo design studies may also be able to fully examine the pharmacologic effect of marijuana on attentional bias independent of marijuana stimulus expectancy (Metrik et al., 2009; Metrik et al., 2012). The participants reported feeling high, enjoying the drug, and experiencing good drug effects, but marijuana did not change subjective positive or negative affect. Despite this, participants attended more to negative images. THC and other cannabinoids may have an anxiolytic role in central mechanisms of fear and anxiety behaviors and provide a justification for examining innovative therapeutic approaches that target the cannabinoid system for anxiety- and social fear-related disorders (Phan et al., 2008).