Baseline assessments of all neuropsychological tests were carried out 1 day before the interventional day (V1).

During the interventional day (V2), all subjects had standardized breakfast and lunch. After breakfast, the first venous blood collection took place, and volunteers received either 4 × 200 mg CBD (>99.8% pure, STI Pharmaceuticals, Brentwood, UK), or corresponding placebo capsules. Thirty minutes later, administration of 2 × 10 mg THC (>98.8% pure, THC pharm, Frankfurt, Germany) or corresponding placebo capsules was scheduled. Approximately 205 min after CBD or placebo administration, blood withdrawal for THC level analysis by LC-MS MS (29) took place, followed by neuropsychological assessments.

For safety reasons, all subjects were clinically examined on the following day (V3), and all neuropsychological tests were repeated.

All neuropsychological tests were carried out in paper-pencil versions.

To capture the participants' emotional state, we used a short version of the adjective mood rating scale (EWL) (30). This questionnaire comprises 60 adjectives and a four-point response format ranging from (0) not at all to (4) strongly. It covers the following six emotional categories: (a) performance-related activity, (b) general inactivation, (c) extroversion, (d) general well-being, (e) emotional excitability, and (f) depressiveness.

Cognitive processing speed was assessed using the Digit Symbol Coding task. Subjects were instructed to replace as many digits as possible in 90 s by given symbols, while the pairing table remained visible. Further, we evaluated working memory performance using the Letter-Number-Sequencing test. After reading an unordered sequence of numbers and letters aloud, subjects were asked to recall the numbers in ascending and the letters in alphabetical order. Both Digit Symbol Coding and Letter-Number-Sequencing are part of the Wechsler Adult Intelligence Scale (31).

In order to quantify changes in concentration and attention, the d2 Test of Attention (32) was used. During this paradigm, participants had to scan 14 test lines with 47 characters (“d” or “p” marked with a different number of dashes) and cross out all “d” characters marked with two dashed while ignoring all other characters. Volunteers were instructed to operate as fast as possible, but to minimize the error rate at the same time. The d2 Test of Attention provides multiple scores, of which we analyzed the following: error-corrected total number, and concentration capacity.

Statistical data analysis was performed using the software R (33).

The sample size of 15 subjects per treatment group, i.e., 60 subjects in total, is sufficient to detect a standardized difference of about 1.1 between any two groups (80% power, two-sided type I error 5%, two-sample Wilcoxon rank sum test, no multiplicity correction).

Group differences of continuous demographic variables were assessed using the Kruskal-Wallis test, while Fisher's exact test was used for the categorical variable “smoking.”

Intraindividual difference scores (investigational day—baseline) were calculated to evaluate the changes from baseline to the interventional day (~205 min post CBD intake) in emotional and attentional state and cognitive performance. The difference scores of the neuropsychological tests were not normally distributed. Thus, changes were analyzed by the non-parametric Kruskal-Wallis test, followed by closed testing, i.e., a stepwise multiple testing procedure that strongly controls the familywise error (34). The significance level was set at α ≤ 0.05 (two-sided). The data are presented as median (mdn) and 0, 25, 75, and 100 percentiles.

At the interventional day, treatment with phytocannabinoids changed subjects' self-evaluation of their emotional state in four of six categories. Ratings of general inactivation and general well-being did not differ between interventional groups.

All subjects showed slightly reduced performance-related activity compared to baseline. Interestingly, subjects who received CBD/THC showed significantly lower performance-related activity difference scores (mdn = −8, −12, −9, −5, −4 [median, percentiles 0, 25, 75, 100]) compared to PLA/PLA (mdn = −3, −12, −5, −1, 2, p = 0.002) or CBD/PLA (mdn = −2, −18, −7, −1,8, p = 0.035, see Figure 2A). The change in performance-related activity was solely lower by trend in CBD/THC. The difference score did not differ from those observed in the PLA/THC group but from those of the PLA/PLA and CBD/PLA group.

In the category depressiveness, CBD/PLA subjects showed similar scores compared to baseline. PLA/THC and CBD/THC led to higher depressiveness scores compared to baseline, although significance was only reached in CBD/THC subjects (mdn = 1, −1, 0, 6, 8) vs. PLA/PLA (mdn = 0, 4, 0, 1, 3, p = 0.015) and vs. CBD/PLA (mdn = 0, −4, 0, 1, 3, p = 0.026, Figure 2B).

A similar pattern was observed for the emotional category extraversion. While after PLA/PLA and CBD/PLA treatment the extraversion scores were comparable to baseline, PLA/THC and PLA/THC treatment resulted in a slight decrease of extraversion, representing a more introverted behavior in this group. However, only CBD/THC treatment led to a significant decline in extraversion scores (mdn = −3, −11, −8, −2, 0) vs. PLA/PLA (mdn = 0, −10, −2.5, 1, 6, p = 0.013) and vs. CBD/PLA (mdn = −1, −7, −3, 0.5, 4, p = 0.017, Figure 2C).

Regarding emotional excitability, the overall group difference did not reach significance (Kruskal-Wallis: p = 0.058). Nevertheless, by trend emotional irritability seemed to be lower after CBD/PLA treatment, while subjects who took CBD/THC showed increased excitability (Figure 2D).

Compared to the first assessment at the baseline visit (V1), subjects receiving PLA/PLA or CBD/PLA showed increased performance in both cognitive tests at the interventional day (Figures 3A,B).

On the other hand, negative digit symbol coding difference scores, indicating impaired cognitive processing, were found after PLA/THC and CBD/THC administration. However, this reduction was only significant in comparison of CBD/PLA (mdn = 6, −3, 5, 12, 20) vs. PLA/THC (mdn = 2, −79, −4.5, 6, 20, p = 0.039) and vs. CBD/THC (mdn = −2, −27, −12.5, 7, 11, p = 0.016; Figure 3A).

In addition, the working memory performance assessed by the Letter Number Sequencing test was significantly lower in the group receiving CBD/THC (mdn = 0, −4, −1.5, 1, 2) compared to CBD/PLA (mdn = 2, −2, 0.5, 3, 4, p = 0.005; Figure 3B).

The d2 Test of Attention was applied three times within 3 days and mean scores averaged over all participants increased continuously. Information processing speed displayed in this test by the error corrected total number (32) augmented from V1 (mdn = 536, 293, 461, 556.25, 618.00) over V2 (mdn = 539, 329.0, 456.5, 583.0, 652) 523.8 to V3 (mdn = 593.5, 414.0, 547, 639.5.653).

Like in the cognitive tests, we observed positive difference scores of concentration capacity in participants treated with PLA/PLA or CBD/PLA, indicating better test results at the interventional day compared to baseline testing. In addition, participants of the PLA/THC group showed higher concentration capacities at V2 than at baseline testing, while subjects receiving CBD/THC were less focused, indicated by negative difference scores. The reduction of attention after CBD/THC administration (mdn = −10, −55, −25, 15, 62) was significant compared to PLA/PLA (25, −20, 16, 33.5, 43, p = 0.005) and CBD/PLA treatment (mdn = 29, −48, 10.5, 35.5, 67, p = 0.010; Figure 3C).

Subjects who received PLA/THC showed slightly lower THC blood levels (mdn = 0.954, 0.0, 0.817, 1.615, 9.690 pmol/ml) compared to subjects who received CBD/THC (mdn = 2.61, 0.614, 1.032, 3.960, 8.350 pmol/ml). However, this difference did not reach statistical significance.

We observed no significant influence of PLA/PLA and CBD/PLA treatment on the six emotional categories (a) performance-related activity, (b) general inactivation, (c) extroversion, (d) general well-being, (e) emotional excitability, and (f) depressiveness. Slight reductions in performance-related activity in both groups could be most likely traced back to the intense investigational day. However, CBD/PLA might favor a slightly more relaxed emotional state compared to PLA/PLA, as the mean difference scores in emotional excitability were lower by trend. These findings are in line with a recent study, showing that a single CBD administration did not affect the three main dimensions of affect (hedonic tone, energetic arousal, and tense arousal) assessed by the University of Wales Mood Adjective Checklist (UMACL) (20). However, it has also been reported that subjects felt more quick-witted and clear-minded (18) after acute CBD treatment, while another study found increased Adjective Mood Rating Scale (Bf-S) scores, indicating a decline in general well-being, 3 h after CBD treatment (19). Hitherto, available literature on the emotional effects of CBD in humans focuses mainly on anxiolytic properties. Results of published studies indicate that CBD does not affect anxiety per se, as it has no effect on baseline anxiety (18, 19, 35, 36). Our results are consistent with these findings in the most general sense, although the used adjective mood rating scale does not retrieve anxiety, but depressiveness experience. However, it has been reported that CBD exhibits anxiolytic properties in experimental anxiogenic settings (18, 37) and seems to be effective in subjects with social phobia and generalized social anxiety disorder (38, 39). Recreational use of cannabis is mostly associated with feeling “high” and relaxing properties (40), while only a subset of naïve users reports dysphoria (41) and even paranoia (40). In the present study, we observed that THC administration reduced performance-related activity and extraversion by trend, while subjects felt slightly more depressive compared to PLA/PLA and CBD/PLA treated subjects. These results are consistent with previous studies, reporting that healthy volunteers acutely treated with THC feel anxious (18) and tense (35). Acute cannabis use has also been associated with an euphoriant effect (“high”), decreased anxiety, depression, and tension, as well as increased sociability if taken in friendly surroundings (40). This seems contradictory, but in line with a proposed biphasic effect of THC. While low doses [around 5–10 mg THC- in a joint (40)] have revealed anxiolytic effects, higher doses can induce dysphoric experiences (40).

As mentioned above, CBD may reduce experimentally induced anxiety. It has been shown that CBD also reverses the acute anxiogenic effects of THC when both cannabinoids were administered simultaneously (18, 35). Thus, we hypothesized that CBD pre-treatment might alleviate THC effects on depressiveness and extraversion. Surprisingly, the administration of THC subsequent to a single dose of CBD had nearly the same effect on emotional perception as the THC treatment following placebo, as reflected by significantly higher depressiveness and significantly lower extraversion scores. This finding is indicative of more complex pharmacodynamic interaction of THC and CBD, depending on the treatment regime and eventually, the dosages, and consistent with the observation that different compositions of THC and CBD content in medicinal cannabis influence anxiety, depression, and stress in different ways (42). Noteworthy, Cuttler et al. (42) also suggested a difference in acute and long-term effects, whereby initially reduced perceived symptoms of negative affect may exacerbate baseline symptoms of depression over time, further demonstrating the complexity of the phenomenon.

We observed that participants treated with PLA/PLA scored higher at the investigational day in both cognitive tasks and the d2 Test of Attention. This improvement might be due to learning- and training effects, even though the tasks were repeated in different versions for repeated measures. Similar results were observed for subjects treated with CBD/PLA, indicating that CBD did negatively influence cognitive processing speed and attention, although ameliorating working memory performance by trend.

PLA/THC administration resulted in reduced cognitive processing speed compared to CBD/PLA treatment, although the performance remained widely unaffected compared to baseline. This difference to CBD/PLA is likely due to the improvement of the latter group on the investigational day, also seen in PLA/PLA. Therefore, it may be speculated if THC reduced a potential learning effect or other factors are contributing to this finding. However, the observed group differences do probably not reflect difficulties in accessing working memory, which remained largely unaffected or long-term memory contents, as Ranganathan and D'Souza (43) showed that THC does not disrupt access to information learned before administration.

Subjects treated with PLA/THC or CBD/THC reported reduced subjective performance-related activity, with CBD/THC even becoming significant vs. PLA/PLA and CBD/PLA as another indicator of more pronounced THC effects in the CBD/THC group, potentially related to the reported higher THC plasma levels. This subjective reduction in performance-related activity is unlikely attributed to the intense investigational day, as this reduction was not apparent in the PLA/PLA and CBD/PLA groups. While a reduced subjective performance-related activity was reflected by the impairments in cognitive processing speed (Digit Symbol Coding Task) and concentration capacity (d2 Test of attention), a discrepancy between subjective experience and objective performance is noteworthy and has been observed before (44).

Interestingly, THC did not influence working memory performance. This finding is consistent with a study reporting that THC did not alter working memory whilst disturbing episodic memory and verbal learning (45). Unfortunately, the latter domains were not assessed by the tasks applied in the present study. However, other studies reported that intravenous as well as pulmonal THC administration induced working memory impairments, reflected by reduced performance in the Digit Span forward and reverse task (20), the Digit Symbol Substitution Task (46, 47), the Paced Auditory Serial Addition Task and the spatial N-back task (48).

Although subjects treated with PLA/THC still showed higher attentional scores than at baseline, this increase was less pronounced compared to PLA/PLA and CBD/PLA treatment. Again, it might be that learning effects superimpose impairments provoked by THC, as it has been previously reported that THC treatment resulted in a significant decrease of d2 test performance compared to placebo (49) and Divided Attention Task performance compared to baseline assessment (47). Furthermore, the synthetic THC analog nabilone led to a dose-dependent deterioration of attention in the Cognitive Drug Research computerized assessment system (50).

As already discussed with regard to emotional states, we hypothesized that CBD is able to counteract the effects of THC on cognition and attention. Interestingly, this was not the case in our setting. Subjects who received CBD/THC treatment showed no improvement in cognitive processing speed, working memory, and attention compared to subjects who received PLA/THC. Probably based on the slightly higher THC levels in the CBD/THC group, the effects of THC were more pronounced. We observed significantly reduced cognitive processing speed, working memory, and attention compared to CBD/PLA and PLA/PLA. At first, this is surprising, as it has been shown that CBD pre-treatment (600 mg orally) is able to diminish impairments of episodic memory induced by intravenously injected THC (20). However, the same study reported that CBD pre-treatment did not attenuate THC induced impairments on immediate recall, digit-span forward, and digit span backward (20). Furthermore, vaporized THC (8 mg) alone and combined with vaporized CBD (16 mg) showed the same level of impairment in episodic memory on prose recall in the story recall task from the Rivermead Behavioral Memory Test when compared to placebo (48). Interestingly, acute treatment with CBD (300 or 600 mg) did also not improve the performance of schizophrenia patients in the Stroop Color Word Test (51), and thus did not affect selective attention and processing speed in patients. It may well be that CBD pre-treatment is more effective on other cognitive domains than those applied in the present study, as it had been suggested that CBD, in particular, protects hippocampal-dependent memory performance from the impact of THC (20). On the other hand, it might also be that repeated CBD treatment is required to activate mechanisms, improving cognitive and attentional performance while acute CBD administration is not sufficient. However, Arkell et al. (46) observed a reduction of attention after a combined CBD/THC administration in the Divided Attention Task, while the task performance was not affected by THC alone.

Although we tried to address several issues affecting studies on THC and CBD in healthy volunteers (52), there are still a number of limitations to this study.

First, we only recruited male subjects due to the initially planned parallel study using radionuclides to investigate cannabinoid receptor availability after cannabinoid intake. However, due to regulatory changes, we had to halt this study indefinitely shortly before the initiation of our trial reported here. Thus, our data lacks generalisability with regard to gender.

Second, we observed substantial interindividual differences. The two major approaches to minimize standard deviations are increasing the number of subjects and homogenizing the sample cohort. The number of 60 participants was already higher than in the majority of comparable studies. Furthermore, our study was designed to reduce sample heterogeneity by stratifying for functional COMT polymorphism to exclude the influence of different dopamine elimination rates, as it has already been shown that cannabis affected cognition in a COMT genotype-dependent manner (27, 28).

Third, we included only subjects with body mass indices ranging from 18 to 30 and provided equivalent meals for all participants throughout the interventional day to achieve a comparable cannabinoid uptake. However, providing individual weight adapted dosing would be even better, but this was not feasible due to limitations of differential dosing per os. Further, oral administration of drugs is accompanied by a delayed digestion-dependent uptake and variable metabolization in the liver due to individual enzyme activities. Inhalative and intravenous applications of cannabinoids bypass the first-pass effect in the liver and result in faster peaks (43, 53), and may also reduce interindividual differences. However, oral administration leads to more consistent long-lasting peak concentrations and is also used in other clinical trials. They represent a route of administration more suitable for medical use. In particular, the latter is essential as CBD has recently been approved as an orphan drug for Dravet and Lennox-Gastault syndrome in children and is currently investigated regarding its beneficial effects in other diseases, e.g., schizophrenia. We have recently demonstrated in rodents that different pharmaceutical preparations of THC can influence its behavioral effects depending on the kinetics of the surge of THC (54), suggesting that the investigation of both oral as well as inhalative/intravenous administration of THC and CBD is justified.

Fourth, we investigated the change from baseline to post-drug intake to reduce general interindividual differences. However, multiple testing resulted in potential learning effects in both cognitive tasks and attention testing, even when using appropriate test/re-test paradigms. These learning effects might have superimposed some cannabinoid effects on cognition and attention.

Fifth, the potential use of caffeine, nicotine, and alcohol has been out of our rigid control between baseline (V1) and the following investigational day as subjects were not kept in a closed environment such as an experimental ward for compliance and funding reasons. However, subjects were thoroughly screened for any history of use or even abuse of these compounds and did neither report more pronounced use nor demonstrate any clinical signs related to it, particularly not during visits or the extensive time they were under direct observation during the trial.

Sixth, we tested only one dose of CBD and THC, respectively. The THC dose was chosen based on the recommended maximum daily dosage of dronabinol and the available oral dosage forms, while the dose of CBD was based on the dosage used in our previous clinical trial in schizophrenia (55), demonstrating the antipsychotic effects of CBD. However, it may be that the effects induced by the high THC dose were too strong to be controlled by CBD. Thus, future studies should investigate the effects of a broader range of CBD on various THC doses.