Participants
Inclusion/exclusion criteria
Participants were required to be fluent in English, 21+ years old, and free of serious neurological and medical conditions (including pregnancy and lactation), learning disabilities, and serious psychiatric disorders. Participants were excluded if they reported heavy alcohol use (> 4 drinks, > 4 times/week), heavy smoking (> 30 cigarettes/week), any illicit drug use in the past 6 months, and/or a history of substance abuse diagnosis or treatment. To reduce risks of adverse reactions, participants had to be experienced cannabis users (i.e., ≥ 50 lifetime uses, current use ≥ 1x/week for ≥ 1/year, experience with both flower and concentrates) and report no serious prior adverse reactions (e.g., panic attacks, psychosis). Participants also had to have access to a computer with a webcam and stable internet connection in a private environment (free of distractions and children) where they could inhale cannabis.
Participant characteristics
Results of a power analysis indicate that in order to achieve power of .80 to detect large-sized effects (i.e., (upeta _{{text{p}}}^{2}) = .13) with alpha set at .05, a total sample size of 80 was required. While 86 participants were recruited, participants who did not complete the protocol (n = 2) and who purchased a product that did not meet the study criteria (n = 4) were excluded. The final sample comprised 80 cannabis users, with 20 assigned to each of the four groups. The sample was well balanced with respect to gender (43.8% women, 56.3% men) and ranged in age from 21 to 44 (M = 23.87; SD = 5.67). The average age of onset of cannabis use was 17.01 (SD = 1.98), and participants had used cannabis for an average of 5.66 (SD = 5.30) years. There were no significant differences in any demographic characteristics or cannabis use patterns across the four groups suggesting random assignment was successful in producing groups equivalent at baseline (see Table 1).
Materials
Weschler test of adult reading (WTAR)
The WTAR is a brief measure of premorbid verbal IQ22. The test consists of a list of 50 words that participants pronounce aloud. They are given 1 point for each correctly pronounced word. This test was used to ensure groups had comparable cognitive abilities at baseline.
Prospective memory tests
Participants’ ability to remember to execute tasks in the future was assessed with two prospective memory tests: the reminder test23,24,25 and difficulty ratings test26. For the difficulties ratings test participants were asked to rate the difficulty of each test they completed immediately after completing each, using a 0 to 10 scale. The percentage of tests each participant provided difficulty ratings for was computed. For the more motivationally salient reminder test, participants were required to remind the experimenter to email them their $40 Amazon e-gift card at the end of the study, during the debriefing. The reminder test was scored in a binary manner (0 = no reminder at appropriate time, 1 = reminder).
Source memory test
For the source memory test26, participants saw 32 basic pictures (e.g., black and white line drawings of common objects and animals such as an envelope, hat and horse) and printed words for 2-s each. After a 10-min retention interval, they freely recalled as many items as possible. Next, they were presented with 64 words and were asked to identify whether each was presented previously as a picture or a word or whether the word was not presented earlier. Source memory discrimination indices were computed based on the single source conditional-source identification measure26,27. Specifically, the discrimination index (DI) for picture memory was calculated as [Correctpic/(Correctpic + False Alarmspic)], and the DI for word memory was calculated as [Correctword/(Correctword + False Alarmsword)] where Correct represents the number of pictures/words that were correctly identified, and False Alarms represent the number of pictures/words that were incorrectly identified. Therefore, the DIs for picture and word memory represent the proportion of total pictures/words that were accurately identified.
Deese–Roediger–McDermott false memory paradigm (DRM)
During the DRM28, participants heard six lists of 12 words that each related to one critical lure word which was not on the list. For example, a list might include the words: tired, pillow, bed, dream, and night, while the critical lure word ‘sleep’ that directly relates to each of the words in the list is never presented. For the free recall trial, participants recalled as many words as they could from each list immediately after hearing each. After a 10-min retention interval, participants were presented with six new lists of seven words which contained two old words, one critical lure, two new related words (i.e., words which were related to the theme of an original list, but not as directly related as the critical lure word), and two new unrelated words (i.e., words that are completely unrelated to the theme of the original list). False identification of critical lures, related words, and unrelated words was used to measure false memory.
Temporal order memory test
To assess memory for the sequential order of previous events26, participants were asked to freely recall all the tasks they completed during the testing session, in the order in which they were completed. The total number of tests they correctly recalled was scored as a measure of free recall. Participants’ responses were then individually scored from 0 to 2. Participants received a score of 2 for each task they recalled in the correct sequence. If they switched the order of a task with an adjacent task or recalled a series of tests in the correct order, but in the incorrect location within the larger list, they received a score of 1 for each test. If participants recalled tests in an order that was completely out of sequence, they received a score of 0 for each test. Individual scores were then summed to create a total temporal order memory recall score. Finally, participants were presented with 10 pairs of tasks that they completed and were asked to indicate which of the two tasks they completed first. Participants received a total temporal order recognition score which reflected the number of questions they answered correctly.
Under/overconfidence test
For the under/overconfidence test of meta-cognition (i.e., participants’ awareness of their own knowledge)29, participants were presented with 15 true/false statements (e.g., True or False: Amman is the capital of Jordan), and were asked to indicate whether each statement is true or false and to rate how confident they were in that decision, using a scale ranging from 50% (just guessing), to 100% (absolutely sure). The percentage of true/false questions that were correctly answered was subtracted from the average confidence rating. As such, negative scores reflect under-confidence, and positive scores reflect over-confidence.
Resistance to framing test
The resistance to framing test29 examines whether decisions are impacted by the manner in which problems are framed. The test is divided into two parts; a gain framing trial and a loss framing trial, the order of which were counterbalanced. For the gain framing trial, participants were presented with seven scenarios with two response options framed in terms of gains. For example, in a gain-framing trial, participants would be presented with hypothetical scenarios, such as: “Imagine that the U.S. is preparing for the outbreak of an unusual disease, which is expected to kill 600 people. Two alternative programs to combat the disease have been proposed. Assume that the exact scientific estimates of the consequences of the programs are as follows: If Program A is adopted, 200 people will be saved. If Program B is adopted, there is a 33% chance that 600 people will be saved, and a 67% chance that no people will be saved.” For the loss-framing trial, participants were presented with the same seven scenarios; however, the two response options were framed in terms of losses. For example, participants were presented with the same scenario, but this time were informed that: “If Program A is adopted, 400 people will die. If Program B is adopted, there is a 33% chance that nobody will die, and a 67% chance that 600 people will die. On both trials, participants were asked to indicate their preference between the two options using a 6-point scale. The absolute difference in responses between paired questions was computed and averaged. Lower scores indicate higher resistance to framing.
Consistency in risk perception test
The consistency in risk perception test29 measures participants’ ability to follow rules of probability while judging the likelihood of broader versus narrower risky events. Participants rated the probability of 10 different events occurring in the next 5 years. Events were paired, with one of each pair being broad in nature (e.g., the probability that someone will steal from you in the next 5 years), and the other being narrower in focus (e.g., the probability that someone will break into your home and steal something from you in the next 5 years). The number of times participants appropriately judged the broader event to be more likely than its more narrowly focused pair was scored, with consistent judgements receiving a score of 2 points per pair.
Resistance to sunk cost test
The resistance to sunk cost test29 assesses participants’ tendency to ignore previous investments when making decisions regarding future circumstances. Participants were presented with a series of scenarios and were asked to rate the likelihood that they would select either (1) a response option that is less desirable, but preserves a previous investment, or (2) a response option that is more desirable but involves a loss of a previous investment. For example, “You are in a hotel for one night and paid $6.95 to watch a movie on pay-per-view TV. You then discover there is a movie you would much rather see on one of the free cable channels. Would you be more likely to watch the movie you paid for, or the one on the free cable channel?” Participants were asked to indicate their preference between the two options on a 6-point scale [1 = absolutely prefer the option that preserves a previous investment (e.g., watching the movie you paid for), 6 = absolutely prefer the option that involves a loss of a previous investment (e.g., watching the free movie)]. Scores were averaged with higher scores indicating higher resistance to sunk cost.
Procedures
Figure 1 displays an overview of the study procedures.
Overview of study procedures. Block A contained the source memory test, the under/overconfidence test of decision making, and the first half of the resistance to framing test (gains framing trials). Block B contained the DRM false memory paradigm, the consistency in risk perception test, and the other half of the resistance to framing test (loss framing trials). Block A and B were completely counterbalanced with 10 participants per group per order in order to control for order effects (e.g., fatigue, strength of drug effects).
Preparation
The study was reviewed by our university’s Office of the Attorney General and was approved by the Washington State University Institutional Review Board. The research was conducted in accordance with the Declaration of Helsinki. Participants were recruited through advertisements posted online (e.g., Facebook, Craigslist) as well as via printed flyers in cannabis dispensaries in Washington state where recreational cannabis is legal. Prospective participants completed an online survey to determine their eligibility, obtain basic demographic and cannabis consumption information. Eligible participants were randomly assigned to one of the four groups and were emailed a Zoom link for their testing session. Those in the three cannabis-using groups were informed that we are interested in studying the effects of specific products and were sent a list of products available at local recreational cannabis dispensaries that met criteria for the group to which they were assigned (e.g., participants randomly assigned to the THC flower group were emailed a list of pre-rolled joints with ≥ 20% THC and 0% CBD available at local dispensaries). Prior to testing, participants in the cannabis-using groups purchased a product off the list using their own funds. All participants were required to abstain from using any cannabis products for a minimum of 8 h prior to testing and those in the sober control group were further required to remain sober during the testing session.
Baseline assessment
After initiating the Zoom session, the researcher obtained informed consent online. Participants then indicated how many hours ago they last used cannabis and provided baseline ratings of intoxication, anxiety, mood, and stress using 0 (not at all) to 10 (extremely) rating scales. To ensure random assignment produced groups equivalent in cognitive ability at baseline, participants also completed the WTAR while sober.
Cannabis consumption
Participants in the three cannabis-using groups were asked to show the researcher the cannabis product they purchased for the study. The brand, strain, and cannabinoid content (%THC and %CBD) were recorded. The vast majority of participants used the list they were sent to purchase the product-type to which they had been randomly assigned to use. As described previously, the small number (n = 4) who purchased a product that did not meet the study criteria were excluded from data analyses.
Participants were shown the paced puff procedure30 using screen sharing and were observed while they inhaled their cannabis product. The researcher recorded the number of puffs and duration of inhalations and holds until the participant self-titrated but did not enforce adherence to the paced puff procedure, due to federal legal restrictions.
Cognitive testing
Participants again rated their intoxication, mood, anxiety, and stress using the same 0 to 10 scales. Then, they received instructions for the prospective memory tests. The majority of the remaining tests were divided into two testing blocks (A and B) that were counterbalanced to control for potential order effects. Ten people in each of the four groups completed each order to achieve perfect counterbalancing. Block A began with the source memory test. During the retention interval for this test, participants completed the under/overconfidence test of decision making, and the first half of the resistance to framing test (gain framing trials). At the end of Block A (which required ~ 25 min), participants again rated their intoxication, mood, anxiety, and stress. Block B began with the DRM false memory paradigm. During the retention interval for this test, participants completed the consistency in risk perception test and the other half of the resistance to framing test (loss framing trials). Next, they completed the resistance to sunk cost measure. At the end of Block B (which required ~ 25 min), participants re-rated their intoxication, mood, anxiety, and stress. After both testing blocks were completed, the temporal order memory test was administered. Finally, participants were debriefed and emailed their compensation in the form of a $40 Amazon e-gift card. For all tests, verbal instructions were provided over Zoom along with written instructions presented via the screen sharing option in Zoom. Screen sharing was also used to display word lists, pictures and other stimuli.
Data analysis
Data were screened for univariate outliers defined as scores > 3.29 standard deviations from the mean31. Eight outliers (< 0.5% of the data) were detected and trimmed to one unit higher/lower than the nearest non-outlying value31. The reminder prospective memory test was scored in a binary manner and analyzed using chi-square. One-way analyses of variance (ANOVAs) and a priori planned pairwise LSD comparisons were used to compare the four groups’ performance on the remaining cognitive tests. Mixed factorial ANOVAs were used to compare changes in intoxication over time across the groups. IBM SPSS (version 26) was used to conduct all analyses, with pairwise deletion used for missing data (1% of the total dataset).
