The health and psychological consequences of cannabis use - chapter 7

THIS DOCUMENT HAS BEEN RESCINDED: Chapter 7.4 Effects of chronic cannabis use on cognitive functioning.

Page last updated: 1994

Please note the information in this publication may no longer be current but is retained on our website for historical or research purposes.

<< Previous Chapter | Contents | Next Chapter>>

7. The psychological effects of chronic cannabis use

7.4 Effects of chronic cannabis use on cognitive functioning

Because cannabis use acutely impairs cognitive processes, a concern has arisen that chronic cannabis use may cause chronic cognitive impairment. Such a chronic effect need not necessarily be permanent, but it would persist beyond the elimination of cannabinoids from the body, and hence would be the result of secondary changes induced by cumulative exposure to cannabinoids. Such chronic effects could produce relatively enduring behavioural deficits which presumably reflect changes in brain function.

This chapter deals with the evidence from a variety of different types of study about the cognitive effects of chronic cannabis use. The caveats mentioned in the introduction must be born in mind whilst critically assessing this evidence: many other factors must be controlled in order to confidently attribute any cognitive effects to cannabis use. Among these, the most important are ensuring that the cognitive impairment did not precede cannabis use, and ensuring that the cognitive effects are not the result of the multiple drug use that is especially common among heavy cannabis users (Carlin, 1986).

7.4.1 Clinical observations

Concerns about the cognitive effects of chronic cannabis use during the early 1970s were first prompted by clinical reports of mental deterioration in persons who had used cannabis heavily (at least daily) for more than one year (Fehr and Kalant, 1983). Kolansky and Moore (1971, 1972), for example, reported cases of psychiatric disorder in adolescents and young adults (38 cases) and among adults (13 cases) who had used marijuana at least twice per week. The clinical picture was one of "very poor social judgment, poor attention span, poor concentration, confusion, anxiety, depression, apathy, passivity, indifference and often slowed and slurred speech" (Kolansky and Moore, 1971). Cognitive symptoms included: apathetic and sluggish mental and physical responses; mental confusion; difficulties with recent memory; and incapability of completing thoughts during verbal communication. These symptoms typically began after cannabis use and disappeared within three to 24 months of abstinence. The course and remission of symptoms also appeared to be correlated with past frequency and duration of cannabis smoking. Those with a history of less intensive use showed complete remission of symptoms within six months; those with more intensive use took between six and nine months to recover; while those with chronic intensive use were still symptomatic nine months after discontinuation of drug use.

These clinical reports, similar observations by Tennant and Groesbeck (1972) among hashish smoking US soldiers in West Germany, and a report of cerebral atrophy in young cannabis users (Campbell et al, 1971) excited substantial controversy about the cognitive effects of chronic cannabis use. Critics were quick to object to the lack of objective measures of impairment and the biased sampling from psychiatric patient populations. It was also difficult to rule out alternative explanations of the apparent association between cannabis use and cognitive impairment, namely, that many of these effects either preceded cannabis use, or were the result of other drug use. Whatever their limitations, these clinical reports alerted the community to the possible risks of using cannabis when it was becoming popular among the young in Western countries; they also prompted better controlled empirical research on the issue.

Top of page

7.4.2 Cross-cultural studies

In response to public anxiety about the increase in marijuana use in the late 1960s, the National Institute on Drug Abuse (NIDA) in the United States commissioned three cross-cultural studies in Jamaica, Greece and Costa Rica to assess the effects of chronic cannabis use on cognitive functioning (among other things). The rationale for these studies was that any cognitive effects of chronic daily cannabis use would be most apparent in cultures with a long-standing tradition of heavy cannabis use. Jamaica

Bowman and Pihl (1973) conducted two field studies of chronic cannabis use in Jamaica, one with a small sample of 16 users and 10 controls from rural and semi-rural areas, and the other with a small urban slum sample of 14 users and controls. Users had consumed cannabis daily for a minimum of 10 years (current use of about 23 high potency joints/day), while controls had no previous experience with cannabis. Tests were selected on the basis of having previously been shown to be sensitive to impairment following chronic heavy alcohol use (Bowman and Pihl, 1973). The groups were matched for age, sex, social class, alcohol use, education and "intelligence", but most subjects were illiterate or semi-literate, with an average age of 30. No differences were found between the users and non-users in either study, even when the rural and urban samples were combined.

Soueif (1976b) argued that a null result would be expected according to his hypothesis that cannabis-induced impairments require a minimum level of literacy to be detected. Bowman and Pihl replied that the controls were sufficiently literate to enable any impairment in the users to manifest. Moreover, their study required only a minimum of four hours abstinence prior to testing, which meant that some subjects were still intoxicated at the time of testing. This possibility would have biased the test results in favour of finding poorer performance among the users.

A more extensive study of 60 working class males in Jamaica (Rubin and Comitas, 1975) compared 30 users and 30 non-users matched on age, socioeconomic status and residence. The users who were aged between 23 and 53 years with a mean age of 34 years, had used cannabis for an average of 17.5 years (range seven to 37 years) at around seven joints per day (range one to 24) containing an estimated 60mg of THC. They had not used any other substances other than alcohol and tobacco. While no control subject had used cannabis heavily in recent years, nine were current "occasional" users of cannabis and all but 12 of the controls had some experience with cannabis.

A battery of 19 psychological tests were administered after three days of abstinence, as part of a six-day inpatient stay. The psychological tests included three tests of intellectual and verbal abilities, and 15 neuropsychological tests measuring abilities previously shown to be affected by acute cannabis intoxication. Comparisons of the users and non-users on 47 test scores failed to reveal any consistent significant differences. There were three statistically significant results which were not easily interpreted and were considered chance findings. There was no strong suggestion of differences that failed to be detected because of a small sample size, since the user group scored better than the non-user group on 29 variables, albeit non-significantly.

The interpretation of these null results must be qualified because several factors may have attenuated differences between users and non-users. First, the tests used were not standardised for use in Jamaica. The authors' arguednerability to Drug Abuse. Washington: Am for both users and controls and therefore would not obscure any group differences (Rubin and Comitas, 1975, p111). Second, the Weschler Adult Intelligence Scale (WAIS) subtests may have been too easy or too difficult to allow detection of group differences. Third, the inclusion of cannabis users in the control group may have
Further reduced the chance of detecting group differences. Fourth, the Jamaican sample were primarily farmers, fishermen and artisans from rural areas, or casual urban labourers. The failure of cannabis to impair their cognitive performance does not exclude the possibility that the long-term use of cannabis may impair the performance of persons required to perform at a cognitively more demanding level.

Top of page Greece

The Greek NIDA study (Stefanis et al 1976, 1977) compared the cognitive performance of a sample of 47 chronic hashish users and 40 controls matched for age, sex, education, demographic region, socioeconomic status and alcohol consumption. The subjects were mostly refugees from Asia Minor, residing in a low income, working class area of Athens. The average duration of hashish use was 23 years of 200mg per day. Most users had smoked hashish on the day before testing, and some had smoked several hours before the test session. Controls were slightly better educated than users.

These researchers administered the Weschler Adult Intelligence Scale (WAIS) and Raven's Progressive Matrices to assess general intelligence and mental functioning (Kokkevi and Dornbush, 1977). Subtests of the WAIS were used to evaluate impairment in specific cognitive and perceptual functions. The Raven's test was considered to be a more culture-free assessment of intelligence and was used for reliability and validity purposes. The groups did not differ in global IQ score on either the WAIS or Raven's Progressive Matrices, but non-users obtained a higher verbal IQ score than users. The users' performance was worse than controls on all but one of the subtests of the WAIS, even if not significantly so. Significant differences in performance between the two groups were obtained in three subtests of the WAIS, indicating possible defects in verbal comprehension and expression, verbal memory, abstraction and associative thinking, visual-motor coordination and memorising capacity, and logical sequential thought.

The interpretation of these results was complicated by the lack of a requirement that subjects abstain from hashish prior to testing. Consequently, it was not clear whether the impairments found on these subtests were related to long-term use of hashish, or were due to the persistence of an acute drug effect at the time of testing. Because the differences between verbal and performance IQ were similar in both groups, the authors argued that there was no evidence of deterioration in mental abilities in the hashish users.

Top of page Costa Rica

The NIDA study of chronic heavy cannabis users in Costa Rica was modelled upon the Jamaican project, but with greater sensitivity to cross-cultural issues. It involved an intensive physiological, psychological, sociological and anthropological study of matched pairs of users and non-users (Carter, 1980). Satz, Fletcher and Sutker (1976) compared 41 male long-term heavy cannabis users (9.6 joints per day for 17 years) with matched controls on an extensive test battery designed to assess the impact of chronic cannabis use on neuropsychological, intellectual and personality variables. The educational level of the Costa Rican sample was slightly higher than that of either the Greek or the Jamaican samples, although more than half of the user group had not completed primary school, and both users and non-users had left school at 12 years of age. The users were working class, mostly tradesmen with lower than average income, who reported that they often used cannabis to improve their work performance.

Despite their long duration and heavy use, the Costa Rican users did not differ significantly from controls on any test. Users scored consistently lower, if not significantly so, than non-users on 11 of 16 neuropsychological tests. Although users' performance was poorer, particularly in the mean number of errors made, learning curves were similar for both groups. The authors concluded that there was insufficient evidence for significant impairment of memory function in the chronic cannabis users. Users performed slightly better on six of the 11 WAIS subtests and had a slightly higher verbal and full-scale IQ. There were no correlations between test results and the level of marijuana use.

A 10-year follow-up of the Costa Rican sample was conducted by Page, Fletcher and True (1988). By the time of follow-up, the users had an average 30 years experience with cannabis, but the sample size had dropped to 27 of the 41 original users and 30 of the 41 controls. The test protocol included some of the original tests, as well as additional tests which measured short-term memory and attention, and which had been selected for their sensitivity in detecting subtle changes in cognitive functioning.

No differences were detected on any of the original tests, but three tests from the new battery yielded significant differences between users and controls. In Buschke's Selective Reminding Test, the user group retrieved significantly fewer words from long-term storage than the non-user group, although the groups did not differ on a measure of storage. Users performed more slowly than non-users in the Underlining Test, with particularly poor performance in the most complex subtest. The Continuous Performance Test also revealed users to be slower than controls on measures requiring sustained attention and effortful processing, although there were no differences in performance.

Page et al (1988) interpreted their results as evidence that long-term consumption of cannabis was associated with difficulties in sustained attention and short-term memory. They hypothesised that such tests require more mental effort than the tests used in the original study, and, as such, that long-term users of cannabis experience greater difficulties with effortful processing. This study differs from previous cross-cultural investigations in that it found differences between users and non-users in tests of information processing, sustained attention and short-term memory. Nevertheless, Page et al (1988) emphasised that the differences they found were "quite subtle" and "subclinical", with only a small number of subjects being clinically impaired. Because the differences are so small and subtle, it was difficult to exclude the alternative explanation that the differences were due to acute intoxication or recent use, since 24-hour abstinence was requested but not verified.

Top of page Egypt

Soueif (1971) studied 850 Egyptian hashish smokers and 839 controls obtained from a male prison population which was poorly educated, largely illiterate and of low socioeconomic status. Significant differences were found between users and controls on 10 out of 16 measures of perceptual speed and accuracy, distance and time estimation, immediate memory, reaction time and visual-motor abilities (Soueif, 1971; 1975; 1976a; 1976b). These differences were more marked in those under 25 years and among the best educated urban users.

Soueif's study was subsequently criticised for methodological reasons (Fletcher and Satz, 1977). A major criticism was that the groups differed on a number of variables that were relevant to cognitive performance, including education (with literate non-users being better educated than illiterate users). There were also higher rates of opiate and alcohol use among the cannabis users. Soueif (1977) later reported that in his sample, differences between users and non-users were not explained by education or polydrug use (Soueif, 1977). The validity of these findings remain under doubt, however, because some of the tests used did not have established neuropsychological validity (Carlin, 1986). India

Agarwal et al (1975) studied 40 subjects who had used bhang (a tea-like infusion of cannabis leaves and stems) daily for about five years. These users were less than 45 years of age, and reasonably well educated: none were illiterate and 65 per cent had completed high school. There was no control group, so scores were compared to normative data on the tests used. By comparison with these norms, 18 per cent of the bhang users had memory impairment, 28 per cent showed mild intellectual impairment on an intelligence test (IQs less than 90) and 20 per cent showed substantial cognitive disturbances on the Bender-Gestalt Visuo-Motor Test. Wig and Varma (1977) substantially replicated these results.

Mendhiratta, Wig and Verma (1978) compared 50 heavy cannabis users (half bhang drinkers, half charas smokers of at least 25 days per month for a mean of 10 years) with matched controls. The entire sample was of low socioeconomic status. Tests were administered after 12 hours abstinence which was verified by overnight admission to a hospital ward. The cannabis users reacted more slowly, and performed more poorly in concentration and time estimation. The charas smokers were the poorest performers, showing impaired memory function, lowered psychomotor activity and poor size estimation. A follow-up of 11 of the original bhang drinkers, 19 charas smokers and 15 controls nine to 10 years later (Mendhiratta et al, 1988) showed significant deterioration on several of the tests.

Ray et al (1978) assessed the cognitive functioning of 30 chronic cannabis users (aged 25-46) who had used bhang, ganja or charas for a minimum of 11 times/month for at least five years. They compared their performance to 50 randomly selected non-user controls of similar age, occupation, socioeconomic status and educational background. Few differences were found on tests of attention, visuomotor coordination, or memory. Cannabis users' performance was impaired on one of the subtests of the memory scale. However, the matching of subjects was not rigorous, and the fact that all subjects were illiterate may have produced a floor effect masking differences between groups.

Varma et al (1988) administered 13 psychological tests selected to assess intelligence, memory and other cognitive functions, to 26 heavy marijuana smokers and 26 controls matched on age, education and occupation. The average daily intake of the cannabis users was estimated as 150mg THC, with a frequency of at least 20 times per month, and a mean duration of use 6.8 years (minimum five years). Twelve hours abstinence was ensured by overnight hospitalisation. Cannabis users were found to react more slowly on perceptuomotor tasks, but did not differ from controls on the tests of intelligence. When the scores of all the memory tests were combined, there was no difference between the total scores of cannabis users and controls, although cannabis users scored significantly more poorly on a subtest of recent memory. There were trends toward poorer performance on subtests of remote memory, immediate and delayed recall, retention and recognition.

Top of page Summary

The results of the cross-cultural studies of long-term heavy cannabis users provided at most equivocal evidence of an association between cannabis use and more subtle long-term cognitive impairments. Given that cognitive impairments are most likely to be found in subjects with a long history of heavy use, it is reassuring that most such studies have found few and typically small differences. It is unlikely that the negative results of these studies can be attributed to an insufficient duration or intensity of cannabis use within the samples studied, since the duration of cannabis use ranged between 16.9-23 years, and the estimated amount of THC consumed daily ranged from 20-90mg daily in Rubin and Comitas's Jamaican study to 120-200mg daily in the Greek sample.

The absence of differences is all the more surprising, since a number of factors may have biased these studies toward finding poorer performance among cannabis users. These include: higher rates of polydrug use, poor nutrition, poor medical care, and illiteracy among users; and the failure in many studies to ensure that subjects were not intoxicated at the time of testing. Given the generally positive biases in these studies, it has been argued that if cannabis use did produce cognitive impairment, then these studies should have shown positive results (Wert and Raulin, 1986b).

The force of this argument is weakened by the fact that most of these studies also suffered from methodological difficulties which may have operated against finding a difference. First, the instruments used have been developed and standardised on Western populations. Second, many of these studies were based on small samples of questionable representativeness. Third, a number of studies failed to include a control group, while others used inappropriate controls. Fourth, generalisation of the results of these studies to users in the West or other cultures is difficult, given the predominance of illiterate, rural, older and less intelligent or less educated subjects in these studies. Fifth, the studies were only capable of detecting gross deficits. Sixth, few attempts were made to examine relationships between neuropsychological test performance and frequency and duration of cannabis use.

Despite all these problems, there was nonetheless suggestive evidence of more subtle cognitive deficits. Slower psychomotor performance, poorer perceptual motor coordination, and memory dysfunction were the most consistently reported deficits. In terms of memory function, four studies detected persistent short-term memory and attentional deficits (Page et al, 1988; Soueif, 1976a; Varma et al, 1988; Wig and Varma, 1977), while three failed to detect such deficits (Bowman et al, 1973; Satz et al, 1976; Mendhiratta et al, 1978). The measures of short-term memory were often inadequate, failing to determine which processes may be impaired (e.g. acquisition, storage, encoding, retrieval) and often excluded higher mental loads and conditions of distraction. A proper evaluation of the complexity of effects of long-term cannabis use on higher cognitive functions requires greater specificity in the selection of assessment methods, as well as the use of more sensitive tests.

Top of page

7.4.3 Studies of young Western users

A number of studies have been conducted on the cognitive performance of American or Canadian cannabis users. These samples have generally been young and well educated college students with relatively short-term exposure to cannabis, by comparison with the long history of use among chronic users in the cross-cultural studies.

In one of the earliest studies, Hochman and Brill (1973) surveyed 1,400 college students and compared the performance of non-users (66 per cent), occasional users (26 per cent) and chronic users (9 per cent: defined as having used three times/week for three years or, had used daily for two years). They found no relationship between either frequency or duration of use and academic achievement. In about 1 per cent of marijuana users there was impaired ability to function. In a follow-up of the original sample over two consecutive years (1971: N=1,133; 1972: N=901), Brill and Christie (1974) compared non-users, occasional users (<2 times per week), frequent (2-4/week), and regular users (˛5/week) by a self-report questionnaire. The majority of users reported no effect of cannabis use on sychosocial adjustment. A small proportion (12 per cent) who reported that their academic performance had declined were likely to have either reduced their frequency of use or quit. There were no significant differences between users, non-users or former users in grade point average.

A series of studies conducted since then has largely confirmed the results of Hochman and Brill's studies. Grant et al (1973), for example, studied the effects of cannabis use on psychological test performance on eight measures from the Halstead-Reitan Battery among medical students. They found no differences between 29 cannabis users (of median duration, four years and median frequency of use, three times per month) and 29 age and intelligence matched non-users on seven of the eight measures. The failure to find any difference in sensory-motor integration or immediate sensory memory was later replicated by Rochford, Grant and LaVigne (1977) in a comparison of 25 users (of at least 50 times over a mean 3.7 years) and 26 controls matched on sex, age and scholastic aptitude scores.

Weckowicz and Janssen (1973) compared eleven male college students who smoked cannabis three to five times per week for at least three years with non-users who were matched on age, education and socioeconomic and cultural backgrounds. They were assessed on a variety of tests of cognitive function. Users performed better than controls on eight of the 11 cognitive tests but performed more poorly on one which suggested that chronic use may affect sequential information processing. Otherwise, there was no evidence of gross impairment of cognitive functioning. Weckowicz, Collier and Spreng (1977) largely replicated these findings in a comparison of 24 heavy smokers (at least daily for three years) belonging to the "hippie subculture" with non-user controls matched for age, education, and social background. Similar results were reported by Culver and King (1974) in a comparison of the neuropsychological performance of three groups of undergraduates (N=14) from classes in two successive years: marijuana users (at least twice/month for 12 months); marijuana plus LSD users (LSD use at least once/month for 12 months); and non-drug users. There were no consistent differences between the groups across the different years.

Top of page

In 1981, Schaeffer et al (1981) reported no impairment of cognitive function in one of the first studies of a prolonged heavy cannabis using population in the United States, who used the drug for religious reasons. They assessed 10 long-term heavy users of ganja, aged between 25 and 36 years, all of whom were Caucasian, and had been born, raised and educated in the USA. All had smoked between 30gm and 60gm of marijuana (>8 per cent THC) per day for a mean of 7.4 years. They had not consumed alcohol or other psychoactive substances. This study was also used a laboratory test to detect recent ingestion of cannabis. Schaeffer et al reported that at the time of testing, all subjects had at least 50ng/ml cannabinoids in their urines. Performance on a series of tests of cognitive ability was compared with the standardised-normative information available for each test. Overall, WAIS IQ scores were in the superior to very superior range, and the scores of all other tests were within normal limits. Despite the heavy and prolonged use of cannabis, there was no evidence of impairment in the cognitive functions assessed, namely, language function, non-language function, auditory and visual memory, remote, recent and immediate memory, or complex multimodal learning.

Carlin and Trupin (1977) assessed 10 normal subjects (mean age 24 years) who smoked marijuana daily for at least two years (mean five years) and who denied other drug use. They administered the Halstead Neuropsychological Test Battery after 24 hours abstinence. No significant impairment was found by comparison with non-smoking subjects matched for age, education and full-scale IQ. Cannabis users performed better on a test sensitive to cerebral impairment than non-users.

Not all studies have produced null results, however. Gianutsos and Litwack (1976), for example, compared the verbal memory performance of 25 cannabis smokers who had used for two to six years and at least twice/week for the last three months, with 25 non-smokers who had never smoked cannabis. Subjects were drawn from an undergraduate university student population and were matched on age, sex, year at university, major and grade point average. Cannabis users reported that they had not smoked prior to testing, although the length of abstinence was not reported. Cannabis users recalled significantly fewer words overall than non-users, and the difference in performance increased as a function of the number of words they were required to learn.

Entin and Goldzung (1973) also found evidence of impairment in two studies of the residual impact of cannabis use on memory processes. In the first study, verbal memory was assessed by the use of paired-associate nonsense syllable learning lists. Twenty-six cannabis users (defined as daily for at least six months) were compared to 37 non-users drawn from a student population. Cannabis users scored significantly more poorly on both free recall (the number of words recalled after a delay) and on acquisition, measured as improvement in recall over repeated trials. In the second study, verbal and numerical memory were tested by the presentation of word lists, interspersed with arithmetic problems prior to recall. Cannabis users (N=37) recalled significantly fewer words than non-users (N=37), but did not differ from controls on arithmetic test scores. These findings were interpreted as residual impairment of both the acquisition and recall phases of long-term memory processes. The authors attributed the impairments to either an enduring residual pharmacological effect on the nervous system, or to an altered learning or attention pattern due to repeated exposure to cannabis.

Top of page Summary

The results of these empirical studies served to allay fears that cannabis smoking caused gross impairment of cognition and cerebral function in young adults. The lack of consistent findings failed to support Kolansky and Moore's (1971, 1972) clinical reports of an organic impairment, although some critics (e.g. Cohen, 1982) argued that the value of these studies was weakened by their small sample sizes and the fact that by studying college students, they had sampled from a population unlikely to contain many impaired persons. On Cohen's hypothesis, the younger, brighter college cannabis users may reflect the survivors, whereas Kolansky and Moore sampled the casualties. Such an hypothesis conflicts with the explanations provided for the failure to find impairment in the cross cultural studies. Soueif's hypothesis, for example, was that the lower the non-drug level of proficiency, the smaller the size of functional deficit associated with drug usage. This would imply maximal differences at the high end of cognitive ability.

A more pertinent explanation for the lack of impairment is that the duration of cannabis use in these samples was quite brief, generally less than five years. It has been argued that cannabis has not been smoked long enough in Western countries for impairments to emerge. Further, when psychometric testing was used as a metric of cognitive function as opposed to self-report questionnaires, sample sizes were often too small to permit the detection of all but very large differences between groups.

Not all studies found negative results. A small number of studies did find significant impairments in their cannabis users. It is noteworthy that these studies selected tests to assess a specific cognitive function (memory), and attempted to determine the specific stages of processing where dysfunction occurred. Entin and Goldzung (1973), for example, found that users were impaired on both verbal recall and acquisition of long-term storage memory tasks, but not on arithmetic manipulations which require short-term storage of information.

Top of page

7.4.4 Controlled laboratory studies

A different approach to the investigation of the cognitive consequences of chronic cannabis use is to examine the cognitive effects of daily cannabis use over periods of weeks to months. Such studies have attempted to control for variation in quantity, frequency and duration of use, as well as other factors such as nutrition and other drug use, by having subjects reside in a hospital ward while receiving known quantities of cannabis. All such studies employed pre- and post-drug observation periods. Because of their expense, sample sizes in these studies have been small and the duration of cannabis administration has ranged from 21 to 64 consecutive days.

Dornbush et al (1972) administered 1g of marijuana containing 14mg THC to five regular smokers (all healthy young students) for 21 consecutive days. The subjects were tested immediately before and 60 minutes after drug administration. Data were collected on short-term memory and digit symbol substitution tests. Performance on the short-term memory test decreased on the first day of drug administration but gradually improved until by the last day of the study, performance had returned to baseline levels. On the post-experimental day baseline performance was surpassed. Performance on the digit symbol substitution test was unaffected by drug administration and also improved with time, suggesting a practice effect.

Mendelson, Rossi and Meyer (1974) reported a 31-day cannabis administration study in which 20 healthy, young male subjects (10 casual and 10 heavy users, mean age 23) were confined in a research ward and allowed 21 days of ad libitum marijuana smoking. Psychological tests were administered during a five-day drug-free baseline phase, the 21 day smoking period and a five-day drug-free recovery phase. Acute and repeat dose effects of marijuana on cognitive function were studied with a battery of psychological tests known to be sensitive to organic brain dysfunction. There was no overt impairment of performance prior to or following cannabis smoking, nor was there any difference between the performance of the heavy and the casual users. Short-term memory function, as assessed by digit span forwards and backwards, was impaired while intoxicated, and there was a relationship between performance and time elapsed since smoking.

Similar failures to detect cognitive effects have been reported by three other groups of investigators. Frank et al (1976) assessed short-term memory and goal directed serial alternation and computation in healthy young males over 28 days of cannabis administration. Harshman et al (1976) and Cohen (1976) conducted a 94-day cannabis study in which 30 healthy moderate to heavy male cannabis users, aged 21-35, were administered on average 5.2 joints per day (mean 103mg THC, range 35-198mg) for 64 days, and were assessed on brain hemisphere dominance before, during and after cannabis administration. Psychometric testing was not employed, but subjects were given two work assignments with financial incentive: a "psychomotor" task involving the addition of two columns of figures on a calculator, and a "cognitive task" of learning a foreign language. No long-term impairments were detected with these somewhat inadequate assessment methods.

Top of page Summary

The experimental studies of daily cannabis usage for periods of up to three months in young adult male volunteers have consistently failed to demonstrate a relationship between marijuana use and neuropsychological dysfunction. This is not surprising given the short periods of exposure to the drug in these studies. Furthermore, since subjects served as their own controls, and had all used cannabis for at least one year prior to the study, it would be surprising if a few additional weeks of cannabis use produced any significant decrements in performance.

7.4.5 Recent research

The equivocal results of the early investigations into long-term effects of cannabis on cognitive function led to something of a hiatus in research on the cognitive effects of cannabis in the 1980s. Although the accumulated evidence indicated that cannabis did not severely affect intellectual functioning, uncertainty remained about more subtle impairments. Their study required advances in methodology and assessment techniques which were made in the field of cognitive psychology and neuropsychology in the 1980s. Modern theories of cognition, memory function and information processing were developed, as were more sensitive measures of cognitive processes. By the late 1980s, interest in the cognitive effects of cannabis revived at a time when cannabis had been widely used for more than 15 years, its use was widespread and initiated at a progressively younger age among young Americans.

Research from the late 1980s through the 1990s improved upon the design and methodology of previous studies by using adequate control groups, verifying abstinence from cannabis prior to testing, and precisely measuring the quantity, frequency and duration of cannabis use. In addition, greater attention was paid to investigating specific cognitive processes and relating impairments in them to the quantity, frequency and duration of cannabis use.

The greater specificity in study focus was made possible by accumulating evidence that cannabis primarily exerts its effect upon those areas of the brain responsible for attention and memory. Miller and Branconnier (1983), for example, reviewed the literature and concluded that impaired memory was the single most consistently reported psychological deficit produced by cannabinoids acutely, and the most consistently detected impairment in long-term cannabis use. Intrusion errors were one of the most robust type of cannabis-induced memory deficits in both recall and recognition (Miller and Branconnier, 1983). Such errors involve the introduction of extraneous items, word associations or new material during free recall of words, or the false identification of previously unseen items in recognition tasks. Miller and Branconnier conjectured that these intrusion errors occurred because cannabis users were unable to exclude irrelevant associations or extraneous stimuli during concentration of attention, a process in which the hippocampus plays a major role. The finding of high densities of the cannabinoid receptor in the cerebral cortex and hippocampus (Herkenham et al, 1990) supports the hypothesis that cannabinoids are involved in attentional and memory processes.

Top of page Studies of long-term adult users

Solowij et al (1991; 1992; 1993) conducted a series of studies of the effects of long-term cannabis use on specific stages of information processing. In keeping with Miller and Branconnier's hypothesis, Solowij et al assessed the integrity of attentional processes in long-term cannabis users using a combination of performance and brain event-related potential measures. Event-related potential (ERP) measures are sensitive markers of covert cognitive processes underlying overt behaviour; the amplitude and latency of various ERP components have been shown to reflect various stages of information processing.

Solowij et al, (1991) studied a small and heterogeneous group of long-term cannabis users (N=9), aged 19-40, who had used cannabis for a mean of 11.2 years at the level of 4.8 days per week. The cannabis users were matched on age, sex, years of education and alcohol consumption with nine non-user controls who had either never used or had limited experience with cannabis (maximum use 15 times). Strict exclusion criteria were applied to any subjects with a history of head injury, neurological or psychiatric illness, significant use of other drugs, or high levels of alcohol consumption. The groups did not differ in premorbid IQ, as estimated by the NART score (Nelson, 1984).

Subjects were instructed to abstain from cannabis and alcohol for 24 hours prior to testing and two urine samples were analysed to ensure that subjects were not acutely intoxicated at the time of testing. Subjects completed a multidimensional auditory selective attention task in which random sequences of tones varying in location, pitch and duration were delivered through headphones while brain electrical activity (EEG) was recorded. They were instructed to attend to a particular ear and a particular pitch, and to respond to the long duration tones with a button press. This procedure enabled an examination of the brain's response to attended and unattended tones.

Cannabis users performed significantly more poorly than controls, with fewer correct detections, more errors and slightly longer reaction times. Analysis of the ERP measures showed that cannabis users had reduced P300 amplitudes compared to controls, reflecting dysfunction in the allocation of attentional resources and stimulus evaluation strategies. Further, cannabis users showed an inability to filter out irrelevant information, while controls were able to reject this irrelevant information from further processing at an early stage. These results suggested that long-term cannabis use impairs the ability to efficiently process complex information.

Solowij et al (1992; 1993) conducted a second study with a larger sample to examine the relationships between degree of impairment and the frequency and duration of use. Thirty-two cannabis users recruited from the general community were split into four groups of equal size (N=8) defined by frequency (light: ˇ twice/week vs heavy: ˛ three times/week) and duration (short: 3-4 years vs long: ˛ five years) of cannabis use. The mean number of years of use for the long duration users was 10.1, and 3.3 for short duration users (range three to 28 years). The mean frequency of use was 18 days per month for the heavy group and six for the light group (range: once/month to daily use). Subjects were matched to a group of non-user controls (N=16) as in the first study, and a similar methodology was employed.

Top of page

Once again cannabis users performed worse than the controls, with the greatest impairment observed in the heavy user group, thereby replicating the earlier ERP findings. In addition, different cognitive processes were differentially affected by frequency and duration of cannabis use. The long duration user group showed significantly larger processing negativity to irrelevant stimuli than did short duration users and controls, who did not differ from each other. There were no differences between groups defined on frequency of use. A significant correlation between the ERP measure and duration of cannabis use indicated that the ability to focus attention and filter out irrelevant information was progressively impaired with the number of years of use, but was unrelated to frequency of use. Frequency of use affected the speed of information processing, as reflected in a delayed P300 latency in the heavy user group compared to light users and controls. There was a significant correlation between P300 latency and increasing frequency of use, while this measure was unrelated to duration of use.

These results suggest that different mechanisms underlie the short-term and long-lasting actions of cannabinoids. The slowing of information processing suggests a chronic build up of cannabinoids, and reflected a residual effect which could be eliminated by reducing the frequency of use. The inability to focus attention and reject irrelevant information possibly reflected long-term changes at the cannabinoid receptor site. The consequences of these impairments may be apparent in high levels of distractability when driving, operating complex machinery, and learning in the classroom situation, and interference with efficient memory and general cognitive functions.

Solowij et al also conducted specific analyses to disentangle the relationship between duration of cannabis use and age. The results of these analyses indicated that impairment was greatest in younger subjects. Further, the studies demonstrated the insensitivity of performance measures to cannabinoid effects, emphasising the need to use more sensitive measures to examine otherwise inaccessible, covert cognitive processes.

Supportive evidence has emerged from a project funded by the National Institute on Drug Abuse (NIDA) in the U.S. (principal investigator F. Struve) that investigated persistent central nervous system sequelae of chronic cannabis exposure. This research, which has focused upon quantitative EEG, has found evidence of larger changes in EEG frequency, primarily in frontal-central cortex, in daily cannabis users of up to 30 years duration compared to short-term users and non-users (e.g. Struve et al, 1993). The results also suggest a dose-response relationship between EEG changes and the total cumulative exposure (duration in years) of daily cannabis use which may indicate organic changes. The major limitation of this research is that changes in frequency of EEG spectra have not been shown to be related to cognitive events.

One study from this research group has used cognitive event-related potential measures. It found smaller P2 and N2 amplitudes in long-term cannabis users (>15 years) compared to moderate users (of three to six years). Cannabis users overall showed significantly smaller auditory and visual P300 amplitudes than controls, but no significant latency differences (Straumanis et al, 1992). Unfortunately, this study has only been reported in abstract form, and results have not been examined as a function of frequency of cannabis use.

This research group has also assessed cognitive functioning by neuropsychological tests (e.g. Leavitt et al, 1991; 1992; 1993). These investigations have been well controlled. Subjects were extensively screened for current or past psychiatric or medical disease or CNS injury, and underwent extensive drug history assessments, with eight weeks of twice weekly drug screens. Groups were matched for age and sex. Daily cannabis users who had at least three years to six years of use were compared to a group who had used for six to 14 years, a group who had used on a daily basis for 15 years or more, and a non-user control group. Sample sizes varied from study to study, but averaged 15 per group.

Top of page

An extensive battery of psychological tests included measures of simple and complex reaction time, attention and memory span, language and comprehension tasks, construction, verbal and visual learning and memory, and mental abilities such as concept formation and logical reasoning. The effects of age and education have been statistically controlled for by multiple regression. Preliminary analyses have shown a dose-response relationship between test performance and intensity of cannabis use, with the best performance characterising controls, followed by the daily cannabis users, and the worst mean scores occurring in the very long-term group (Leavitt et al, 1991; 1992; 1993; Leavitt, personal communication). Tests sensitive to mild cortical dysfunction were those most affected in the long-term user groups.

The authors acknowledge that small sample sizes dictate caution, and that there were no data available to assess premorbid cognitive capacity of these subjects. Nevertheless, the results suggested that long duration users seem to process some kinds of information more slowly than non-users, and that the effects of long-term cannabis use are most likely to surface under conditions of moderately heavy cognitive load.

One crucial requirement for evaluating the performance of chronic marijuana users is comparison with an appropriately matched group of non-using subjects. Although the studies described have made substantial progress in this regard, one concern remains that some of these impairments may have been present in the cannabis users prior to their cannabis use. Block et al (1990) used scores on the Iowa Tests of Basic Skills collected in the fourth grade of grammar school as a measure of premorbid cognitive ability. Block et al matched their user and non-user samples on this test to ensure that they were comparable in intellectual functioning before they began using marijuana. The study aim was to determine whether chronic marijuana use produced specific cognitive impairments, and if so, whether these impairments depend on the frequency of use. Block and colleagues assessed: 144 cannabis users, 64 of whom were light users (one to four/week for 5.5 years) and 80 heavy users (˛five/week for 6.0 years), and compared them with 72 controls. Subjects were aged 18-42. Twenty-four hours of abstinence were required prior to testing.

Subjects participated in two sessions. In the first session they completed the 12th grade version of the Iowa Tests of Educational Development, which emphasise basic, general intellectual abilities and academic skills and effective utilisation of previously acquired information in verbal and mathematical areas. In the second session, subjects were administered computerised tests that emphasise learning and remembering new information, associative processes and semantic memory retrieval, concept formation and psychomotor performance. These tasks had been previously shown to be sensitive to the acute and chronic effects of cannabis, and to relevant skills required in school and work performance. The results showed that heavy users who were matched to controls on fourth-grade Iowa scores, showed impairment on two tests of verbal expression and mathematical skills when tested on the 12th-grade Iowa test. No results have been reported to date from the computerised tests.

Top of page Studies in children and adolescents

A very different approach to assessing the long-term consequences of exposure to cannabis has been taken in a well controlled longitudinal study of children who were exposed to cannabis in utero (Fried, 1993). The levels of exposure to cannabis in the sample were approximately as follows: 60 per cent of the mothers used cannabis irregularly, 10 per cent reported smoking two to five joints per week, and 30 per cent smoked a greater amount during each trimester of pregnancy. Prenatal exposure to cannabis was associated with high pitched cries, disturbed sleep cycles, increased tremors and exaggerated startles in response to minimal stimulation in newborn to 30-day-old babies. The babies showed poorer habituation to visual stimuli, consistent with the sensitivity of the visual system to the teratogenic effects of cannabis demonstrated in rhesus monkeys and rats. Fried hypothesised that exposure to cannabis may affect the rate of development of the central nervous system, slowing the maturation of the visual system. This hypothesis was supported by visual evoked potential studies of the children at four years of age, when children who had been exposed to cannabis in utero showed greater variability and longer latency of the evoked potential components, indicating immaturity in the system.

From one to three years of age, no adverse effects of prenatal exposure were found. At two years it appeared that the children were impaired on tests of language comprehension, but this effect did not persist after controlling for other factors such as ratings of home environment. At four years of age, however, the children of cannabis using mothers were significantly inferior to controls on tests of verbal ability and memory. The explanation for the failure to detect impairments in the preceding age group was that the degree and types of deficits observed may only be identifiable when cognitive development has proceeded to a certain level of maturity.

At five and six years of age, the children were not impaired on global tests of cognition and language. By age six, however, there was a deficit in sustained attention on a task that differentiated between impulsivity and vigilance. Fried proposed that "instruments that provide a general description of cognitive abilities may be incapable of identifying nuances in neurobehavior that may discriminate between the marijuana-exposed and non-marijuana exposed children" (p332). He suggested the need for tests which examine specific cognitive characteristics and strategies, such as the test of sustained attention. Fried concluded that cannabis "may affect a number of neonatal behaviours and facets of cognitive behavior under conditions in which complex demands are placed on nervous system functions".

Top of page

The effects of long-term cannabis use on adolescents have not been adequately addressed. This issue is of greater relevance with an increase in the prevalence of cannabis use among adolescents and young adults in Western society. In the first study of its kind with adolescents, Schwartz et al (1989) reported the results of a small controlled pilot study of persistent short-term memory impairment in 10 cannabis-dependent adolescents (aged 14-16 years). Schwartz's clinical observations of adolescents in a drug-abuse treatment program suggested that memory deficits were a major problem, which according to the adolescents persisted for at least three to four weeks after cessation of cannabis use. His sample was middle-class, North American, matched for age, IQ and history of learning disabilities with 17 controls, eight of whom were drug abusers who had not been long-term users of cannabis, and another nine whom had never abused any drug. The cannabis users consumed approximately 18g per week, smoking at a frequency of at least four days per week (mean 5.9) for at least four consecutive months (mean 7.6 months). Subjects with a history of excessive alcohol or phencyclidine use were excluded from the study. Cannabinoids were detected in the urines of eight of the 10 users over two to nine days.

Users were initially tested between two and five days after entry to the treatment program, this length of time allowing for dissipation of any short-term effects of cannabis intoxication on cognition and memory. Subjects were assessed by a neuropsychological battery which included the Wechsler Intelligence Scale for Children, and six tests "to measure auditory/verbal and visual/spatial immediate and short-term (delayed) memory and praxis (construction ability)" (p1215). After six weeks of supervised abstinence with bi-weekly urine screens for drugs of abuse, they were administered a parallel test battery.

On the initial testing, there were statistically significant differences between groups on two tests: cannabis users were selectively impaired on the Benton Visual Retention Test and the Wechsler Memory Scale Prose Passages. The differences were smaller but were still detectable six weeks later. Cannabis users committed significantly more errors than controls initially on the Benton Visual Retention Test for both immediate and delayed conditions, but differences in the six-week post-test were not significant. Users scored lower than controls on both immediate and delayed recall in the Wechsler Memory Prose Passages Test in both test sessions. The fact that there was a trend toward improvement in the scores of cannabis users suggests that the deficits observed were related to their past cannabis use and that functioning may return to normal following a longer period of abstinence.

Schwartz's study was the first controlled study to demonstrate cognitive dysfunction in cannabis-using adolescents with a relatively brief duration of use. The implications of these results are that young people may be more vulnerable to any impairments resulting from cannabis use. Unfortunately, like many of its predecessors, Schwartz's team made little effort to interpret the significance of the selectivity of their results. There was nothing to indicate which specific elements of memory formation or retrieval were disrupted.

Top of page

7.4.6 Discussion

Previous reviewers have generally concluded that there is insufficient evidence that cannabis produces long-term cognitive deficits (e.g. Wert and Raulin, 1986a; 1986b). This is a reasonable conclusion when gross deficits are considered. However, the findings from recent, more methodologically rigorous studies provide evidence of subtle cognitive impairments which appear to increase with duration of cannabis use. The evidence suggests that impairment on some neuropsychological tests may become apparent only after 10-15 years of use, but very sensitive measures of brain function are capable of detecting specific impairments after five years of use.

Impairments appear to be specific to the organisation and integration of complex information, involving various mechanisms of attention and memory processes. The similarity between the kinds of subtle impairments associated with long-term cannabis use and frontal lobe dysfunction is becoming more apparent (e.g. short-term memory deficits, increased susceptibility to interference, lack of impairment on general tests of intelligence or IQ). Frontal lobe function is difficult to measure, as indicated by the fact that patients with known frontal lobe lesions do not differ from controls on a variety of neuropsychological tests (Stuss, 1991), so the difficulty of assessing frontal lobe functions is not unique to research into the long-term effects of cannabis.

One of the functions of the frontal lobes is the temporal organisation of behaviour, a key process in efficient memory function, self-awareness and planning. The frontal lobe hypothesis of impairments due to long-term use of cannabis is consistent with the altered perception of time demonstrated in cannabis users and with cerebral blood flow studies which demonstrate greatest alterations in the region of the frontal lobes (see Section 7.5 brain damage).

The equivocal results of previous studies may be due primarily to poor methodology and insensitive test measures. Wert and Raulin (1986b) had rejected the possibility that tests used previously were not sensitive enough to detect impairments, on the grounds that the same tests have demonstrated impairment in alcoholics and heavy social drinkers. However, the cognitive deficits produced by chronic alcohol consumption may be very different to those produced by cannabis. The mechanisms of action of the two substances are different, with cannabis acting on its own specific receptor, and Solowij et al (1993) showed that the attentional impairments detected in their long-term cannabis users were not related to their alcohol consumption. Furthermore, tests may have been selected inappropriately because they were previously shown to be affected by acute intoxication, when the consequences of chronic use may be very different. A priority for future research is the identification of specific mechanisms of impairment by making direct comparisons with the effects of a variety of other substances.

Top of page

Recent research has aimed at identifying specific cannabis effects by using strict exclusion criteria, and carefully matching control groups to ensure that any deficits observed are attributable to cannabis. However, interactions between the effects of long-term cannabis use concurrent with use of other substances need to be further explored. Subjects have also been excluded if they have had a history of childhood illness, learning disabilities, brain trauma or other neurological or psychiatric illness. The effects of long-term cannabis use on such individuals may be worthy of further investigation.

Cognitive deficits may not be an inevitable consequence of cannabis use. The long-term effects of cannabis on healthy individuals may differ from those in individuals with co-existing mental illness or pre-existing cognitive impairments. On the other hand, some individuals appear to function well even in cognitively demanding occupations despite long-term cannabis use. Wert and Raulin (1986b) suggested that some individuals may adapt and overcome some forms of cognitive impairment by a process of relearning.

When users and non-users are compared, differences may not always reach statistical significance due to large individual variability, particularly when small sample sizes are used. Carlin (1986) proposed that "studies that rely upon analysis of central tendency are likely to overlook impairment by averaging away the differences among subjects who have very different patterns of disability". Individual differences in vulnerability to the acute effects of cannabis are well recognised and are likely to be a factor in determining susceptibility to a variety of cognitive dysfunctions associated with prolonged use of cannabis.

There has been no research designed to identify individual differences in susceptibility to the adverse effects of cannabis. A susceptibility may be due to structural, biochemical or psychological factors, or as Wert and Raulin suggested, to lack of the "cerebral reserve that most of us call on when we experience mild cerebral damage", for example, after a night of heavy drinking. Wert and Raulin suggested that prospective studies are the ideal way to identify those subjects who show real impairment in functioning by comparing pre- and post- cannabis performance scores. However, even in a retrospective design it is possible to compare the characteristics of subjects who show impairment with those who do not, thereby identifying possible risk factors. Insufficient consideration has been given to gender, age, IQ and personality differences in the long-term consequences of cannabis use.

Top of page

Almost all of the studies reviewed have been retrospective studies of naturally occurring groups (users vs. non-users). Although the matching of control groups has become more stringent, and attempts to obtain estimates of premorbid functioning have increased, prospective studies where each subject is used as his/her own control would eliminate the possibility of cannabis users having demonstrated poorer performance before commencing their use of cannabis. A longitudinal study in which several cohorts at risk for drug abuse are followed over time would be the best way to assess the detrimental effects of long-term cannabis use on cognition and behaviour.

Given the growing prevalence of cannabis use, and proposals to reduce legal restrictions on cannabis use, it is essential that research into cognitive effects of long-term cannabis use continues. according to US survey data (Deahl, 1991), more than 29 million people in the United States may be using cannabis, and more than seven million of these use on a daily basis. While there is some controversy surrounding the issue, it seems likely that the potency of cannabis has increased over the years, as more potent strains have been developed for the black market. Increased THC potency combined with decreased age of first use may result in the more marked cognitive impairments in larger numbers of individuals in the future.

Future research should adhere to rigorous methodology. This should include the use of the best available techniques for detecting cannabinoids in the body to provide greater precision in the investigation of the effects of abstinence on performance. This would permit a distinction to be made between those impairments which are residual, and likely to resolve with abstinence over time, from those of a more enduring or chronic nature, which would be associated with cumulative exposure.

Given that recent research has identified cognitive impairments that are associated with cumulative exposure, it is a priority to investigate the extent and rate of recovery of function following cessation of cannabis use. Furthermore, the parameters of drug use require careful scrutiny in terms of evaluating how much cannabis must be smoked and for how long, before impairments are manifest in different kinds of individuals. One of the problems in assessing the cannabis literature is the arbitrariness with which various groups of users have been described as "heavy", "moderate" or "light", "long-term", "moderate" or "short-term".

The use of very sensitive measures of cognitive function is important for the detection of early signs of impairment, which may permit a harm minimisation approach to be applied to cannabis use. With further research, it may be possible to specify levels of cannabis use that are "safe", "hazardous" and "harmful" in terms of the risk of cognitive impairment. Further research examining the consequences of cannabis use in comparison to other substances could provide users with the ability to make an informed decision about whether or not to use the drug, and if they use, how much and how often to use.

Top of page

7.4.7 Conclusion

The weight of evidence suggests that the long-term use of cannabis does not result in any severe or grossly debilitating impairment of cognitive function. However, there is clinical and experimental evidence which suggests that long-term use of cannabis produces more subtle cognitive impairments in specific aspects of memory, attention and the organisation and integration of complex information. While these impairments may be subtle, they could potentially affect functioning in daily life. The evidence suggests that increasing duration of use leads to progressively greater impairment. It is not known to what extent such impairment may recover with prolonged abstinence.

It is apparent that not all individuals are affected equally by prolonged exposure to cannabis. Individual differences in susceptibility need to be identified and examined. For those who are dysfunctional, there is a need to develop appropriate treatment programs which address the subtle impairments in cognition and work toward their resolution. There has been insufficient research on the impact of long-term cannabis use on cognitive functioning in adolescents and young adults, or on the effects of chronic use on the cognitive decline that occurs with normal aging. Gender differences have not been examined to date and may be important, given that such differences have become apparent in differential responses to alcohol.

Future research should aim to identify with greater specificity those aspects of cognitive functioning which are affected by long-term use of cannabis, and to examine the degree to which they are reversible. There is converging evidence that dysfunction due to chronic cannabis use lies in the higher cognitive functions that are subserved by the frontal lobes and which are important in organising, manipulating and integrating a variety of information, and in structuring and segregating events in memory.

Until better measures have been developed to investigate the subtleties of dysfunction produced by chronic cannabis use, cannabis may be viewed as posing a lesser threat to cognitive function than other psychoactive substances such as alcohol. Nevertheless, the fact remains that in spite of its illegal status, cannabis use is widespread. We therefore have a continuing responsibility to minimise drug-related harm by identifying potential risks, subtle though they may be, and communicating the necessary information to the community.