Hyper-priming in cannabis users: A naturalistic study of the effects of cannabis on semantic memory function

Abstract

Psychotic symptoms have theoretically been linked to semantic memory impairments in patients with schizophrenia. Little is known of the effects of cannabis, the world's most popular illicit drug, on semantic memory and whether they are linked to the psychotomimetic states elicited by the drug. Thirty-six cannabis users were tested whilst under the influence of cannabis. They were then tested again when not intoxicated and compared with 38 non-drug using controls. Semantic memory was assessed using a semantic priming task with a long and short stimulus onset asynchrony (SOA) to differentiate automatic and controlled processing. Under the influence of cannabis, users showed increases in both automatic semantic priming and schizotypal symptoms compared with controls. When abstinent, cannabis users exhibited hyper-priming at long SOAs. Cannabis users did not differ from controls in either trait schizotypy or state schizotypy when not intoxicated. Acute cannabis use increases schizotypyal symptoms and may increase automatic semantic priming in recreational users of this drug. When drug-free, cannabis users did not differ from controls in schizotypy but did show hyper-priming at the long SOA. The acute increase in automatic semantic priming may be one factor contributing to the psychotomimetic effects of cannabis.

Introduction

Cannabis is the most widely used illegal recreational drug: amongst all recreational drugs its use is second only to that of alcohol. The British Crime Survey reported that amongst young people (aged 16–24), 30% of men and 18% of women said they had used cannabis during 2004/5 (British Crime Survey, 2006). There have been growing concerns recently over the consequences of long-term cannabis use for cognition and mental health, particularly in light of claims that the drug may induce psychosis in some susceptible individuals (e.g. Di Forti et al., 2007). Acute cannabis intoxication is known to produce memory impairments (for a review, see Ranganathan and D'Souza, 2006). There are many chemical compound constituents of the cannabis plant, the most psychoactively potent of which has been identified as Δ-9-tetrahydrocannabinol (THC). The memory-impairing effects of an acute dose of cannabis are thought to relate to the action of THC at the Cannabinoid 1 (CB1) receptor, which is found at particularly high densities in the hippocampus, cerebellum and basal ganglia. Evidence of memory deficits following long-term cannabis use is less convincing, with verbal memory deficits found in some studies (Dafters et al., 2004, Messinis et al., 14-3-2006, Block and Ghoneim, 1993) but not others (Pope et al., 2001, Verdejo-Garcia et al., 2005, Fried et al., 2005). Reasons that have been suggested to account for these discrepancies are time since abstinence, age of onset and extent of use (Di Forti et al., 2007).

Despite numerous investigations of the effect of acute and chronic cannabis administration on what is termed episodic memory, or ‘time-locked’ memory for events and occurrences, another area of memory has been largely neglected: semantic memory. Semantic memory refers to our memory for facts and general knowledge of the world, including memory for meanings and language. Semantic memory in cannabis users is particularly interesting as semantic memory impairments are suggested to underlie some of the symptoms observed in psychosis such as thought disorder (Spitzer et al., 1993). A recent study found no impairment in speed of semantic processing in cannabis users (Wadsworth et al., 2006), although some previous studies have found evidence for an impairment in category fluency, where participants are required to generate as many category exemplars as they can within a given interval (e.g. Messinis et al., 2006), which is suggestive of a semantic memory impairment.

A more valid method of assessing semantic processing than category fluency is the semantic priming paradigm, a lexical-decision task in which participants must respond to a target word that is preceded by a prime word that is either related to the target or not (Meyer and Schvanevedlt, 1971). Semantic priming is a facilitated by responding to concepts that are semantically related. When a prime word is related to the target word (e.g. chair, table), people generally respond faster than when they are unrelated (e.g. chair, fish), and this decrease in reaction time is the ‘priming effect’. The mechanistic explanation for this describes the semantic system as organised in a map-like network of nodes, representing concepts (Neely and Keefe, 1989). When one node is activated, spreading of activation occurs so that connected nodes are also activated. This spreading of activation facilitates faster responding to related nodes in the network.

Semantic priming is a popular methodology for investigating semantic memory as the standard priming task can be manipulated to investigate different semantic mechanisms. Stimulus onset asynchrony (SOA, the time between the onset of the prime and the target) manipulations are used to assess effects on either automatic or controlled processes. SOAs shorter than 250 ms are considered to tap into automatic, pre-attentional processing, whereas controlled processing, which is slower and thought to involve attention, is tapped by SOAs longer than 700 ms.

Recent claims have related cannabis use to psychosis, with implications that there may be a causal link between the two (Ferdinand et al., 2005, Fergusson et al., 2005). Disturbance in the organisation and processing of knowledge has long been considered a central feature of schizophrenia and, more broadly, psychosis (Bleuler, 1911/1950). As a result of this, semantic processing tasks are at the forefront of much research on cognitive deficits associated with psychosis and schizophrenia. Whilst normal semantic priming is associated with normal language processing, it is thought that overactive priming (faster and further spreading of activation) may relate to schizophrenic thought disorder and language disruptions. A faster spreading of activation in semantic networks in schizophrenia may result in spurious nodes becoming activated and interfering with the normal train of thoughts. In patients with schizophrenia, research has supported this hypothesis with increased automatic (i.e. at a short SOA) priming or 'hyper-priming' found in some studies (Moritz et al., 2001, Moritz et al., 2003). However, the results of research on semantic priming in schizophrenia have been far from consistent. A review of research in this field concluded that hyper-priming was more robustly found in conditions of ‘indirect’ priming (Minzenberg et al., 2002). Indirect priming refers to priming where, unlike ‘direct’ priming in which the words are directly semantically related (e.g. lemon-sour), the two words presented have a mediator word that is related to both of them but not presented (e.g. lemon-sweet, not presented mediator = sour). We therefore included an indirect as well as a direct priming condition in the current study, which aimed to examine the effects of cannabis, both acutely and sub-acutely, on semantic memory.

The present study set out to compare a naturalistic sample of cannabis users with controls on semantic processing and schizophrenia-like symptoms, both under the influence of the drug and then again when not intoxicated. We investigated semantic priming with a computer-based, lexical-decision task, which manipulated SOA. State schizotypy was assessed using a novel state schizotypy questionnaire (Mason et al., 2008). As little research has investigated the long-term effects of cannabis on semantic memory, this aspect of the study was exploratory. We hypothesised that intoxicated cannabis users would show a pattern of semantic priming similar to one that has been shown in thought-disordered schizophrenic patients, i.e. hyper-priming at a short SOA, and that this would be related to schizophrenia-like symptoms.