Psilocybin

Psychiatry’s next top model: cause for a re-think on drug models of psychosis and other psychiatric disorders

This survey study (2013) compared five drugs on their ability to model a variety of psychiatric conditions. It found, among other results, that psilocybin was the best model for positive psychotic symptoms.

Authors

  • Brugger, S.
  • Carhart-Harris, R. L.
  • Nutt, D. J.

Published

Journal of Psychopharmacology
individual Study

Abstract

Despite the widespread application of drug modelling in psychiatric research, the relative value of different models has never been formally compared in the same analysis. Here we compared the effects of five drugs (cannabis, psilocybin, amphetamine, ketamine and alcohol) in relation to psychiatric symptoms in a two-part subjective analysis. In the first part, mental health professionals associated statements referring to specific experiences, for example ‘I don’t bother to get out of bed’, to one or more psychiatric symptom clusters, for example depression and negative psychotic symptoms. This measured the specificity of an experience for a particular disorder. In the second part, individuals with personal experience with each of the above-listed drugs were asked how reliably each drug produced the experiences listed in part 1, both acutely and sub-acutely. Part 1 failed to find any experiences that were specific for negative or cognitive psychotic symptoms over depression. The best model of positive symptoms was psilocybin and the best models overall were the acute alcohol and amphetamine models of mania. These results challenge current assumptions about drug models and motivate further research on this understudied area.

Unlocked with Blossom Pro

Research Summary of 'Psychiatry’s next top model: cause for a re-think on drug models of psychosis and other psychiatric disorders'

Introduction

Drug models have been widely used in psychiatry to generate hypotheses about the biology of mental disorders and to provide experimental platforms for testing treatments. Carhart-Harris and colleagues note that psychosis in particular has attracted attention because several psychoactive substances (cannabis/THC, serotonergic psychedelics such as psilocybin and LSD, amphetamines, and dissociatives like PCP and ketamine) can produce symptoms resembling psychosis. Despite this, few studies have compared different drug models directly, and uncertainty remains about which substances most closely and reliably mirror particular psychiatric symptom clusters. To address that gap, the investigators designed a two-part pilot, subjective study to compare five drugs (high-potency cannabis, psilocybin-containing mushrooms, amphetamine, ketamine and alcohol) on two dimensions: how specifically first‑person experiential statements map onto psychiatric symptom clusters, and how reliably each drug produces those experiences acutely and in the sub-acute period. The stated aim was to identify which drugs serve as the best models for particular psychiatric states and to use the findings to motivate more rigorous controlled work.

Methods

The study employed a two-part web-based survey. Part 1 targeted mental health professionals (psychiatrists and clinical psychologists) who were presented with 58 first‑person statements describing subjective experiences (for example, 'I feel sad' or 'my thoughts stop as if they are blocked'). Respondents could ascribe each statement to up to nine categories: positive psychotic symptoms, negative psychotic symptoms, cognitive symptoms of psychosis, thought disorder, depression, anxiety, mania, spiritual experiences, or none. Seventy respondents were solicited via direct email and snowballing; 63 completed the survey. The degree of association between a statement and a symptom cluster was calculated as the proportion of ascriptions to that cluster relative to ascriptions to other clusters, expressed as a percentage. From the part 1 results, 17 statements were selected for part 2 on the basis of their specificity to symptom clusters, while ensuring coverage across different clinical phenomena (the investigators retained some items with lower specificity where necessary to represent less-specific clusters). Part 2 surveyed experienced drug users recruited via online forums (including bluelight.ru, drugs-forum.org and maps.org). To participate respondents had to report prior experience with all five target drugs; 224 individuals completed the survey. For each of the 17 statements, respondents rated how often they experienced the phenomenon after taking at least a moderate dose of each drug, separately for the acute (immediate) effects and the sub-acute (comedown/withdrawal) effects, using the categories 'rarely/never', 'sometimes' or 'often/always'. Guidance doses for a 'moderate' exposure were provided (e.g. >3 pints alcohol, >60 mg snorted ketamine, ~1 g psilocybe cubensis), based on consultation with experienced users. Ethical approval was obtained from Imperial College Research Ethics Committee and the surveys were implemented using Bristol Online Surveys. The primary analytic approach for part 2 quantified a drug's reliability for inducing a symptom cluster as the mean percentage of respondents endorsing 'often/always' across the statements associated with that cluster. For example, the percentage of 'often/always' endorsements for mania-related items under a given drug were averaged to produce that drug's mania-reliability score.

Results

Part 1: Mental health professionals showed substantial co‑association across several clusters. No first‑person statements were uniquely associated with negative psychotic symptoms or with cognitive symptoms of psychosis; these domains overlapped heavily with depression and thought disorder. Mania-related statements were the most selectively associated, with a mean exclusivity of about 70%. Thought disorder had the lowest mean specificity among the non-empty clusters (about 38.5%). The statement 'I can't concentrate on things' illustrated the overlap problem: although 90% of respondents linked it to depression, many also linked it to mania (37%), anxiety (49%), thought disorder (30%), negative symptoms (17%) and cognitive symptoms (43%), yielding a relatively low unique association score. Part 2: Across the recreational user sample, classic drug models of psychosis generally performed poorly in terms of reliably producing positive psychotic symptoms. Acute psilocybin emerged as the best single model of positive symptoms but with low reliability (reported as 13% of respondents saying they 'often/always' experienced positive symptoms after psilocybin). Thought disorder was best modelled by acute ketamine (27% reliability), with acute and sub-acute alcohol showing moderate associations with thought disorder (22% and 18%, respectively). The spiritual-type experience was most closely modelled by sub-acute psilocybin; the extracted text reports this reliability as approximately 37% in one place and 39% in another, an inconsistency in the extraction. Acute ketamine and acute psilocybin were also reported to have appreciable associations with spiritual-type items (around the low 30%s). One particular item, 'I feel a profound inner peace', had the strongest unique association with the spiritual cluster (reported as 77% association in part 1), and over half the respondents reported experiencing it acutely under psilocybin (reported as 53% acute and 48% sub-acute in the extracted text). Manic symptoms were most reliably produced acutely by alcohol (46% reliability) and closely by amphetamine (45%). Sub-acute depressive symptoms were most associated with alcohol and amphetamine (reported as 24% and 25%, respectively). None of the five drugs produced anxiety-related symptoms with notable reliability; the highest single values reported for anxiety were 15% for acute cannabis and 15% for sub-acute amphetamine. Demographics reported in the extracted text indicate 63 clinician respondents in part 1 and 224 drug-user respondents in part 2; part 2 respondents were predominantly male (79%), about 50% American and 23% British.

Discussion

Carhart-Harris and colleagues interpret the findings as challenging assumptions about the face validity and reliability of common drug models of psychosis and other mental states. They highlight that, in this naturalistic sample of experienced users, none of the classic psychotomimetic drugs produced positive psychotic symptoms with high reliability; psilocybin was the best among them but only yielded frequent positive symptoms in a small minority (about 13%). The investigators contrast this with laboratory studies where positive psychotic-like symptoms are commonly reported, and they outline several possible explanations for the discrepancy: experienced users may be resilient or reluctant to report adverse effects, selection and reporting biases on the recruitment forums may under-represent those who had severe psychotomimetic reactions, laboratory dosing and participant naivety may differ, and context exerts a powerful influence on drug responses. The authors also emphasise the importance of defining what aspect of a disorder a drug is intended to model. They suggest that some features of acute psychosis (for example, auditory-verbal hallucinations) may be transient or require chronic exposure to model, whereas anxious states or aberrant salience-related phenomena may be more amenable to acute drug provocation. Regarding negative symptoms, the study found those first‑person experiences overlapped substantially with depressive symptoms in clinicians' judgements, so the investigators could not identify statements uniquely indexing negative or cognitive psychotic symptoms; this undermines their ability to evaluate claimed models of negative symptoms such as ketamine in this paradigm. Unexpectedly, alcohol performed strongly as an acute model of mania (46% reliability) and, together with amphetamine, as a sub-acute model of depression. The authors note that alcohol's association with loss of inhibition contrasts with amphetamine's association with racing thoughts, and they flag the alcohol–bipolar relationship as worthy of further study. Psilocybin's sub-acute association with spiritual-type experiences — and the high endorsement of 'profound inner peace' acutely and sub-acutely — is taken as consistent with experimental findings that single high-dose psilocybin sessions can reliably evoke lasting spiritual-type states with therapeutic correlates. The investigators acknowledge important limitations: the web-based, self-report design lacks experimental control and verifiable respondent characteristics; the naturalistic context and sample composition (predominantly male, experienced users) limit generalisability; some potentially relevant items or demographic/use-frequency data were not collected to avoid overlong surveys; and extraction/context inconsistencies in the present text complicate precise reporting of some values. They recommend follow-up controlled within-subject studies that manipulate drug, dose and setting, and that include standardised scales for targeted symptom clusters to more rigorously define drugs' psychotomimetic properties. The authors conclude that the pilot findings question some contemporary assumptions about drug models and should motivate more systematic, controlled research.

View full paper sections

INTRODUCTION

Drug models of mental illness are useful in psychiatry as they help to inform hypotheses on the biology of pathological states and provide a valuable testing ground for novel medications. Psychosis has proved a popular domain for drug models because it is a difficult-to-treat disorder with some distinctive symptoms that have long been known to be replicable by intoxication with different psychoactive drugs. Psychotomimetic (psychosis-mimicking) drugs that have received particular attention are: cannabis or delta-9-tetrahydrocannabinol (THC), serotonergic psychedelics such as psilocybin and LSD, amphetaminesand the dissociative anaesthetic agents phencyclidine (PCP) and ketamine. However, few studies have compared the relative merits and shortcomings of different drug models in a controlled manner, and consequently, it is difficult to say which drug offers the most complete model of psychosis. If this matter is not addressed, then considerable resources may be wasted on sub-optimal models of psychosis. Serotonergic psychedelics continue to be considered candidate drug models of psychosis; however, despite considerable historical interest, they have fallen out of favour in recent years, as focus has shifted more on to their therapeutic potential. Amphetamine has also lost some appeal as an acute model of psychosis, as although it reliably produces manic symptoms in acute doses, repeated dosing is required to produce frank psychotic symptoms. THC has attracted special interest recently however, due to its putative psychotogenic properties (Dand evidence that it can produce marked positive symptoms in some people. However, ketamine is currently arguably the most popular drug model of psychosis in humans, largely due to the notion that it can model both positive and negative symptoms of schizophreniaand evidence for the role of glutamatergic dysfunction in this disorder. In addition to pathological phenomena, it has been claimed that certain drugs can reliably produce spiritual-type experiences that are indistinguishable from spontaneously occurring spiritual experiences. However, clinicians sometime identify such phenomena as pathological. Thus, this study sought to address this matter by enquiring about: 1) the specificity of apparent spiritual experiences over psychopathological symptoms, and 2) the reliability with which such states can be induced by different drugs. A comparison of the relative strengths and weaknesses of different drug models of mental states is fraught with difficulty. For example, the same drug may induce different effects in different people (e.g. drug-naïve research volunteers versus experienced drug users) and in different social contexts (e.g. in a clinical study versus at home). Two factors are especially important when evaluating a model of a disorder: 1) How closely the features of the model resemble the symptoms of a specific disorder, and 2) how reliably these features occur in the model. In the context of drug models of psychiatric disorders, an optimal model should reliably produce an experience that is typical of a specific disorder. This would effectively imply that the druginduced and endogenous states are so similar and distinctive that strong inferences can be made about the endogenous state based on the induced state. It is relatively straightforward to determine the reliability with which a drug can produce an endogenous state. However, determining the degree to which a particular experience is specifically associated with a particular disorder or symptom cluster is more complicated. For example, if the experience of finding it difficult to concentrate is common to a range of disorders, then a drug that makes it difficult for someone to concentrate will not be useful in making predictions about one disorder compared with another. As will be discussed below, the issue of the degree to which a drug model is specifically associated with a given illness has been neglected in previous model evaluations. Therefore, in this two-part study, using novel methodologies, we sought to address an important and unanswered question: what are the best drug models of psychiatric illness? This subjective study was conceived as a pilot study to motivate and inform the design of a larger controlled study.

BASIC DESIGN

To address the question of the relative merits of different drug models, we devised a subjective pilot study in two parts. In the first part, mental health professionals were contacted and asked to complete a web-based survey. In this survey they were presented with 58 of first-person statements and asked to rate how closely they resembled particular mental states or symptom clusters. They had nine options and could choose more than one: 1) positive psychotic symptoms, 2) negative psychotic symptoms, 3) cognitive symptoms of psychosis, 4) thought disorder, 5) depression, 6) anxiety, 7) mania, and 8) spiritual experiences, or 9) none of these. In the second part of the study, we devised a survey to be completed by drug users with personal experience with five different drugs: 1) high-potency cannabis, 2) psilocybe magic mushrooms, 3) ketamine, 4) amphetamine and 5) alcohol. Respondents were presented with a subset of the statements from part 1 (17 of them, see Figure) and asked to say whether and how often they experienced the referred to phenomena, when under the acute influence of a drug, and in the sub-acute period following its use. The possible categorical responses were: 1) 'rarely/never', 2) 'sometimes' or 3) 'often/always'. This study received ethical approval from Imperial College Research Ethics Committee. The surveys for each part were built using the Bristol Online Surveys service (www.survey.bris.ac.uk) and can be found in the supplementary material.

PARTICIPANTS

Part 1. A link to the part 1 survey was emailed to psychiatrists and clinical psychologists known to us who were asked to complete the survey. We also asked them to forward it onto eligible colleagues (to be eligible, respondents simply had to be mental health professionals). In all, 63 mental health professionals completed the survey. Demographic data is presented in Table. Part 2. In total, 224 experienced drug users completed the part 2 survey. The survey was advertised on the drug user forums: www.bluelight.ru and www.drugs-forum.org, as well as the website www.maps.org which is an organisation concerned with promoting clinical research with psychedelic drugs; 224 individuals responded to the survey. In order that the participant's responses be properly informed, each had to declare previous experience with all of the following drugs: alcohol, high-potency cannabis, amphetamine, psilocybin-containing magic mushrooms and ketamine. Demographic data is presented in Table.

SURVEY FORMAT AND PROCEDURE

Part 1. In the part 1 survey, respondents (mental health professionals) were presented with 58 first-person statements referring to experiences or symptoms a psychiatric patient might describe (e.g. 'I feel sad', 'my thoughts stop as if they are blocked' and 'I get worried that other people are plotting against me'; see supplementary material to view all 58 statements). Seven additional statements were included that are characteristic of spiritual-type experiences, based on the available literature on its phenomenology. The statements themselves were generated by a consultant psychiatrist and psychosis specialist (JMS), based on his own training and experience with patients, in discussion with RCH and DJN. Respondents were instructed to ascribe each statement to up to eight different disorders or symptom clusters: 1) mania, 2) anxiety, 3) depression, 4) positive psychotic symptoms, 5) negative psychotic symptoms, 6) cognitive symptoms of psychosis and 7) formal thought disorder. Respondents could also select: 8) spiritual experiences and 9) none of these. More than one of the nine options could be selected for each of the 58 statements, for example: 'I feel worried or frightened' might be ascribed to anxiety and to positive psychotic symptoms. The schedule of statements was designed to cover experiences associated with a range of psychiatric disorders, plus the spiritual experience. Seventeen statements from part 1 were selected for inclusion in part 2 based on how specifically they were associated with a particular symptom cluster (i.e. those with the highest % association were chosen). However, to ensure we included items that covered a range of clinical phenomena and not just, for example, all statements of relevance to positive psychotic symptoms, some items with lower % association values but relevance to a disorder that generally has low % association values (e.g. negative psychotic symptoms) were chosen. This explains why, for example, there were six statements for positive psychotic symptoms whereas the others had fewer (i.e. because experiences of relevance to positive psychotic symptoms are relatively specific to that disorder and not generalizable to others). The number of statements had to be reduced for part 2 because of the length of time required to complete the part 2 survey. We calculated a statement's degree of association with a particular mental state as the number of ascriptions to that state over the number of ascriptions to other symptom clusters. This was then expressed as a percentage. For example, for the statement: 'I become more talkative', which was most closely associated with mania, its degree of association was calculated by summing the number of ascriptions to mania (61) over the total number of ascriptions to any mental state (35) + 61. That is, 61/96 * 100 = 64%, or 'I become more talkative' had a 64% association with mania. The 17 first-person statements that were selected for the part 2 survey are shown in Figure. Part 2. For each statement, respondents were instructed to select (from options 'rarely/never', 'sometimes' or 'often/ always'), the frequency with which they experienced the effect described by each statement after taking at least a moderate dosage of each drug. Respondents were asked to do this for both the acute drug effects -i.e. the immediate 'high' -and separately for the post-acute effects -i.e. the comedown or withdrawal period. The following doses were provided as guidance of what was meant by a 'moderate' dose (these doses were determined by consultation with self-proclaimed experienced drug users on druguser forums): Cannabis -at least one well-loaded spliff/joint, deeply inhaled and shared between no more than two people, or at least one 'bowl' in a bong or pipe shared by no more than two people; alcohol ->3 pints of medium to high-strength beer or cider, or at least three large glasses of wine; psilocybe mushrooms -at least 12 fresh liberty cap magic mushrooms or at least 1 g of psilocybe cubensis; amphetamine -at least 500 mg snorted or bombed amphetamine; ketamine -over 60 mg snorted.

DATA ANALYSIS FOR THE PART 2 SURVEY

We quantified a drug's ability to induce a particular symptom cluster or mental state as the mean percentage of part 2 respondents selecting 'always/often' for statements associated with a particular symptom cluster. For example, 77% of respondents said that acute amphetamine always or often made them more talkative, 24% said it always/often made them less worried about spending money or doing dangerous things and 34% said it always/often makes their thoughts go so quickly others can't keep up with them. Thus, the reliability with which acute amphetamine induces mania-related symptoms is (77+24+34)/3 or 45%. The results of the part 2 survey are shown in Figure.

PART 1 RESULTS

Figuredisplays each symptom cluster and the statements that had the greatest degree of (specific) association with them. None of the statements had any significant unique association with negative psychotic symptoms or cognitive symptoms of psychosis. This was mainly due to a high co-association between four symptom clusters: depressive symptoms, thought disorder, negative psychotic symptoms and cognitive symptoms of psychosis. For example, the statement 'I can't concentrate on things' had a unique association of only 28% for its most closely related symptom cluster, depression, because although 90% associated this statement with depression, respondents also associated it with mania (37%), anxiety (49%), thought disorder (30%), negative symptoms of psychosis (17%) and cognitive symptoms of psychosis (43%). Indeed, even if negative symptoms and cognitive symptoms were collapsed into one symptom cluster (on the basis that they both relate to functional deficits), no new statements were identified that had specificity for this single symptom cluster. The statements relevant to mania were the most uniquely or exclusively associated (70% mean association). Excluding negative and cognitive symptoms, for which no statements were uniquely associated, the symptom cluster with the least uniquely associated statements was thought disorder, with a mean association of 38.5%.

PART 2 RESULTS

Drug models of positive psychotic symptoms. The results of the part 2 survey are displayed in Figure. The first notable result was that none of the classic drug models of psychosis produced positive psychotic symptoms with any reliability. The best model of positive symptoms was acute psilocybin; however, even this had a low reliability of just 13% (i.e. 13% of respondents reported often or always experiencing positive symptoms after psilocybin), suggesting that although it was the best drug model of positive symptoms, it is not an especially reliable model.

DRUG MODELS OF THOUGHT DISORDER.

The best drug model of thought disorder was acute ketamine (27% reliability). Acute and sub-acute alcohol had a relatively high association with thought disorder at 22% and 18%, respectively. Drug models of the spiritual experience. The best drug model of the spiritual experience was sub-acute psilocybin with 37% reliability, followed by acute ketamine (33%) and psilocybin (30%). Acute and sub-acute psilocybin (53% and 48% reliability, Table. Demographic data for the sample of respondents to both surveys: A total of 63 mental health professionals completed the part 1 survey and 224 experienced drug users completed part 2 survey. In the part 1 survey, respondents could select more than one speciality. Most respondents were psychiatrists (87%) and the remainder were psychologists; 76% were British. Most of the respondents in part 2 were male (79%); 50% were American and 23% were British. respectively) were closely associated with the statement 'I feel a profound inner peace'. In part 1, this statement was the most exclusively associated with a 'symptom cluster' or mental state, with an association of 77% for the spiritual experience (Figure). Drug models of mania. The best drug model of mania was acute alcohol (46% reliability), closely followed by acute amphetamine (45%). The reliability with which acute alcohol and amphetamine could produce manic symptoms was the highest of all of the drug-symptom associations. Drug models of depression. The two drugs that reliably produced manic symptoms when taken acutely were also those that most reliably produced depressive symptoms sub-acutely. Subacute alcohol and amphetamine's degree of association with depression was 24% and 25%, respectively. Drug models of general anxiety. None of the five drugs produced symptoms of anxiety with any reliability. Acute cannabis (15%) and sub-acute amphetamine (15%) had the highest likelihood to induce anxiety-related effects.

DISCUSSION

A number of interesting findings emerged from this novel analysis. Firstly, none of the four classic drug models of psychosis produced positive psychotic symptoms with any reliability. Psilocybin emerged as the best model of positive symptoms, which is consistent with the finding of a previous study that compared the psychotomimetic properties of the psychedelic drug dimethyltryptamine with those of the dissociative ketamine. Nevertheless, only 13% of psilocybin users claimed to always or often experience positive psychotic symptoms after this drug. Indeed, drugs were less likely to produce positive psychotic symptoms than any of the other symptom clusters. This is in contrast to acute administration studies in a laboratory setting, in which positive psychotic symptoms are commonly reported under psychotomimetics. There are many possible reasons for this discrepancy. One is that the survey was completed by experienced drug users who may be more resilient to psychotomimetic effects than drug-naïve healthy subjects. These individuals may be more reluctant to disclose adverse responses due to positively biased perspectives on these drugs, as well as 'bravado' in the case of males, who dominated the sample (80%). Alternatively, individuals who experience psychotic effects with these drugs may be less likely to take them recreationally and to frequent the websites on which this survey was advertised. It is also possible that the doses used in human laboratory studies have been higher, and the participants less experienced, or that these studies were affected by implicit biases, especially where clinician-administered scales are used. However, perhaps the most likely explanation for the discrepancy is that reactions to psychotomimetic drugs are known to be highly sensitive to the context in which they are taken. For example, the same drug, self-administered in a comfortable setting is less likely to elicit paranoid reactions than if it is administered by an unfamiliar person in an unfamiliar setting. A key question when evaluating a drug's ability to model a multifarious condition like psychosis is: what specific aspect of the disorder is the drug supposed to be modelling? Auditory-verbal hallucinations are common in acute psychosis but very rare in drug states, whereas anxious states engendering false inferences are much more common. This is supported by the sensitivity of psilocybin for the statement: 'I think that things that I see and hear that I would not normally notice have been put there to give me a message' (35% reported often/always experiencing this after psilocybin) which speaks to both the false inferenceand aberrant salience) models of psychosis. Auditory hallucinations can be thought of as transients in a mechanistic sense, i.e. shortlived events occurring periodically. It may be that acute drug states are incapable of modelling such phenomena. It may be more realistic to consider only how a drug can model a specific acute state, and transients may be merely epiphenomena of prolonged states. We might speculate that there would be a greater likelihood of auditory hallucinations if a psychotomimetic drug was taken chronically because repeated use might kindle emergent phenomena. This hypothesis is consistent with the psychotogenic potential of cannabis, a drug commonly used over long periods (in contrast to psychedelic drugs, which are typically taken irregularly). Fixed symptoms such as auditory hallucinations may depend on an extended set of experiences -thus explaining why they are so difficult to model acutely. It is often claimed that the popular ketamine model of psychosis is superior to other drug models because it models both positive and negative symptoms of psychosis. However, some have questioned the construct validity of negative symptoms as they are not specific to schizophrenia. Negative symptoms refer to such things as social withdrawal, emotional withdrawal, blunted affect and poverty of thought or speech. Statements included in the initial 58 that we thought might resonate with negative symptoms were: 'I become emotionally withdrawn', 'I lose interest in doing things', 'I don't bother to get out of bed', 'I don't want to socialise with other people' and 'I would rather not be around other people'. However, the mental health professionals who responded in the first survey were more likely to attribute these experiences to depression. Acknowledging that cognitive symptoms of psychosis might load onto the same factor as negative symptoms, we collapsed these into a single cluster to see if an improvement in specificity could be achieved -but still no statements could be identified that were unique to this new symptom cluster over depression. Importantly, these results do not imply that negative symptoms do not exist in schizophrenia, nor that they are seriously disabling; indeed, evidence suggest they are among the most prevalent and disabling symptoms. However, they do demonstrate that there is a great deal of overlap between self-reported negative and depressive symptoms and, from the present study, it is not possible to make any claims about the validity of currently proposed drug models of negative symptoms. One possible reason why we were unable to identify any experiences that were uniquely associated with negative symptoms is that statements referring to blunted affect or to feelings of spontaneity were not included in part 1. It is possible that these statements may have been more closely associated with negative symptoms than depression by mental health professionals. Perhaps the most surprising finding of the present study was the relative reliability with which acute alcohol produced symptoms of mania. Some 46% of respondents said they often or always experienced manic symptoms after alcohol, and 45% said that they often or always experienced these after amphetamine. While amphetamine is an established drug model of mania, alcohol is not, and yet it produced symptoms with the highest exclusivity for any symptom cluster (70% specificity for mania). In comparing the alcohol versus amphetamine models of psychosis, it is interesting to note that alcohol was especially associated with loss of inhibition about doing dangerous things, whereas amphetamine was especially associated with racing thoughts. It was also interesting that the best drug models of depression were sub-acute alcohol and amphetamine, with 24% of respondents reporting that they often or always experienced depressive symptoms during the alcohol hangover and 25% of respondents saying they experienced this during amphetamine withdrawal. Depression during amphetamine withdrawal has been documented before, as has depression post alcohol use, and acute and sub-acute amphetamine has been proposed as a model of bipolar disorder, but to our knowledge the same has not been said of alcohol. Exaggerated mania has been described in manic patients presenting with concurrent alcohol misuse. Thus, the relationship between alcohol use and bipolar disorder may be worth investigating. A particularly novel feature of the present survey was its inclusion of first-person statements relevant to the spiritual experience. This area is receiving growing interest after it was found that profound spiritual-type experiences could be reliably elicited in healthy volunteers after a single high dose of psilocybin. Moreover, these experiences tended to have a lasting impact, improving emotional wellbeing and increasing trait 'openness' over 12 months after the acute experience. These findings are consistent with the results of the present study since the best model of the spiritual experience was sub-acute psilocybin, with 39% of respondents saying that they often or always have spiritual-type experiences in the period following psilocybin use. Interestingly, the statement with the strongest association for any 'symptom cluster' was 'I felt a profound inner peace' (77% association with the spiritual experience). Over half of the respondents (53%) said they always or often experience this in the acute psilocybin state and 48% said they experience it in the sub-acute period following psilocybin. The unique association of statements for the spiritual experience psychotic symptom clusters suggests that the spiritual experience (as it is described by these two statements) is non-pathological and distinct from psychosis. This study has some important limitations. Data were derived from web-based surveys, and although this is an efficient means of data collection that introduces unique possibilities, it also requires the surrender of many experimental controls. We have no control over who completes the survey and no indication about the accuracy of their replies. Moreover, the respondents may have held biases, and this is a problem with subjective assessments in general. Also, respondents gave their assessments based on recreational use in naturalistic settings, whereas controlled studies administer drugs in a very different context. That context can have such a marked influence on subjective responses to a drug is important, and it implies that we should be cautious about making general inferences about a drug's pharmacology based on a set of subjective reports because very different experiences may be produced by the same drug in a different context. Future research might try to assess drug by environment interactions in a controlled manner to better understand this phenomenon. Another possible criticism is that we could have collected more data. For example, we could have enquired about respondents' ages and, in the case of part 2, the frequency of their drug use. However, the advantages of extending the surveys had to be balanced against the disadvantages of losing participants due to boredom or irritation with the length of the survey. In summary, the results of the present study suggest that the best drug models of psychiatric disorders are the acute alcohol and amphetamine models of mania, since these reliably produce symptoms that are relatively exclusive to a specific disorder. The results also suggest that drug models of positive psychotic symptoms are comparatively poor, at least when the drugs are taken naturalistically by experienced users. Psilocybin emerged as the best model of positive symptoms, but models of negative symptoms could not be assessed because we failed to find any experiences that were specific for this symptom cluster over depression. Lastly, sub-acute psilocybin appeared to be a reliable model of non-pathological spiritual-type experiences -which may have implications for therapeutic applications of this drug. It is hoped that this pilot study will motivate a larger controlled study involving administration of a range of psychotomimetic drugs to healthy individuals. For example, in a within-subjects design, with sufficient time separating administrations, it would be interesting to compare the relative psychotomimetic properties of different drugs at different doses. It would also be interesting to incorporate rating scales for other disorders and symptom clusters (e.g. depression and specifically early psychotic symptoms) so that a drug's psychotomimetic properties can be more accurately and specifically defined. To finish, it is hoped the present study has stimulated new thoughts about an understudied area: the relative strengths of different drug models of psychosis and other mental states. Some of its implications challenge contemporary assumptions and therefore require follow-up by controlled research.

Study Details

Your Library