Language as a Window Into the Altered State of Consciousness Elicited by Psychedelic Drugs

This review (2022) explores how the acute effects of psychedelic drugs impact speech organization regardless of its semantic content, and how to characterize the subjective effects of psychedelic drugs by analyzing the semantic content of written retrospective reports. It is suggested that researchers studying psychedelics can considerably expand the range of their potential scientific conclusions by analyzing brief interviews obtained before, during and after the acute effects.

Abstract

“Psychedelics are drugs capable of eliciting profound alterations in the subjective experience of the users, sometimes with long-lasting consequences. Because of this, psychedelic research tends to focus on human subjects, given their capacity to construct detailed narratives about the contents of their consciousness experiences. In spite of its relevance, the interaction between serotonergic psychedelics and language production is comparatively understudied in the recent literature. This review is focused on two aspects of this interaction: how the acute effects of psychedelic drugs impact speech organization regardless of its semantic content, and how to characterize the subjective effects of psychedelic drugs by analyzing the semantic content of written retrospective reports. We show that the computational characterization of language production is capable of partially predicting the therapeutic outcome of individual experiences, relate to the effects elicited by psychedelics with those associated with other altered states of consciousness, drawing comparisons between the psychedelic state and the symptomatology of certain psychiatric disorders, and investigate the neurochemical profile and mechanism of action of different psychedelic drugs. We conclude that researchers studying psychedelics can considerably expand the range of their potential scientific conclusions by analyzing brief interviews obtained before, during and after the acute effects. Finally, we list a series of questions and open problems that should be addressed to further consolidate this approach.”

Author: Enzo Tagliazucchi

Summary

INTRODUCTION

Humans and other animals display a natural tendency to consume drugs that transiently modify their behavior, cognition and overall state of consciousness. Some humans have adopted an exploratory attitude towards drugs, experiencing their effects and then communicating to others the nature of their subjective experiences.

Due to the complexity of the effects induced by psychedelic drugs, it is almost impossible for users to communicate the nature of their subjective experience without resorting to language or other form of explicit reporting.

Language plays an important role in the investigation of serotonergic psychedelics, yet studies of the interaction between these drugs and language production are relatively underrepresented in the literature. In this review, we will discuss the use of natural language in the study of psychedelics.

HOW SHOULD PSYCHEDELIC DRUGS BE STUDIED

Psychedelic research must necessarily include human participants, yet what is the adequate methodology to investigate these participants and their subjective experiences?

While some studies have been done on the neural correlates of self-reported subjective effects, most have been done on tasks that evaluate specific domains of human cognition. Self-reported questionnaires have been widely used to study how psychedelic compounds affect cognition, conscious perception, thought processes, beliefs, attitudes, and personality traits in humans. These results have been correlated with objective measurements of brain activity obtained using neuroimaging techniques.

In contrast to paradigms based on constrained tasks and self-reported measures assessed via questionnaires, unconstrained speech produced either spontaneously or in response to simple queries is uncommon. However, this limitation can also be considered a potential advantage, since unstructured text data is difficult to extract meaningful objective and quantitative data from.

The Experience Vaults at Erowid contain thousands of reports of drug-induced experiences across a wide range of chemicals and their combinations. The usefulness of this data is hindered by several unknowns.

Psychedelic drugs can modify certain features of language produced by an individual under their acute effects, including both semantic and non-semantic features. This opens a new dimension of analysis beyond the possibilities of psychometric questionnaires. As we will discuss in this review, certain language alterations may be characteristic of specific neuropsychiatric disorders, and relating this body of research to the speech abnormalities observed under the effects of psychedelics could be informative.

FUNDAMENTALS OF NATURAL LANGUAGE PROCESSING

To analyze natural language reports of psychedelic drugs, a necessary first step is to extract reliable and quantitative information. This is more challenging than the analysis of questionnaires, for which standardized numerical outcomes are readily available.

Analysis of written reports can be subdivided into semantic and non-semantic analyses, with semantic analyses being concerned with inferring the meaning intended by the subjects.

The semantic similarity between two words depends on their statistical co-occurrence in similar linguistic contexts. The similarity of neural representations of words correlates with semantic similarity, although neurobiological support has not yet been obtained.

Commonly, texts are first pre-processed to reduce the proliferation of different but semantically related terms, and to remove very rare or very frequent terms. Semantic analyses usually start from a representation of the data given by a term-document matrix.

In practice, NLP methods are replaced by more sophisticated methods that take into account two important obstacles, such as sparse entries in the term-document matrix and the possibility that two terms might never co-occur in the documents but frequently occur together with a third term, thus being semantically related.

Latent semantic analysis (LSA) reduces the number of linearly independent rows in the term-document matrix, i.e. by lowering its rank, and then estimates the similarity between pairs of documents by computing the cosine distance or the correlation coefficient between the corresponding columns of the term-document matrix.

Semantic similarity can also be estimated using word embeddings, which can be learned from data using a shallow neural network. These methods are behind some of the most promising NLP-based markers of altered thought processes in psychiatric patients.

Non-semantic methods transform texts into sequences of tokens arranged according to syntactical rules, regardless of their meaning. The topological properties of the resulting graphs contain information useful to characterize drug-induced language alterations, as well as abnormalities specific to certain neuropsychiatric disorders.

Semantic methods are useful to investigate retrospective subjective reports in terms of their content, and non-semantic methods are useful to determine how the overall structure of verbal expression is modified during the acute effects.

LANGUAGE PRODUCTION UNDER THE ACUTE EFFECTS OF PSYCHEDELICS

The analysis of speech production has shown great promise to detect and predict psychotic episodes, and psychedelics have been shown to render speech less predictable and enhance free-association. However, the characteristics of unconstrained speech produced during the acute effects of psychedelics remain relatively underexplored.

Sanz and colleagues applied semantic and non-semantic methods to interviews conducted at two different time points after intravenous infusion of 75 g of LSD. They found that the effects of the drug increased Shannon’s entropy.

The entropic brain hypothesis explains the similarity between LSD-induced acute changes in speech and the psychotomimetic hypothesis, which predicts that psychedelics have a net scrambling or disorganizing effect on brain activity at multiple spatial and temporal scales, including those associated with perception, cognition, and the production of language. Sanz et al. investigated LSD, a drug presenting biphasic effects with a gradual shift from serotonergic to dopaminergic action.

The results of the experiment by Sanz et al. seem to contradict those of Amarel and Cheek (1965), who showed that LSD reduced the predictability of speech and reduced the total number of spoken words.

Natural speech produced during or immediately after the acute effects of psychedelics can be used to compare 5-HT2A receptor activation with other altered states of consciousness, such as rapid eye movement (REM) sleep.

Language-related cortical areas are characterized by specific neurotransmitter receptor fingerprints, and the analysis of natural language could be useful to characterize the effects and mechanism of action of novel psychoactive drugs.

ANALYSIS OF RETROSPECTIVE REPORTS

Huxley’s books glowed with brighter colors, a profounder significance, and were bound in white jade, agate, aquamarine, yellow topaz, and lapis lazuli.

The semantic content of a retrospective subjective report can be used to estimate the subjective effects elicited by a drug, and thus can be used to hypothesize that the drug-induced experiences were similar.

Coyle et al. used supervised machine learning methods to separate between reports of different drugs based on the associated vocabulary frequency vectors, showing that different families of compounds were linked to narratives with distinctive semantic content.

Sanz and colleagues applied a similar framework to investigate the relationship between pharmacological mechanism of action and shared semantic content of Erowid reports. Figure 5 shows the unsupervised classification of several drugs based on the semantic similarity of their associated subjective reports. The classification respects traditional categories such as antidepressants and antipsychotics, psychedelics, dissociatives, entactogens, stimulants, and sedatives, among others.

Using retrospective reports, it is possible to compare subjective effects of drugs to the content of dreams. The closest comparison is between psychedelic drugs and dreams of high lucidity, with LSD being the closest across all drugs in the Erowid corpus.

LSA was used to demonstrate that ketamine, a glutamatergic dissociative agent, has subjective effects most similar to those reported after “near death experiences”, although the acute effects of ketamine and related dissociative agents might negatively affect the survival of the organism when faced with situations that require a rapid and precise response.

These studies suggest that a taxonomy of conscious states may be possible, based on data-driven similarity metrics, which are constructed using natural language reports of subjective experiences. However, disagreements exist about what these dimensions should be.

The type of analysis represented in Figure 5 might represent an interim solution to the open problem of describing and characterization of conscious states, since it is possible to estimate the similarity of effects based on the shared semantic content of natural language reports.

SEMANTIC SIMILARITY PARALLELS NEUROCHEMICAL AND PHARMACOLOGICAL SIMILARITY

The mechanism of action of different drugs is well-characterized at the molecular and cellular levels, but less is known about the downstream effects on large-scale activity patterns that correlate with cognition and conscious experience.

Zamberlan and colleagues explored how the action of psychedelic drugs at other binding sites can nuance the resulting subjective effects, and found that the closest two drugs are at this level (i.e. the most similar their binding affinity profiles are), the closest those drugs should be in terms of the semantic content of their associated subjective reports.

The relationship between subjective effects and neurochemical action of psychedelic compounds can be further informed by natural language processing applied to subjective reports. This approach can be used to determine the most specific subset of receptors to target for eliciting certain subjective effects.

Taking this analysis one step further, it is possible to decompose the subjective reports into topics, which are summarized by sets of terms that tend to co-occur through the documents. These topics can be linked to high binding affinity at different sites, which can be used to inform the subjective effects.

More than 6.000 reports of drug-induced experiences were mapped to 40 neurotransmitter receptor subtypes in the brain via gene transcription levels from invasive tissue probes.

NATURAL LANGUAGE REPORTS AND THE THERAPEUTIC USE OF PSYCHEDELICS

Studies have shown that psychedelics may be useful in treating psychiatric disorders such as depression and anxiety. The subjective effects induced by psychedelics appear to influence the outcome of the intervention, although the role of mystical-type experiences was brought into question.

Carrillo and colleagues showed that baseline interview data can predict which patients will respond to psilocybin for treatment-resistant depression. Cox and colleagues used natural language narratives to predict who among more than 1,000 individuals would quit or reduce using drugs following a psychedelic experience.

Machine learning methods applied to linguistic features obtained using NLP could be used to identify patients who could benefit from psychedelic therapy, and to screen for mental health conditions that could present problematic interactions with psychedelics.

The use of natural language processing in the early stages of psychedelic-assisted therapy is not an isolated trend, and the automated analysis of natural speech will consolidate into a valuable tool to assist clinicians in the design and implementation of therapy sessions assisted by psychedelic compounds.

The potential ethical implications of developing and implementing NLP-based tools to assist with clinical decision making are discussed. These issues include false negatives and the possibility of “self-fulfilling prophecies” where predictions of positive outcomes are concentrated in participants with certain characteristics.

LIMITATIONS AND FUTURE DIRECTIONS

We have reviewed several key studies illustrating how the analysis of natural language can assist in the investigation of psychedelic compounds.

The most obvious limitation of this approach is the unconstrained nature of natural language reports. Subjects might omit important details or fail to clearly express the most relevant information, and some personality traits might be manifest in their reports. Standardized psychometric questionnaires have the advantage of uniformly guiding the participants through the points that are considered most relevant by the researchers.

The analysis of large online databases (e.g., Erowid’s Experience Vaults) presents additional problems associated with unknown or underinformed variables, such as the precise nature of the compounds that were consumed, their dosage, whether drugs were consumed alone or in combination with others, subject demographics, mental health status and past history of drug use.

Psychedelics impair different aspects of memory while increasing the vividness of affectively intense memories, which raises the possibility of inducing false memories. Furthermore, changes in the capacity to produce spoken reports throughout the acute effects of a psychedelic drug could induce systematic sampling biases.

In spite of some potential shortcomings, natural language analysis can still be considered a promising tool to tackle research questions about the nature of subjective experience.

When used to predict the outcome of psychedelic treatments, natural language reports have higher accuracy than questionnaires.

CONCLUSION

Language is our main everyday vehicle for the expression of ideas, emotions, plans, and subjective, inner feelings. Natural language processing can be leveraged for the scientific exploration of serotonergic psychedelics.