Glutamatergic Model Psychoses: Prediction Error, Learning, and Inference

This paper (2010) reviewed studies using intravenous ketamine administration to develop a model of individual variability in response to ketamine.

Abstract

“Modulating glutamatergic neurotransmission induces alterations in conscious experience that mimic the symptoms of early psychotic illness. We review studies that use intravenous administration of ketamine, focusing on interindividual variability in the profundity of the ketamine experience. We will consider this individual variability within a hypothetical model of brain and cognitive function centered upon learning and inference. Within this model, the brains, neural systems, and even single neurons specify expectations about their inputs and responding to violations of those expectations with new learning that renders future inputs more predictable. We argue that ketamine temporarily deranges this ability by perturbing both the ways in which prior expectations are specified and the ways in which expectancy violations are signaled. We suggest that the former effect is predominantly mediated by NMDA blockade and the latter by augmented and inappropriate feedforward glutamatergic signaling. We suggest that the observed interindividual variability emerges from individual differences in neural circuits that normally underpin the learning and inference processes described. The exact source for that variability is uncertain, although it is likely to arise not only from genetic variation but also from subjects’ previous experiences and prior learning. Furthermore, we argue that chronic, unlike acute, NMDA blockade alters the specification of expectancies more profoundly and permanently. Scrutinizing individual differences in the effects of acute and chronic ketamine administration in the context of the Bayesian brain model may generate new insights about the symptoms of psychosis; their underlying cognitive processes and neurocircuitry.”

Authors: Philip R. Corlett, Garry D. Honey, John H. Krystal & Paul C. Fletcher

Summary

A Brief History of Psychotomimesis and NMDA Blockade

Psychotomimetic drugs have a long history in neuropsychopharmacology research, with the first use of the term in the 1840s. The central ideas of this review are generalizing across a number of different psychotomimetic drugs, with a focus on those whose mode of action involves noncompetitive NMDA receptor antagonism.

We aim to outline a model for how psychotomimetic compounds can induce disordered subjective experience, and to extrapolate from perturbed synaptic function to disordered subjective experience.

Phencyclidine (or PCP) and ketamine bind noncompetitively to NMDA receptors and produce a dissociative anesthesia. However, subjects treated with PCP experienced psychotomimetic side effects, including subjective thought block, anhedonia, catatonia, and waxy flexibility.

Subjects administered with a low subanesthetic dose of the drug experienced disturbances in body image and a difficulty distinguishing self and nonself, and described a peculiar state (described by Luby et al as Hypnagogic).

Ketamine was first synthesized by Parke-Davis in 1963 and was found to induce emergence reactions including marked changes in mood and affect, and frank hallucinatory episodes.

NMDA receptor blockade can induce both positive and negative symptoms in rats, providing a more complete model of psychosis than other drugs. However, the positive symptoms are not related to NMDA receptors, suggesting some non-NMDA component to the generation of positive symptoms.

Ketamine has been used to induce psychotic symptoms in many studies, and the intravenous route is almost exclusively used. Bolus injections are followed by a maintenance dose, and the infusion is often controlled by a computer-controlled pump.

The pharmacokinetic parameters for computerized pumps for psychotomimetic ketamine came from early anesthetic studies by Domino et al, 1982. These models were compartment based and often inaccurate, but newer models have a shorter time to peak concentration and a more accurate maintenance of ketamine plasma levels.

Ketamine has two enantiomers, (S) and (R) ketamine, which have different effects on brain metabolism. There is significant interindividual variability in the ketamine response even within a single study using one batch of ketamine.

Exploring Individual Variability: Relating Tasks to Symptoms

Modern neuroscientific techniques enable us to dissociate the individual variability in symptoms from the variations in dose received. We can use these techniques to test hypotheses about the cognitive and neural bases of specific symptoms.

Previous studies have found links between latent behavioral processes and positive and negative symptoms. This study used functional neuroimaging to provide an assay for prediction error.

Cognitive neuropsychiatric models of psychotic symptoms can be tested using fMRI data to examine the relationship between magnitude of brain response and behavioral competence during cognitive tasks. This relationship is critical in interpreting any correlation between brain responses and ketamine-induced symptoms.

Exploring Individual Variability: Using Tasks to Predict Symptoms

Prediction error is a driving force in learning that has been inferred but not observed in many years. However, recent studies have revealed the presence of prediction error in the midbrain, striatum, and frontal cortex.

The prediction error signal in the brain can be estimated even though it does not have any explicit behavioural correlates. This signal has proven useful in exploring the brain basis for the anomalous perceptions and beliefs that occur in association with acute ketamine administration.

A weaker auditory mismatch negativity (MMN) response at baseline (in the absence of ketamine) was predictive of more severe positive symptoms under ketamine. The study used a predictable pattern of tones with a deviant occurring once every ten tones, but subjects with a smaller average MMN response might have an increased sensitivity to prediction error and learn the predictability of their sensory inputs more effectively.

Negative symptoms are associated with a reduction in processing capacity of prefrontal cortex, which leads to difficulties in sustaining concentration and maintaining task set. Subjects who had inefficient prefrontal function during the attention and working memory tasks showed increased vulnerability to negative symptoms under high-dose ketamine.

Predicting thought disorder and pseudo-hallucinations can be done by exploring the relationship between self-monitoring and positive symptoms. Self-monitoring is relevant to the genesis of auditory hallucinations, delusions of passivity, and formal thought disorder. We explored the idea that ketamine may produce thought disorder and hallucination-like perceptual aberrations. We found that subjects who required the greatest level of activation to carry out a task engaging self-monitoring were those most vulnerable to these particular effects of the drug.

The prediction error model, which considers brain responses predictive of delusional ideas, may also be useful in considering brain responses predictive of ketamine-induced symptoms and experiences.

Individual Variability and the Bayesian Brain: a Model that Links Cognitive Variability to Symptom Susceptibility

We will consider whether it is possible to use baseline measures of task-specific neural response to predict how an individual will respond to ketamine, and whether this might show us how cognition and symptoms emerge from the same systems whose variable responsiveness predicts this vulnerability.

The model conceives of the brain as a system that learns to predict its subsequent inputs with increasing certainty. This concept has been formally developed to account for reinforcement and causal learning, as well as the receptive field properties of the visual system.

Implementing Bayes in the Brain

The Bayesian hierarchical model predicts that NMDA receptors are involved in prediction and AMPA receptors are involved in prediction errors. These predictions are sent upward through the hierarchy and the prediction errors are sent downward through the hierarchy.

The distribution of NMDA and AMPA receptors on the surface of neocortical neurons suggests that NMDA receptors near the soma might regulate the amplification of synaptic signals resulting from AMPA receptor activation on remote dendritic sites.

Striatal GABAergic interneurons control the output of the striatum to the frontal cortex via the direct and indirect pathways. Fast spiking interneurons have large rectifying AMPA-mediated currents, but no detectable NMDA-mediated component, while low-threshold spiking interneurons have small rectifying AMPA currents and an NMDA receptor-mediated component.

The glutamatergic synapse and its regulation is complex and includes high steady-state glutamate levels in the extrasynaptic space, active exchange mechanisms, and the modulatory role of mglur2 receptors on feedforward driving inputs.

Another metabotropic receptor, subtype 5 (mglur5), interacts with the NMDA receptor via postsynaptic density scaffolding proteins homer and shank, and is associated with risk for schizophrenia. Mglur5-knockout mice display schizophrenia-like phenotypes, and mglur5 antagonists are psychotomimetic in human subjects.

Slower NeuromodulatorsFPrecision, Uncertainty, and Learning

Rapid glutamatergic and GABAergic neurotransmission represent prediction error, while slower neuromodulators encode the precision of prediction errors. Inferences and learning are intimately connected, but dissociable to some extent.

Dopamine modulates signal to noise response properties of neural units encoding prediction error, and is also involved in maintaining surprise in working memory and updating expectancies through changes in synaptic function.

The general model of learning, inference, perception, and consciousness is an oversimplification, but it provides a framework for considering symptoms of psychosis.

Prediction Error and Psychosis

Positive symptoms of psychosis involve gross misrepresentations of reality, and involve a prediction error dysfunction. Passivity phenomena occupy a hinterland between hallucinations and delusions, and involve a prediction error dysfunction as well.

Psychotic symptoms include changes in the intensity of perceptual experience, inappropriate relatedness between external and internal stimuli and events, and a sense of vividness and novelness that renders even mundane experiences vivid, novel, and important.

Ketamine, a psychotomimetic drug, induces transient psychotic states by blocking NMDA receptors and enhancing AMPA receptor signaling. This causes the world to become highly unpredictable, and delusions arise as explanatory schemes.

Predictive Learning, Negative Symptoms, and Cognitive Control

Negative symptoms include deficits in motivated behavior, a lack of pleasure for previously enjoyed acts, events, and experiences, and working memory dysfunction, which may be related to a reduction in behavioral engagement and output, as well as an inability to logically order thoughts and produce coherent and communicative speech.

We contend that negative symptoms arise when no viable predictions can be produced or maintained, and that social isolation (presumably related to negative symptoms in at least some patients) appears to incubate the generation of psychotic symptoms.

We have considered a prediction error-based model of schizophrenia and shown how observations of neural responses may help us to understand ketamine-induced psychosis.

EXTENDING THE MODEL IN TIME: LEARNING FROM THE EFFECTS OF CHRONIC AND REPEATED KETAMINE USE

The acute ketamine model is not without its critics, but changes in synaptic function can happen that rapidly and ketamine effects curtail rapidly. There is no evidence of adverse effects subsequent to a single acute ketamine challenge in healthy subjects.

Ketamine abusers show escalating cognitive changes and symptom experiences, despite a lack of sensitization across a small number of administrations, widely spaced in time and in the experimental setting.

CHRONIC KETAMINE ABUSE

Ketamine was first used recreationally in the 1970s in North America. John C Lilly and Marcia Moore both published books on their experiences with ketamine in 1978, and both spiraled toward larger and larger consumption, before dying from exposure in a forest near their home.

Ketamine users have cognitive impairments that correlate with their degree of exposure, and they engage in predictive responding toward irrelevant stimuli, which is a superstitious responding style. Ketamine use increases the severity of delusional ideation when subjects are followed up longitudinally for 1 year.

Genetic Effects

Genetic variability between individuals might influence response to drugs. This idea was described in 1959 and may underpin the individual differences we observed.

Family history of alcoholism blunts the psychotomimetic effects of ketamine, with fewer dysphoric effects, fewer positive symptoms, and milder negative symptoms compared to individuals without such a family history.

Apolipoprotein E4 is involved in lipid metabolism and may play a role in the psychotomimetic effects of ketamine. However, the relationship between the presence of the epsilon 4 allele and the severity of ketamine-induced psychosis in healthy volunteers has yet to be established empirically.

Studies into emergence phenomena following ketamine anesthestia explored personality as an explanatory factor, but found inconsistent results. The lack of a significant association between schizotypal personality and ketamine responses suggests that schizotypy and acute ketamine-induced psychosis are mediated by different neurochemical mechanisms.

Ketamine abuse may engender more convincing delusions than acute administration, but it is impossible to carry out a placebo-controlled study. It is possible that more schizotypal subjects are drawn to engage in repeated ketamine use.

Future work should triangulate the relationship between ketamine-induced psychopathology, personality measures, and cortical neurochemistry, and explore other relevant personality measures.

Other Candidate Mechanisms for Individual Variability in Ketamine Response

There are numerous biological mechanisms through which the observed variability in ketamine response may be mediated, including gene copy number repeat variation and alternative splicing.

Alternative splicing involves the formation of multiple messenger RNAs from the same gene, and has been implicated in the generation of interindividual differences in drug responses. For example, alternatively spliced NMDA receptors are more sensitive to ketamine in vitro.

The subunit composition of NMDA receptors may affect ketamine response, but the Balb/c mouse strain does not have any significant alteration in mRNAs.

Alternative splicing of mRNAs might influence ketamine responses through interactions with microRNAs. Variations in miRNA function might also be involved in the individual differences in ketamine response, in particular those effects that involve aberrant neural learning.

CONSILIENCE?

We aim to provide a consilient account of acute and chronic NMDA antagonist administration in terms of the Bayesian model we introduced. The brains function as Helmholtz machines, and predictions are made by NMDA signaling, AMPA and GABA receptor signaling.

Inference and learning are both glutamatergically mediated, but acute and chronic ketamine exposure differentially affects these two processes. We posit that persistent changes in glutamatergic function alter the parameters of dopaminergic uncertainty coding and thus affect subsequent inferences.

Acute ketamine impairs the specification of top-down priors and enhances presynaptic glutamate release, engendering aberrant prediction error responses. Lamotrigine, which blocks presynaptic glutamate release, reverses the psychotomimetic effects of ketamine.

Ketamine increases cortical cholinergic function, which may be a mechanism for the perceptual aberrations and attentional capture that characterize its effects. Furthermore, formal learning theories and preclinical behavioral neuroscience tell us that increasing cortical acetylcholine engages new explanatory learning and tunes attention to stimuli with unexpected and unpredictable consequences.

Ketamine-treated subjects retain insight into their experiences, but do not experience longer-term effects of the drug once it has been metabolized. Ketamine also affects the homeostatic regulation of glutamatergic synapses, which may explain why chronic ketamine use is associated with delusions rather than hallucinations.

Ketamine use is associated with delusions that worsen over time, but not hallucinations. The crystallization process involves sensitized dopamine function in the basal ganglia and glutamatergic reinforcement of new explanatory priors.

Dopamine antagonism may ameliorate the psychopathology generated by chronic ketamine use, despite having no effect on the psychopathology induced by acute ketamine administration.

Repeated ketamine use is associated with a commonality of experience across exposures, which may be explained by changes in how the genome can be expressed. This process may be mediated in part by proteins called histones that combine with DNA to form chromatin.

Ketamine use may alter brain structure more macroscopically, such as by engendering excitotoxic cell death, and may be a mediator of the delusions associated with chronic self-administration.

In the Bayesian hierarchical model, negative symptoms result from a failure to specify coherent expectations with sufficient certainty in order to justify motivated behaviors. Genes may predispose individuals to particularly severe negative symptoms on ketamine.

CLINICAL UTILITY OF THE MODEL

NMDA antagonist model psychoses inspired the development of the first antipsychotic drug without a direct effect on dopamine function, an mglur2/3 agonist, and other adjunct treatments, such as lamotrigine, that reverse the neural and phenomenological effects of ketamine.

The Bayesian analysis suggests that propanolol may be a novel treatment option for psychosis. Propanolol attenuates the excessive glutamate release engendered by NMDA antagonists and has been used to treat psychosis with varying results.

FUTURE DIRECTIONS

The Bayesian conception of psychosis provides a framework for considering the co-occurrence of positive and negative symptoms, and may provide a useful step in developing novel antipsychotic medications. The Bayesian model of psychosis may be used to target specific psychotic symptoms both theoretically and empirically, and may allow for the use of fMRI and cognitive science in clinical decision-making and perhaps targeted, personalized medicine.

CONCLUSION

NMDA receptor antagonists can cause both positive and negative psychotic symptoms. With repeated chronic administration, delusions can develop.

We used a hierarchical Bayesian model of cognition, comportment, and brain function to explain why ketamine does not engender hallucinations, why the psychotomimetic effects of ketamine are not reversed by haloperidol, and why highly schizotypal individuals may be particularly susceptible to chronic ketamine use.