Subanaesthetic ketamine and altered states of consciousness in humans

Abstract

Background

Despite its designation as a ‘dissociative anaesthetic,’ the dissociative and psychoactive effects of ketamine remain incompletely understood. The goal of this study was to characterise the subjective experiences and accompanying EEG changes with subanaesthetic doses of ketamine.

Methods

High-density EEG was recorded in 15 human volunteers before, during, and after subanaesthetic ketamine infusion (0.5 mg kg⁻¹ over 40 min), with self-reported measures of altered states of consciousness obtained after ketamine exposure. Sensor- and source-level EEG changes were analysed with a focus on spectral power and regional changes.

Results

Ketamine-induced altered states were characterised predominantly by dissociative experiences such as disembodiment and ego transcendence; sensory disturbances were also common. Ketamine broadly decreased low-frequency power, with mean reductions largest at alpha (8–12 Hz) in parietal (−0.94 dB, P<0.001) and occipital (−1.8 dB, P<0.001) channel clusters. Significant decreases in alpha were identified in the precuneus and temporal-parietal junction.

Conclusions

Ketamine induces altered states of consciousness during periods of reduced alpha power in the precuneus and temporal-parietal junction. Modulation of these temporal-parietal loci are candidate mechanisms of the psychoactive effects of ketamine, given that this region is involved in multisensory integration, body representation, and consciousness.

Editor's key points.

• •The dissociative and psychogenic effects of ketamine are poorly understood.
• •The effects of a subanaesthetic dose of racemic ketamine to produce altered states of consciousness were assessed using a validated questionnaire in 15 healthy volunteers monitored with high-density EEG.
• •Ketamine-induced altered states were characterised by dissociative experiences such as disembodiment and ego transcendence.
• •Ketamine decreased low-frequency EEG power, with significant decreases in alpha identified in the precuneus and temporal-parietal junction.
• •Modulation of these temporal-parietal loci are candidate mechanisms for the psychoactive effects of ketamine.

Ketamine has anaesthetic,1, 2 analgesic,1, 2 antidepressant,³ and psychoactive⁴ properties, including perceptual distortions, cognitive impairment, and feelings of disconnection from the body and environment.⁵ Given the phenomenology and behavioural features associated with ketamine, it was originally called a ‘dissociative anaesthetic.’1, 6 Although ketamine use has remained popular in a variety of clinical settings,⁷ the dissociative nature of ketamine remains incompletely understood. Studying validated, self-reported measures of altered states of consciousness and accompanying neurophysiologic changes may provide more precise characterisation of dissociative states related to ketamine and advance understanding of the neural mechanisms of ketamine's dissociative properties.

Recent data have provided mechanistic insights into pharmacologically induced dissociative experiences and other altered states of consciousness. In particular, alpha power correlates inversely with the experience of ego-dissolution during exposure to psilocybin and lysergic acid diethylamide.8, 9, 10 This correlation might account for changes in ego integrity as synchronised alpha rhythms have been posited to contribute to self-awareness during normal, waking consciousness.8, 10, 11, 12 Furthermore, alpha rhythms have been suggested to play a central role in conscious orientation to space and time via coordinated, oscillatory activity that subserves neural information processing.¹³ Clinically, disrupted alpha rhythms have been observed in association with postoperative delirium¹⁴ and hepatic encephalopathy.15, 16 Ketamine also suppresses alpha rhythms,17, 18, 19 suggesting the possibility that a reduction in alpha power is a neural mechanism by which ketamine induces dissociative states.

To test this hypothesis, we conducted a study of subanaesthetic doses of ketamine in healthy volunteers with high-density EEG and validated measures of altered states of consciousness.²⁰ We hypothesised that ketamine-induced dissociative states, characterised by validated altered states of consciousness scales, occur concurrently with reduced EEG alpha power. Although we previously examined dose-dependent effects of ketamine during subanaesthetic, anaesthetic, and recovery states,¹⁷ we did not report altered states of consciousness data or source-level neuroanatomical findings, as the purpose of our prior work was to investigate anaesthetic-specific effects.

Methods

This study was approved by the University of Michigan Medical School Institutional Review Board (HUM00061087), and written informed consent was obtained from all participants. Study procedures were conducted at the University of Michigan Medical School. Healthy volunteers (seven males, eight females, age 20–35 yr) were recruited using flyers posted throughout the medical school and main hospital. We published a distinct EEG analysis for 10 of these participants that examined dose-dependent effects of ketamine on oscillatory and connectivity patterns, with particular focus on anaesthetic dosing.¹⁷ This previous publication included neither EEG source analysis nor self-reported altered states of consciousness scales, which were reserved for the current manuscript.

Study population

Participants were eligible if they were ASA physical status 1, aged 20–40 yr, with BMI <30 kg m⁻² and had no predictors of a difficult airway. Exclusion criteria included cardiovascular disease, cardiac abnormalities, hypertension, obstructive sleep apnoea, asthma, respiratory illness, gastro-oesophageal reflux, history of drug use (or positive drug screen before experiment), family history of problems with anaesthesia, neurologic disorders, psychiatric disorders, or current pregnancy.

Sample size justification

This was an exploratory study investigating neurophysiologic changes accompanying ketamine-induced self-reported altered states. The sample size of 15 subjects is similar to or larger than related studies investigating neurophysiologic and neuroanatomical associations with ketamine.21, 22

Experimental protocol

Participants fasted from food and drink for 8 h before the experiment. A complete medical history and physical examination were conducted at the start of the study protocol. A minimum of two anaesthesiologists and two researchers were present throughout each experiment. Standard monitors were applied and i.v. catheters inserted before EEG recording. Procedures occurred in a well-lit operating room with participants lying supine. The first period (baseline) was 5 min of rest with eyes open and 5 min of rest with eyes closed. The second period was 40 min with eyes closed during continuous infusion of subanaesthetic (0.5 mg kg⁻¹ total) racemic ketamine (Ketamine Hydrochloride, Hospira, Inc., Lake Forest, IL USA), followed by a brief physical examination and ondansetron (8 mg i.v.) for nausea prophylaxis. We chose this dosing regimen because of its common use in psychiatry. This was followed by 5 min of rest with eyes open and 8–10 min for completion of the altered states of consciousness questionnaire (Fig. 1). Within 48 h of the study period, participants completed online questionnaires assessing altered states of consciousness experiences.

Fig 1.

Schematic representation of the study protocol. The protocol began with a baseline, 5-min eyes-closed period, followed by a 40-min ketamine infusion (0.5 mg kg⁻¹ total dose). After completion of the infusion, ondansetron was given for nausea and vomiting prophylaxis. The altered states of consciousness questionnaire was then administered after a short rest period. EEG data were recorded throughout the entire protocol as outlined above. ASC, altered states of consciousness.

Altered states of consciousness psychometrics

A 71-item altered states of consciousness questionnaire was used with 62 questions indexing 11 altered states of consciousness subscales²⁰ including the following: experiences of unity, spiritual experience, blissful state, insightfulness, disembodiment, impaired control and cognition, anxiety, complex imagery, elementary imagery, audiovisual synaesthesia, and changed meaning of percepts. An additional two scales (nine items)—transcendence of time and space and ineffability—were included from the Revised Mystical Experiences Questionnaire.²³ Terms and definitions are outlined in Table 1. For all items, the response scale was from 0 (no, not more than usual) to 10 (yes, very much more than usual) with 11 total discrete response options. Scale scores were the average of items within each scale. Cronbach's alpha (α) was used to assess intra-scale reliability for each altered state of consciousness scale, with higher α values reflecting better internal scale reliability.²⁴ The study scale questionnaire was completed at the end of ketamine infusion (study score) and then again 48 h after experiment completion (lifetime history score).

Table 1.

Altered states of consciousness terms and definitions are presented. *The term ‘definition’ describes phenomenological content of each respective metric. Terms generally correlate with positive or negative experiences and moods20, 23

Term	Definition*	Experience/mood
Experiences of unity	Eternal oneness, beyond contradictions, merging of self and environment	Positive
Spiritual experience	Religious cognition, sense of awe, presence of a higher power	Positive
Blissful state	Experiences of boundless pleasure, which may include bliss, peace and love	Positive
Insightfulness	Profound, clear, original thoughts	Positive
Complex imagery	Vivid complex visual patterns such as scenes and imagery; from past experiences or fantasy, occurring with eyes closed or total darkness	Positive
Elementary imagery	Seeing regular patterns with eyes closed or in total darkness	Positive
Audiovisual synaesthesia	Audiovisual abnormalities including shapes, colours of things, or both changing with sounds and noises	Positive
Changed meaning of percepts	Everyday things gain a special and strange meaning; things get more emotionally engaging	Positive
Disembodiment	Floating, being out-of-body, not having a body	Negative
Impaired control and cognition	Cognitive difficulty and disorganisation, decreased agency, paralysis, isolation	Negative
Anxiety	Fear, terror, distortion, threat, strangeness	Negative
Transcendence of time and space	Loss of usual sense of time, space, and current location, including being outside of time, no spatial boundaries, and timelessness	–
Ineffability	The experience cannot be adequately described or done justice to with words	–

EEG acquisition and processing

Data were acquired with 128-channel EGI Hydrocel Nets (Eugene, OR, USA) digitised continuously at 500 Hz with a vertex reference. Channel impedance was kept at <50 kΩ as recommended by the manufacturer. Processing was performed with EEGLAB,²⁵ Chronux (http://chronux.org/),26, 27 and custom MATLAB (MathWorks, Natick, MA USA) scripts. For each recorded session, data were band-passed filtered [eegfiltnew, zero-phase, Hamming-windowed FIR filter, 3301 points (6.6 s), 0.5 Hz transition, 0.5 55⁻¹ passband, −6 dB cutoff: 0.25 55.25⁻¹]. Electrodes on the lowest parts of the face and head were removed to obtain a cleaner signal and better decomposition of neural independent components, leaving 98 channels. All artefact-detection procedures for channels, time periods, and independent components included partially automated and expert visual review (T.B.). Bad or noisy channels were detected and removed by visual inspection, leaving 86–91 channels per session, and the data were average referenced. Two periods of interest were extracted: baseline rest with eyes closed (5 min) and subanaesthetic infusion (40 min). The subanaesthetic period was broken into three sequential 12-min blocks: block 1 (2–13 min), block 2 (14–25 min), and block 3 (26–37 min). Brief periods (1–10 s) with large artefacts (movement, multi-channel artefacts) were removed. Remaining data were segmented into 3-s trials. Mean remaining trial counts were 66 (nine) for the baseline period, and for the subanaesthetic period 170 (19) for block 1, 176 (14) for block 2, and 171 (17) for block 3. After independent component analysis (pop_runica, default settings), non-neural components (eye, muscle, non-dipolar, single-trial or single-channel focus) were removed.²⁸ A mean of 14 neural independent components (range=8–30) was retained per participant. These numbers are consistent with the mean number of valid, independent neural components generally retained (≤15) with high-density, high-quality data.29, 30 Channels were interpolated to the full 98 channel subset.

For statistical analyses of the subanaesthetic period, only block 3 trials were used because we assumed the ketamine infusion reached pharmacological steady-state conditions by that point. As part of a pilot study in human volunteers, we assessed plasma concentration using the same infusion dose (0.5 mg kg⁻¹). After a 40-min infusion (i.e. at the end of block 3), plasma ketamine concentration was approximately 180 ng ml⁻¹ compared with approximately 50 ng ml⁻¹ in block 1.

Based on the lowest remaining trial count of one subject for the baseline condition (50 trials), graphics and analyses were computed based on a random subset of trials drawn, for each participant, from the baseline (50 trials) and block 3 periods (50 trials). The power spectral density was calculated for each trial with Chronux for the two experimental periods at each of five frequency bands (delta: 1–4 Hz, theta: 4–7 Hz, alpha: 8–12 Hz, beta: 15–30 Hz, and gamma: 30–48 Hz). All channel-data figures consist of the median of EEG power spectral metrics across all participants. Grand-average spectrograms and baseline-adjusted spectrograms are for a midfrontal channel cluster and for a posterior cluster including both occipital and parietal channel clusters. Data trials represented in the spectrograms and topomaps are sequential but not contiguous in time because of trials that were removed because of artefacts and the random selection of a subset of trials.

Differences in subanaesthetic power (relative to baseline) were computed by subtracting the mean power of the baseline period from the subanaesthetic period, then dividing by the power of the baseline period and transforming by 10×log10 (i.e. decibels, dB). Power spectral density was converted to decibels to allow direct comparisons between participants, channels, and frequency bands by minimising the effects of power-law scaling and within-subject variation in baseline power. Baseline conversion was done for each channel, sample, and frequency bin, and baseline bin-specific estimates were averaged across all available samples.

EEG channel-level analyses

We tested for differences between experimental periods by entering single-subject spectral power values into 2×4 repeated-measures analysis of variance (anova) with a condition factor (baseline, subanaesthetic) and a location factor (anterior, central, parietal, and occipital). Based on a priori expectations31, 32 and channel-level results,17, 18, 19 we focused only on midline occipital and parietal channel clusters at alpha and beta, and the midline anterior and parietal theta channel clusters. The median was computed for each of four channel clusters (Supplementary Fig. S1). A separate anova was computed for each frequency band. Factor effects with P<0.05 were considered significant. Post hoc pairwise comparisons were Bonferroni-corrected. Baseline-adjusted power was submitted to anova analyses with frequency band as a single five-level factor.

EEG source analysis

Standardised low-resolution, distributed-source imaging of current density (A m⁻²) was computed for the same five frequency bands as in channel analyses using default settings in the standardised Low Resolution Brain Electromagnetic Tomography (sLORETA) toolbox,³³ which generates a three-dimensional solution space classified as volumetric. These analyses are restricted to the cortical grey matter and hippocampal regions. A voxel was labelled as grey matter if it met the following three conditions: its probability of being grey matter was higher than that of being white matter, its probability of being grey matter was higher than that of being cerebrospinal fluid, and its probability of being grey matter was >33%. Only grey matter voxels that belonged to cortical grey matter and hippocampal regions were used. Analysis resulted in 6239 voxels at 5 mm resolution restricted to cortical grey matter. A subanaesthetic vs baseline paired-groups t-test contrast was computed on ‘voxel-wise’ normalised images (equivalent to a relative power transformation) resulting in the log of F-ratio for each voxel. Statistical non-parametric mapping (5000 randomisations) was applied to generate critical thresholds and P-values.³⁴ Significant voxels were those that passed below 0.05 (single-tailed, ketamine < baseline as per a priori expectations; two-tailed testing was not undertaken for this analysis). Significant voxels (described in Results) were reduced to four non-overlapping 5-mm radius voxel clusters by merging significant voxels that were within 5 mm of each other, and then taking their centre voxel as the centre of the cluster. For computation of source-estimated difference metrics for correlation with altered states of consciousness, single-subject solutions for each condition were voxel-normalised and logged, and the difference was computed by subtracting baseline values from subanaesthetic values.

Exploratory correlation analysis

Associations of altered states of consciousness with EEG spectral power and source estimates were computed via Spearman's correlations to mitigate effects of non-normal and outlier data points while retaining all data. Correlations were computed between altered states of consciousness study scores and the baseline-adjusted ketamine EEG metrics. Only correlations ≥0.5 in magnitude are presented, as previously reported for exploratory neurophysiologic analyses involving ketamine and other psychedelics.³⁵ Corrected P-value thresholds were calculated by dividing the initial significance value by the number of hypotheses tested (Bonferroni's method). Statistics were computed with IBM SPSS 22 software (IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY, USA: IBM Corp.).

Results

Channel-level analyses

Grand-average power spectra, spectrograms, and topoplots are presented in Fig 2. Global decreases in beta, alpha, and delta were readily apparent with ketamine (Fig. 2A). Relative power increases were only observed in the gamma bandwidth; otherwise, power was reduced throughout all low-frequency channel clusters (Fig. 2B). Decreases in alpha power were the most prominent, and this was demonstrated in both frontal and posterior channel clusters (Fig. 2B–D). In terms of quantitative analysis, there was a significant effect of condition factor, location factor, and condition-location interaction for each bandwidth (Table 2). The largest absolute power reductions were found in occipital channel alpha (−1.8 dB, P<0.001) and parietal channel alpha (−0.94 dB, P<0.001) (Table 3). Post hoc tests of the condition by location interactions indicate that no statistically significant power differences were found between anterior and posterior channels for both alpha baseline and ketamine conditions, arguing against an amplitude-driven phenomenon (Supplementary Table S1). Delta and beta power were also uniformly reduced across all channel clusters (Table 3, Fig. 2B). Topoplots further demonstrate anterior and posterior power reductions throughout low-frequency bandwidths (Fig. 2E). See Supplementary Figure S2 for single-subject spectrograms across baseline and ketamine periods.

Fig 2.

Grand average power spectrum, spectrograms, and topomaps (n=15). (A) The power spectrum values for baseline and subanaesthetic conditions (block 3). (B) Differences between baseline and subanaesthetic periods (block 3) at five frequency bands and four channel clusters. Negative values represent mean relative decreases in power from baseline, and positive values represent mean relative increases from baseline. Error bars represent standard deviation of these relative means. Lines with stars show significant post hoc comparisons between frequency bands across all channels. (C) Frontal and posterior channel-cluster spectrograms for baseline and subanaesthetic periods (blocks 1–3). (D) Frontal and posterior channel-cluster baselined spectrograms for subanaesthetic periods (blocks 1–3). (E) Topomaps for baseline (row 1) and subanaesthetic block 3 (row 2) periods.

Table 2.

Channel-level analysis of variance (anova) effects and interactions for condition (subanaesthetic vs baseline), frequency band (delta, theta, alpha, beta, gamma), and electrode clusters (anterior, central, parietal, occipital)

	Factor	df	F	P	eta2	Power
Delta	Condition	1,14	19.56	0.001	0.58	0.98
	Location	3,42	13.26	0.000	0.48	1
	Interaction	3,42	8.84	0.000	0.38	0.99
Theta	Condition	1,14	3.94	0.067	0.22	0.46
	Location	3,42	8.14	0.000	0.36	0.99
	Interaction	3,42	4.34	0.009	0.23	0.85
Alpha	Condition	1,14	14.85	0.002	0.51	0.95
	Location	3,42	6.47	0.001	0.31	0.96
	Interaction	3,42	6.85	0.001	0.32	0.97
Beta	Condition	1,14	22.42	0.000	0.61	0.99
	Location	3,42	15.78	0.000	0.53	1
	Interaction	3,42	5.4	0.003	0.27	0.91
Gamma	Condition	1,14	5.74	0.031	0.29	0.61
	Location	3,42	20.07	0.000	0.58	1
	Interaction	3,42	6.46	0.001	0.31	0.95

Table 3.

Descriptive statistics by condition, band, and channel cluster. Stars depict significant post hoc pairwise comparisons between conditions for each specific region. P-values are Bonferroni-corrected within each band. *P<0.05,**P<0.01,***P<0.001,****P<0.0001. sd, standard deviation

Bandwidth	Channel cluster	Baseline		Ketamine
		Mean	sd	Mean	sd
Delta	Anterior	0.39	0.21	0.21****	0.09
	Central	0.40	0.26	0.24***	0.15
	Parietal	0.38	0.23	0.20***	0.08
	Occipital	0.73	0.56	0.35***	0.20
Theta	Anterior	0.35	0.21	0.32	0.23
	Central	0.39	0.35	0.35	0.30
	Parietal	0.40	0.43	0.23	0.14
	Occipital	0.70	0.68	0.42**	0.26
Alpha	Anterior	1.13	1.08	0.34***	0.43
	Central	0.81	0.85	0.34***	0.48
	Parietal	1.28	1.13	0.34***	0.39
	Occipital	2.88	3.65	1.08***	1.59
Beta	Anterior	0.078	0.044	0.039****	0.020
	Central	0.071	0.049	0.037***	0.026
	Parietal	0.080	0.052	0.038****	0.022
	Occipital	0.133	0.095	0.074***	0.050
Gamma	Anterior	0.012	0.004	0.013	0.004
	Central	0.011	0.005	0.012*	0.005
	Parietal	0.010	0.003	0.011	0.004
	Occipital	0.017	0.008	0.023*	0.013

Source estimation metrics

Significant effects in the voxel-wise normalised current source estimates were found only for the alpha band. These included significant decreases in 22 voxels mainly at the right lateralised temporal-parietal junction (TPJ) and superior parietal regions, summarised in Fig 3A and Supplementary Table S2. Significant voxels were reduced to four small non-overlapping 5 mm clusters with Brodmann area (BA) designated accordingly (Fig. 3B): medial temporal gyrus (BA 39, X=55, Y=−60, Z=10), inferior parietal lobe (BA 40, X=55, Y=−45, Z=25), supramarginal gyrus (BA 40, X=50, Y=−50, Z=20), and precuneus (BA 39, X=20, Y=−75, Z=50). Non-thresholded solutions for alpha (Supplementary Fig. S3) suggest deactivations at anterior cingulate, mid-cingulate, midline premotor, and dorsolateral prefrontal regions.

Fig 3.

Alpha-band source estimates of differences between subanaesthetic and baseline periods. (A) Consists of current source estimates, with significant voxels (P<0.05) in dark blue, kernel scaling=7. From left to right, (A) includes a right hemisphere lateral view, a posterior view, and an inflated posterior view of the right hemisphere. (B) Depicts the four source clusters. Black squares on smaller images are zoomed in within the larger black squares. The clusters are shown on the right hemisphere only. sLORETA, standardised Low Resolution Brain Electromagnetic Tomography.

Altered states of consciousness psychometrics

Altered states of consciousness scores were of good intra-scale reliability (α>0.7) except for six scales, which had lower reliabilities (α>0.6): spiritual experience, complex imagery, elementary imagery, changed meaning of percepts, insightfulness, and experiences of unity. One item was dropped from each scale, based on low (α<0.2) or negative item-scale correlations, which brought reliability for each scale to 0.6 or higher. Overall, altered states of consciousness study scores were higher than lifetime history scores on most scales, except for spiritual experience (see Table 4 for a summary of study and lifetime mean scores, reliabilities, difference scores, and t-test values). Both study and lifetime score differences were highest for disembodiment, transcendence of time and space, and complex imagery (Table 4).

Table 4.

Altered states of consciousness study and lifetime scales. Scale labels, number of items per scale, reliabilities, descriptive statistics, and t-test (study vs lifetime) differences. **P<0.01,***P<0.001. α, Cronbach's scale reliability alpha; M, mean; sd, standard deviation

Scale name	Items	Study			Lifetime			Difference	t-test	P-value
		α	M	sd	α	M	sd
Experiences of unity	4	0.64	4.76	2.12	0.87	0.95	0.95	3.81	5.81	***
Spiritual experience	2	0.72	1.87	1.92	0.86	2.32	1.84	-0.45	−0.71
Blissful state	2	0.81	5.77	2.72	0.74	2.57	1.62	3.20	3.66	**
Insightfulness	2	0.71	5.53	2.29	0.31	2.14	1.50	3.39	4.81	***
Disembodiment	3	0.72	6.72	2.07	0.72	0.52	0.57	6.20	10.97	***
Impaired control and cognition	7	0.77	4.46	1.81	0.81	1.21	1.00	3.24	6.79	***
Anxiety	5	0.90	2.12	2.01	0.78	1.01	0.79	1.11	1.93
Complex imagery	2	0.62	7.73	2.02	0.92	2.54	2.22	5.20	7.30	***
Elementary imagery	2	0.68	5.46	2.87	0.96	2.04	1.66	3.43	4.67	***
Audiovisual synaesthesia	3	0.77	5.56	2.72	0.82	0.88	1.15	4.67	6.64	***
Changed meaning of percepts	2	0.75	3.57	2.50	0.85	1.39	1.29	2.17	3.34	**
Transcendence of time and space	6	0.84	6.62	2.08	0.83	1.21	0.99	5.41	8.75	***
Ineffability	3	0.79	6.53	2.58	0.91	2.12	1.77	4.41	4.48	**

Exploratory analysis of altered states of consciousness correlations with channel and source metrics

Correlations between altered states of consciousness study scores in relation to channel and source metrics are presented in Fig 4, Fig 5. EEG channel-altered states of consciousness correlation analyses were only performed in alpha and beta, as corrected comparisons revealed significant spectral differences primarily within these bandwidths (Table 3). Although significant reductions in delta power were also noted, ocular artefact can contribute to delta activity³⁶; so we deferred delta correlation analysis. The strongest EEG channel-altered states of consciousness correlations (Spearman's rho magnitude ≥0.5) occurred between elementary imagery scores and central alpha, parietal alpha, anterior beta, and parietal beta (Fig. 4). In terms of alpha source metrics, each of the four voxel clusters (i.e. TPJ and precuneus regions) negatively correlated with audiovisual synaesthesia, and the precuneus negatively correlated with transcendence of time and space (Fig. 5). However, these correlations did not reach statistical significance after correction for multiple comparisons (88 correlations for channel-level analyses and 44 source-level comparisons). There were also possible negative correlation trends for disembodiment and transcendence of time and space with the TPJ (Supplementary Table S3).

Fig 4.

Correlation coefficients (Spearman's rho) presented for exploratory EEG channel-altered states of consciousness correlation analysis. Only correlation values with magnitude ≥0.5 are presented. None of these correlations reached statistical significance when correcting for multiple comparisons (P=0.05/88 correlations=0.00057). Disemb, disembodiment; AVSyn, audiovisual synaesthesia; ICC, impaired control and cognition; CI, complex imagery; EI, elementary imagery; CMP, changed meaning of percepts; Insight, insightfulness; Bliss, blissful state; Unity, experiences of unity; TST, transcendence of time and space; Ineff, ineffability.

Fig 5.

Spearman's rho correlations presented for exploratory EEG source-altered states of consciousness analyses. Correlation values with magnitude ≥0.5 are presented. None of these correlations reached statistical significance when correcting for multiple comparisons (P=0.05/44 correlations=0.0011). Disemb, disembodiment; AVSyn, audiovisual synaesthesia; ICC, impaired control and cognition; CI, complex imagery; EI, elementary imagery; CMP, changed meaning of percepts; Insight, insightfulness; Bliss, blissful state; Unity, experiences of unity; TST, transcendence of time and space; Ineff, Ineffability; MTG, medial temporal gyrus; PCUN, precuneus; SMG, supramarginal gyrus; IPL, inferior parietal lobe.

Discussion

We assessed the neurophysiologic effects of subanaesthetic ketamine during altered states of consciousness. Ketamine broadly decreased alpha power in parallel with self-reported measures of dissociation (e.g. disembodiment, transcendence of time and space), supporting our original hypothesis and suggesting a common signature and potential mechanism of altered states of consciousness for psychoactive drugs (e.g. psilocybin) with distinct molecular mechanisms. Furthermore, source analysis revealed significant alpha current reduction at the precuneus and right TPJ. These findings are significant, as the temporal-parietal region may be part of a posterior cortical ‘hot zone’ of neuroanatomical correlates of consciousness.³¹ In fact, altered temporal-parietal connectivity correlates with similar dissociative states induced by classic (i.e. serotonergic) psychedelic drugs.9, 10 These findings also fit with the current understanding of the TPJ, which is known to play a role in body ownership, multisensory integration and—when lesioned—out of body experiences.37, 38, 39 Thus, functional lesions involving specific temporal-parietal loci are candidate mechanisms of ketamine-induced altered states of consciousness that require further investigation.

Our results align with previously described neurophysiologic effects of ketamine. Resting-state magnetoencephalography and EEG studies also demonstrated low-frequency reductions, particularly in the alpha and beta bandwidths.18, 19 Such decreases in alpha might cause dysfunction in alpha-based gating or inhibition of bottom-up sensory processing,40, 41 and beta reductions might represent dysfunction of hierarchical processing and top-down predictive signaling.⁴¹ Indeed, altered visual integration and processing were found in relation to both alpha and beta power (i.e. inverse correlations with elementary imagery). Gamma increase, also found in this study, might reflect cortical disinhibition and bottom-up prediction error.41, 42 These altered oscillations might collectively result in disrupted information processing and discoordination of higher-order networks, including the default mode network.21, 40

Source analysis revealed reduced alpha current density within the TPJ and precuneus, posterior cortical regions that are posited to have significant roles in consciousness. The precuneus is implicated in self-centred imagery, spatial cognition, multimodal processing, and episodic memory;37, 38 precuneus deactivation has been previously reported in magnetoencephalography studies of subanaesthetic ketamine.¹⁸ Exploratory correlation analyses suggest that transcendence of time and space and audiovisual synaesthesia inversely correlate with alpha current in the precuneus, further implicating this region in ketamine-induced altered states. As mentioned, alpha current was also significantly reduced in the right TPJ, which is a principal brain hub associated with bodily self-consciousness, out-of-body experiences, multisensory and vestibular integration, and peripersonal space.37, 39, 43 Disruption of the right TPJ in particular impairs environmental perception and beliefs.⁴⁴ Correlation analyses demonstrated a possible inverse correlation with the right TPJ and audiovisual synaesthesia, and inverse correlation trends were noted for disembodiment and transcendence of time and space (Supplementary Table S3). Interestingly, ketamine administration has been shown to increase bilateral temporal-parietal functional connectivity,⁴⁵ which is similar to connectivity changes associated with lysergic acid diethylamide-induced ego-dissolution.¹⁰ Chronic ketamine use also increases TPJ white matter abnormalities.⁴⁶ Thus, functional modulation of temporal-parietal networks, specifically involving regions such as the precuneus and TPJ, may contribute to ketamine-induced altered states of consciousness. Such modulation might be related to the suppression of alpha rhythms, which, in the context of altered states of consciousness, are postulated to a play a role in maintaining ‘ego integrity’.⁸ That is, alpha power positively correlates with self-awareness and introspection,¹² whereas disruption of alpha correlates with experiences of ego disintegration during psychedelic-induced altered states of consciousnesss.⁸ The current study adds to the literature by demonstrating that suppression of alpha preferentially occurs in key temporal-parietal loci—namely, the TPJ and precuneus.

These findings also have implications for surgical patients who experience psychological side-effects from ketamine. The PODCAST trial recently reported that ketamine can induce negative experiences (e.g. nightmares, hallucinations) perioperatively with routine clinical dosing.⁴⁷ In the current study, audiovisual sensory changes were reported with ketamine; exploratory analysis revealed that such changes might inversely correlate with parietal alpha and beta power, and audiovisual synaesthesia might negatively correlate with alpha at specific TPJ loci. Thus, visual and sensory disturbances experienced during exposure to ketamine could relate to temporal-parietal dysfunction, particularly at the TPJ region and involving the alpha bandwidth. Interestingly, these changes are not necessarily characterised as negative.²⁰ With regard to phenomena such as nightmares, we did not assess sleep disturbances after this study.

Limitations

The correlation findings should be interpreted as exploratory and hypothesis-generating. The sample size was small and statistical significance was not reached after correcting for multiple comparisons. However, these results can now be used to design targeted follow-up correlation analyses with specific brain regions of interest (i.e. TPJ loci). The lack of a placebo condition or placebo questionnaire is also a significant limitation; we did not include a placebo condition because the intent of the original EEG study was to identify anaesthetic-specific effects of ketamine. Given the lack of placebo group, it is also difficult to account for meditative and other internally related responses that could have occurred during the 40-min eyes-closed period. However, similar studies with placebo groups revealed significant altered states of consciousness changes only during ketamine administration.19, 22, 48 Random EEG trial sampling with replacement (i.e. bootstrapping) leading to range estimates was not performed, as the purpose of the current study was not to examine the variability of estimates across trials within single-subjects or between the two conditions. However, spectral estimates derived from steady-state conditions have been found to produce high internal reliability.49, 50, 51, 52 Other limitations include potential self-selection biases in the sample, cognitive impairment during study assessment (which might lead to misunderstanding of the altered states of consciousness questions), and a possible confounding influence of collecting lifetime altered states of consciousness scores after rather than before the study experience. Furthermore, relatively low dosage, bright lighting, and lack of privacy might have led to weaker than expected ego-dissolution and ‘peak’ effects,⁵³ especially considering that ego-dissolution effects are dose-dependent⁵⁴ and the lack of an initial drug bolus can lead to fewer altered states of consciousness experiences.³² Altered states of consciousness dimensions and ego-dissolution require continued refinement in terms of psychometric validity and reliability, temporal effects, pharmacological modulation, and individual differences in altered states of consciousness propensity.54, 55 Dissociative states (e.g. disembodiment, ego transcendence) were the main focus of this study, though we acknowledge that these experiences can be more broadly characterised as altered states. Differentiating dissociative states from other phenomena requires further neuropsychological and neurophysiologic investigation.

Conclusions

The current study yielded novel findings regarding parallel neural and phenomenological effects of subanaesthetic ketamine dosing, demonstrating reduced alpha and beta power during dissociative experiences and audiovisual perturbations. Ketamine suppressed alpha activity around the right TPJ, a region of known relevance for bodily consciousness and multisensory integration. A continued multi-dimensional approach (e.g. self-reported questionnaires, neurophysiologic analysis, neuroimaging assessment) is warranted to advance our mechanistic understanding of the psychoactive effects of ketamine.

Authors' contributions

Study design: P.E.V., T.B., R.E.H., G.A.M.

Managed study protocol and operations: P.E.V., T.B., E.J., V.T., P.P., G.A.M.

Data acquisition: P.E.V., T.B., E.J., V.T., P.P., G.A.M.

Data analyses: P.E.V., T.B., A.N., K.C., E.S., R.E.H., G.A.M.

Statistical analyses: P.E.V., T.B., A.N., K.C., E.S.

Critically reviewed the manuscript for intellectual content and approved the final version: all authors.

Complete access to all study data and confirm responsibility and integrity for data analyses: all authors.

Declaration of interest

R.E.H. has previously received grant support from Pfizer, Inc., for work involving functional neuroimaging. The other authors declare that they have no conflicts of interest.

Funding

National Institutes of Health (T32GM103730 to P.E.V. as trainee and G.A.M. as PI; R01GM111293 to G.A.M. and R.E.H.). Also supported by the Department of Anesthesiology, University of Michigan.

Handling editor: H.C. Hemmings Jr

Editorial decision: March 19, 2018

Appendix A. Supplementary data

The following are the supplementary data related to this article:

figs1.

figs2.

figs3.