Introduction

Neural oscillations in the delta (1–4 Hz) and slow (<1 Hz) frequency bands are generally indicators of unconsciousness1,2,3 or severely reduced consciousness4, especially when activity is diffuse and of high amplitude. Nonetheless, some pharmacological agents are known to enhance activity at these frequencies while sparing consciousness. Such paradoxical pharmacological dissociations (PPDs) are useful for falsifying putative spectral electroencephalogram (EEG) markers of conscious and unconscious states, e.g., if a substance can induce cortical slow waves during wakeful consciousness, then slow waves are not a universal indicator of loss of consciousness. Furthermore, understanding how such PPDs are possible is crucial for understanding how neural activity relates to consciousness.

A recent review5 described a range of conditions in which high amplitude delta oscillations (HADOs) appear in the awake and conscious state including, most notably, Angelman syndrome, a rare genetic condition characterized by diffuse HADOs during conscious wakefulness in children6. However, the above review focused largely on neurological conditions and less so on pharmacological manipulations. Here, we continue reviewing paradoxical EEG patterns by describing pharmacological drug challenges that induce HADOs without loss of consciousness. Our mini-review comprehensively covers literature in both non-human animals and human participants from over 70 years ago to the present day. By doing so, we illuminate PPDs in which the oscillatory regime exhibited by the EEG may be safely and reversibly switched independently of consciousness in the laboratory, e.g., as a means of validating candidate biomarkers of consciousness. Given their mechanistic similarities and the lack of meaningful boundaries between frequency bands7, we treat both delta and slow oscillations as belonging to the delta band for the purpose of this mini-review; thus, our principal focus is on oscillations occurring at frequencies at or below 4 Hz. Due to the inverse relationship between oscillatory frequency and spatial extent7, delta activity is often extensive across the scalp. For this reason, our mini-review does not particularly emphasize the scalp location of delta (but see the column “Spatial topography” in Table 1).

Table 1 Key EEG and MEG studies of pharmacologically-induced delta activity in humans.
Full size table

Depending on the location of the reference electrode, either the peak or trough of each oscillatory cycle in the delta band corresponds to a down state of cortical silence8. During these periodic down states, cortical neurons switch to a more hyperpolarized resting membrane potential, increasing the difficulty of spiking and communicating with other neurons9,10. Conversely, up states result in a widespread spiking of cortical neurons at the opposite phase of the delta cycle. This dynamical regime can be represented by a simplified two-state model (described as the “monolithic brain” by Tononi11) in which the cortex is either up (spiking) or down (not-spiking).

Theories that attribute consciousness to the information-richness of cortical dynamics, such as integrated information theory in both its strong12 and weak13 formulations, explain the loss of consciousness that typically occurs under these conditions in terms of the loss of information integration. According to this view, HADOs may cause such loss of information by restricting the number of possible states visited by the brain. Yet even in rare genetic disorders where EEG oscillations are incongruent with the typical oscillatory regime seen during consciousness, informational measures (such as permutation entropy) reliably indicate the presence or absence of consciousness14. To learn how consciousness and complexity may persist under such unusual condition, one may use pharmacological manipulations that reversibly induce abnormal EEG patterns in healthy adults, thus allowing for comparisons of normal and abnormal cortical dynamics within the same individuals. The aim of such research should be to find a common denominator of consciousness14 across both normal and abnormal cortical dynamics.

Below, we review pharmacological manipulations that might be used for this purpose. Although the level consciousness is not entirely unimpaired by some substances (e.g., tiagabine) reviewed below, participants do not lose consciousness all together in the studies described, and in some cases, certain dimensions of consciousness are even enhanced (e.g., using psychedelic tryptamines). In the latter context, recent evidence15 shows that awake EEG delta activity correlates with global functional connectivity as assessed by concurrent functional magnetic resonance imaging (fMRI). Some substances which we review have potential as valuable experimental tools for revealing the common denominator of consciousness which persists even as an abnormal oscillatory regime emerges.

Antiepileptic drugs

Antiepileptic drugs carbamazepine and tiagabine are reported to induce delta oscillations during wakeful consciousness, albeit while inducing mild sedation. Carbamazepine’s main mechanism of action is inhibition of voltage-gated sodium channels16. An early report of the delta-enhancing effects of carbamazepine came from a small study of patients who took a 400 mg dose over a period of 35 days, which reported increases in EEG delta and theta power relative to baseline17. Soon after, a similar effect was also inferred from an observational study of schizophrenia patients, many of whom regularly took carbamazepine with their antipsychotic medications and showed diffuse delta activity18. Although delta activity was diffuse, the authors noted that it was sometimes frontally predominant, similar to an earlier report19 of intermittent frontal rhythmic delta activity in eight schizophrenia patients taking antipsychotic drugs (exact medications not specified). However, a more recent study of acute carbamazepine (600 mg) in 15 healthy volunteers only found significant increases in resting EEG power in the theta, alpha, and beta bands20, and it is thus unclear if carbamazepine’s effect on EEG delta power is seen outside of chronic dosing.

Unlike carbamazepine, tiagabine is a gamma-aminobutyric acid (GABA) reuptake inhibitor21. Besides enhancing delta power during wakefulness in young adult volunteers, as seen diffusely both in sensor space22 and source space23, tiagabine also induces episodes of hypersynchronous delta activity during wakefulness in rodents24, and in elderly humans, it increases delta power not only during NREM sleep, but also during REM sleep, i.e., when episodes of dreaming, and therefore some degree of consciousness, are common25. The degree of delta power enhancement induced by tiagabine during wakefulness in healthy volunteers, referenced to placebo, is highly variable, with some individuals exhibiting magnetoencephalogram (MEG) delta power increases >1000%26—comparable to the degree of delta enhancement in the Angelman syndrome EEG27—while others show much more modest changes (Fig. 1). Tiagabine is thus an attractive drug for studying PPDs that resemble the Angelman syndrome EEG phenotype, but its usefulness may be limited by tolerability issues28 (see below).

Fig. 1: MEG power change (averaged across all sources and epochs) induced by tiagabine (15 mg) in 14 healthy volunteers.
figure 1

Data are from publicly available26 source-localized resting-state MEG recordings23. The experiment was approved by the UK National Research Ethics Service (South East Wales) and all participants gave written informed consent. One outlier participant was excluded due to an extremely large power increase for which we could not confirm that tiagabine was the sole cause. a Tiagabine induces clear increases in cortical oscillatory delta activity, as evidenced by the MEG power changes relative to placebo. The mean power change (red) induced by tiagabine is greatest at 3.3 Hz, featuring a more than 5-fold (430%) power increase from placebo. The largest power change we observed (after excluding one outlier, not shown here) was a 1200% power increase at 3.1 Hz. b The majority of participants (8/14) showed maximum power changes in the delta band (<4 Hz). The mean frequency of the maximum power change was 4.3 Hz, and the median frequency was 3.4 Hz. Source data supporting panel b are presented in Supplementary Data 1.

Full size image

A recent small study by Darmani et al.20 investigated the “degraded state of consciousness” induced by tiagabine. Prior to each of four transcranial magnetic stimulation (TMS)-EEG sessions, 15 healthy volunteers were administered tiagabine, carbamazepine, brivaracetam (another antiepileptic drug), or placebo, experiencing each condition once. Similar to another report of partial sedation, confusion, disorientation, hallucinations, and amnesia caused by 15 mg tiagabine28, three participants had difficulty tolerating tiagabine and did not complete the subsequent TMS-EEG session. Tiagabine was found to significantly increase resting power in all frequency bands examined (delta, theta, alpha, and beta). Using the perturbational complexity index (PCI), a highly accurate measures of one’s level of consciousness based on the cortical response to TMS29,30,31, tiagabine was found to induce a PCI value (computed using the state transition algorithm30) intermediate between consciousness and unconsciousness. Similarly, other studies have shown that ketamine (not an antiepileptic, but rather an N-methyl-D-aspartate receptor antagonist) yields a PCI value that is intermediate between that typically seen in wakeful consciousness and unconsciousness, though it should be emphasized that PCI values obtained with ketamine were not significantly different from wakefulness in a small n = 6 sample31,32 (see the “other substances” section below for further discussion). By comparison, the other drugs investigated by Darmani et al.20 (brivaracetam and the ostensible delta-enhancer carbamazepine) did not significantly alter delta power or PCI values, though carbamazepine significantly increased power in other bands (theta, alpha, and beta). While both carbamazepine and tiagabine are generally known to increase delta EEG power (as shown in Fig. 1 for tiagabine), it is unclear whether results in the Darmani et al. study truly corresponded to delta-specific effects or rather non-specific increases in broadband power.

GABAB receptor agonists

Gamma hydroxybutyrate (GHB), sodium oxybate (i.e., the sodium salt of GHB), and baclofen are GABAB receptor agonists that induce large increases in EEG delta power during sleep and wakefulness33. Unlike baclofen, GHB is also an agonist of the eponymous GHB receptor, an endogenous neurotransmitter34, and both a precursor and a metabolite of GABA35. One of the oldest published reports of pharmacologically induced HADOs is from a 1966 study of GHB’s effects in adult volunteers36. Volunteers given large doses (35–63 mg/kg) of GHB were awakened from an unconscious state induced by GHB and displayed HADOs while behaviorally responsive36. However, consciousness could only be maintained briefly (10–15 s) due to GHB’s strongly sedating effects. More recently, in an attempt to better understand the neural underpinnings of this PPD, GHB has been studied at lower doses that allow for uninterrupted wakefulness (20–35 mg/kg) in healthy male volunteers using sophisticated analyses such as source localization, global omega complexity, and functional connectivity analysis37. Findings of this analysis suggest that low-frequency oscillations induced by GHB emerge from the posterior cingulate cortex, a region that is sometimes implicated in consciousness due to its role in the default mode network, and a later study by the same group also reported that GHB increased delta power in posterior cingulate cortex, as well as medial prefrontal cortex, parahippocampal gyrus, and fusiform gyrus, during NREM sleep38. Results of both studies also suggested that GHB sedation may be related to increased lagged phase synchronization between posterior cingulate cortex and other regions37,38.

Likely because the above study of GHB chose a relatively low GHB dose, it only detected a trend-level increase in EEG delta power induced by GHB compared to placebo37. However, other studies from the 2010s33,38 have reinforced the results of older literature36,39, reproducing GHB-induced increases in relatively low-frequency EEG power during NREM sleep, REM sleep, and wakefulness. Evidence of enhanced activity at similar frequencies has also been obtained for baclofen—a similar GABABergic compound—in both humans33,40 and mice41,42, though most studies have focused on NREM sleep and suggest an effect in the theta band more so than the delta band. As with tiagabine20, GHB and baclofen are likely to be useful pharmacological tools for validating PCI and other biomarkers of consciousness under conditions of abnormal cortical dynamics, and both drugs may be better tolerated than tiagabine28, even when TMS is combined with pharmaco-EEG to compute PCI43,44. Underscoring the potential usefulness of these substances, GABAergic inhibition is a key parameter of perturbational complexity, as has been revealed by experiments that estimated PCI using electrical stimulation in ferret cortical slices while synchronous states (low PCI) were induced using both GABAA and GABAB receptor antagonists45. Interestingly, the above study induced low-frequency cortical oscillations by blocking GABAB receptors, whereas baclofen and GHB induce low-frequency cortical oscillations while activating GABAB receptors; thus, tuning the ratio of excitation to inhibition in either direction away from its optimum may induce hypersynchronous HADOs. However, because these experiments were performed in cortical slices, effects of GABAB receptor antagonism on consciousness could not be studied.

Anticholinergics

During states of wakefulness and arousal, acetylcholine acts as a neuromodulator and desynchronizes cortical activity46, thus diminishing delta oscillations. Conversely, anticholinergic drugs enhance the presence of delta oscillations. This has been noted for decades using the muscarinic antagonist atropine, which was first observed in 1952 to induce a PPD in dogs47. Several years later, a report also noted EEG slowing with 10 mg atropine in a small group of human volunteers, although the effect was reported to be more in the 5–8 Hz theta range48. Slowing effects have been noted using the muscarinic antagonist scopolamine in two small human studies that noted increased delta power49,50,51,52. This delta band effect was, however, not replicated by at least one EEG study53. More recently, a small study of healthy elderly participants found evidence of enhanced MEG connectivity (phase locking value) in the delta band under scopolamine54.

Because anticholinergic substances derived from the genus Datura, including atropine and scopolamine, are known to produce deliriant effects, cognitive impairment, and tachycardia, human research is highly limited. For instance, one study of scopolamine50 in healthy adult volunteers noted that following a 0.75 mg dose, participants “complained about subjective symptoms which were definitely unpleasant” and “were restless and confused, with mild muscular incoordination and evident drowsiness when not adequately aroused.” Due to these concerns, anticholinergic drugs—unlike the substances reviewed in previous sections—are not recommended for validating biomarkers of consciousness in humans. However, careful research using anticholinergic substances in non-human animals may reveal useful insights into the compatibility of delta oscillations with consciousness. A recent study of atropine and scopolamine in cats55, for example, found that the EEG pattern induced by anticholinergics, while resembling NREM sleep with respect to delta oscillations and sleep-like spindles, had a distinct profile in the gamma band, featuring gamma bursts and greater gamma coherence than NREM sleep. The authors concluded55 that these gamma EEG features, sometimes associated with consciousness56,57, might “explain why the animals remain awake in spite of the presence of slow waves and spindles.”

Tryptamines

Of all substances that enhance delta oscillations while sparing consciousness, perhaps the most puzzling are the psychedelic tryptamines that actually increase the richness and intensity of visual imagery, imagination, and perceptual meaning58,59 while also inducing delta activity. This crucially demonstrates that, unlike some prior examples, delta-enhancing drugs do not necessarily induce drowsiness and partial-sedation. Most notable of these tryptamines is N,N-dimethyltryptamine (DMT), a principal ingredient in ayahuasca, a psychedelic brew made by the indigenous people of the Amazon basin60. DMT is also an endogenous neurotransmitter61, though its function as such is poorly understood. The first ever rigorous, placebo-controlled EEG study of DMT62 in humans recently revealed a surprising increase in delta power that occurred with intravenous DMT administration in 13 participants during eyes closed EEG. Moreover, this increase in delta power after subtracting the 1/f background of the EEG power spectrum correlated positively with subjective ratings of both the visual and overall intensity of the DMT experience. EEG signal complexity measured using the Lempel-Ziv algorithm63 (a common proxy for the level of consciousness, e.g., used to compute PCI) increased despite the introduction of delta oscillations which might otherwise be expected to reduce signal complexity by increasing the regularity and predictability of the signal. The above EEG findings were recently replicated by the same research team in a newer study using 17 participants each given 20 mg of intravenous DMT during simultaneous fMRI while wearing an eye mask15. Notably, increases in frontal delta power following DMT infusion correlated with increases in global fMRI connectivity. This replication also demonstrated that EEG delta power in the experiment did not significantly correlate with visual analog scales of drowsiness (in fact, the sign of the correlation coefficient was always negative). Separately, a study of inhaled DMT in naturalistic settings outside of the laboratory using 35 participants64 also reported an increase in delta power induced by DMT compared with an eyes closed resting baseline condition (as in the work by Timmerman et al.62, participants kept their eyes closed through the experiment). Interestingly, DMT enhanced gamma power and coherence relative to the baseline recording, thus suggesting a similar mechanism as that hypothesized for scopolamine55, whereby gamma-band activity sustains consciousness despite background delta activity. Finally, DMT in a naturalistic setting64 also recapitulates the laboratory finding of increased Lempel-Ziv complexity with DMT compared to the eye closed baseline EEG.

Other tryptamines have also been reported to increase delta power despite their psychedelic effects. In mice65, the DMT derivative 5-methoxy-DMT enhances theta and gamma power in medial prefrontal cortex and delta power in primary visual cortex. The latter finding supports the hypothesis5 that enhanced delta power observed in powerful psychedelic states relates to hyperpolarization of primary visual cortex when decoupled from the external sensory environment66, as may also occur in non-pharmacological dream states during REM sleep67,68,69 (though the recent findings of Timmermann et al.15 may point instead toward global hyperconnectivity). Additionally, 4-hydroxy-DMT (better known as psilocin, the pharmacologically active agent of psychedelic mushrooms) enhances 4 Hz power in mice during both wakefulness and REM sleep70; the latter effect was interpreted by the study’s authors as “a bleeding of NREM-like activities (delta waves, spindles, reduced gamma) into REM sleep.” Despite this, psilocybin (the prodrug of psilocin) does not appear to enhance delta MEG power23 or alter PCI values71 in human volunteers.

Other substances

Besides the substances reviewed above, we also wish to highlight possible effects of other miscellaneous substances on delta EEG activity, albeit with caveats.

Subanesthetic doses of ketamine, a dissociative drug72, are known to induce a 1–3 Hz rhythm in layer 5 of retrosplenial cortex in mice (also demonstrated with the dissociative compound phencyclidine; here, the term “dissociative” refers to detachment or dissociation from one’s body or surroundings, rather than a PPD)73. Similarly, a 3–4 Hz rhythm in posterior cingulate and isthmus cingulate cortex was recently demonstrated using subanesthetic ketamine in human epilepsy patients74. Note that, as with ketamine, posterior cingulate cortex has also been identified as a plausible source of delta activity induced by GHB37,38. Delta frequency rhythms induced by ketamine, while linked to dissociation in both mice73 and humans74, appear limited to deep posteromedial areas observable only with invasive recordings—no enhancement of delta activity is seen in humans with subanesthetic doses of ketamine using noninvasive recordings, e.g., EEG75,76 or MEG23. In fact, at least one study reported a significant delta power decrease caused by subanesthetic ketamine in human EEG77. Studies which do report delta EEG activity in scalp recordings under ketamine have used anesthetic doses78,79,80,81.

Next, basmisanil, a selective GABAA-α5 negative allosteric modulator, enhances 4 Hz EEG power during wakefulness in patients with Down syndrome, while also suppressing power in the beta band82. However, in healthy participants, EEG power enhancement with basmisanil occurs more so in the theta band (6–9 Hz)83. By the same token, benzodiazepines (i.e., GABAA positive allosteric modulators) generally have the opposite EEG effects—EEG suppression at delta/theta frequencies83,84,85,86 (but see an exception for delta here77) and enhancement at beta frequencies77,83,84,87,88—accompanied by partial or full sedation89. This demonstrates a very different PPD in which depressant and hypnotic effects occur without delta oscillations and are instead marked by high frequency activity more typical of cortical activation.

Another relevant substance is the acetylcholinesterase inhibitor donepezil. In healthy elderly adults, one study reported that an acute dose (5 mg) of donepezil caused a significant increases in EEG delta power alongside memory impairments90. On the other hand, chronic treatment with donepezil improves cognition and memory91,92 and reduces delta power during both wakefulness93 and REM sleep94 in patients with Alzheimer’s disease. While donepezil may have no obvious effect on waking consciousness, both it95 and another acetylcholinesterase inhibitor, galantamine96, affect sleeping consciousness by stimulating lucid dreaming, but this effect has not yet been studied in relation to EEG delta activity using acetylcholinesterase inhibitor compounds.

Near the end of our list are two sedating compounds, gabapentin and gaboxadol. Gabapentin, an analgesic and antiepileptic drug, is known to influence delta EEG activity. In rats, gabapentin normalizes deficits in delta power during slow-wave sleep induced by ethanol97. In humans, chronic daily use of gabapentin increases the percentage of delta power relative to a control group that received no drug or placebo98. However, it remains unknown whether gabapentin acutely enhances delta EEG power during wakefulness, especially at lower doses that do not require gradual titration. Next, gaboxadol is a muscimol derivative99 that has, like gabapentin, been shown to enhance low-frequency EEG power during NREM sleep100. In mice, gaboxadol administration leads to slow wave activity during wakefulness and REM sleep101. In humans, gaboxadol has been shown to enhance low-frequency power (including delta) during overnight wakefulness in the context of sleep restriction102. Delta enhancement has also been noted alongside enhancement of the theta and alpha bands in resting-state MEG recordings from male volunteers challenged with gaboxadol22. However, the amplitude of this gaboxadol-induced delta activity in humans is unclear and should be characterized by further studies.

Finally, opioids have been shown to increase delta EEG power103,104, but it is unclear to what extent this occurs in humans outside of anesthetic doses, and, moreover, this effect is inconsistent, with some studies reporting attenuation of delta activity by opioids105,106.

The effects of these and other substances on delta oscillations and consciousness warrant further investigation, as in many of the above cases, delta enhancement appears limited to a particular anatomical region, arousal state, or chronic dosing.

Outlook and perspectives

Until recently, large delta and slow waves were commonly interpreted as evidence of unconsciousness. This interpretation is challenged by many documented exceptions, both from neurological disorders (reviewed previously5) and pharmacological manipulations. At first glance, some studies we have reviewed may appear to show that delta-enhancing drugs (e.g., tiagabine, GHB, scopolamine) simply induce drowsiness and partial-sedation. However, this seemingly uncomplicated relationship is contradicted by studies of psychedelic tryptamines such as DMT, which induce hypervivid experiences59 that often feel “more real than everyday normal consciousness”107,108. Future work is necessary to determine under what circumstances a gradual increase in delta activity may also indicate a gradual loss of consciousness and why this is not always the case (Box 1).

Although most pharmaco-EEG/MEG studies we cited used small samples (N ≤ 20, see Table 1), their findings are nonetheless sufficient to demonstrate that delta waves are in principle possible during consciousness. Because these substances allow PPDs to be observed in healthy volunteers, they may prove invaluable for future pharmaco-EEG experiments that seek to validate EEG biomarkers of conscious state (e.g., PCI) in neurologically-typical adults. In particular, GABABergic substances such as GHB may prove to be safe and well-tolerated models for abnormal cortical dynamics manifesting in the delta band. Following recent work that has used rare neurogenetic disorders to model abnormal cortical dynamics for the purpose of validating EEG biomarkers of consciousness14, such substances could be administered in healthy adult individuals to further validate EEG markers of consciousness such as the entropy of spontaneous EEG signals or perturbational complexity.