More than meets the eye: The role of sensory dimensions in psychedelic brain dynamics, experience, and therapeutics

This pre-print (2022) reviews the current body of evidence surrounding the effects psychedelics have on low-level sensory dimensions of experience, and corresponding brain dynamics, particularly visual dimensions. It is proposed that psychedelic-induced alterations in low-level sensory dimensions of experience are not entirely causally reducible to alterations in high-level dimensions, but rather co-occur in a dialogical interplay and play a causally relevant role in determining high-level alterations and therapeutic outcomes.

Authors

Aqil, M.
Roseman, L.

Published

November 10, 2022

Neuropharmacology

meta Study

Links

Read Paper DOI Google Scholar

Abstract

Psychedelics are undergoing a major resurgence of scientific and clinical interest. While multiple theories and frameworks have been proposed, there is yet no universal agreement on the mechanisms underlying the complex effects of psychedelics on subjective experience and brain dynamics, nor their therapeutic benefits. Despite being prominent in psychedelic phenomenology and distinct from those elicited by other classes of hallucinogens, the effects of psychedelics on low-level sensory - particularly visual - dimensions of experience, and corresponding brain dynamics, have often been disregarded by contemporary research as ‘epiphenomenal byproducts’. Here, we review available evidence from neuroimaging, pharmacology, questionnaires, and clinical studies; we propose extensions to existing models, provide testable hypotheses for the potential therapeutic roles of psychedelic-induced visual hallucinations, and simulations of visual phenomena relying on low-level cortical dynamics. In sum, we show that psychedelic-inducead alterations in low-level sensory dimensions 1) are unlikely to be entirely causally reconducible to high-level alterations, but rather co-occur with them in a dialogical interplay, and 2) are likely to play a causally relevant role in determining high-level alterations and therapeutic outcomes. We conclude that reevaluating the currently underappreciated role of sensory dimensions in psychedelic states will be highly valuable for neuroscience and clinical practice, and that integrating low-level and domain-specific aspects of psychedelic effects into existing nonspecific models is a necessary step to further understand how these substances effect both acute and long-term change in the human brain.

Unlocked with Blossom Pro

Research Summary of 'More than meets the eye: The role of sensory dimensions in psychedelic brain dynamics, experience, and therapeutics'

Introduction

Psychedelics such as psilocybin, DMT, LSD and mescaline produce a wide range of effects across the brain functional hierarchy, from low-level sensory alterations to changes in mood and high-level cognition. Contemporary explanatory frameworks commonly emphasise a reduction in top-down control and increased bottom-up influence of sensory information, and many models centre on the role of 5-HT2A receptor activation. Nevertheless, current dominant accounts—exemplified by the REBUS model and by thalamic-gating proposals—tend to treat psychedelic sensory effects, especially visual phenomena, as nonspecific or epiphenomenal, and therefore underplay modality-specific cortical dynamics and their potential causal role in higher-level changes and therapeutic outcomes. Aqil and colleagues set out to reassess the role of low-level sensory dimensions—particularly vision—in psychedelic brain dynamics, subjective experience, and therapeutics. The paper reviews evidence from neuroimaging, pharmacology, questionnaires and clinical studies, presents computational considerations and simulations relevant to visual phenomena, and proposes testable hypotheses about how sensory alterations might causally interact with high-level processes and contribute to therapeutic effects. The authors argue for integrating domain-specific, low-level mechanisms into existing nonspecific models to better understand both acute psychedelic phenomenology and longer-term clinical outcomes.

Methods

This paper is a narrative synthesis rather than an empirical trial. Aqil and colleagues draw on multiple evidence streams: neuroimaging findings (fMRI, EEG/travelling-wave analyses), pharmacological data (PET, autoradiography, electrophysiology), psychometric questionnaire studies (notably the Altered States of Consciousness [ASC] and Mystical Experience Questionnaire [MEQ]), clinical outcome reports, and computational models/simulations related to retinotopic and harmonic cortical dynamics. The authors integrate these sources to evaluate how low-level sensory activity could both co-occur with and causally influence high-level alterations. The extracted text does not report a formal systematic search strategy, databases searched, date ranges, or explicit inclusion/exclusion criteria; the methodology therefore appears to be a targeted, conceptual review that brings together representative empirical findings and theoretical work. Computational arguments are invoked to link retinotopic cortical structure and oscillatory dynamics to visual phenomenology (for example, how pairs of cortical harmonics can produce apparent motion). Where empirical studies are described, the authors report key observed patterns and statistical associations from those studies as presented in the primary literature, but do not report pooled quantitative meta-analytic methods.

Results

Across the domains reviewed, several convergent findings are reported that support a substantive role for sensory cortex activity in psychedelic states. Pharmacological evidence shows high 5-HT2A receptor binding and mRNA expression in primary visual cortex (V1) across cortical layers, with particular V1-specific patterns (for example, high expression in layer IV within V1). Rodent electrophysiology implicates 5-HT2A receptors in modulating V1 responses to visual input. Neuroimaging studies under psychedelics report increased cerebral blood flow and BOLD signal in visual cortex, increased functional connectivity (FC) between primary visual areas and the rest of the brain, and patterns of activity resembling responses to spatially localised visual stimuli during eyes-closed imagery tasks. EEG and travelling-wave analyses show shifts in occipital oscillatory dynamics: a reduction in dominant alpha power and altered travelling-wave properties, with some psychedelic-induced dynamics resembling those induced by intense visual stimulation. The authors note that alpha suppression can be interpreted either as reduced top-down inhibition or as secondary to the presence of high-frequency, content-encoding activity in visual cortex. Whole-brain analyses indicate a pattern of hypo-connectivity in high-level association networks alongside hyper-connectivity in sensory networks; these patterns are correlated such that stronger sensory coupling accompanies stronger associative decoupling. Specific correlations reported include a significant association between DMN disintegration and ego-dissolution (p = 0.03) and a near-significant correlation of ego-dissolution with the visual medial network (p = 0.07). One notable fMRI finding is a significant correlation between increased V1 FC under LSD and subjective ratings of both elementary and complex visual hallucinations; in some analyses increased V1 FC also correlates with several high-level subjective dimensions (unity, spiritual experience, insightfulness) and with ego-dissolution, suggesting non-exclusive relationships between low-level activity and high-level phenomenology. Questionnaire-based findings show that high-level dimensions (mystical-type experiences, oceanic boundlessness) more strongly predict therapeutic outcomes than generic measures of perceptual intensity. However, the ASC family of instruments displays an asymmetry: high-level factors are phenomenologically specific and carry embedded emotional valence, whereas sensory factors are broader, often valence-neutral, and may conflate distinct sensory contents. This asymmetry can obscure the contribution of sensory content to therapeutic processes. The authors also cite a study of ayahuasca where intensity of perceptual experience predicted remission among responders, indicating that sensory intensity can relate to clinical outcomes in some contexts. On the modelling side, the authors describe how retinotopic and harmonic analyses can account for classic psychedelic visuals (for example, geometric motifs and 'breathing walls') via excitation-induced oscillatory dynamics shaped by lateral connectivity and cortical eigenmodes. They also contrast simple (abstract, geometric) hallucinations—argued to arise largely from bottom-up cortical dynamics—with complex (figurative, memory-laden) hallucinations that likely recruit top-down, culturally and personally shaped content. Taken together, the reviewed evidence is consistent with: (1) a direct pharmacological action of psychedelics in low-level sensory cortex; (2) sensory-cortex dynamics that can be broadcast widely and influence global brain activity; and (3) meaningful associations between low-level activity and high-level subjective states and therapeutic-relevant outcomes, although high-level dimensions often show stronger predictive power in questionnaire studies.

Discussion

Aqil and colleagues interpret the assembled evidence as supporting a dialogical model in which low-level sensory alterations and high-level changes interact bidirectionally rather than one strictly causing the other. They argue that contemporary nonspecific models—those that chiefly characterise psychedelics as relaxing high-level priors or disinhibiting thalamic gating—do not fully capture modality-specific features of psychedelic phenomenology, particularly the strikingly visual character of many psychedelic experiences. The authors position their synthesis relative to earlier frameworks by noting points of complementarity and limitation. REBUS-type models emphasise relaxation of high-level priors and increased bottom-up signalling, but the review highlights that 5-HT2A receptor density in primary visual cortex and observed occipital dynamics provide plausible mechanisms for primary visual excitation that could themselves drive or amplify high-level changes. Similarly, thalamic-gating accounts point to reduced subcortical filtering but do not explain the visual specificity; integrating cortical, modality-specific mechanisms could resolve some of these explanatory gaps. Computational and retinotopic models provide mechanistic accounts for specific visual motifs that are otherwise difficult to derive from purely top-down or subcortical gating explanations. The paper acknowledges key uncertainties and limitations. The causal direction between sensory-cortex activity and high-level network changes remains empirically unsettled; resting-state paradigms and FC measures lack directionality, complicating inference about broadcasting versus receiving roles of sensory regions. Questionnaire evidence is limited by measurement asymmetries that may underrepresent sensory content and valence; the extracted text does not report a systematic search methodology for the review, which constrains claims about comprehensiveness. The authors also caution that not all sensory phenomena will be therapeutically relevant, and that high-level processes—especially interpretation and integration—remain crucial for clinical benefits. In terms of implications, the authors propose testable hypotheses and research directions: more balanced psychometric instruments and qualitative analyses to capture sensory content and valence; neuroimaging and electrophysiological studies designed to probe directionality (for example, causal influence of occipital sources on association networks); experimental work differentiating abstract versus figurative visuals and their distinct mechanistic bases; and computational modelling linking retinotopy, cortical harmonics and subjective content. Clinically, they suggest that appreciating sensory dimensions could inform therapeutic protocol design, set and setting considerations, and the evaluation of candidate non-hallucinogenic analogues. Overall, the authors recommend extending existing models to explicitly incorporate modality-specific, low-level cortical dynamics alongside high-level mechanisms.

Conclusion

The authors conclude that psychedelic-induced alterations of low-level sensory dimensions—especially vision—are neither merely epiphenomenal nor fully reducible to high-level changes. Instead, sensory-cortex dynamics and high-level processes likely coexist in reciprocal interaction, and accounting for both causal directions is necessary to advance mechanistic models. They argue that integrating specific, modality-dependent cortical mechanisms with existing nonspecific frameworks will be valuable for neuroscience and for the design of future clinical research and therapeutic practice.

View full paper sections

INTRODUCTION

Psychedelics -serotonergic hallucinogens such as psilocybin, DMT, LSD, and mescaline -are undergoing a major resurgence of scientific, clinical, and popular interest. These substances elicit an extremely varied array of effects, spanning the entire brain functional hierarchy, from low-level sensory, to mood, and high-level cognitive alterations. Multiple theoretical frameworks are available describing how psychedelics achieve their characteristic effects on brain dynamics and subjective experience. While there is no universal agreement, perhaps one of the few widely accepted characterizations of the psychedelic state is as one of reduced top-down cognitive control, and increased bottomup influence of sensory information. However, besides the involvement of the 5-HT2A receptor, it is still not fully clear how this state is achieved, and what the relations between low-level and high-level alterations would be, if any. One of the most prominent existing models, the RElaxed Beliefs Under pSychedelics [REBUS] model, suggests that the relaxation of top-down expectations ("high-level priors" in the language of predictive coding) encoded by high-level regions and networks, in particular the Default Mode Network, may be the defining characteristic of psychedelic brain dynamics and subjective experience. Complementary or alternative models have focused on subcortical structures such as the thalamus and the claustrum, which are proposed to have roles of gating sensory inputs, or orchestrating brain dynamics, respectively. Interestingly, despite differences in implementation, all of these models take conceptual inspiration from, and the historical qualifications of psychedelics as generic "mind-manifesting", or "nonspecific amplifiers", or "meaning-enhancing", or "suggestibility-inducing" substances. These nonspecific frameworks do not take into account that, particularly in the visual domain, psychedelic phenomenology presents highly specific characteristics, distinct from other classes of hallucinogens, such as deliriants or dissociatives; as such, the psychedelicinduced visionary state cannot easily be captured as a nonspecific amplification. Notably, there is currently also no universal agreement on the mechanisms underlying the therapeutic potential of psychedelics. Supported by studies in animal models, some have proposed neurobiological mechanisms, such as increased neuroplasticity; others, supported by clinical studies in humans, have proposed "mystical-type" experiences as a key driver of the therapeutic benefits of psychedelics. Interestingly, alterations in low-level sensory dimensions have been regarded by both camps as not particularly important, or possibly even counterproductive for therapeutic outcomes, and hence largely relegated to a status of 'epiphenomenal byproducts' in clinical research. However, this notion may represent a throwing out the baby with the bathwater, partially due to the evolution of narratives around psychedelics, which in recent years moved away from the historically-charged and negativelyladen 'hallucinogenic' characterization, and towards the more positively-laden (and more easily marketable) 'entheogenic' or 'neuroplastogenic' characterizations. This disregard for low-level dimensions of experience also sits squarely within broader trends in brain sciences. For most of the 20th century, research on low-level sensory-motor dimensions has constituted most of neuroscience and psychology research, since it was essentially all that available techniques allowed access to; only in the last decades of the 20th century, advances in neuroimaging methods have allowed unprecedented investigations into the high-level cognitive functions of the human brain. The conjunction of this broader trend with the shift in narrative around psychedelics is likely to have contributed to further removing the interest of psychedelic researchers away from the low-level bodily and sensory dimensions, and towards high-level cognitive ones. Nonetheless, many researchershave, to varying degrees, recognised the potential roles of low-level sensory dimensions of experience in causally influencing alterations of high-level dimensions and therapeutic outcomes; and neuroimaging studies have robustly measured effects of psychedelics in sensory brain areas, such as increased functional connectivity of the primary visual cortex with the rest of the brain, increased bottom-up traveling waves, increased retinotopic coordination in visual areas, and changes in oscillatory activity in occipital brain regions. The emerging picture suggests that the present conceptualizations of psychedelics might be underestimating the role played by low-level sensory dimensions in psychedelic brain dynamics and phenomenology. To fill in this gap, here we review findings from neuroimaging, pharmacology, questionnaires, clinical studies, propose extensions to existing models, provide computational simulations of psychedelicinduced visual phenomena, and propose testable hypotheses for the potential therapeutic roles of visual hallucinations. In sum, we find that psychedelic-induced alterations in low-level sensory dimensions of experience 1) are not entirely causally reconducible to alterations in high-level dimensions, but rather co-occur in a dialogical interplay with the latter, and 2) play a causally relevant role in determining high-level alterations and therapeutic outcomes. We conclude that integrating low-level and domain-specific aspects of psychedelic effects into existing models is a necessary step to improve the current understanding of how these substances effect both acute and long-term change in the human brain. The considerations presented here are relevant for the fundamental understanding of psychedelic phenomenology and pharmacology, but also for the design of future clinical protocols and therapeutic practices.

MODELS AND NEURAL DYNAMICS

In recent years, the framework of hierarchical predictive coding has arguably become the most common approach to model neuroscientific and psychological phenomena. An invaluable contribution of predictive coding to contemporary brain sciences has been emphasising the critical role of Bayesian "priors" in brain function and behavior, i.e. the brain's ability to construct and leverage expectations, predictions, and internal models of the world to inform subsequent perception and action; little doubt is left about this insight which we owe to predictive coding. Far from lacking explanatory power, one notable challenge of this framework is perhaps its overability to explain phenomena. It is not uncommon to find diverging, but internally consistent, predictive-coding accounts for the same observations, particularly for complex phenomena such as psychedelics. Indeed, multiple such accounts of psychedelics have been proposed, suggesting overly detailed priors, weakened or relaxed priors, or bothas an explanation for psychedelic effects. The most prominent and widely accepted account, the RE-BUS model, posits that the crucial component of psychedelic states is the relaxation of high-level priors, i.e. a reduced top-down influence exerted by cognitive processes sitting at the highest level of the brain functional hierarchy; in particular, the DMN is most often identified as the key neural correlate of these highlevel processes. In turn, this process is said to imply an increased flow of bottom-up information from low-level sensory regions, 'unsuppressed' by the original priors. It is important to remark that, despite the emphasis placed on the causal primacy of (reduced) top-down control, the authors maintain a degree of openness as to the possibility that bottom-up components may be concurrent. For example, by acknowledging that "The exceptionally high expression of 5-HT2ARs in the brain's highest hierarchical levels provides a solid anatomic basis for this [the REBUS] assumption, although we recognize expression is also particularly high in the primary visual cortex". In more speculative sections, the authors leave the door open for a possible causal role of bottom-up information flow, although this role is always implicitly characterised as occurring after top-down prior relaxation: "How these acute brain changes and associated mind states trigger the relevant long-term changes in beliefs is, at this stage, not fully understood-but increased bottom-up information flow (i.e., from lower-level intrinsic systems such as the limbic system) impacting on sensitised high-level priors (e.g., high-level networks and their dynamics) are logical places to look.". Going further in the direction of this intuition, we argue that taking stock of the specific effects of psychedelics on low-level sensory brain areas, and of the impact of sensory content on top-down priors and high-level dimensions of psychedelic experience is a necessary step towards a more satisfactory description of psychedelic brain dynamics and phenomenology. An existing line of work combines models of neural dynamics with retinotopic considerations, or more recently, connectome graphs. This approach has already provided many fruitful insights into psychedelic phenomena. Of particular relevancy in this context, consider 'simple' visual hallucinations -geometric, abstract visual patterns of concentric spirals, circles, tunnels, honeycombs, webs -which are among the most widely recognized features of psychedelic states; they are specific to psychedelics, in the sense that they are very rarely reported in other classes of hallucinogens, such as deliriants, which crucially also differ from psychedelics in terms of the sense of reality that people attribute to the experience) (more on this in Sec 3); and they appear cross-culturally, suggesting a potential biological underpinning. The cortical-retinotopic mapping implies that subjective experience of geometric visual hallucinations should correspond to regular patterns of neural activity in the visual cortex. Indeed, existing mathematical models have shown that such regular patterns of neural activity can emerge when the excitatory-inhibitory balance, coupled with the anatomical lateral connectivity of visual cortex, is pushed away from its normal stable equilibrium and into oscillatory dynamics. A natural question at this points concerns the source of this instability. Could the relaxed priors serve this purpose? The REBUS authors do not discuss this possibility directly, perhaps because such an argument presents a risk of circularity, since the relaxed priors would then become both the explanans and the explanandum. Given that the high density of 5-HT2A receptors in primary visual areas, it is possible to hypothesize that a direct, local activation of 5-HT2A receptors in these areas could cause an increase in excitability, leading to oscillatory dynamics, which would take shape and be experienced based on the local retinotopic architecture. Importantly, a direct effect of the 5-HT2A receptor in visual areas would also provide an explanation for why psychedelic phenomenology presents such specific and prominent visual characteristic, whereas the effect of a generic relaxation of high-level priors could be expected to be largely modality-independent. The REBUS model provides, as an example, an explanation for the classic psychedelic visual phenomenon of 'breathing walls', i.e. the psychedelic-induced illusory perception of regular motion in inanimate objects, which can be in-terpreted as a 'breathing' of sorts. The authors state that a psychedelic-induced reduction of confidence in the prior assumption that 'walls don't breathe' could force extrastriate visual areas to negligently transmit farther up the hierarchy an "increased signalling arising from lower-level units", which would then be interpreted as breathing walls. In this example, it remains unclear what the source of "increased signalling" would be; and more importantly, it is left unsaid why this specific prior should be so robustly relaxed by psychedelics, and not one of the many others that might be available regarding walls (for example, why not 'walls don't emit peculiar smells' or 'walls don't emit peculiar sounds'). Notably, these objections may be addressed by integrating the REBUS model with considerations on visual-spatial geometry and oscillatory neural dynamics. The structural harmonics of an object are its fundamental spatial patterns of resonance, much like the modes of vibrations of a guitar string; mathematically, the harmonics are the basis of the Fourier transform and the eigenfunctions of the Laplacian operator. In one-dimension, such as on a guitar string, they are sinusoidal waves; in higher dimensions or more complex objects, such as connectome graphs, they take more complex shapes. Consider a retinotopic region of interest such as V1, approximated as a flat surface, it is possible to show that temporal oscillations of pairs of harmonics would determine neural dynamics indistinguishable from the representation of spatially moving objects (Supplementary Materials); even though there are no actual moving objects, but only simple temporal oscillations of structural harmonics. Is it plausible to propose that psychedelics would alter the harmonic-temporal patterns of brain dynamics? Indeed, existing evidence shows that psychedelics robustly alter temporal oscillations in occipital brain regions, shifting the relative balance of temporal power away from the normally dominant alpha, and towards other frequencies; as well as the harmonic spectrum of global brain activity, shifting the balance of spatial power in favour of higher harmonics, i.e. harmonics of higher spatial frequencies. Hence, a parsimonious bottom-up account of the 'breathing walls' phenomenon could be provided by a psychedelic-induced excitation of specific harmonic-temporal oscillations in visual areas, which would be experienced as illusory motion, and could be interpreted by high-level systems as a 'breathing' of sorts when superimposed to the visual input of inanimate objects such as walls. Another highly influential model describing the effect of psychedelics on brain dynamics, focusing on the role of the thalamus in the 'gating' sensory stimuli, is the corticalstriatal-thalamic-cortical (CSTC) model. Briefly, in the CSTC model, a reduced thalamic gating is posited to allow sensory information to inundate the cortex, leading in turn to the cascade of effects that comprise psychedelic phenomenology. While this model may appear antithetical to REBUS, since one chiefly focuses on subcortical levels and the other on the highest cortical levels, both share their roots in Huxley's hypothesis that psychedelics would generically relax constraints on the brain as a 'reducing valve'. The main difference is in the placement of the valve: at the highest levels, in the form of top-down priors, in one model, and at the lowest levels, in the form of thalamic gating, in the other. Hence, the issue of unspecificity also appears to be present the CSTC model: the specific phenomenology of psychedelic visionary states implies that they cannot simply be described as 'nonspecific amplifications' of ungated external sensory input. Additionally, the CSTC model was developed in an explicit analogy between psychedelic states and schizophrenic psychosis; in the past decades, this analogy has been largely abandoned. Other classes of substances, such as amphetamines or anticholinergic hallucinogens, are now considered more fitting analogues for the sensory symptoms of schizophrenic psychoses, which tend to be far more often of an auditory nature, in contrast with the specificity and predominance of visual alterations in psychedelic phenomenology. Hence, while we share with the proponents of the CSTC model the proposal for a role of low-level sensory dimensions in causally influencing high-level dimensions, we suggest that this model, much like REBUS, could benefit by integrating modality specific aspects of psychedelic phenomenology, and corresponding low-level cortical dynamics. This is not to say that all psychedelic sensory phenomena arise entirely via bottom-up cortical dynamics, nor that topdown processes do not impact subjective experience: somewhat in contrast with the REBUS proposal, the influence of cultural factors, beliefs, and personal expectations on psychedelic experience has long been recognized (La. The key point here is that bottom-up effects of psychedelics are unlikely to be entirely byproducts of top-down processes, but rather might be caused by a direct involvement of psychedelics in low-level sensory areas; and that the two are likely to exist in a dialogical interplay, where low-level processes also play a causally relevant role in influencing high-level effects. In this sense, it is important to note that psychedelic alterations in low-level sensory dimensions of experience occur temporally earlier and at lower doses than alterations in highlevel dimensions, and even earlier than emotional alterations. What could one expect to be the impact on of an extremely intense, novel, unpredictable, low-level sensory experience on pre-existing high-level priors? More specifically, what about an experience of immersive visual content of highly non-naturalistic (abstract) character, not relayed by subcortical structures, but rather instantiated directly at the cortical level, such as is expe-rienced during a high-dose psychedelic state? Given that high-level priors are built via a lifetime of naturalistic sensory stimuli, one could reasonably expect such an experience to lead to a profound alteration of the brain's model of what visual input is 'supposed to look like'; and given the particularly significant impact of visual input for the development, maintenance, and revision of these constructs, of what the world is; as a consequence, such an experience might trigger a revision of beliefs even at more abstract levels, bordering into the metaphysical, as is indeed the case for psychedelics. High-level systems would indeed plausibly react to such a shock of novelty by broadening and flattening their expectations on the range of possible experiences; and by attempting (while not necessarily succeeding at) the crucial task of integrating the newly available information into a coherent world-model narrative, as the experience progresses and subsumes. In other words, psychedelic-induced activity in sensory regions might, to some extent, causally underlie the relaxation of high-level priors described by the REBUS model; and in particular, the specific characteristics (e.g. non-naturalism) of information encoded in psychedelic-induced low-level sensory activity may play a crucial role in this sense (see also Sec. 3). Furthermore, this may not be the only way in which psychedelicinduced sensory alterations impact high-level priors in a bottomup fashion: for example, the increased sensitivity to sensory stimuli (i.e. the 'setting') effected at medium and low doses, might significantly increase the correlation between the external and internal state models, and by this virtue underlie the psychedelic sense of connectedness (when positively interpreted) or feelings of anxious ego dissolution (when negatively interpreted), with the direction of interpretation likely depending on existing priors (i.e. the 'set') in top-down manner. In other words, we suggest that top-down and bottomup cortical processes amalgamate to produce the complexity and variety of psychedelic experience. Disentangling the specific underlying brain dynamics is an open and challenging empirical question; to begin addressing it, the study of the visual system may represent a valuable beachhead. For example, an important dimension in the space of psychedelic-induced hallucinations ranges from abstract to figurative content. While both may feature in psychedelic states, they are likely to rely on different mechanisms, and hence may be differentially represented in brain dynamics. In particular, abstract visual hallucinations are more likely to arise bottom-up, due to the intrinsic anatomical and functional properties of neural dynamics in low-level visual areas, and indeed are reported cross-culturally. On the other hand, figurative hallucinations often include content and themes that are specific to the individual (e.g. personal life events) and their culture (e.g. deities or spirits of the local religion). Memories, beliefs, expectations, intentions, sociocultural factors can shape the content and interpretaexample the visual phenomena described in the previous section and simulated in Supplementary Materials, would be gateless. tion of figurative hallucinations, implying the involvement of top-down mechanisms of attention (preferentially directed to aspects of visual experience that are deemed important for the individual and their culture), and the persistence of (at least some) high-level priors. Hence, investigating abstract and figurative visual hallucinations might provide new insights into the processes and brain areas involved in the interplay of top-down and bottom-up cortical dynamics in psychedelic experience. A potential route in this sense may be the combination of detailed phenomenology of visual experience, neuroimaging data, and recent computational approaches which have allowed, to an extent, the reconstruction of subjective visual experience from brain activity in different areas. In sum, we contend that psychedelic-induced alterations of low-level sensory dimensions of experience are unlikely to be entirely a consequence of top-down effects, but rather coexist in a dialogical interplay with, and in some cases may causally underlie, alterations in high-level dimensions of experience. Hence, alongside considering the top-down>bottomup direction of causality, accounting for the converse bottom-up>top-down direction, including the visual specificity of psychedelic phenomenology, appears as a necessary step for the extension of existing models.

QUESTIONNAIRES AND THERAPEUTICS

The Altered States of Consciousness [ASC] questionnaire is one of the most widely used instruments for the measurement of alterations of subjective experience. Importantly, the ASC spans a variety of dimensions of subjective experience, including both low-level sensory and highlevel cognitive components; other questionnaires, such as the Mystical Experience Questionnaire [MEQ], do not allow a direct comparison between these different dimensions, since they focus on a singular construct. Clinical studies employing the ASC show that high-level dimensions of experience provide good predictions for therapeutic outcomes, better so than low-level dimensions. This is supported by studies employing the MEQ, which have found positive correlations of mystical experience dimensions with therapeutic outcomes. Low-level ASC dimensions still show positive correlations with therapeutic outcomes, albeit more weakly than high-level dimensions; and interestingly, one study investigating the efficacy of ayahuasca in the treatment of depression and employing a different questionnaire (the Hallucinogen Rating Scale) found intensity of perceptual experience to be a significant predictor of remission in people that responded to treatment. Hence, the existing evidence speaks in favour of the primacy of high-level dimensions for the prediction of therapeutic outcomes; but also hints at a possible concurrent role of low-level dimensions. Despite the valuable dimensional inclusivity of the ASC, a notable asymmetry is present between its low-level and high-level dimensions: high-level dimensions have far greater phenomenological specificity, and crucially, they have clearly distinct and built-in emotional valence, as opposed to lowlevel dimensions, which are broad in scope and emotionally uncharged. The high-level dimensions of the 5D-ASC, pertaining mainly to alterations in the sense of self, are dubbed 'oceanic boundlessness' and 'dread of ego dissolution'. These dimensions attempt to capture very specific aspects of experience; and the distinction of opposite emotional valence is stark and directly embedded in the dimensions. Conversely, the main sensory dimensions ('visionary restructuralization' and 'auditory alterations') do not attempt to distinguish specific kinds or aspects of experience, but rather measure the strength of a generic alteration of a dimension of experience; and crucially, they lack any distinction based on emotional valence. In the 11D-ASC, a more recent and improved variant of the original 5D-ASC, the first issue is ameliorated by an increased phenomenological specificity of sensory dimensions, which come to include distinct 'simple' and 'complex' dimensions for visual hallucinations; but even in the 11D-ASC, there is no additional possibility to capture specific content, themes, and crucially, emotional valence is not embedded in sensory dimensions. On the other hand, the high-level dimensions of the 11D-ASC ('spiritual experience', 'insightfulness', 'experience of unity', 'impaired control and cognition') are still more phenomenologically diverse than low-level dimensions, and are still embedded with a clear emotional valence. This asymmetry has important consequences for the interpretation of results based on the ASC questionnaire. Subjective experience is integrated: a person engrossed in the vivid sensory experience of a complex, realistic, nightmarish hellscape would not, in most cases, be simultaneously reporting a blissful experience of oceanic boundlessness. However, because of the asymmetry in the ASC dimensions, this will not be reflected by the questionnaire results: two people who experienced visual hallucinations of similar intensity, but opposite in terms of content and emotional valence, will not be distinguished by the ASC questionnaire on the lowlevel dimensions; however, they will be clearly separated on the high-level dimensions, which carry distinct emotional valence. Furthermore, many of the high-level dimensions usually have visual components: for example, spiritual experiences can take the form of visions of angels, entities, the universe as a whole, or simply the white light; emotional, cathartic moments can be sparked by vividly re-experiencing personal memories; insights can involve a strong moment of imagination in which the insight is seen, before being consciously understood. Many of these lowlevel aspects will be diluted in ASC results, and often implicitly integrated in high-level dimensions. Interestingly, in the construction of the 11D-ASC, Studerus et al. noted that "the "oceanic boundlessness" and "visionary restructuralization" factors could be combined on a high level of the construct hierarchy", hinting at potential unexplored relationships between low-level and high-level dimensions of experience; and other authors have already pointed out that high-level constructs such as the sense of self arise via a multidimensional integration which includes low-level sensory aspects. Aside from development of more balanced questionnaires, one route to address these limitations is via qualitative and quantitative analyses of narrative self-reports; indeed, existing studies in this sense have reported a prominent presence of sensory themes. In sum, a critical examination of existing questionnaire studies provides an additional line of evidence for the underappreciated role of low-level sensory dimensions in psychedelic experience and therapeutic outcomes, which may be addressed by employing methods that are more faithful to the phenomenological richness of psychedelic experiences.

TESTABLE HYPOTHESES FOR THE POTENTIAL THERAPEUTIC ROLES OF PSYCHEDELIC-INDUCED VISUAL HALLUCINATIONS A NATURAL

question at this point is whether there is any potential causal role for low-level sensory experiences in psychedelic therapeutics. While the ongoing development of novel 'nonhallucinogenic psychedelics' may certainly be an appealing business model and lead to valuable outcomes, we contend that causal role of sensory dimensions in psychedelic therapeutics is far from exhaustively understood. In particular, some authors have suggested the sense of awe as a possible key mediator of psychedelic therapeutics; in the absence of pharmacological agents, the sense of awe is often elicited by sensoryparticularly visual -stimuli. Hence, rather than occurring "in a vacuum", it seems plausible that the sense of awe may also be elicited by low-level sensory alterations in psychedelic experience. For example, a highly intense and immersive experience of geometric visual hallucinations might trigger a feeling of aesthetic, and even spiritual, awe; it is perhaps not surprising to note in this sense the presence of abstract, concentric motives -reminiscent of psychedelic-induced geometric hallucinations -in many traditions of religious art. The experience of awe is also known to be characterized by a reduction in the sense of self, a feature of psychedelic experience which is thought to be of major therapeutic value, but also generally considered unrelated and independent from low-level sensory dimensions; instead, we suggest that psychedelic-induced sensory alterations might to some extent causally underlie alterations of high-level constructs, such as the sense of self. Interestingly, the perception of awe-inspiring visual stimuli includes stronger DMN deactivation among its neural correlates. Furthermore, considering their non-naturalistic and abstract character, geometric hallucinations of sufficient intensity (such as those experienced during a DMT peak) may directly produce a disruption of the sense of reality. Indeed, the feeling of being transported to a realm 'other' than everyday reality, 'seeing through the matrix', or 'lifting of the veil' experiences, are common in psychedelic trip reports. Notably, such experiences are often reported to feel 'more real' than everyday reality; in a sense, this intuition might be correct. Not in the literal sense that psychedelics would allow us to peer at an external, supernatural reality; rather, in the sense that they allow us to peer at the structure of our own systems of perception, revealing the nature of our everyday reality as far more contingent and less objective than we normally think; in turn, this process could trigger reflections on high-level concepts pertaining to the relations between self and world, reaching the highest metaphysical levels. It is interesting to note that some reports describe the psychedelic experience as less real than everyday reality -like a movie or a cartoon. It seems plausible that, when interpreting psychedelic visual experiences, top-down effects of culture, expectations, and prior beliefs may still play a key role, leading to diverging interpretations. Crucially, note that in both these seemingly opposite interpretations ('more real' or 'less real'), the psychedelic experience is still most often judged as other than and distinct from everyday reality. Unlike the experience induced by deliriant hallucinogens, rarely (if ever) a psychedelic experience is confused with everyday reality. A parsimonious explanation for this may be that the low-level sensory content of psychedelic experience -in particular geometric hallucinations -is unlikely to be confused with the percepts of everyday reality, because of its intrinsically non-naturalistic aspect. In other words, the specific content and visual characteristics of low-level dimensions of psychedelic experience might play a key role in determining alterations in the sense of awe (increased), self (reduced), and reality attributed to the experience ('other than everyday reality'); while the integration and interpretation of these experiences (e.g. 'more real' or 'less real') is likely to be influenced, in a top-down fashion, by prior beliefs and expectations. Complex hallucinations -non-abstract visions of entities, objects, people, animals -can also be elicited by psychedelics. Note that 1) the binary distinction between simple and complex hallucinations is a simplification, and a scale varying in degree would be more appropriate; 2) 'simple' hallucinations are only simple insofar as they are of an abstract nature: they can be of high intricacy and complexity in their own right; 3) the 'realism' of psychedelic complex hallucinations does not imply that they are taken at face value and confused with the percepts of everyday reality, possibly because they often still contain a degree of the characteristic geometric patterns of simple visual hallucinations. It is possible to hypothesise a role for complex visual hallucinations in the therapeutic effects of psychedelics; to some extent, this has been proposed before, e.g., but it is worth revisiting this idea in light of recent findings. Complex hallucinations elicited by psychedelics often contain subjectively meaningful elements: for example, they might involve autobiographical memories of traumatic experiences, loved ones that passed away, childhood memories. The vivid experience of personally relevant sensory content, in the form of complex hallucinations, in combination with altered senses of reality, self, and awe, which we hypothesised may be elicited by simple visual hallucinations, could represent an important therapeutic mode of psychedelics. Of course, the subsequent integration of these profound experiences into highlevel narratives would also be a crucial step for therapeutic outcome, as well-documented by psychedelic therapy practice; but the role of sensory experience in the acute phase should not be disregarded. In this context, it is important to mention existing evidence suggesting that psychedelics do increase the attribution of meaning to otherwise neutral or meaningless stimuli, an effect which could be attributed to a top-down process. However, Preller et al. also report that, under the influence of psychedelics, stimuli that were personally meaningful before the experience remain far more meaningful than ones that were previously meaningless or neutral. Hence, it seems likely that the experience of personally-relevant sensory content would have a higher chance to trigger therapeutically valuable revisitation, reappraisal, and subsequent integration, compared to personally-irrelevant content. In sum, we contend that the constellation of therapeutic processes set in motion by psychedelics are unlikely to rely exclusively on mechanisms independent of subjective experience, or on nonspecific top-down processes, such as an increased attribution of meaning or relaxed priors. We argue that the sensory phenomena elicited by psychedelics are far from being 'epiphenomenal byproducts' and rather may constitute, if not the engine, at least a valuable propellant for psychedelic therapeutics. As an important caveat, we note that this is not to say that any and all sensory phenomena elicited by psychedelics will be relevant for therapeutic outcomes; nor that high-level processes do not play a crucial role, particularly at the stage of interpretation and integration of experience; our argument here is only that the available evidence does not warrant disregarding all sensory phenomena elicited by psychedelics, in particular visual hallucinations, as irrelevant for therapeutic outcomes. Specifically, we have hypothesised that simple (abstract, geometric) visual hallucinations may contribute to psychedelic therapeutics by eliciting the sense of awe, and by triggering changes in the sense of reality and sense of self, relaxing the normal bounds on these constructs; while complex hallucinations may allow the revisitation of personally relevant sensory content, which may be reappraised during a critical period of shifted perspective; both contributing significantly to the revision of high-level beliefs and narratives of self and world, bordering into the metaphysical.

NEUROIMAGING AND PHARMACOLOGY

At this point, it becomes crucial to ask whether neuropharmacological evidence is consistent with a direct action of psychedelics in low-level sensory areas. PET and autoradiography studies have found high 5-HT2AR binding and mRNA expression in the primary visual cortex. 5-HT2AR binding and mRNA expression were found in all cortical layers with high expression in layer V pyramidal neurons. Within the primary visual cortex (V1), highest expression of the 5-HT2A receptor was found in layer IV (and not in layer V pyramidal neurons, as in the majority of other regions), the main input layer in the cortex, receiving direct sensory input from the lateral geniculate nucleus of the thalamus, which in turns receives input from the eye; furthermore, three hours of monocular deprivation (covering the eye) can lead to downregulation of 5-HT2AR mRNA in layer IV of V1, suggesting V1 specific expression patterns of receptor mRNA are sustained by the ongoing visual activity. Additionally, electrophysiological studies in rodents have indicated the involvement of the 5-HT2A receptor in modulating responses to visual stimuli in the primary visual cortex. Neuroimaging of the visual cortex under the influence of psychedelics has revealed increased cerebral blood flow, increased correlations with the rest of the brain, patterns of brain activity analogous to those recorded during the viewing of spatially localised stimuli, and increased BOLD signal during an eyes-closed mentalimagery task (but not during eyes-open viewing of natural images). Together, these findings provide a solid neuropharmacological basis for direct action of psychedelics in low-level sensory (and specifically visual) regions. Next, we ask whether neuroimaging evidence is consistent with the underappreciated roles of psychedelic action in low-level sensory areas which we described in the previous sections. In this context, it is important to note that 'resting-state' is a term commonly used in neuroscience to denote brain imaging or recordings performed in the absence of a specific external sensory stimulus or cognitive task. The resting-state characterization is often also applied to recordings carried out under the influence of psychedelics, when there is no external stimulus or task. However, the psychedelic state is one of rich sensory content: while an external stimulus may not be present, vivid and intense subjectively experienced sensory content most often is; importantly, any psychedelic-induced neural activity at lower levels of the brain's functional hierarchy would likely be interpreted by higher regions, and thus experienced as, content-encoding activity. The presence of neural activity encoding sensory representations is generally not taken into account by resting-state models and analyses, but it is an important caveat to be kept in mind when attempting to disentangle which aspects of high-level function are altered by the direct action of psychedelics, and which may be indirectly altered by activity in low-level sensory areas, alongside other important methodological considerations. The increase in functional connectivity (FC) between primary visual areas and the rest of the brain under the influence of psychedelics is a broadly replicated finding from psychedelic neuroimaging. In, this finding is interpreted as "a far greater proportion of the brain contributes to visual processing in the LSD state than under normal conditions". However, since FC does not have a directionality, one could have also interpreted this finding as "visual processing contributes to a far greater proportion of brain dynamics in the LSD state than under normal conditions". In other words, it is plausible to ask whether it is activity in early visual regions that becomes increasingly 'broadcasted' throughout the brain during a psychedelic experience, increasing its influence over global brain activity, rather than the other way round. This interpretation also seems to square better with the reduced influence of top-down priors prescribed by the REBUS model: assuming relaxed priors, the rest of the brain should presumably contribute less, not more, to low-level visual processing, while the converse should be true. Indeed,found that, after ayahuasca administrations, oscillations in occipital regions increased their influence over signals measured at frontal locations, while oscillations at frontal sources decreased their influence over activity at occipital sites. An fMRI study bycontrasted LSD, with placebo, and LSD+ketanserin (a 5-HT2A receptor antagonist). The main pattern observed is one of hypo-connectivity in high-level association networks, and a converse hyper-connectivity in low-level sensory and somatomotor networks; furthermore, notably, the two patterns are correlated: "participants with the highest LSD-induced coupling within sensory and somatomotor networks also showed the strongest LSD-induced de-coupling in associative networks. This suggests that LSD-induced alterations in information flow across these networks". In, activity in visual cortices is found to be more dependent on its intrinsic anatomical (retinotopic) organisation during psychedelic effects, consistent with neural activity encoding anatomically-constrained, excitationdriven visual hallucinations. Interestingly, brain-wide analyses have indicated that on a global level, brain-activity is less dependent on its anatomical constraints during the effects of psychedelics. This may be explained by proposing that activity in sensory (particularly visual) cortices, driven by 5-HT2A activation and intrinsic anatomical constraints, becomes more widely broadcasted throughout the brain, which would then become globally less tethered to their own anatomy. In other words, the psychedelic state could be characterised as one of 'seeing with the brain', instead of 'seeing with the visual cortex', as we do under normal circumstances. Taken together, the available findings are consistent with the hypothesis that neural dynamics in sensory cortices during the psychedelic experience might exert far-ranging effects throughout the brain. The decrease in power of alpha oscillations generated by occipital sources in the brain is another robust finding from 'resting state' psychedelic neuroimaging. In the absence of pharmacological alterations, alpha oscillations are prominent in occipital regions during eyes-closed resting-state, and are thought to subserve an inhibitory function; hence, the finding of psychedelic-induced reduction in alpha oscillations has often been interpreted as a reduced influence of topdown inhibitory processes in the visual cortex. However, the relationship between alpha oscillations and sensory content in the visual cortex is mutual: alpha oscillations suppress incoming inputs, but incoming inputs also strongly suppress alpha oscillations. In other words, the presence of highfrequency neural activity encoding sensory content in the visual cortex suppresses the otherwise dominant alpha oscillations. Hence, it is important to at least question whether alpha power reduction under the influence of psychedelics is best modelled as a reduction of top-down effects, or whether the presence of neural activity encoding visual content could be the causal driver of suppression of occipital alpha oscillations. FMRI of the visual cortex under the influence of psychedelics has found it to be processing eyes-closed psychedelic imagery as if a visual stimulus was present (de Araujo et al., 2012) and spatially localized; research on ayahuasca also highlighted a specific connection between suppression of alpha oscillations and the visual effects of psychedelics; and analysis of travelling brain waves during DMT experience found remarkable similarity with the dynamics observed during visual stimulation, arguably making the DMT state more similar to one of highly intense and immersive visual stimulation, than to a resting-state with abolished alpha oscillations; if the psychedelic state was only or solely one of reduced top-down influence causing more nonspecific bottom-up activity to permeate into consciousness, perhaps one could expect a more clearly marked distinction with respect to the dynamics of traveling waves induced by regular bottom-up visual stimulation. Together, these findings suggest that the alpha power reduction observed in psychedelic states may be a consequence of the presence of content-encoding neural activity in visual areas. The deactivation or 'disintegration' of the default mode network (DMN), and its correlation to dissolution of ego (or more broadly, reduction in the sense of self), is perhaps the most well-known finding from psychedelic neuroscience. Carhart-Harris et al. report a significant correlation of DMN disintegration with egodissolution (at a 0.05 level, with a p-val=0.03). Notably, the correlation of ego-dissolution with the visual medial network is also close to significance at the same level (with a p-val=0.07), closer to significance than all other RSNs aside from the DMN itself, and all three visual networks considered in the analysis show correlations with ego-dissolution in the same direction(Supplementary Materials), begging the question of whether the partitioning of visual areas into three separate networks might have diluted the correlation significance. Another milestone result ofis a very significant correlation between the LSD-induced increase in V1 functional connectivity, and subjective ratings of both elementary and complex hallucinations. This is a seemingly straightforward finding to interpret, since classically visual areas of the brain are expected to be involved in subjective visual experience. To further understand the relations between imaging measures and subjective experience, in table S7 the authors evaluate the specificity of the correlations between changes in functional connectivity and different dimensions of subjective experience. Interestingly, no significant difference is found between the correlation of increased V1 FC with elementary imagery and that with several high-level dimensions of the ASC questionnaire. In other words, the correlation between increased V1 FC and elementary hallucinations is not significantly different from the correlation between increased V1 FC and high-level dimensions of subjective psychedelic experience such as experience of unity, spiritual experience, and insightfulness. This is echoed by an independent analysis of the VAS scale: the correlation between increased V1 FC and elementary visual hallucinations is found to be significantly different from the correlation between increased V1 FC and complex hallucinations, positive mood, emotional arousal; but it is not found to be significantly different from its correlation with ego-dissolution. Taken together, these findings suggest that the effects of psychedelics on the DMN (and subjectively, on the sense of self) may be related to alterations in low-level sensory areas. Could the effects of psychedelics at the highest level of the brain's functional hierarchy be causally influenced by low-level alterations? In light of the considerations presented in the previous sections, and of the characteristics of the DMN, this is perhaps not a completely implausible proposal. The DMN is a demand-negative network: it deactivates in the presence of tasks or sensory stimuli. Ordinary tasks and ordinary sensory stimuli deactivate the DMN in the ordinary, functional way that we are all subjectively familiar with from everyday life, temporarily diminishing one's narrative, selfcentred, ruminating focus. What would an astoundingly intense, highly unpredictable, highly novel, gatelesssensory experience do to the DMN? Perhaps, not just ordinarily deactivate it, but effectively disrupt it to a much more profound degree, leading not to a brief decrease, but to a unique disintegration, and possibly to a complete loss of the sense of self. In other words, psychedelic-induced low-level sensory activity may causally contribute to the DMN deactivation, and related alterations of the sense of self; it is perhaps then not surprising to note that sensory overload has been em-ployed in ancient and modern routes to loss of self, from shamanic traditions to rave parties. Furthermore, interestingly, recent empirical work on large-scale fMRI datasets has revealed a visual-spatial organisation of the DMN (Szinte and Knapen, 2020); and recent theoretical work has provided hypotheses of ways in which low-level sensory experience could affect high-level constructs like the sense of self, from an information-theoretic perspective. Together, these findings suggest that future models could benefit from taking into account alterations in lowlevel sensory areas when attempting to describe psychedelic effects on global brain dynamics and high-level dimensions of experience. In sum, the available evidence from pharmacology and neuroimaging provides a solid basis for a direct and specific effects of psychedelics in sensory (particularly visual) brain areas, and is consistent with the hypothesis of causal relations between brain dynamics in low-level sensory regions, and high-level dimensions of psychedelic experience. In particular, activity in low-level areas may causally contribute to key features of psychedelic states, such as DMN deactivation and reduction in sense of self.

CONCLUSION

Psychedelics are currently thought to induce a mode of brain activity and subjective experience characterised by reduced top-down, and increased bottom-up dynamics. However, there is no universal agreement on how this state is achieved, nor on the mechanisms underlying their therapeutic benefits. In Sec. 2, we argued that influential contemporary models present largely nonspecific accounts of psychedelic effects, which do not fully account for the specificity of psychedelic phenomenology, and we provided considerations pertaining to the role of neural dynamics in visual cortices (see also Supplementary Materials); in Sec. 3, we discussed how asymmetries in existing questionnaires may have confounded the causal influence of low-level sensory dimensions on highlevel dimensions and therapeutic outcomes, and presented hypotheses for the potential therapeutic roles of psychedelicinduced visual hallucinations; in Sec. 4, we examined evidence from pharmacology and neuroimaging, showing that it is consistent with an underappreciated role of sensory regions and networks in psychedelic brain dynamics. In sum, we contend that psychedelic-induced alterations of seemingly distinct dimensions of experience do not occur "in a vacuum" with respect of each other, but rather are intertwined in a dialogical interplay. As influential contemporary models have described the influence of high-level alterations on low-levels, so we argue that low-level alterations are also likely to causally influence high-levels, and that extending existing models will require explicitly accounting for both causal directions. In particular, we contend that reevaluating the currently underappreciated role of sensory dimensions in psychedelic states will be highly valuable for neuroscience and clinical practice. We conclude that integrating the specific and nonspecific aspects of psychedelic dynamics, as well as unravel-ing causal connections between seemingly unrelated dimensions of experience, is a necessary step to further understand how these remarkable compounds effect both acute and long-term therapeutic (or possibly harmful) changes in the human brain.