ArticlePDF Available

Abstract and Figures

Entropy is a dimensionless quantity that is used for measuring uncertainty about the state of a system but it can also imply physical qualities, where high entropy is synonymous with high disorder. Entropy is applied here in the context of states of consciousness and their associated neural dynamics, with a particular focus on the psychedelic state. The psychedelic state is considered an exemplar of a primitive or primary state of consciousness that preceded the development of modern, adult, human, normal waking consciousness. Based on neuroimaging data with psilocybin, a classic psychedelic drug, it is argued that the defining feature of ‘primary states’ is elevated entropy in certain aspects of brain function, such as the repertoire of functional connectivity motifs that form and fragment across time. It is noted that elevated entropy in this sense, is a characteristic of systems exhibiting ‘self-organised criticality’, i.e., a property of systems that gravitate towards a ‘critical’ point in a transition zone between order and disorder in which certain phenomena such as power-law scaling appear. This implies that entropy is suppressed in normal waking consciousness, meaning that the brain operates just below criticality. It is argued that this entropy suppression furnishes consciousness with a constrained quality and associated metacognitive functions, including reality-testing and self-awareness. It is also proposed that entry into primary states depends on a collapse of the normally highly organised activity within the default-mode network (DMN) and a decoupling between the DMN and the medial temporal lobes (which are normally significantly coupled). These hypotheses can be tested by examining brain activity and associated cognition in other candidate primary states such as REM sleep and early psychosis and comparing these with non-primary states such as normal waking consciousness and the anaesthetised state.
Content may be subject to copyright.
published: 03 February 2014
doi: 10.3389/fnhum.2014.00020
The entropic brain: a theory of conscious states informed
by neuroimaging research with psychedelic drugs
Robin L. Carhart-Harris1*, Robert Leech2, Peter J. Hellyer2, Murray Shanahan3, Amanda Feilding4,
Enzo Tagliazucchi5, Dante R. Chialvo6and David Nutt1
1Division of Brain Sciences, Department of Medicine, Centre for Neuropsychopharmacology, Imperial College London, London, UK
2C3NL, Division of Brain Sciences, Department of Medicine, Imperial College London, London, UK
3Department of Computing, Imperial College London, London, UK
4The Beckley Foundation, Beckley Park, Oxford, UK
5Neurology Department and Brain Imaging Center, Goethe University, Frankfurt am Main, Germany
6Consejo Nacional de Investigaciones Científicas y Tecnológicas (CONICET), Buenos Aires, Argentina
Edited by:
Nikolai Axmacher, University of
Bonn, Germany
Reviewed by:
Samantha J. Brooks, Uppsala
University, Sweden
Katherine MacLean, Johns Hopkins
University School of Medicine, USA
Robin L. Carhart-Harris, Division of
Brain Sciences, Department of
Medicine, Centre for
Neuropsychopharmacology, Imperial
College London, Burlington Danes
building, Du Cane Rd., W12 0NN
London, UK
e-mail: r.carhart-harris@
Entropy is a dimensionless quantity that is used for measuring uncertainty about the state
of a system but it can also imply physical qualities, where high entropy is synonymous
with high disorder. Entropy is applied here in the context of states of consciousness
and their associated neurodynamics, with a particular focus on the psychedelic state.
The psychedelic state is considered an exemplar of a primitive or primary state of
consciousness that preceded the development of modern, adult, human, normal waking
consciousness. Based on neuroimaging data with psilocybin, a classic psychedelic drug,
it is argued that the defining feature of “primary states” is elevated entropy in certain
aspects of brain function, such as the repertoire of functional connectivity motifs that
form and fragment across time. Indeed, since there is a greater repertoire of connectivity
motifs in the psychedelic state than in normal waking consciousness, this implies that
primary states may exhibit “criticality,” i.e., the property of being poised at a “critical”
point in a transition zone between order and disorder where certain phenomena such
as power-law scaling appear. Moreover, if primary states are critical, then this suggests
that entropy is suppressed in normal waking consciousness, meaning that the brain
operates just below criticality. It is argued that this entropy suppression furnishes normal
waking consciousness with a constrained quality and associated metacognitive functions,
including reality-testing and self-awareness. It is also proposed that entry into primary
states depends on a collapse of the normally highly organized activity within the default-
mode network (DMN) and a decoupling between the DMN and the medial temporal lobes
(which are normally significantly coupled). These hypotheses can be tested by examining
brain activity and associated cognition in other candidate primary states such as rapid eye
movement (REM) sleep and early psychosis and comparing these with non-primary states
such as normal waking consciousness and the anaesthetized state.
Keywords: serotonin, default mode network, criticality, entropy, 5-HT2A receptor, metastability, consciousness,
REM sleep
The main aim of this paper is to introduce a new theory of
conscious states that incorporates principles of physics, neurobi-
ology, and psychoanalysis. The theory is intended to assist our
understanding of the makeup of the human mind, addressing
such questions as: “how does the normal waking consciousness
of healthy adult humans relate to other states of consciousness?”
“how does the human brain maintain its normal state of waking
consciousness?” and “what happens to the human brain’s func-
tionality when non-ordinary states such as rapid eye movement
(REM) sleep/dreaming, early psychosis and the psychedelic state
At its core, the entropic brain hypothesis proposes that
the quality of any conscious state depends on the system’s
entropy1measured via key parameters of brain function. Entropy
is a powerful explanatory tool for cognitive neuroscience since it
provides a quantitative index of a dynamic system’s randomness
or disorder while simultaneously describing its informational
character, i.e., our uncertainty about the system’s state if we were
to sample it at any given time-point. When applied in the context
of the brain, this allows us to make a translation between mecha-
nistic and qualitative properties. Thus, according to this principle,
1Entropy in its purest information theoretical sense is a dimensionless quan-
tity that is used for measuring uncertainty or ignorance about the state of
a system. By implication, entropy/uncertainty is greater the more random a
system is. Thus, entropy is most strictly a measure of uncertainty but it also
reflects the degree of randomness or disorder in a system (Ben-Naim, 2012).
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |1
Carhart-Harris et al. The entropic brain
increased subjective uncertainty or “puzzlement” accompanies
states of increased system entropy. These ideas are consistent
with Karl Friston’s free-energy principle2and readers interested
in Bayesian inference and the mechanisms by which the brain
is hypothesized to minimize free-energy/surprise should consult
this work (Friston, 2010).
System entropy, as it is applied to the brain, is related to
another current hot-topic in cognitive neuroscience, namely
“self-organized criticality”3(Chialvo et al., 2007). The phe-
nomenon of self-organized criticality refers to how a complex
system (i.e., a system with many constituting units that displays
emergent properties at the global-level beyond those implicated
by its individual units) forced away from equilibrium by a regu-
lar input of energy, begins to exhibit interesting properties once
it reaches a critical point in a relatively narrow transition zone
between the two extremes of system order and chaos. Three prop-
erties displayed by critical systems that are especially relevant to
the present paper are: (1) a maximum number of “metastable” or
transiently-stable states (Tognoli and Kelso, 2014), (2) maximum
sensitivity to perturbation, and (3) a propensity for cascade-like
processes that propagate throughout the system, referred to as
“avalanches” (Beggs and Plenz, 2003). There is growing evidence
that brain activity, like much of nature, displays critical behav-
ior (Beggs and Plenz, 2003)—and this raises some interesting
questions: e.g., does the brain activity of healthy-adult-humans
exhibit characteristics of criticality during normal waking con-
sciousness, or are there other states of consciousness in which
these characteristics are even more pronounced?
Another major topic that is covered in this paper is the psy-
choanalytic model of the structure of the mind (i.e., Freud’s
“metapsychology”). Specifically, we discuss some of the most fun-
damental concepts of Freudian metapsychology, with a special
focus on the ego 4. We focus on the ego because it is one of
Freud’s less abstract constructs and it is hypothesized that its dis-
integration is necessary for the occurrence of primary states. The
ego can be defined as a sensation of possessing an immutable
2The free energy principle is an extension of predictive coding (Dayan et al.,
1995) and tries to explain how biological systems, such as the brain, main-
tain their order by developing inferences and behaviors that serve to minimize
surprise and uncertainty. Free-energy is formally related to entropy in the
information theoretical sense, where entropy (uncertainty) is the average of
free-energy (surprise) (Friston, 2010). See also (Friston et al., 2012a)fora
philosophical discussion of free-energy.
3Self-organized criticality (SOC) is a property of certain systems that gravi-
tate toward a ‘critical’ point in a transition zone between order and disorder.
Critical systems display certain characteristics such as power-law scaling (Bak
et al., 1987). SOC is typically observed in slowly driven non-equilibrium sys-
tems, with many units that interact in a non-linear fashion (Bak et al., 1987;
Jensen, 1998).
4The ego can be defined as a sensation of possessing an integrated and
immutable identity, i.e., “this is me” or “I am like this.” It is equivalent there-
fore with one’s sense of self. In psychoanalytical theory however, the ego is also
a system which works in concert with and against other processes in the brain
to determine the quality of consciousness. It is worth noting that Freud’s term
for the ego was ‘the I’ and it was only in the standard translation from German
that the term “the ego” became associated with Freud (1927). In everyday (lay)
usage, “ego” has become synonymous with exaggerated self-confidence or an
inflated ego/sense of self.
identity or personality; most simply, the ego is our “sense of
self.” Importantly however, in Freudian metapsychology, the ego
is not just a (high-level) sensation of self-hood; it is a fundamen-
tal system that works in competition and cooperation with other
processes in the mind to determine the quality of consciousness.
It is because Freud described “the ego” in this mechanistic sense
that it can be considered a useful complement to the more widely
used concept of “the self.” Effectively, the terms “ego” and “self”
are synonyms, except that “the ego” has a background in Freudian
Finally, the shared topic that connects all of the above
and offers a unique potential for their empirical study is the
psychedelic drug state. In the following section we make the
case that scientific research with psychedelics has considerable
potential for developing aspects of psychoanalytic theory and
for studying human consciousness more generally. Citing recent
neuroimaging findings involving the classic psychedelic drug,
psilocybin, the psychedelic state is described as a prototypical
high-entropy state of consciousness (i.e., higher than normal
waking consciousness). Intriguingly, we show evidence that the
brain exhibits more characteristics of criticality in the psychedelic
state than are apparent during normal waking consciousness.
Moreover, this leads to the proposal that the brain of mod-
ern adult humans differs from that of its closest evolutionary
and developmental antecedents because of an extended capac-
ity for entropy suppression, implying that the system (i.e., the
brain) gravitates away from criticality proper toward a state
of slight sub-criticality. The psychological counterpart of this
process is the development of a mature ego5and associated
metacognitive functions (see below for relevant definitions of
these terms). Specifically, we propose that within-default-mode
network (DMN) 6resting-state functional connectivity (RSFC)7
and spontaneous, synchronous oscillatory activity in the poste-
rior cingulate cortex (PCC), particularly in the alpha (8–13 Hz)
frequency band, can be treated as neural correlates of “ego
integrity.” Evidence supporting these hypotheses is discussed in
the forthcoming sections.
Before beginning it is important to address an initial point
of potential ambiguity. The view taken here is that the human
brain exhibits greater entropy than other members of the animal
kingdom, which is equivalent to saying that the human mind pos-
sesses a greater repertoire of potential mental states than lower
animals (see Giulio Tononi’s information integration theory of
consciousness cited below). Thus, if referring to human evo-
lution beyond our closest surviving relatives then it would be
misleading to suggest that entropy-suppression is the defining
property of the human brain—indeed, it might be more accurate
to speak of entropy-expansion. The evolution of human con-
sciousness may have occurred through a process of relatively rapid
5A fully-developed, adult ego or sense of self.
6A network of functionally and structurally connected brain regions that show
high spontaneous or “on-going” metabolism yet a relative deactivation during
goal-directed cognition (Raichle et al., 2001).
7Functional connectivity is defined as temporal correlations between spatially
distinct neurophysiological events (Karl Friston). Resting-state refers to task-
free, unconstrained experimental conditions (typically sitting or lying still
with eyes closed).
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |2
Carhart-Harris et al. The entropic brain
entropy-expansion (with a concomitant increase in system disor-
der) followed by entropy-suppression (or system re-organization
and settling). Thus, the proposal that normal waking conscious-
ness in healthy, adult, modern humans depends on entropy sup-
pression implies that there was a state relatively proximal to this
(e.g., in archaic homo-sapiens and in infants) in which entropy
was relatively elevated, as it is in primary states. The point is that
the brain of adult modern-humans is in a settling rather than
expanding phase.
“It does not seem to be an exaggeration to say that psychedelics,
used responsibly and with proper caution, would be for psychiatry
what the microscope is for biology and medicine or the telescope
is for astronomy. These tools make it possible to study important
processes that under normal circumstances are not available for
direct observation.” (Grof, 1980)
In 1953, the British research psychiatrist Humphrey Osmond was
investigating the psychotomimetic (psychosis mimicking) effects
of mescaline, a psychedelic drug derived from the peyote cactus.
The British author Aldous Huxley learned of Osmond’s work and
struck up a correspondence, requesting that Osmond supervise
a personal psychedelic experience. Huxley’s subsequent mesca-
line experience would become the subject of his famous book
The Doors of Perception”(Huxley, 1954). Like many before and
after him, Huxley was profoundly affected by his experiences with
psychedelics and in 1956 sought with Osmond a satisfactory term
for this class of drugs. At the time, “psychotomimetics” and “hal-
lucinogens” were popular, but both men felt that these referred to
mere aspects of the drug experience and not its essential character.
Huxley suggested “phanerothyme,” intending to mean “bring-
ing forth the spirit or soul” (Huxley et al., 1977), and Osmond
offered “psychedelic” combining the Greek words for “mind” or
“soul” (psych¯
e) with “d¯
e,” meaning “to manifest. While it was
Osmond’s “psychedelic” that would stick, it is telling that both
men were searching for a word that could denote the same essen-
tial property, i.e., psychedelic’s ability to make manifest latent
aspects of the mind.
In 1943, Swiss chemist Albert Hofmann discovered the
extraordinary psychological properties of lysergic acid diethy-
lamide (LSD) (Hofmann, 1980) and the first reports on its effects
appeared in scientific journals in the late 1940s. These papers
immediately highlighted LSD’s potential to be psychologically
agitative. The first English language publication was released in
1950 and here the authors reported: “the effect of LSD was a
transitory toxic state, disturbing the barrier of repression and per-
mitting a re-examination of significant experiences of the past that
were sometimes relived with a frightening realism.” ( Busch and
Johnson, 1950) In the following years, psychedelics became one of
the most researched classes of psychoactive drug in science, with
several hundred relevant publications (Grinspoon and Bakalar,
1979). During these years, the focus shifted from psychedelics as
psychotomimetics to psychedelics as psychotherapeutic adjuncts,
with major international conferences on the topic (Grinspoon
and Bakalar, 1979) and even the construction of purpose-built
psychedelic treatment centers (Sandison, 2001). Political pressure
in the late 1960s led to the illegalization of psychedelics and this
had a significant negative impact on legitimate scientific research
(Grinspoon and Bakalar, 1979; Lee and Shlain, 1985)—a prob-
lem that continues today (Nutt et al., 2013). Despite this however,
there has been a resurgence of scientific interest in psychedelics
in recent years (Vollenweider et al., 1998; Nichols, 2004; Griffiths
et al., 2006, 2008; Moreno et al., 2006; Gonzalez-Maeso et al.,
2007; Grob et al., 2011; Carhart-Harris et al., 2012a).
The dominant theoretical and therapeutic approach dur-
ing the early era of psychedelic research was psychoanalytic.
Psychedelics were used therapeutically under the rationale that
they work to lower psychological defenses to allow personal con-
flicts to come to the fore that can then be worked through with a
therapist (Cohen, 1972). A related model was that the relinquish-
ment of “ego” enabled profound existential or “peak” experiences
to occur that could have a lasting positive impact on behav-
ior and outlook (Savage, 1962). Innumerable cases of apparent
spontaneous insights about “self” or “nature” exist in the litera-
ture on psychedelics (Cattell, 1954; Sandison, 1954; Sandison and
Whitelaw, 1957; Denber, 1958; Hausner and Dolezal, 1965; Torda,
1969; Cohen, 1972; Grof, 1982) and reports of “ego-dissolution
or “disintegration” are commonplace among those who have
experienced the effects of these drugs (Carhart-Harris and Nutt,
2010; Carhart-Harris et al., 2012b). Some psychiatrists even
believed that psychedelics could provide the necessary scientific
evidence for major psychoanalytic hypotheses (Sandison, 1954;
Cohen, 1972; Grof, 1982). For example, one enthused: “The phe-
nomenology of the psychodynamic experiences in LSD sessions is to a
large extent in agreement with the basic concepts of classical psycho-
analysis...Observations from LSD psychotherapy could be consid-
ered laboratory proof of the basic Freudian premises.” (Grof, 1982).
Psychoanalytic theory dominated psychiatry in the 1950s but
after influential critiques (Eysenck, 1973), the cognitive revolu-
tion (Neisser, 1967) and significant pharmacological develop-
ments in psychiatry (Ban, 2001a,b; Fink, 2010), its influence
significantly waned. As illustrated in Figure 1, despite over a cen-
tury since its inception, psychoanalysis has failed to establish itself
as a science of the mind. This may be because its hypotheses
are hollow (Webster, 1995) or because they do not easily lend
themselves to controlled experiment. In contrast, cognitive psy-
chology is a mechanistic framework for describing observable
phenomena that has become the natural bedfellow for human
neuroscience. In comparison with the spectacular success of cog-
nitive psychology, what should we make of the relative stagnancy
of psychoanalysis? Is psychoanalysis scientifically redundant? Its
fiercest critics claim that it is a belief system, a tautology with
untestable hypotheses (Webster, 1995) but others claim that it has
considerable explanatory value but could benefit from a closer
integration with cognitive neuroscience (Kandel, 1999; Carhart-
Harris and Friston, 2010; Panksepp and Solms, 2012). The present
article takes this latter view and argues that the most realistic way
forward for psychoanalysis as a science is for its most tangible
hypotheses to be simplified and applied within the framework of
cognitive neuroscience. Here we take the view that this is a neces-
sary concession for psychoanalysis if it is to develop its credibility
as a model of the mind.
In what follows, a roadmap is presented for how scien-
tific research with psychedelics can assist the integration of
psychoanalysis with cognitive neuroscience in order to further
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |3
Carhart-Harris et al. The entropic brain
FIGURE 1 | Relative publications on psychoanalysis and cognitive
psychology. Annual publications in major medical science journals
referenced in the leading database, PubMed. Articles were retrieved by
entering the search terms “psychoanalysis” and “cognitive psychology” in
separate searches using the default search parameters of PubMed. It is
worth noting that psychoanalysis has a different publication culture to
cognitive neuroscience. Articles on psychoanalysis are not always available
via PubMed and many psychoanalytic writings are published in books rather
than academic journals. With these caveats entered however, the publication
count shown above still helps to illustrate the general point that
psychoanalysis has failed to gain a significant foothold in mainstream
analytical science.
our understanding of human consciousness. This is motivated
by the view that psychoanalysis can contribute something sub-
stantial to the mind sciences because it bridges an explanatory
gap that has been left vacant by cognitive psychology. This gap
only exists because cognitive psychology (rightly) focuses on phe-
nomena that can be observed and manipulated by controlled
experiment but crucially, without psychedelic drugs, it is virtually
impossible to bring the core phenomena of psychoanalytic theory
into an observable space, namely the “unconscious mind.
Freud famously said of dreams that they provide privileged
access to the workings of the unconscious mind (Freud, 1937)but
research on dreaming is fraught with difficulties because [despite
the phenomenon of lucid dreaming (Ogilvie et al., 1982)] the
dream experience cannot be easily reflected on and reported in
real-time, and neither can its onset and offset be easily controlled.
Thus, Freud’s cherished “royal road” has not proved particu-
larly regal and a more practical alternative is required if key
psychoanalytic theories are to be incorporated into the mind
“If, as Freud said, dreams are the royal road to the unconscious,
is it possible that psychedelic drugs are a superhighway to the
unconscious?” (Holden, 1980)
This article argues that controlled studies with psychedelics are
capable of providing major new insights into the nature of the
mind and how it arises from brain activity. This is because
become cognizant of its constituents and how they interact to
give rise to global phenomena. The unique scientific value of
psychedelics rests on their ability to selectively target processes
that appear to be critical for the maintenance of normal waking
consciousness. In addressing the action of psychedelic drugs on
the brain, this article begins at the cellular level before progressing
to the systems level. The intention is to offer a comprehen-
sive account of how psychedelics alter brain function to alter
Somewhat uniquely, psychedelics can be studied at a range of
epistemological levels; from molecular pharmacology (Gonzalez-
Maeso and Sealfon, 2009) to psychoanalytic psychology (Cohen,
1964; Grof, 1982), few topics can engage scientists from as wide
a range of disciplines. This reflects not only the special research
value of psychedelics but also the immensity of the challenge
involved in understanding them; especially, if the intention is to
develop a comprehensive account of how psychedelics affect the
brain to alter consciousness. The present article should therefore
be read with an acknowledgement that this quest is on-going.
Before we begin, it is necessary to enter some important
caveats. Firstly, it needs to be stated that those looking for evi-
dence for the authenticity of aspects of Freudian theory will be left
dissatisfied by this article. Categorically, this is not its aim. This
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |4
Carhart-Harris et al. The entropic brain
challenge requires a thorough review of the phenomenology of
relevant altered states of consciousness (e.g., the psychedelic state)
and this is something that has been attempted before (Carhart-
Harris, 2007; Carhart-Harris and Friston, 2010). Thus, due to
space limitations, this article’s treatment of the relevant phe-
nomenology is relatively superficial. Instead it places its focus on
the system-level mechanics of the psychedelic state as an exem-
plar of a regressive8style of cognition that can also be observed in
REM sleep and early psychosis.
Some proponents of psychoanalysis may feel that this mech-
anistic approach has little relevance to psychoanalysis in its
hermeneutic or interpretative guise. However, the inherent sub-
jectivity of this aspect of psychoanalysis means that it is difficult
to see how it can ever significantly impinge on the scientific
study of the mind and brain. Indeed, Freud acknowledged that
it was his “metapsychology” that had the most to offer science
(Freud, 1949), and at least as a first step, this is where psycho-
analytic theory (rather than psychoanalytic practice) should look
to develop its scientific credibility. Briefly, for readers who are
unfamiliar with Freudian metapsychology and wish to under-
stand it better, his original material should be read (e.g., Freud,
1927, 1949; Freud et al., 1957) and the following review articles
may be useful (Carhart-Harris et al., 2008; Carhart-Harris and
Friston, 2010). For those interested in the rich phenomenology
of the psychedelic experience and how this relates to Freudian
and/or Jungian descriptions of “the unconscious mind,” the fol-
lowing references may be of interest (Sandison and Whitelaw,
1957; Huxley, 1959; Cohen, 1964; Grof, 1982; Merkur, 1998;
Sandison, 2001). Lastly, it is necessary to state that questions
related to the safety of scientific research with psychedelics will
not be addressed here. However, evidence strongly supports the
position that, conducted with appropriate caution, research with
psychedelics presents a low risk of harm to study participants
(Johnson et al., 2008; Morgan et al., 2010; Carhart-Harris and
Nutt, 2010; Studerus et al., 2011; van Amsterdam et al., 2011).
Before introducing the focal topic of this paper, i.e., entropy and
its relation to key brain imaging parameters, it is important to
provide a brief introduction to the pharmacology of psychedelics.
By definition, all classic psychedelic drugs are agonists at the sero-
tonin 2A receptor (5-HT2AR) (Glennon et al., 1984). There is a
strong positive correlation between a psychedelic’s affinity for the
5-HT2AR and its psychedelic potency (Glennon et al., 1984). For
example, LSD has a very high affinity for the 5-HT2A Randis
remarkably potent, being psychoactive in doses as small as 20 µM
(Hintzen and Passie, 2010). Blockade of the 5-HT2ARwiththe
5-HT2AR antagonist ketanserin, attenuates the principal hallu-
cinogenic effects of psilocybin in humans (Vollenweider et al.,
1998). The 5-HT2AR is primarily expressed in the cortex (Pazos
et al., 1987). In humans, the distribution of 5-HT2ARs is generally
high throughout the cortex but is densest in high-level association
regions such as the PCC and lowest in the primary motor cor-
tex (Erritzoe et al., 2009; Carhart-Harris et al., 2012a). This may
8The quality of returning to an earlier state of development.
by psychedelics whereas motor action is generally not. In terms
of the cortex’s laminar organization, 5-HT2ARs are most densely
expressed postsynaptically on the apical dendrites of layer 5 pyra-
midal neurons (Weber and Andrade, 2010). These large excitatory
neurons are the primary source of output from a cortical region,
projecting to hierarchically subordinate cortical and subcortical
regions (Spruston, 2008). 5-HT2AR stimulation depolarizes the
host cell, making it more likely to fire (Andrade et al., 2011)and
this effect has been demonstrated in layer 5 pyramidal neurons in
rodents (Aghajanian and Marek, 1997).
Beginning in 2009, our research team embarked on a
series of studies with the classic psychedelic, psilocybin
(Carhart-Harris et al., 2012a), culminating in a recent MEG
study (Muthukumaraswamy et al., 2013). Our first study utilized
arterial spin labeling (ASL), an fMRI technique that measures
changes in CBF. Specifically, we compared CBF before and after
intravenous (i.v.) administration of 2mg psilocybin and placebo
(Carhart-Harris et al., 2012a). The onset of the subjective effects
of psilocybin is rapid when it is administered intravenously,
commencing within seconds of the end of the infusion (Carhart-
Harris et al., 2011). The infusion occurred over 60 s, beginning
6 min into an 18 min resting state scan. Drug-induced changes in
CBF were modeled based on psilocybin’s rapid pharmacodynam-
ics (Carhart-Harris et al., 2011). Fifteen healthy volunteers were
scanned and the results revealed decreased CBF after psilocybin
and no increases. The decreases were localized to high-level
association cortices, including key regions of the DMN (see
Some background on the default mode network (DMN) for an
overview of this system) and subcortical hub structures such as
the putamen and thalamus (Carhart-Harris et al., 2012a).
These findings were later replicated using the classic BOLD sig-
nal of fMRI. Another 15 healthy volunteers were scanned using a
similar placebo-controlled design, with 60 s i.v. infusions begin-
ning midway through two separate 12min eyes-closed resting
state scans on different days. Again, only signal decreases were
observed after drug infusion. Moreover, the location of the BOLD
signal decreases was consistent with the CBF decreases, e.g., in
midline cortical nodes of the DMN (Muthukumaraswamy et al.,
In addition to modeling changes in the direction of the BOLD
signal post-infusion of psilocybin, we also measured changes in
brain network integrity using resting-state functional connectiv-
ity. Three regions of interest were chosen for separate seed-based
resting state functional connectivity (RSFC) analyses: a medial
prefrontal cortex (mPFC) seed, a right middle frontal gyrus
(mFG) seed, and a bilateral hippocampal seed. Decreased con-
nectivity was observed within the DMN using the mPFC and hip-
pocampal seeds and in a major task-positive network (TPN), the
dorsal attention network (DAN), using the mFG seed (Figure 2).
In our third and most recent study, we used MEG to investigate
the effects of psilocybin on neural activity. Broadband decreases
in oscillatory power were observed after psilocybin, and again,
these were localized to association cortices, including key regions
of the DMN, such as the PCC (Raichle et al., 2001; Greicius et al.,
2003)—see Figure 2.
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |5
Carhart-Harris et al. The entropic brain
FIGURE 2 | The effect of psilocybin on fMRI and MEG measures of
brain activity. (A) Decreased CBF post-psilocybin. (B) Ventromedial PFC
(red) resting state functional connectivity (RSFC) at baseline (top, orange)
and decreases post-psilocybin (bottom, blue). (C) Dorsolateral PFC (red)
RSFC at baseline (top, orange) and decreases post-psilocybin (bottom,
blue). (D) Hippocampal (red) RSFC at baseline (top, orange) and decreases
post-psilocybin (bottom, blue). (E) Decreases in oscillatory power (purple)
post-psilocybin measured with MEG. All spatial maps were whole-brain
cluster corrected Z>2.3. p<0.05.
These studies provide some useful clues about the mechanisms
by which psychedelics alter brain function to alter consciousness.
They imply that cerebral blood flow, BOLD signal, functional
connectivity and oscillatory power are decreased in brain regions
that are normally highly metabolically active, functionally con-
nected and synchronous/organized in their activity. These results
provided the kernel for our subsequent thinking about increased
entropy in the psychedelic state. Although none of the analyses
formally measured entropy, they spoke to a general principle that
psychedelics alter consciousness by disorganizing brain activity.
The DMN has become one of the most discussed topics in cognitive
neuroscience over the last decade and there are several reasons
why it is justified to consider it important (Guldenmund et al.,
2012). DMN regions receive more blood flow (Zou et al., 2009)
and consume more energy (Raichle and Snyder, 2007)thanother
brain regions. Indeed, CBF and metabolic rate are approximately
40% higher in the PCC than the average of the rest of the
brain (Raichle et al., 2001). The magnitude of the DMN’s energy
consumption dwarfs the comparatively trivial energy changes
induced by stimulus cues (Raichle, 2006, 2010). DMN regions are
centers of dense connectivity (Hagmann et al., 2008), implying that
theyserveasimportantconnectorhubs for information integration
and routing (van den Heuvel et al., 2012). Consistent with this,
a major node of the DMN, the PCC, can be spatially segmented
into sub-components that functionally couple to different brain
networks(Leechet al.,2012).Similarly,duringtransientwindows of
especially high internal coupling (functional connectivity) within
the DMN, coupling between the DMN and other brain networks
is also markedly increased (de Pasquale et al., 2012). Importantly,
this functional centrality of the DMN is not shared by other brain
networks (de Pasquale et al., 2012; Braga et al., 2013), implying
that, as the highest level of a functional hierarchy (Carhart-Harris
and Friston, 2010), it serves as a central orchestrator or conductor
of global brain function. Functionally, the DMN is relatively
removed from sensory processing (Sepulcre et al., 2012)andis
instead engaged during higher-level, metacognitive operations
such as self-reflection (Qin and Northoff, 2011), theory-of-mind
(Spreng and Grady, 2010) and mental time-travel (Buckner and
Carroll, 2007)—functions which may be exclusive to humans.
DMN connectivity increases through development from birth to
adulthood (Fair et al., 2008; Gao et al., 2009) and DMN regions
have undergone significant evolutionary expansion (Van Ess en a nd
Dierker, 2007). Despite our knowledge of these things however, it is
poorly understood why the DMN consumes so much of the body’s
energy (Raichle and Mintun, 2006). This uncertainty regarding
the nature of the DMN’s disproportionate energy consumption
has led to loose analogies being made between it and the dark
energy of cosmology (Raichle, 2006, 2010). It is consistent with
the hypotheses of this paper to suggest that this apparent excess
energy of apparently unknown function, residing in the DMN, is
in fact the physical counterpart of the narrative-self or ego—much
of which is indeed unconscious or implicit.
This article proposes that states such as the psychedelic state,
REM sleep, the onset-phase of psychosis and the dreamy-state
of temporal lobe epilepsy are examples of a regressive style of
cognition that is qualitatively different to the normal waking con-
sciousness of healthy adult humans. We will refer to this mode of
cognition as “primary consciousness”9and the states themselves
as “primary states.” To enter a primary state from normal wak-
ing consciousness, it is proposed that the brain must undergo
a “phase transition” (Zeeman, 1973; Waddington, 1974), just
human consciousness with the relatively rapid development of
the ego and its capacity for metacognition10. This implies that the
relationship between normal waking consciousness and “primary
consciousness” is not perfectly continuous.
9A regressive, pre-ego style of consciousness characterized by unconstrained
brain dynamics and cognition. The psychedelic state is an exemplar of primary
consciousness and REM sleep and early psychosis are others.
10A cognitive capacity unique to adult humans and dependent on the for-
mation of a mature ego. Metacognition means “cognition about cognition.”
Examples include: self-reflection, theory-of-mind and mental time-travel.
(Fleming et al., 2012).
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |6
Carhart-Harris et al. The entropic brain
Freud was a great admirer of Darwin and made several refer-
ences to him throughout his work (Freud et al., 1953). Indeed,
Freud considered his own hypotheses to be natural deductions
from evolutionary theory. He argued that dreaming and psychosis
typify a primitive style of thinking that is dominant in human
infancy11 and dominated the cognition of primordial man12,pre-
ceding the development of the ego of modern adult humans.
Primitive thinking is fundamentally different to the style of cogni-
tion possessed by healthy adult humans. This is because in healthy
adults, the formation of a mature ego endows the mind with a
capacity for metacognition i.e., an ability to reflect on one’s own
thoughts and behavior (Shimamura, 2000; Fleming et al., 2012).
is appropriate to clarify them here. A distinction is being made
between two fundamentally different styles of cognition, one that
is associated with the consciousness of mature adult humans, and
another that is a mode of thinking the mind regresses to under
certain conditions, e.g., in response to severe stress, psychedelic
drugs and in REM sleep. The style of cognition that is dominant
in normal waking consciousness will henceforth be referred to as
secondary consciousness13 and the (pre-ego) style of cognition that
is associated with primitive states will be referred to as primary
before (Edelman, 2004) but their meaning in the present context
is largely independent.
Consistent with Karl Friston’s free-energy principle (Friston,
2010), this article takes the view that the mind has evolved (via
secondary consciousness upheld by the ego) to process the envi-
ronment as precisely as possible by finessing its representations
of the world so that surprise and uncertainty (i.e., entropy) are
minimized. This process depends on the ability of the brain to
organize into coherent, hierarchically-structured systems (Bassett
et al., 2008; Friston, 2010), critically poised between order and
disorder (Friston et al., 2012b; Schwartenbeck et al., 2013). In
contrast, in primary states, cognition is less meticulous in its sam-
pling of the external world and is instead easily biased by emotion,
e.g., wishes and anxieties.
Later we finesse this basic model, arguing that secondary
consciousness actually depends on the human brain having devel-
oped/evolved a degree of sub-criticality in its functionality, i.e.,
an extended ability to suppress entropy and thus organize and
constrain cognition. It is argued that this entropy-suppressing
function of the human brain serves to promote realism, foresight,
careful reflection and an ability to recognize and overcome wish-
ful and paranoid fantasies. Equally however, it could be seen as
exerting a limiting or narrowing influence on consciousness.
This paper argues that the underlying neurodynamics of
primary states are more “entropic” than secondary states
i.e., primary states exhibit more pronounced characteristics
of criticality and perhaps supercriticality than normal waking
consciousness—implying that the latter is slightly sub-critical,
11The term “infancy” is used here in an extended sense to mean the period
from birth to two years of age.
12The earliest humans, i.e., archaic homo-sapiens.
13The style of consciousness of healthy, adult, modern humans during wak-
ing. Secondary consciousness is “constrained” relative to primary conscious-
ness, both in a mechanistic and qualitative sense.
if not perfectly critical. Secondary consciousness pays deference
to reality by carefully sampling the world and learning from its
encounters (Friston, 2010), whereas primary consciousness does
this more haphazardly. Mechanistically, whereas the brain strives
toward organization and constraint in secondary consciousness,
processes are more flexible in primary consciousness. Freud out-
lined these ideas in his writings on “the reality principle” (Freud,
1927) and they are recast here in a more mechanistic form, tied to
modern cognitive neuroscience.
The phenomenon of “magical thinking”14 (Frazer, 1900;
Subbotskii, 2010; Hutson, 2012) is a potential product of primary
consciousness. Magical thinking is a style of cognition in which
supernatural interpretations of phenomena are made. Magical
thinking is more likely in situations of high uncertainty because
there is a greater opportunity for dreaming up explanations that
lack an evidence base (Friston, 2010). Wishful beliefs are a clas-
sic product of magical thinking because they interpret the world
according to what an individual wants to be true (in Freudian
terms, they adhere to the pleasure principle). Wishful inferences
are quick-fixes that reduce uncertainty but via simplistic explana-
tions that satisfy fancies or desires before careful reason. Another
example of magical thinking is paranoia; in this case, an indi-
vidual jumps to negative conclusions about a situation, even in
the face of contradictory evidence, because to do so effectively
suspends uncertainty while providing some narcissistic satisfac-
tion. The popularity of magical thinking also suggests that there
is some enjoyment in uncertainty, perhaps because it promotes
imaginative and creative thinking—and that this is associated
with positive affect.
In the forthcoming section we discuss the relationship between
medial temporal lobe (MTL—i.e., specially the hippocampus and
surrounding parahippocampal structures) activity and primary
consciousness, highlighting a specific change in activity that may
serve as a marker of primary states.
Recording directly from MTL circuits in different altered states
presents a significant challenge for cognitive neuroscience, but
not one that should deter us from trying to expand its reach
into areas of relevance to psychoanalytic theory. Pioneering surgi-
cal interventions for epilepsy and Alzheimer’s disease (Axmacher
et al., 2008, 2010; Fell et al., 2011, 2012; Laxton and Lozano,
2012) are opening up new possibilities for depth recordings, and
although it would be a challenge to defend the administration
of psychedelics to such patients, recording from MTL circuits in
other primary states, such as REM sleep, might be more feasible
(Cantero et al., 2003).
Another way to circumvent the problem of recording directly
from limbic regions is to use non-invasive imaging with high spa-
tial resolution. We recently used fMRI to investigate the involve-
ment of the MTLs in the mechanism of action of psychedelics,
performing a hippocampal functional connectivity analysis using
14A style of cognition in which causal relationships between phenomena
are assumed despite an absence of supportive scientific evidence. Examples
include superstition, and metaphysical beliefs. Crucially, such beliefs usually
honor an emotional sentiment such as a wish (or fear) that something is true.
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |7
Carhart-Harris et al. The entropic brain
the same psychophysiological interaction (PPI) model used in our
previous analyses with psilocybin (Carhart-Harris et al., 2012a).
A combined bilateral hippocampal and parahippocampal mask
was created on a standard brain and time-series were extracted
from these regions for each subject and regressed against
their functional data, with the pharmacodynamics of intra-
venous psilocybin modeled as an interaction term. Remarkably,
decreases in functional coupling were observed after psilocy-
bin that were selectively localized to the cortical nodes of the
DMN (Figure 2D), entirely consistent with the hypothesis that
decreased MTL-DMN coupling underlies phase transitions to
primary consciousness.
In a separate analysis, we looked at changes in BOLD signal
variance (i.e., amplitude fluctuations) after psilocybin and found
significant increases in variance that were almost exclusively local-
ized to the bilateral hippocampi and parahippocampal gyri. This
result was important because it reinforced the impression given by
the RSFC analysis that under psilocybin, the hippocampi become
decoupled from the DMN. However, perhaps even more inter-
estingly, the increase in MTL signal variance was consistent with
some early depth electrode work with psychedelics that impli-
cated the MTLs in their mechanism of action. This work is
reviewed below.
Human depth recordings involving the insertion of electrodes
into structures located deep in the brain (Ramey and O’Doherty,
1960) were not uncommon in the 1950s and 60s. Remarkably,
some intracranial recordings were carried out during this period
in individuals administered LSD and mescaline (Schwarz et al.,
1956; Monroe et al., 1957; Monroe and Heath, 1961). The relevant
reports document unusual phasic activities in the MTLs during
the drug state that were difficult to detect at the scalp. Moreover,
in separate studies, temporal lobectomy was found to abolish the
effects of LSD in humans (Serafetinides, 1965) and chimpanzees
(Ramey and O’Doherty, 1960) and frontal lobotomy was found to
augment them (Keup, 1970). Further support for the involvement
of MTLs in the mechanism of action of psychedelics comes from
reports of similar phasic limbic activity in other altered states of
consciousness that show phenomenological similarities with the
psychedelic state, namely: REM sleep, acute psychosis and the
so-called “dreamy-state” of temporal lobe epilepsy and electri-
cal stimulation of the MTL (Carhart-Harris, 2007; Carhart-Harris
and Friston, 2010). Importantly, these states are all characterized
by a particular style of cognition that is fundamentally different
from that of normal waking consciousness.
It is proposed here that coupling between the MTLs and
the cortical regions of the DMN is necessary for the mainte-
nance of adult normal waking consciousness, with its capacity for
metacognition. Moreover, a breakdown in hippocampal-DMN
coupling is necessary for a regression to primary consciousness.
These hypotheses are motivated by our finding that DMN-
hippocampal coupling is decreased under psilocybin (Figure 3D),
and while DMN activity becomes desynchronous and there-
fore disorganized (Figure 3E), the amplitude of BOLD signal
fluctuations increases in the hippocampus (Figure 2).
Layer 5 pyramidal neurons densely express 5-HT2A receptors
(Weber and Andrade, 2010). These cells are an important target
FIGURE 3 | Increased variance/amplitude fluctuations in the
hippocampus post-psilocybin. The charts on the left show the complete
time series from the hippocampus (left in blue, right in green) in 3 different
individual subjects during the 12min scan in which they received psilocybin.
The transparent red vertical line indicates the beginning and duration of the
60 s infusion of psilocybin. The images on the right show the right
hippocampal region where the increases in variance were especially
of psychedelics (Aghajanian and Marek, 1997) and are known
to fire with an intrinsic alpha frequency (Silva et al., 1991; Sun
and Dan, 2009). Alpha oscillations are thought to be related to
temporal framing in perceptual processing (Lorincz et al., 2009;
Klimesch et al., 2011) but more intriguingly given the focus of
the present article, a positive relationship has been found between
self-reflection and alpha power (Knyazev et al., 2011)andalpha
synchronization during rest and Blood Oxygen Level Dependent
(BOLD) activity in regions of the DMN (Jann et al., 2009).
Evidence implicates the DMN in self-reflective and introspective
functions (Qin and Northoff, 2011) and the phase of fluctuating
activity in the DMN is often inversely correlated (or “anticor-
related”) with fluctuating activity in networks concerned with
task-focused attention (task-positive networks, TPNs) (Fox et al.,
2005). Like the DMN, alpha oscillations mature developmentally
and evolutionarily (Basar and Guntekin, 2009), tempting spec-
ulations that these rhythms have developed to reduce “entropy”
[i.e., disorder or uncertainty (Ben-Naim, 2008)] by increasing
mutual information among neuronal ensembles (Tononi et al.,
1994; Basar and Guntekin, 2009). With this in mind, it was
remarkable that we recently found a highly significant positive
correlation between the magnitude of alpha power decreases in
the PCC after psilocybin and ratings of the item “I experienced
a disintegration of my ‘self’ or ‘ego.” Scores on this item also
correlated positively with decreases in delta, theta, beta, and low
gamma power, although alpha explained the most variance (a
considerable 66%) see Figure 4. Twenty three subjective items
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |8
Carhart-Harris et al. The entropic brain
FIGURE 4 | Decreased PCC alpha power predicts ego-disintegration
and magical thinking after psilocybin. Top: Decreased PCC alpha power
vratings of ego-disintegration. Bottom: Decreased PCC alpha power vs.
ratings of magical/supernatural thinking. Both correlations were significant
after correction for multiple comparisons (0.05/23 =0.002). These charts
are derived from data discussed in Muthukumaraswamy et al. (2013).
were rated after psilocybin but the one enquiring about ego-
disintegration showed the closest relationship with the decreases
in alpha power, surviving the conservative Bonferroni correc-
tion for multiple comparisons. Interestingly, the only other item
that survived correction for multiple comparisons referred to
the promotion of magical thinking, i.e., “the experience had a
supernatural quality.” It is a central hypothesis of this paper that
psychedelics induce a primitive state of consciousness, i.e., “pri-
mary consciousness” by relinquishing the ego’s usual hold on
reality (DMN control on MTL activity).
The organizing influence of alpha applies more generally to
oscillatory rhythms in the brain (Salinas and Sejnowski, 2001;
Buzsaki and Draguhn, 2004). Harmonics are known to exist
between the brain’s oscillatory rhythms, with higher frequency
oscillations “nested” within lower frequencies (Jensen and Colgin,
2007). For example, intracranial recordings from the ventral
PCC in humans revealed a dominant presence of theta oscilla-
tions. The phase of these oscillations modulate the amplitude of
high-gamma oscillations and the magnitude of this coupling fluc-
tuates at a frequency that is consistent with that of spontaneous
BOLD signal fluctuations (i.e., 0.1 Hz) observed in resting-
state networks (RSNs) such as the DMN (Foster and Parvizi,
2012). Theta oscillations are a canonical rhythm of hippocam-
pal circuits, at least in rodents (Buzsaki, 2002), and MTLs are
known to be strongly connected to the PCC (Parvizi et al., 2006)
and DMN more generally (Kahn et al., 2008). Thus, it is feasi-
ble that a function of PCC theta oscillations is to constrain the
activity of limbic circuits, which reciprocally input to the PCC.
Evidence that MTL activity exerts a driving influence on PCC
activity comes from a recent report on deep brain stimulation for
Alzheimer’s disease. Chronic stimulation of the fornix, an impor-
tant component of hippocampal circuitry, was associated with
significantly increased glucose metabolism in the PCC (Laxton
et al., 2012).
In summary, interaction between different oscillatory rhythms
introduces a structured quality to brain activity (Rumsey and
Abbott, 2004), constraining the naturally stochastic firing of
individual pyramidal neurons (Rolls and Deco, 2010)andso
providing ideal conditions for the emergence of “complexity”
(Tononi et al., 1994) or “self-organized criticality” (Jensen, 1998).
A key hypothesis of this article is that it is through the devel-
opment of self-organized activity in the DMN [and concomitant
entropy/uncertainty/disorder minimization (Friston, 2010)] that
a coherent sense of self or “ego” emerges (Carhart-Harris and
Friston, 2010). This process of maturational settling succeeds
an earlier state of elevated entropy (primary consciousness) and
psychedelic drugs induce a regression to this entropic brain state
via the mechanisms outlined above.
With these foundations laid, the following hypotheses can be
proposed: (1) coupling within the DMN, and especially between
the MTL and DMN, is a characteristic of maturational settling
that is necessary for secondary consciousness and the develop-
ment of an integrated sense of self; (2) a relative decoupling
within the DMN and specifically between the MTLs and DMN
occurs when secondary consciousness abates and there is a recip-
rocal increase in the influence of primary consciousness; (3)
decreased MTL-DMN coupling allows the MTLs to function
more independently of the DMN and this can result in unusual
MTL activities such as have been recorded with depth electrodes
in primary states (see above and Grof, 1982; Bassett et al., 2008;
Axmacher et al., 2010)andmayhavebeendetectedintheBOLD
signal amplitude increases in the MTL post-psilocybin (Figure 3);
(5) unconstrained/disinhibited/anarchic MTL activity is a princi-
pal characteristic of primary states and the occurrence of these
activities is consistent with a system at criticality; (6) brain activ-
ity in primary consciousness is closer to criticality-proper than
it is during normal waking consciousness (which may be slightly
sub-critical rather than perfectly critical).
DMN resting-state functional connectivity correlates positively
with ratings of internal awareness (Vanhaudenhuyse et al., 2011),
depressive rumination (Berman et al., 2011)andtraitneuroti-
cism (Adelstein et al., 2011). DMN connectivity increases dur-
ing mental time-travel (Andrews-Hanna et al., 2010; Martin
et al., 2011) and activity in the medial prefrontal node of
the DMN is reliably elevated in depression (Farb et al., 2011;
Lemogne et al., 2012). These findings strongly implicate the
DMN in introspective thought and suggest that hyper activity and
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |9
Carhart-Harris et al. The entropic brain
connectivity in the DMN is related to a certain style of concerted
To step back, one of the primary hypotheses being devel-
oped here is that metacognition, and in particular, the human
capacity for self-reflection, is an advanced behavior that rests on
self-organized activity in the DMN and between the DMN and
the MTLs. Thus, if the DMN is hyper-active and connected in
depression, does this imply that mild depression is an evolution-
arily advanced state? The phenomenon of “depressive realism” has
been recognized for several decades (Dykman et al., 1989; Haaga
and Beck, 1995) and sits comfortably with the idea that a primary
function of the DMN is to support metacognition (Fleming et al.,
2010). The suggestion is that increased DMN activity and connec-
tivity in mild-depression promotes concerted introspection and
an especially diligent style of reality-testing. However, what may
be gained in mild depression (i.e., accurate reality testing) may
be offset by a reciprocal decrease in flexible or divergent thinking
(and positive mood).
The proposal that increased DMN activity and connectivity
is a key functional correlate of concerted introspection, such
as is seen in depression, may seem inconsistent with the asso-
ciation between DMN activity and mind-wandering (Mason
et al., 2007) but this is a conceptual problem that can be eas-
ily resolved. The positive relationship between increased BOLD
signal in the DMN and the frequency of mind-wandering dur-
ing task-performance (Mason et al., 2007) tells us nothing about
the nature or style of the cognition in the off-task state, it simply
tells us that the mind has drifted off-task. It is known however,
that the strength of inverse coupling between activity in the DMN
and TPNs is increased when task performance is more consis-
tent (Kelly et al., 2008), implying increased focus and a relative
decrease in off-task attentional lapses. DMN-TPN inverse cou-
pling is decreased in patients with attention deficit/hyperactivity
disorder (ADHD) (Hoekzema et al., 2013) and increased after
administration of the attention-enhancers modafinil (Schmaal
et al., 2013) and nicotine (Cole et al., 2010). Thus, it is too sim-
plistic to regard increased BOLD signal in the DMN as a correlate
of freely-wandering cognition, and decreased inverse coupling
between the DMN and TPN is probably a more informative index
of this. As will be discussed later, this point is reinforced by
findings that inverse coupling between the DMN and TPNs is
decreased under psilocybin, and DMN activity and connectiv-
ity is also decreased. This is important because unconstrained,
explorative thinking is a hallmark of the psychedelic state (see
Figure 5).
In the next section we cite direct evidence for increased
entropy in brain networks in psychedelic state and use this to sup-
port a general principle: that the transition from normal waking
consciousness to primary consciousness is marked by an increase
in system entropy.
There is an emerging view in cognitive neuroscience that the
brain self-organizes under normal conditions into transiently sta-
ble spatiotemporal configurations (Sporns et al., 2004; Shanahan,
2010; Deco and Corbetta, 2011; Tagliazucchi et al., 2012)and
FIGURE 5 | Psilocybin promotes unconstrained thinking and decreases
blood flow, venous oxygenation and oscillatory power in the DMN.
This chart shows the average (+SE) ratings for the item “my thoughts
wandered freely” in 3 neuroimaging studies, each involving the
administration of psilocybin and placebo to 15 healthy volunteers. Ratings
were given within 30 min of the end of the relevant resting state scans.
This particular item was one of the highest rated items in all 3 studies and
nicely communicates the quality of cognition that predominates in the
psychedelic state. The brain image on the left displays the mean regional
decreases in CBF post-psilocybin in the ASL study; the central image
displays the mean regional decreases in BOLD signal post-psilocybin in the
BOLD study; and the image on the right displays the mean regional
decreases in alpha power post-psilocybin in the MEG study. All images
were derived using a whole brain corrected threshold of p<0.05.
that this instability is maximal at a point where the global sys-
tem is critically poised in a transition zone between order and
chaos (Tononi et al., 1994; Shanahan, 2010; Deco and Jirsa, 2012;
Tagliazucchi et al., 2012). In the present context, the “metastabil-
ity” (Tognoli and Kelso, 2014)ofabrainnetworkisameasure
of the variance in the network’s intrinsic synchrony over time.
That is, if the signal in all of the voxels within a given network
deviates little from the network’s mean signal, then variance is
low, whereas if the signal in voxels fluctuate erratically, then vari-
ance is high. Thus, using the data from the BOLD fMRI study
with psilocybin, we recently looked at changes in the variance of
intra-network synchrony over time in nine canonical resting-state
networks (Smith et al., 2009) pre and post placebo and psilocy-
bin. Results revealed significantly increased network variance in
high-level association networks after psilocybin but not in sen-
sory specific and motor networks, and there were no changes after
placebo. These results imply that activity in high-level networks
becomes relatively disorganized under psilocybin, consistent with
the entropic brain hypothesis.
To translate this result into a formal measure of entropy, we
discretized the time course of intra-network synchrony over time
into equal sized bins where each time-point could be entered into
a bin depending on the variance in the network’s synchrony at
that time point. Doing this for each network, we built probabil-
ity distributions of the variance of the intra-network synchrony
across time from which we could then calculate the Shannon
entropy for each network. Not surprisingly, increased entropy
was observed in the networks in which there was increased
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |10
Carhart-Harris et al. The entropic brain
variance post-psilocybin i.e., the high-level association networks
(See Figure 6).
To further assess entropy changes after psilocybin, we took
a slightly different approach. Four regions were chosen from a
limbic/paralimbic system based on prior knowledge that BOLD
signal variance was increased in these regions under psilocybin.
The regions included: the left and right hippocampi and the left
and right anterior cingulate cortex (ACC). A threshold was set
for connection strength such that only connections above a par-
ticular strength survived and were therefore said to “exist.” This
allowed functional connectivity motifs (connectivity graphs) to
be identified at each time point in the time series. With 4 nodes,
there were 64 possible connectivity motifs or graphs at any given
time point. The results revealed a greater repertoire of motifs
FIGURE 6 | Changes in network metastability and entropy
post-infusion of psilocybin. (A) This chart displays the mean variance of
the internal synchrony of 9 brain networks for the sample of 15 healthy
volunteers, as a percentage change post vs. pre-infusion. A post-infusion
increase in metastability for a specific network indicates that the mean
signal in that network is a poor model of the activity in its constituent
voxels, implying that the network is behaving more “chaotically”
post-infusion than pre. Bonferonni correction gave a revised statistical
threshold of p<0.006 (0.05/9). One-sample (2-tailed) t-tests were
performed, comparing the % change against zero. The significant networks
are labeled with an asterisk. (B,C) These probability distributions were
derived from data from the same single subject, by discretizing a measure
of the internal synchrony of the DMN across time into bins. These bins
reflect the distance a data point is from the mean and this gives a
probability distribution of the variance of internal synchrony within a
network for a given time period (e.g., a 5 min period of scanning). The
probability distributions shown in Chart B were produced from placebo
data where it is clear that prediction of internal network synchrony of the
DMN across time is similar before and after infusion (i.e., the blue and
green curves). The probability distributions shown in Chart C were derived
using psilocybin data and here it is evident that following infusion of
psilocybin (i.e., the green curve), prediction of internal network synchrony
within the DMN is more difficult compared to pre infusion (the blue
curve). When the entropy change was calculated for the group,
significantly greater increases in entropy were found in the same networks
highlighted in (A) (post-psilocybin vs. pre) vs. (post-placebo vs. pre).
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |11
Carhart-Harris et al. The entropic brain
under psilocybin than either at baseline or after placebo. Indeed,
a number of motifs were exclusive to the psilocybin condition.
The entropy of a time series could then calculated by assess-
ing the entropy of a sequence of motifs over a period of time
(i.e., how easy/difficult is it to predict a sequence of motifs in
a given state?). This is the same procedure one would follow in
order to calculate the entropy of a transcribed passage of speech
for example (i.e., the likelihood of certain words occurring in a
coherent passage is not random, e.g., some words, such as “I,
occur much more often than others). Thus, it was found that the
sequence of motifs had significantly greater entropy under psilo-
cybin than at baseline, meaning that a more random sequence
of motifs played-out in the psychedelic state. This result implies
that it is harder to predict the sequence of connectivity motifs in
the psychedelic state because it is more random. This outcome
is entirely consistent with the entropic brain hypothesis, which
states that brain activity becomes more random and so harder to
predict in primary states - of which the psychedelic state is an
The DMN appears to have a consistently high level of activity, e.g.,
even when the DMN is relatively deactivated during goal-directed
cognition, it is still receives more blood flow than elsewhere in
the brain (Pfefferbaum et al., 2011). Thus, it can be inferred that
one reason why the DMN is so highly and persistently active, is
that it receive regular endogenous input from internal drivers.
One such driver may be MTL activity (Laxton et al., 2010)and
another may be input from brainstem nuclei such as the seroton-
ergic raphe nuclei. Irrespective of what the specific drivers of the
DMN are, its enduring presence fits comfortably with the idea
that it is the seat of the ego (Carhart-Harris and Friston, 2010),
as in healthy waking consciousness, one’s sense of self is never far
from consciousness:
“Normally, there is nothing of which we are more certain than the
feeling of our self, of our own ego.” (Freud, 1930)
So how does the phenomenon of primary consciousness fit in
here? The first thing to say is that primary consciousness may
be a sub-optimal mode of cognition that has been superseded
by a more reality-bound style of thinking, governed by the ego.
However, if primary consciousness is a psychological atavism,
and the psychedelic state is an exemplar of it, then how does
this explain the putative utility of the psychedelic experience
e.g., as an adjunct to psychotherapy (Moreno et al., 2006; Grob
et al., 2011) and why do some people report being so profoundly
affected by such experiences (and often seemingly for the better)
(Griffiths et al., 2008; Carhart-Harris and Nutt, 2010; MacLean
et al., 2011)?
The phenomenon of depression can help us here. Cognition
during an episode of depression is characteristically inflexible;
the patient’s focus is almost entirely inward and self-critical, and
he/she is unable to remove him/herself from this state (Holtzheimer
and Mayberg, 2011). In the previous section, depressive realism
was discussed in relation to hyper activity and connectivity within
the DMN; however, in severe depression, cognition cannot be said
to be optimal. Depressed patients typically perceive themselves
and their world through an unyielding pessimism (Styron, 1992).
Depressed patients’ cognitive style may become too fixed, such
that the patient loses the ability to think and behave in a flexible
manner. Underlying this phenomenon may be a decrease in
metastability, such that one particular state, e.g., the introspec-
tive default-mode, comes to dominate cognition. The aggressive
self-critical focus that accompanies a loss or abandonment of
object-cathexis in depression (i.e., interest in or focus on objects
in the world, such as work and people) quite naturally leads to
suicidal thoughts and acts (Carhart-Harris et al., 2008). In consid-
eration of these things, narrow-mindedness is to pessimism what
openness (MacLean et al., 2011) is to optimism and strategies that
promote the latter may be effective treatments for depression (see
MacLean et al., 2011).
This article proposes that primary consciousness rests on more
metastable dynamics than secondary consciousness, i.e., brain
sub-states are less stable in primary consciousness. Thus, by
implication, a broader repertoire of transient states may be vis-
ited in primary consciousness because the system is closer to
criticality-proper. Moreover, it is the ability of psychedelics to
disrupt stereotyped patterns of thought and behavior by dis-
integrating the patterns of activity upon which they rest that
accounts for their therapeutic potential. This principle implies
that a brain at criticality may be a “happier” brain. The schematic
below (Figure 7) illustrates differences between primary and sec-
ondary consciousness. The model describes cognition in adult
modern humans as “near critical” but “sub-critical”—meaning
that its dynamics are poised in a position between the two
extremes of formlessness and petrification where there is an
optimal balance between order and flexibility. However, because
of maturational settling, the brain in secondary consciousness
gravitates toward “order” and thus, the dynamics in this state
are more accurately, (slightly) sub-critical. Psychedelics may be
therapeutic because they work to normalize pathologically sub-
critical styles of thought (such as is seen in depression, OCD
or addiction/craving for example) thereby returning the brain
to a more critical mode of operating. Indeed, if the principle
holds that a critical brain is a happy brain, then it would fol-
low that psychedelics could be used to enhance well-being and
divergent thinking, even in already healthy individuals. One neg-
ative consequence of this however could be the neglect of accurate
Recent work has indeed supported the notion that brain
activity is slightly sub-critical in normal waking consciousness
(Priesemann et al., 2013). One reason why it may be advanta-
geous for the brain to operate just below criticality is that by
doing so, it can exert better control over the rest of the natu-
ral world—most of which is critical. This may take the form of
suppressing endogenous processes within the brain or interacting
with the environment in order to shape it and thereby control
it. Indeed, if control is the objective, then it makes sense that
the brain should be more ordered than that which it wishes to
The idea that the brain is closer to criticality in the psychedelic
state than in normal waking consciousness (Figure 7)has
some intuitive appeal as some of the signatures of criticality,
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |12
Carhart-Harris et al. The entropic brain
FIGURE 7 | Spectrum of cognitive states. This schematic is intended
to summarize much of what this paper has tried to communicate. It
shows an “inverted u” relationship between entropy and cognition such
that too high a value implies high flexibility but high disorder, whereas
too low a value implies ordered but inflexible cognition. It is proposed
that normal waking consciousness inhabits a position that is close to
criticality but slightly sub-critical and primary states move brain activity
and associated cognition toward a state of increased system entropy
i.e., brain activity becomes more random and cognition becomes more
flexible. It is proposed that primary states may actually be closer to
criticality proper than secondary consciousness/normal waking
such as maximum metastability, avalanche phenomena and
hypersensitivity to perturbation are consistent with the phe-
nomenology of the psychedelic state. For example, if we consider
just one of these: hypersensitivity to perturbation, it is well known
that individuals are hypersensitive to environmental perturba-
tions in the psychedelic state, which is why such emphasis is
placed on the importance of managing the environment in which
the psychedelic experience unfolds (Johnson et al., 2008). Indeed,
one explanation for why some people celebrate and romanti-
cize the psychedelic experience and even consider it “sacred”
(Schultes, 1980; McKenna, 1992), is that, in terms of critical-
ity, brain activity does actually become more consistent closer
with the rest of nature in this state i.e., it moves closer to
criticality-proper and so is more in harmony with the rest of
A final speculation that is worth sharing, is that the claim
that psychedelics work to lower repression and facilitate access
to the psychoanalytic unconscious, may relate to the brain mov-
ing out of a sub-critical mode of functioning and into a critical
one, enabling transient windows of segregation or modularity
to occur (e.g., with “anarchic” MTL activity) because of the
breakdown of the system’s hierarchical structure. Indeed, repres-
sion may depend on the brain operating in a sub-critical mode,
since this would constrain consciousness and limit its breadth.
Phenomena such as spontaneous personal insights and the com-
plex imagery that often plays out in psychedelic state (Cohen,
1967) and dreaming, may depend on a suspension of repres-
sion, enabling cascade-like processes to propagate through the
brain [e.g., from the MTLs to the association cortices (Bartolomei
et al., 2012)]. Such processes may depend on a reduction of DMN
control over MTL activity.
Many psychiatrists working with psychedelics in the 1950s and
60s expressed great enthusiasm about their therapeutic potential
(Crocket et al., 1963; Abramson, 1967; Grinspoon and Bakalar,
1979; Grof, 1980) but there was an unfortunate failure to substan-
tiate these beliefs with properly controlled studies. Subsequent
reviews and meta-analyses have suggested an impressive efficacy,
especially in relation to the use of LSD in the treatment of alco-
hol dependence (Mangini, 1998; Dyck, 2005; Krebs and Johansen,
2012) and modern trials have lent some support to this sen-
timent (Moreno et al., 2006; Grob et al., 2011). For example,
a single high dose of psilocybin produced profound existential
experiences in healthy volunteers that had a lasting beneficial
impact on subjective well-being (Griffiths et al., 2006, 2008)anda
moderate single dose of psilocybin administered to patients with
advanced-stage cancer significantly reduced anxiety and depres-
sion scores for months after the acute experience (Grob et al.,
2011). In another study, symptoms of obsessive compulsive dis-
order (OCD) were significantly reduced after psilocybin (Moreno
et al., 2006). Supplementing these controlled studies, we surveyed
over 500 recreational drug users, and found that 67% of LSD
users and 60% of psilocybin users claimed that use of these drugs
had produced long-term positive effects on their sense of well-
being (Carhart-Harris and Nutt, 2010), consistent with the results
of the aforementioned controlled studies (Griffiths et al., 2006,
2011). To place this in a context, only 6% of alcohol users claimed
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |13
Carhart-Harris et al. The entropic brain
such improvements from alcohol use (Carhart-Harris and Nutt,
2010). One of the most remarkable properties of psychedelics
is their potential to have a lasting impact on personality and
outlook (McGlothlin and Arnold, 1971; Studerus et al., 2011).
Personality traits are known to be relatively fixed by adulthood
(Costa and McCrae, 1997; McCrae and Costa, 1997), however,
the personality trait “openness” was found to be significantly
increased over 14 months after a single controlled administration
of psilocybin (MacLean et al., 2011). Moreover, neuroimaging
studies (Carhart-Harris et al., 2012a) have found decreased activ-
ity and connectivity after psilocybin in brain regions (e.g., the
mPFC) and networks (e.g., the DMN) that are over-engaged
in depression (Greicius et al., 2007; Berman et al., 2011)but
normalized by a range of effective treatments (Goldapple et al.,
2004; Mayberg et al., 2005; Kennedy et al., 2007; Deakin et al.,
Classic psychedelics are all agonists at the serotonin 2A
receptor (Glennon et al., 1984; Vollenweider et al., 1998)
and 5-HT2AR antagonism blocks the positive mood effects
of psilocybin (Kometer et al., 2012) and MDMA (van Wel
et al., 2012). 5-HT2AR expression is upregulated in depres-
sion (Bhagwagar et al., 2006) likely because of low synap-
tic 5-HT (Cahir et al., 2007). Positron emission tomography
(PET) studies in humans found positive correlations between
5-HT2AR binding and trait neuroticism (Frokjaer et al., 2008)
and pessimism (Meyer et al., 2003). This may imply that
5-HT2AR upregulation, due to low synaptic 5-HT, reflects
a state of chronically deficient post-synaptic 5-HT2AR stim-
ulation that contributes to inflexible patterns of (negative)
thought such as are seen in depression. 5-HT2AR-stimulation
may therefore work to reverse this, effectively “lubricating”
Given our knowledge of the biological effects psychedelics, a
comprehensive model can be presented in which psychedelics:
(1) stimulate the 5-HT2A receptor (Glennon et al., 1984), (2)
depolarize deep-layer pyramidal neurons (Andrade, 2011), (3)
desynchronize cortical activity, (4) “disintegrate” brain networks
(Carhart-Harris et al., 2012a), (5) increase network metastabil-
ity and (6) increase the repertoire of connectivity motifs within a
limbic/paralimbic network. The net effect of these processes is an
increase in system entropy (formally reflected in points 5 and 6)
as the system enters criticality-proper.
Thus, in summary, it is hypothesized that there is a basic mech-
anism by which psychedelics can be helpful in psychiatry, whether
they be used to treat depression, OCD (Moreno et al., 2006)
or addiction (Krebs and Johansen, 2012). Specifically, it is pro-
posed that psychedelics work by dismantling reinforced patterns
of negative thought and behavior by breaking down the stable
spatiotemporal patterns of brain activity upon which they rest.
An important caveat however, is that in order for this process
to be beneficial, the drug-induced transitions to, and the return
from primary consciousness, must be mediated by appropriate
therapeutic care (Johnson et al., 2008). Moving the brain out
of sub-critical modes and into unfamiliar terrain may present
some risks (e.g., loss of contact with reality and persistent mag-
ical/delusional thinking) if not properly managed (Johnson et al.,
“If we consider contemporary accounts of the mystical conscious-
ness, we can see that the individuality, the “I,” disappears and is in
a sense “annihilated.” (Stace, 1961)
In the psychology of religion, one of the most remarkable find-
ings has been that it is possible, by way of a single high dose
of psilocybin, to reliably induce profound spiritual experiences
in healthy volunteers that are effectively indistinguishable from
spontaneously-occurring spiritual experiences (Griffiths et al.,
2006). Perhaps this finding should not be so surprising, psilocybin
containing mushrooms have been used for centuries in shamanic
“healing” ceremonies (Hofmann, 1980), and in a famous study
in the 1960s, high-dose psilocybin was administered to theol-
ogy students partaking in a religious service on Good Friday
and emphatic spiritual experiences were reported (Doblin, 1991).
The so-called “entheogenic” (generating the divine) properties of
psilocybin appear to be shared by the other classic psychedelics
such as LSD and DMT but not the “psychedelic-like” compounds,
MDMA and cannabis (Carhart-Harris and Nutt, 2010; Lyvers and
Meester, 2012). It is intriguing that entheogenic properties appear
to be specific to 5-HT2AR agonist classic psychedelics and this
suggests a key role for this receptor in their genesis.
In William James’ famous lectures on the psychology of reli-
gion he proposed that spiritual experiences depend on the emer-
gence of what he referred to as the “subconscious” or “subliminal”
mind into consciousness (James, 1968). Referring to what psycho-
analysis calls “the unconscious.” James said: “[T]his is obviously
the larger part of each of us, for it is the abode of everything that
is latent and the reservoir of everything that passes unrecorded or
unobserved...It is the source of our dreams... In it arise whatever
mystical experiences we may have...It is also the fountain-head of
much that feeds our religion. In persons deep in the religious life—
and this is my conclusion—the door into this region seems unusually
wide open.” (James, 1968).
James’ ideas are consistent with those of Carl Jung; however,
Jung extended them, arguing that the unconscious hosts the psy-
chological remnants of our phylogenetic ancestry. In dreams,
psychosis and other altered states, archetypal themes shaped by
human history emerge into consciousness (Jung, 1982a). Jung’s
account of the “collective” unconscious fits comfortably with the
phenomenology of the psychedelic experience. Archetypal themes
feature heavily in user “trip reports” (Masters and Houston,
1966; Shanon, 2002), as they do in religious iconography. For
Jung, religion is a manifestation of the collective unconscious,
expressed in a symbolic and ritual form: “The brain is inherited
from its ancestors; it is the deposit of the psychic functioning of
the whole human race. In the brain, the instincts are preformed,
and so are the primordial images which have always been the
basis of man’s thinking—the whole treasure-house of mythological
motifs...Religious symbols have a distinctly “revelatory” character;
they are usually spontaneous products of unconscious psychic activ-
ity...they have developed, plant-like, as natural manifestations of
the human psyche.” (Jung, 1982b).
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |14
Carhart-Harris et al. The entropic brain
Jung’s ideas offer an appealing explanation for the content
of religious experiences, as well as the content of high-dose
psychedelic experiences; however, a more systematic treatise on
the spiritual experience was provided by Walter Stace in 1960
(Stace, 1961). Stace’s work is particularly useful because his ideas
resonate with the findings of recent neuroimaging studies relevant
to the neurobiology of spiritual experiences. Based on a thor-
ough review of first-person accounts derived from individuals
from a variety of different faiths, Stace identified seven univer-
sal components of the spiritual experience: 1) diminished spatial
and temporal awareness, 2) diminished subjectivity (equivalent
to increased objectivity), 3) feelings of profound joy and peace, 4)
a sense of divinity, 5) paradoxicality (where two opposing things
appear to be true), 6) ineffability (the difficulty of expressing the
experience in words) and 7) a sense of oneness with the world,
otherwise known as “the unitive experience.
Importantly, in Stace’s synopsis, he identified the unitive expe-
rience as the core characteristic of the spiritual experience. Freud
referred to the same phenomenon as the “oceanic state” (Freud,
1930). Stace explained that in profound spiritual experiences
the complex multiplicity of normal consciousness collapses into
a simpler state where a sense of an all-encompassing unity or
“oneness” with others, the world and/or “God” is felt. He empha-
sized that there is a collapse in the most fundamental dualities
of consciousness (i.e., self vs. other,subject vs. object and inter-
nal vs. external) in the unitive state. Moreover, he also showed
that reports of unitary consciousness are consistent throughout
the different religions—emphasizing its universality and cultural
independence (Stace, 1961).
Freud had some important things to say about the unitive state
that are directly relevant to the entropic brain hypothesis. For
example, when discussing his friend’s description of an “oceanic
feeling” when in religious practice, Freud says: “Pathology has
made us acquainted with a great number of states in which the
boundary lines between the ego and the external world become
uncertain... Further reflection tells us that the adult’s ego-feeling
cannot have been the same from the beginning. It must have gone
through a process of development...(For example,) an infant at the
breast does not as yet distinguish his ego from the external world;
he gradually learns to do so. Our present ego-feeling is therefore
only a shrunken residue of a much more inclusive—indeed, an all-
embracing feeling, which (early in development] corresponded to a
more intimate bond between the ego and the world. If we assume
that there are many people in whose mental life this primary ego-
feeling has persisted to a greater or less degree, it would exist in them
side by side with the narrower and more sharply demarcated ego feel-
ing of maturity, like a counterpart to it. In that case, the ideational
contents appropriate to it would be precisely those of limitlessness
and of a bond with the universe—the same ideas with which my
friend elucidated the “oceanic feeling.”” (Freud, 1930).
Thus, Stace and Freud’s descriptions of the spiritual experi-
ence are entirely consistent with the view of the primary state as
being regressive. Moreover, they are also consistent with view that
the human brain developed through ontogeny and phylogeny to
minimize disorder/uncertainty (Friston, 2010). In the schematic
presented in Figure 7, primary consciousness is depicted as being
more supercritical than normal waking consciousness. Indeed, at
the extreme end of supercriticality is maximum entropy, which is
equivalent to formlessness or complete disorder. Formerly, there
is no difference between entropy in this thermodynamic sense
(depicted as complete disorder) and entropy in the information
theory sense, where there is maximum uncertainty about the sys-
tem - because there is no order on which to base any predictions.
Extending this, the mechanics underlying the onset of true pri-
mary states (for which the spiritual experiences is an example)
can be viewed in relation to the second law of thermodynamics.
Explicitly, in the absence of a regular driving input, the system
(i.e., self-organized brain activity) will inevitably degrade or col-
lapse toward formlessness or maximum entropy. The interesting
question that follows therefore is: what is the driving input that
ceases in primary states? This paper proposes that regular MTL
activity is a crucial driver of the DMN. Although, interestingly,
there is also evidence that the usual clock-like firing of sero-
tonin neurons in the dorsal raphe nuclei completely ceases in
both the psychedelic state (Aghajanian et al., 1968; Aghajanian
and Vandermaelen, 1982) and REM sleep (Trulson and Jacobs,
1979) and there is some indirect evidence that the DMN may be
(at least partially) a serotonergic system coupled to dorsal raphe
activity (Zhou et al., 2010).
During secondary consciousness, the brain can enter a multi-
plicity of different states and microstates (Tononi, 2010)butdue
to “winner-takes all,” or more strictly, “winnerless” (Rabinovich
et al., 2001) competition between states [“winnerless” because
critical instability or metastability dictates that a state’s “vic-
tory” is transient (Friston et al., 2012b)] the global system only
ever entertains one winning state at any one time (Baars, 2005).
However, according to the entropic brain hypothesis, in primary
states, the potential multiplicity of possible states is not obliter-
ated but rather extended because the selectivity and conspicuity of
a winning state is reduced, and so more transient states may be
visited. In dynamical terms, this would be expressed as attractor
basins or valleys (defining transient states) becoming shallower in
primary states, i.e., the attractor landscape is flattened in primary
Conversely, in depression, OCD and addiction, specific states
(e.g., the default-mode in depression) may be frequented more
regularly than others—and this may be observed as certain mental
states (e.g., introspection in depression or craving in addiction)
or behaviors (e.g., compulsive acts in OCD) being habitually
revisited in a stereotyped fashion. Moreover, these states may be
relatively stable i.e., their basins of attraction are relatively steep
since the patterns of activity upon which they rest have become
entrenched. In such scenarios, uncertainty about the system is
minimized because we know that it possesses a particular char-
acter. It is intriguing to consider therefore that disorders such
as depression, OCD and addiction could be functional in some
sense, perhaps working to resist a more catastrophic collapse to
primary consciousness (with the onset of a psychotic episode for
example) by reinforcing stable patterns of activity.
The following example may help to illustrate what is meant
by competition between conscious states—and the loss of it in
primary consciousness. Functional brain imaging has identified
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |15
Carhart-Harris et al. The entropic brain
distinct brain networks that subserve distinct psychological func-
tions. For example, the DMN is associated with introspective
thought (Andrews-Hanna et al., 2010) and a dorsal frontoparietal
attention network (DAN) is associated with visuospatial attention
(Corbetta et al., 1998; Fox et al., 2006) and is a classic example
of a “task positive network” (TPN)—i.e., a network of regions
that are consistently activated during goal-directed cognition. If
the brain was to be sampled during a primary state (such as
a psychedelic state) we would predict that the rules that nor-
mally apply to normal waking consciousness will become less
robust. Indeed, we recently found this to be so when analysing
the degree of orthogonality or “anti-correlation” between the
DMN and TPN post-psilocybin. Post-drug there was a significant
reduction in the DMN-TPN anticorrelation, consistent with these
networks becoming less different or more similar (i.e., a flatten-
ing of the attractor landscape). The same decrease in DMN-TPN
anticorrelation has been found in experienced meditators dur-
ing rest (Brewer et al., 2011) and meditation (Froeliger et al.,
2012). Moreover, decreased DMN-TPN inverse coupling is espe-
cially marked during a particular style of meditation referred to
as “non-dual awareness” (Josipovic et al., 2011). This is interest-
ing because this style of meditation promotes the same collapse of
dualities that was identified by Stace (and Freud) as constituting
the core of the spiritual experience. The DMN is closely associated
with self-reflection, subjectivity and introspection, and task pos-
itive networks are associated with the inverse of these things, i.e.,
focus-on and scrutiny of the external world (Raichle et al., 2001).
Thus,itfollowsthatDMNandTPNactivitymust be competitive
or orthogonal in order to avoid confusion over what constitutes
self,subject and internal on the one hand, and other,object and
external on the other. It is important to highlight that disturbance
in one’s sense of self, and particularly one’s sense of existing apart
from one’s environment, is a hallmark of the spiritual (Stace,
1961) and psychedelic experience (Carhart-Harris et al., 2012b).
Moreover, as in the psychedelic state (Carhart-Harris et al., 2012a;
Carhart-Harris et al., 2012b), a number of studies have found
decreased DMN activity (Farb et al., 2007; Brewer et al., 2011;
Hasenkamp et al., 2012) as well as decreased DMN-TPN inverse
coupling in meditation (Brewer et al., 2011; Josipovic et al., 2011;
Froeliger et al., 2012).
The contravention or corruption of important rules about
brain organization may explain the sense of confusion and uncer-
tainty that accompanies a transition from secondary to primary
consciousness. In the information theoretical sense, “uncertainty”
is a synonym for entropy (Friston, 2010; Ben-Naim, 2012)—and
disorder and uncertainty are effectively equivalents. Entropy in
information theory is reflected in the shape of a probability distri-
bution (Ben-Naim, 2012), i.e., we have less confidence (or more
uncertainty) about something when the distribution is broader
or more evenly spread. This is because it is more difficult to
predict what the outcome of an individual sampling trial would
be because the system behaves relatively randomly (Ben-Naim,
2012). Conversely, a probability distribution with a sharp peak
would reflect a well-ordered system or high-precision, confidence
or assuredness (Friston, 2010). In the specific context of the DMN
and the psychedelic state, just as there is increased variance in
parameters defining the DMN (e.g., coupling between the nodes
of the DMN or rhythmic alpha oscillations in the PCC), so there is
uncertainty about ones sense of self—typically described as “ego-
disintegration.” Thus, according to the entropic brain hypothesis,
just as normally robust principles about the brain lose definition
in primary states, so confidence is lost in “how the world is” and
“who one is” as a personality.
In addition to the word “uncertainty,” other terms that have
been used as synonyms for entropy include: “freedom,” “disorder”
and “expansion.” The example of a gas expanding post release
of constraints is often used as a metaphor to help explain what
is meant by entropy increasing [e.g., in relation to the second
law of thermodynamics (Ben-Naim, 2012)] See Figure 8.Inthe
information theoretical sense, entropy/uncertainty increases as
the gas expands because with greater expansion, it is more dif-
ficult to predict the spatial location of a single molecule. It is
probably not coincidental that these physical principles resonate
with popular descriptions of the psychedelic experience (Huxley,
1959; Bowers and Freedman, 1966; Masters and Houston, 1972;
Grinspoon and Bakalar, 1981; Merkur, 1998). For example, the
term “consciousness-expansion” is often used to describe the
psychedelic experience—and this may be an inadvertent reference
to increased system entropy in the psychedelic state.
To develop the construct validity of primary consciousness, it will
be necessary to show that the identified parameters of primary
FIGURE 8 | Gas expansion post-release of constraints as a metaphor
for increased entropy in primary states. (A) Entropy is low while the gas
is constrained. (B) Entropy increases once constraints are released. In an
information theoretical sense, entropy/uncertainty is increased
post-expansion because it is more difficult to predict the spatial location of
a single molecule. In primary vs. secondary states, it is hypothesized that
the biological parameters known to define key brain states (e.g., the
default-mode) become more variant or less predictable, thus causing the
subject to become less certain in themselves and their experience of the
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |16
Carhart-Harris et al. The entropic brain
consciousness have high internal and convergent validity (i.e.,
properties of primary consciousness must be shown to exist in
a range of different primary states) and sufficient discriminant
validity (i.e., these properties must be shown to be sufficiently spe-
cific to primary states, i.e., to be absent in non-primary states). To
assist this undertaking, it will be important to optimize subjec-
tive rating scales designed for assessing primary consciousness.
For example, first it will be necessary to identify key experi-
ences that are considered characteristic of primary states (i.e.,
visual analog scale items such as “my imagination was extremely
vivid” or “the experience had a dreamlike quality” that can be
rated during or after a candidate primary state) and then it will
be important to test whether different candidate primary states
(e.g., dreaming, onset-phase psychosis, the near death experi-
ence, the sensory-deprived state and the psychedelic drug state)
score highly on these items—and that candidate non-primary
states (e.g., normal waking consciousness, the anaesthetized or
sedated state, and the depressed state) do not. This will enable
the convergent and discriminant validity of these measures to
be tested and developed. Similarly, by identifying neurobiologi-
cal characteristics hypothesized to be essential to primary states
(e.g., decreased DMN-MTL coupling, disinhibited MTL activity,
decreased DMN-TPN anti-correlation, decreased alpha power in
the PCC and increased metastability in resting-state networks),
it will be important to determine those that most reliably and
specifically identify primary over non-primary states.
As discussed in The research value of psychedelics, psychedelic
drugs are especially useful tools for studying primary states as they
allow for primary consciousness to be “switched on” with a rela-
tively high degree of experimental control (e.g., with intravenous
infusion of a classic psychedelic). Hypotheses about the neuro-
biological character of primary states can therefore be effectively
tested by psychedelic drugs. However, in order to test and develop
the generalizability of these hypotheses, research with alternative
primary states are required. For example, it would be interest-
ing to carry out simultaneous fMRI-EEG or MEG work with a
focus on REM sleep, or to study patients exhibiting early-phase
psychotic symptoms with these techniques. Longitudinal analy-
ses looking at brain maturation would also be relevant, where
infant consciousness is hypothesized to be reflective of primary
As outlined above, a key distinction between the primary and sec-
ondary modes of cognition is that secondary consciousness pays
deference to reality and diligently seeks to represent the world as
precisely as possible, whereas primary consciousness is less firmly
anchored to reality and is easily misled by simple explanations
motivated by wishes and fears. One way this distinction could
be tested would be to utilize a measure of metacognitive accu-
racy (Fleming et al., 2010). As outlined above, metacognition,
and specifically the ability to reflect upon one’s own introspec-
tion, is a particularly advanced behavior associated with the DMN
(Fleming et al., 2010). For example, a behavioral paradigm could
be designed that requires a participant’s friend to rate the partic-
ipant’s personality, e.g., using a standard personality inventory.
Then, during scanning, the participant could be asked to pre-
dict their friend’s ratings—and crucially, to provide an additional
confidence rating for their own predictions. This could be done
under a psychedelic drug and under placebo in a within-subjects
design with 2 different friends for each condition, counterbal-
anced for key factors (e.g., familiarity, intimacy, fondness, dura-
tion of relationship etc). This task would provide a behavioral
index of a high-level metacognitive function associated with the
ego (theory-of-mind). The hypothesis would be that participants
would be less confident in their predictions of their friend’s rat-
ings post-psilocybin and that the accuracy of their predictions
would also be compromised. Biologically, one would hypothesize
decreased within-DMN coupling during the prediction process
and a reduction in induced alpha-oscillations in the PCC.
This article has argued that scientific research with psychedelic
drugs can have a revitalizing effect on psychoanalysis and an
informing influence on mainstream psychology and psychi-
atry. Rather than discuss the content and interpretation of
psychoanalytically-relevant material, we have adopted a mech-
anistic approach, in keeping with the mainstream cognitive
neuroscience. This article proposes that a distinction can be
made between two fundamentally different modes of cogni-
tion: primary and secondary consciousness. Primary conscious-
ness is associated with unconstrained cognition and less ordered
(higher-entropy) neurodynamics, whereas secondary conscious-
ness is associated with constrained cognition and more ordered
neurodynamics (i.e., that strikes an evolutionarily advantageous
balance between order and disorder - that may or more not
be perfectly “critical”). It is hoped that this mechanistic model
will help catalyze a synthesis between psychoanalytic theory and
cognitive neuroscience that can be mutually beneficial to both
It is a fair criticism of this paper that it has given insufficient
consideration to the phenomenological content of the relevant
altered states of consciousness, and to the specifics of Freudian
theory, and so by neglecting this, has failed to present a suffi-
ciently compelling case that these states have anything to do with
psychoanalytic theory. To some extent, this charge can be con-
ceded; however, as outlined in the introduction, the intention of
this paper was to develop a mechanistic account of altered states
of consciousness based on the quantity of entropy, and this task
has demanded a substantial amount of space. A more thorough
discussion of the phenomenology of primary states is required to
develop the case that they show characteristics that are consistent
with Freudian accounts of “the unconscious” or “Id.” The reader
should be made aware however, that this has been attempted
before (Carhart-Harris, 2007; Carhart-Harris and Friston, 2010).
To conclude, it is perhaps not surprising that with only
dreaming and psychosis at its disposal, psychoanalysis has failed
to convince the scientific community that the psychoanalytic
unconscious exists (Hassin et al., 2005). From a neuroscientific
perspective, dreaming and psychosis are notoriously difficult to
study. The occurrence of dreaming in sleep impedes experimen-
tal control and psychosis is an especially complex and variegated
phenomenon. However, for those brave enough to embrace it,
Frontiers in Human Neuroscience February 2014 | Volume 8 | Article 20 |17
Carhart-Harris et al. The entropic brain
research with psychedelics could herald the beginning of a new
scientifically informed-psychoanalysis that has the potential to
influence modern psychology and psychiatry. The unique scien-
tific value of psychedelics rests in their capacity to make con-
sciously accessible that which is latent in the mind. This paper
takes the position that mainstream psychology and psychiatry
have underappreciated the depth of the human mind by neglect-
ing schools of thought that posit the existence an unconscious
mind. Indeed, psychedelics’ greatest value may be as a remedy for
ignorance of the unconscious mind.
“He who would fathom the psyche must not confuse it with con-
sciousness, else he veils from his own sight the object he wishes to
explore.” (Jung, 1961)
“Man’s worst sin is unconsciousness.” (Jung, 1969)
We would like to thank the reviewers for their useful comments
on previous versions of this manuscript. Robin L. Carhart-Harris
would also like to acknowledge that Robert Leech coined the
term “the entropic brain” when in conversation with him over
the topic of his and David Nutt’s psilocybin fMRI results. Robin
L. Carhart-Harris is currently supported by the Medical Research
(1957). PSYCHODYNAMIC and therapeutic aspects of mescaline and lyser-
gic acid diethylamide: round table. J. Nerv. Ment. Dis. 125, 423–424. doi:
Abramson, H. A. (1967). The Use of LSD in Psychotherapy and Alcoholism.
Indianapolis, IN: Bobbs-Merrill.
Adelstein, J. S., Shehzad, Z., Mennes,M., De young, C. G., Zuo, X. N., Kelly, C., et al.
(2011). Personality is reflected in the brain’s intrinsic functional architecture.
PLoS ONE 6:e27633. doi: 10.1371/journal.pone.0027633
Aghajanian, G. K., Foote, W. E., and Sheard, M. H. (1968). Lysergic acid diethy-
lamide: sensitive neuronal units in the midbrain raphe. Science 161, 706–708.
doi: 10.1126/science.161.3842.706
Aghajanian, G. K., and Marek, G. J. (1997). Serotonin induces excitatory
postsynaptic potentials in apical dendrites of neocortical pyramidal cells.
Neuropharmacology 36, 589–599. doi: 10.1016/S0028-3908(97)00051-8
Aghajanian, G. K., and Vandermaelen, C. P. (1982). Intracellular recordings from
serotonergic dorsal raphe neurons: pacemaker potentials and the effect of LSD.
Brain Res. 238, 463–469. doi: 10.1016/0006-8993(82)90124-X
Andrade, K. C., Spoormaker, V. I., Dresler, M., Wehrle, R., Holsboer, F.,
Samann, P. G., et al. (2011). Sleep spindles and hippocampal functional
connectivity in human NREM sleep. J. Neurosci. 31, 10331–10339. doi:
Andrade, R. (2011). Serotonergic regulation of neuronal excitabil-
ity in the prefrontal cortex. Neuropharmacology 61, 382–386. doi:
Andrews-Hanna, J. R., Reidler, J. S., Huang, C., and Buckner, R. L. (2010). Evidence
for the default network’s role in spontaneous cognition. J. Neurophysiol. 104,
322–335. doi: 10.1152/jn.00830.2009
Axmacher, N., Cohen, M. X., Fell, J., Haupt, S., Dumpelmann, M., Elger, C. E.,
et al. (2010). Intracranial EEG correlates of expectancy and memory formation
in the human hippocampus and nucleus accumbens. Neuron 65, 541–549. doi:
Axmacher, N., Elger, C. E., and Fell, J. (2008). Ripples in the medial temporal
lobe are relevant for human memory consolidation. Brain 131, 1806–1817. doi:
Baars, B. J. (2005). Global workspace theory of consciousness: toward a cog-
nitive neuroscience of human experience. Prog. Brain Res. 150, 45–53. doi:
Bak, P., Tang, C., and Wiesenfeld, K. (1987). Self-organized criticality:
an explanation of the 1/f noise. Phys. Rev. Lett. 59, 381–384. doi:
Ban, T. A. (2001a). Pharmacotherapy of depression: a historical analysis. J. Neural
Transm. 108, 707–716. doi: 10.1007/s007020170047
Ban, T. A. (2001b). Pharmacotherapy of mental illness–a historical analysis.
Prog. Neuropsychopharmacol. Biol. Psychiatry 25, 709–727. doi: 10.1016/S0278-
Bartolomei, F., Barbeau, E. J., Nguyen, T., McGonigal, A., Regis, J., Chauvel, P., et al.
(2012). Rhinal-hippocampal interactions during deja vu. Clin. Neurophysiol.
123, 489–495. doi: 10.1016/j.clinph.2011.08.012
Basar, E., and Guntekin, B. (2009). Darwin’s evolution theory, brain oscillations,
and complex brain function in a new “Cartesian view.Int. J. Psychophysiol. 71,
2–8. doi: 10.1016/j.ijpsycho.2008.07.018
Bassett, D. S., Bullmore, E., Verchinski, B. A., Mattay, V. S., Weinberger, D. R.,
and Meyer-Lindenberg, A. (2008). Hierarchical organization of human cor-
tical networks in health and schizophrenia. J. Neurosci. 28, 9239–9248. doi:
Beggs, J. M., and Plenz, D. (2003). Neuronal avalanches in neocortical circuits.
J. Neurosci. 23, 11167–11177.
Ben-Naim, A. (2008). A farewell to Entropy: Statistical Thermodynamics Based on
Information: S=logW. London: World Scientific, Hackensack, N.J.
Ben-Naim, A. (2012). Entropy and the Second Law: Interpretation and Misss-
Berman, M. G., Peltier, S., Nee, D. E., Kross, E., Deldin, P. J., and Jonides, J. (2011).
Depression, rumination and the default network. Soc. Cogn. Affect. Neurosci. 6,
548–555. doi: 10.1093/scan/nsq080