Being You: A New Science of Consciousness by A. Seth

 

Context

Bayesian probability

Anil Seth’s work is deeply shaped by Bayesian ideas, especially through the predictive-processing framework that treats the brain as a prediction engine. Thomas Bayes (1702-1761) was a Presbyterian theologian and mathematician who was the first to use probability inductively and who established a means of calculating, from the frequency with which an event has occurred in prior trials, the probability that it will occur in future trials.

Bayesian probability is an interpretation of the concept of probability in which, instead of frequency or propensity of some phenomenon, probability is interpreted as a reasonable expectation representing a state of knowledge or as a quantification of a personal belief.

Seth argues that perception is not a passive recording of the world but an active process of hypothesis testing: the brain continuously generates prior expectations and compares them to incoming sensory data, updating its beliefs to minimise prediction error. This “controlled hallucination” metaphor directly invokes Bayesian updating, where priors and likelihoods are weighted to produce perceptual experience.

Seth places particular emphasis on the role of priors and their precision (the confidence assigned to predictions) in determining conscious content. Changes in prior strength or precision can account for attention, perceptual biases, and many illusions. They also offer explanatory leverage for some psychiatric and perceptual disorders where weighting between expectation and sensory evidence is disrupted. This focus on precision mirrors Bayesian schemes that weigh prior beliefs against new evidence according to their relative certainty.

Seth extends Bayesian reasoning beyond exteroceptive perception to interoception and the sense of self. In Seth’s view the feeling of being a self emerges from hierarchical inference about bodily states. The brain infers the causes of internal signals just as it infers external causes, producing a coherent self-model. This inference-based account reframes selfhood as a maintained hypothesis about bodily and environmental causes rather than a static entity.

Seth also adopts and discusses active inference, a Bayes-flavoured approach in which actions are chosen to fulfil predictions and reduce expected surprise. Active inference ties perception and action together in an embodied loop, offering mechanistic ways to explain sensorimotor coordination, body ownership, and adaptive behaviour. He integrates these ideas into empirical and computational work, using models and experiments (for example, to analyse multisensory integration and body-ownership illusions) to test how Bayesian-like processes can generate conscious experience.

While Seth is broadly sympathetic to Bayesian and predictive-processing frameworks, he stresses pluralism and empirical grounding. Bayesian ideas are powerful heuristics and modelling tools, but they must be married to mechanistic neuroscience and tested against data rather than treated as metaphysical truth.

Kant

Seth often echoes Kant’s core claim that we never perceive things‑in‑themselves but only brain‑mediated appearances. As Seth writes, 

“Perception is not a passive ‘mirror’ of the world but an active process of hypothesis testing — the brain is continually predicting the causes of its sensory inputs.” 

This formulation maps onto Kant’s distinction between noumena and phenomena: Seth frames conscious experience as the brain’s constructed appearance rather than direct access to reality.

Seth explicitly links this lineage to Helmholtz and predictive accounts of perception: 

“Perception and action are best understood as forms of probabilistic inference: the brain infers the most likely causes of its noisy sensory data.” 

That probabilistic‑inference picture is a modern instantiation of Kant’s idea that the mind brings organising principles to experience. He also gestures toward ethical implications related to Kantian moral concerns when discussing attribution of consciousness. While not invoking Kant’s categorical imperative directly, Seth’s emphasis on how we judge and treat potentially conscious systems resonates with Kantian moral psychology.

Andy Clark

Andy Clark's concepts influenced Anil Seth's vision of consciousness through the extended mind theory. Clark's theory suggests that the mind is not confined to the brain but extends into the environment through tools and technologies. This perspective aligns with Seth's views on how consciousness may involve interactions with external systems, including AI.

Clark also emphasised connectionism, which states that cognitive processes are underpinned by neural networks rather than symbolic processing. Seth's research reflects this approach by focusing on how consciousness arises from complex neural interactions.

Clark's work engages with various philosophical ideas, including functionalism and the nature of subjective experience. Seth incorporates these philosophical discussions into his exploration of consciousness, particularly in distinguishing between intelligence and conscious experience.

The philosophy of mind

Philosophical concepts about representation, selfhood, and illusion from thinkers such as Daniel Dennett, Thomas Metzinger, and enactivist/debate traditions shape Seth's framing of selfhood as a controlled hallucination.

Daniel Dennett shaped Anil Seth’s overall scientific stance toward consciousness by promoting a naturalistic, anti‑mysterian approach that treats subjective experience as a phenomenon to be explained rather than believed. Seth adopts Dennett’s emphasis on functional and representational accounts of mind, favours interdisciplinary methods, and echoes Dennett’s scepticism of dualist intuition. This is visible in Seth’s insistence that consciousness can be explained by identifiable mechanisms rather than irreducible metaphysical properties.

Thomas Metzinger influenced Seth more directly at the theoretical level through the Self‑Model Theory of Subjectivity: the idea that the self is a constructed, transparent representational model. Seth incorporates Metzinger’s core claim that the sense of being a self is produced by neural representations, and he extends that framework by embedding the self‑model in predictive processing and interoceptive accounts, showing how bodily prediction and affect contribute to the feeling of presence.

Seth synthesises these influences by combining Dennett’s programmatic naturalism with Metzinger’s formal self‑modeling: he frames conscious experience as a “controlled hallucination” generated by predictive brain mechanisms that construct both perception and selfhood. This integration steers Seth’s empirical agenda—using perception, interoception, anaesthesia, and clinical disorders to test and refine self‑model and predictive accounts—while preserving Dennett’s insistence on scientific, testable explanations.

Phenomenology

First-person perspectives on experience (for instance Husserl and Merleau‑Ponty's emphasis on lived experience) influence Seth's attention to subjective reports and how experience feels.

Anil Seth’s work on consciousness is notably shaped by phenomenology’s emphasis on first-person experience, even as he grounds his theories in neuroscience and computational approaches. Phenomenology’s core idea that conscious experience has a structured, describable subjective character informs Seth’s insistence that any scientific theory of consciousness must account for the qualitative, experiential aspects (the “what it is like”) rather than only objective behaviour or neural correlates. This leads him to treat perception as an inherently embodied, situated process: experience is not a passive mirroring of the world but an enacted, predictive process shaped by an organism’s body, goals, and prior expectations. This is an outlook resonant with phenomenological thinkers who stress the lived body and intentionality.

Phenomenology’s methodological respect for careful description of experience also appears in Seth’s use of perceptual illusions and controlled phenomenology in experiments. He often leverages subjective reports and manipulations of appearance (e.g., body‑ownership illusions, altered sensory feedback) to reveal how brain mechanisms construct conscious content. Moreover, phenomenology’s critique of naive realism, the idea that perception simply reveals the world as it is, parallels Seth’s predictive processing framework, which treats perception as Bayesian inference where the brain generates best‑guess models of the causes of sensory input. This alignment frames consciousness as a controlled hallucination: experience arises from brain predictions constrained by sensory data, a formulation that echoes phenomenological accounts of the constructive, sense‑making character of experience.

Seth also integrates phenomenological concerns about selfhood and embodied subjectivity into his discussions of the minimal self and the sense of presence. He acknowledges that experiences of body ownership, self‑location, and agency are not mere epiphenomena but central phenomena that any theory must explain, thus bridging phenomenology’s descriptive richness with computational and neuroscientific mechanisms. While Seth does not adopt phenomenology completely, he frequently dialogues with it, borrowing its insights about lived experience and intentionality, translating them into testable hypotheses and using empirical methods to probe and refine the concepts that phenomenology focuses on.

Anil Seth’s work on consciousness has been shaped strongly by mechanistic theories that prioritise explanation in terms of parts, processes, and their interactions. Rather than stopping at functional or phenomenological description, he seeks concrete neurocomputational mechanisms that produce conscious experience, linking levels from neuronal circuits through network dynamics to embodied behaviour. This commitment steers his move from abstract accounts to testable, implementational hypotheses about how perception and selfhood arise.

Mechanistic determination

Predictive processing figures centrally in Seth’s mechanistic outlook. He treats hierarchical generative models and predictive coding not merely as metaphors but as plausible circuit architectures, with feedforward prediction errors and feedback predictions realized by specific neural motifs. Key mechanistic parameters, most notably precision-weighting, are cast in neurobiological terms (e.g., synaptic gain, neuromodulation), providing a bridge between Bayesian algorithms and physiological substrates of attention and perceptual inference.

Mechanistic thinking also underpins Seth’s account of the self as a “controlled hallucination.” He argues that selfhood emerges from ongoing mechanisms that integrate interoceptive and exteroceptive signals with action to minimize prediction error, and he points to concrete interoceptive pathways (visceral afferents, insular cortex, hypothalamic interactions) as substrates for affect and the felt sense of being. This grounds abstract notions of selfhood in identifiable bodily and neural processes.

Empirically, mechanistic commitments lead Seth to favor experiments and interventions that can adjudicate causal roles — manipulating sensory priors, perturbing precision, altering interoceptive input, or using stimulation and pharmacology — rather than relying solely on correlational evidence. Computational modeling plays a dual role as both explanatory framework and mechanistic probe, showing how simple rules can generate perceptual and self-related phenomena and suggesting specific neural implementations to test.

Seth’s approach reflects mechanistic pluralism and constraint. He accepts that multiple mechanisms across circuits, neuromodulators, and timescales contribute to different aspects of consciousness and he uses mechanistic evidence to temper grand philosophical claims. Mechanistic proposals therefore shape not only his explanatory targets but also his methods, driving a research programme that emphasizes concrete, multi-level, and experimentally tractable accounts of perception and selfhood.

David Chalmers on consciousness

David Chalmers and Anil Seth approach consciousness from sharply different starting points, producing complementary but often contested perspectives. Chalmers is best known for formulating the “hard problem” of consciousness: explaining why and how physical processes give rise to subjective experience — the qualitative, felt aspect known as phenomenality or “what it is like.” For Chalmers this problem resists standard mechanistic explanation; even a complete account of brain function might leave open why there is something it is like to undergo those processes. He treats consciousness as a fundamental feature of the world or as requiring novel principles (e.g., property dualism or panpsychism) unless a convincing reductive account appears.

Anil Seth, by contrast, reframes the scientific target away from Chalmers’ metaphysical worry toward what Seth calls “the real problem”: how particular conscious experiences arise from specific biological and computational mechanisms. Seth emphasises predictive processing, embodied cognition, and interoception, arguing that perception and selfhood are model-based constructs — “controlled hallucinations” — produced by the brain’s predictive inference about causes of sensory and bodily signals. For Seth, consciousness is a biological phenomenon to be explained by identifying mechanisms that map onto experiential contents. The pressing task is empirical and mechanistic rather than metaphysical.

The two positions overlap insofar as both take subjective experience seriously and seek theories that connect mind and brain. Chalmers accepts that neuroscience and cognitive science can and should map functions, reportability, and neural correlates of consciousness. He just insists that such mappings do not by themselves solve why experience exists. Seth likewise works to identify neural and computational correlates, but treats those accounts as directly explanatory of how experiences take their particular forms. Where Chalmers suspects a conceptual gap that might point to new ontological categories, Seth treats that gap as methodological: clarify and narrow the target (from “why experience?” to “how experiences are constructed?”) and approach it with experiments and models.

Philosophically, Chalmers foregrounds conceptual analysis, metaphysics, and the limits of physicalist explanation; Seth emphasises empirical neuroscience, predictive-processing theory, and embodied approaches. This leads to different attitudes toward potential solutions: Chalmers is open to non-reductive or novel foundational principles (and has explored panpsychism as one option), while Seth is optimistic about reductive, mechanistic accounts grounded in brain–body–world dynamics.

Their debate is productive rather than merely adversarial: Chalmers’ hard problem highlights why consciousness demands special attention and conceptual clarity, while Seth’s real-problem programme demonstrates how framing the question in mechanistic, experimentally tractable terms can make rapid scientific progress. Taken together, their work frames the field’s twin tasks: map and model the mechanisms that produce specific experiences, and at the same time grapple with whether and how those mechanisms close the conceptual gap about why subjective experience exists at all.

Interoception research

Work on bodily signals (heartbeat, breathing, gut) and their role in emotion and selfhood (e.g., research by Hugo Critchley, Manos Tsakiris) is heavily cited by Seth.

Critchley and Tsakiris influenced Anil Seth by supplying empirical paradigms and physiological grounding that anchored his predictive-processing account of consciousness in concrete bodily signals. 

Critchley’s work on cardiac and autonomic interoception — using heartbeat-detection tasks, heartbeat-evoked potentials, and autonomic monitoring — gave Seth testable mechanisms linking internal bodily states to feeling and affect, supporting his claim that conscious experience is deeply embodied. 

Tsakiris’s experiments on multisensory body representation, including variants of the rubber hand illusion, demonstrated how exteroceptive and interoceptive cues are dynamically integrated to shape body ownership, reinforcing Seth’s view that the self is an inferential, malleable construct shaped by multisensory prediction and precision-weighting. 

Together, their methods and findings moved Seth’s theories from abstract Bayesian models toward empirically grounded hypotheses showing how hierarchical prediction across interoceptive and exteroceptive streams generates subjective selfhood and affective experience.

Computer science 

Computer science and artificial intelligence have provided Seth with both conceptual frameworks and practical tools for his research. Central to his thinking is the predictive processing framework, a computational view that treats perception as Bayesian inference, where the brain continuously generates and updates probabilistic predictions about sensory input. This perspective, rooted in ideas from machine learning and probabilistic modelling, underpins Seth’s characterisation of perception as “controlled hallucination,” emphasising that conscious experience arises from the brain’s predictive simulations rather than a passive reception of sensory data.

Machine learning and generative models from AI have also shaped Seth’s methodological approach. Formal models and computational simulations allow him and collaborators to implement and test precise hypotheses about perception, attention, and the neural correlates of conscious states. Techniques from information theory and algorithmic analysis inform his engagement with measures of complexity and integrated information, tools used to probe candidates for consciousness correlates, while he remains critical of specific formulations that overreach or misapply the concepts. At the same time, Seth uses AI as a clarifying metaphor but cautions against equating current artificial systems with biological brains, highlighting differences in embodiment, evolutionary history, and active inference mechanisms.

Beyond theory and methods, advances in AI have fostered interdisciplinary dialogue that Seth leverages to refine his ideas and design experiments. The rise of neural networks and probabilistic modelling has opened channels for exchange with computer scientists, enabling new experimental paradigms and computational tests of consciousness theories. Additionally, progress in AI prompts Seth to address philosophical and ethical questions about attributing experience to artificial systems. He argues for careful conceptual analysis rather than premature attributions, stressing that computational similarity alone does not settle whether a system is conscious.

Commentary

Being You: A New Science of Consciousness by Anil Seth was published in 2021.

The Real Problem

Anil Seth reframes the core mystery of consciousness away from the traditional “hard problem” (why subjective experience exists) toward what he calls “the real problem”: explaining how conscious experiences are generated and why they have the particular qualities they do. He emphasises that consciousness should be studied as a biological, predictive, and dynamic process rooted in brain-body-world interactions rather than as an inexplicable metaphysical leftover. Seth argues that asking why there is experience at all is less scientifically tractable than asking how specific experiences arise from specific mechanisms.

Seth outlines a predictive-processing account in which the brain is a hypothesis-testing organ that continuously generates perceptual predictions and minimises prediction error. Conscious content, in this view, corresponds to the brain’s best current model of the causes of its sensory inputs. He writes, “Perception is controlled hallucination,” meaning that what we experience is the brain’s prediction constrained by incoming sensory data. This reframing makes room for mechanistic explanations: by identifying computational and neural processes that produce particular experiential qualities, neuroscience can address the “real problem.”

He stresses the role of embodiment and interoception (the brain’s modelling of the body’s internal states) in shaping conscious experience and selfhood. Seth suggests that the sense of self arises from predictive models that integrate bodily signals with external sensory information, producing a stable, embodied perspective. He notes, “The self is a controlled hallucination,” paralleling his claim about perception and highlighting that subjective selfhood is a constructed, model-based phenomenon rather than a simple given.

Seth also defends an empirical programmme: use careful experiments, computational models, and converging measures to map mechanisms to experiences. He urges moving beyond metaphysical standoffs and toward testable hypotheses that link brain activity, bodily dynamics, and reported experience. As he puts it, the aim is “to explain how each particular aspect of experience arises”, a shift that makes consciousness a tractable scientific project rather than an intractable mystery.

Measuring Consciousness

Seth argues that consciousness must be measured indirectly by triangulating behaviour, reports, and neural data, and that we should distinguish between global levels of consciousness (how conscious a system is) and specific contents (what it experiences). He writes, “We need to separate level and content: they are different empirical targets that require different methods,” stressing that conflating them leads to confusion in both research and clinical practice.Seth emphasises that a verbal report is the gold standard for content but is often unavailable, so researchers rely on objective tasks (forced-choice, signal-detection) and “no-report” paradigms as alternatives. He notes, “Reports are invaluable but not always possible — we must develop reliable proxies,” especially for infants, nonhuman animals, and patients with impaired communication. On neural measures, Seth reviews candidate signatures such as widespread recurrent activity, frontoparietal dynamics, and metrics of integration and differentiation, while cautioning that correlation does not equal explanation. He highlights information-theoretic approaches and practical indices like the Perturbational Complexity Index (PCI), saying that “complexity measures can serve as useful proxies for the brain’s capacity for conscious states.”

Crucially, Seth promotes causal perturbation methods (TMS–EEG, optogenetics) to reveal whether a brain can produce integrated, differentiated responses to evidence more diagnostic than passive observation. He writes, 

“Perturbation plus measurement gives us causal leverage: it shows whether the system’s dynamics can support conscious states.”

Seth also discusses applied contexts: assessing disorders of consciousness, monitoring anaesthesia, and the challenges of comparing consciousness across development, species, and machines. He warns of ethical stakes, observing that “measurement choices have moral consequences” and that uncertainty requires cautious, conservative judgements.His practical prescription is pluralistic and pragmatic: “Triangulate: use reports, behaviour, neural signatures and perturbations” to make the most reliable inferences about conscious states, because no single metric suffices.

Phi

In July 2006, Anil Seth finds himself in Las Vegas, engaging in a heated discussion with Giulio Tononi regarding the Integrated Information Theory (IIT) of consciousness. IIT posits that consciousness arises from specific patterns of information processing and is an intrinsic property of systems that integrate information beyond the sum of their parts.

IIT identifies consciousness with a measure called Φ (phi), which quantifies the level of integrated information. A system's consciousness is proportional to its Φ value: high Φ indicates significant consciousness, while zero Φ indicates the absence of consciousness. The theory suggests that both the information contained in a system and how that information is integrated are crucial for conscious experience.

Challenges arise in testing IIT's claims, particularly the assertion that Φ equates to consciousness. Measuring Φ is complicated, requiring knowledge of the system's potential states rather than just its observable behaviours. This intrinsic approach contrasts with observer-relative measures of information, making empirical validation difficult.

IIT leads to provocative implications, including the idea that information itself has ontological significance, echoing theories that posit everything in existence is derived from information. The theory suggests a restrained form of panpsychism, where consciousness exists wherever integrated information is found, but not pervasively throughout the universe.

While IIT is ambitious and intellectually stimulating, its testability remains a major concern. Recent initiatives aim to develop alternative measures of integrated information that employ observer-relative information, thereby seeking practical applications. The author's collaborative work focuses on refining these measures to align theoretical insights with empirical reality, acknowledging the complex relationship between consciousness and its underlying mechanisms.

The exploration of consciousness through IIT highlights the intricate interplay between subjective experience and quantifiable phenomena, suggesting pathways for future research in consciousness science. The concept of personal identity is intricately linked to the theories of consciousness, particularly in how we understand ourselves as conscious beings. Exploring personal identity can reveal how our experiences and the integrated information that constitutes our consciousness shape our sense of self. Thus, the study of Integrated Information Theory not only sheds light on consciousness but also provokes deeper questions about the nature of personal identity in relation to our subjective experiences.

Perceiving from the Inside Out

Seth opens with a vivid description of his surroundings in a cypress forest in California, setting the stage for a discussion on perception and consciousness. He introduces the concept of consciousness as a form of "controlled hallucination," suggesting that what we perceive is constructed by our brains rather than directly experienced from the external world.

The author explains that the brain operates in complete darkness and silence, relying on a barrage of electrical signals that are not inherently labelled. These signals are interpreted, leading to the perceptions we experience. He contrasts the common "bottom-up" view of perception, where sensory information flows into the brain to create our perception, with a "top-down" understanding, where the brain's predictions influence how we interpret sensory signals.

He critiques the traditional view of perception, which posits a direct relationship between external reality and our sensory experiences. Instead, Seth emphasises that perception is generated by the brain's interpretations, influenced by prior experiences and expectations. He cites historical figures such as Wittgenstein and Kant to illustrate the philosophical foundations of this idea.

The concept of "controlled hallucination" suggests that perceptual experiences are not merely reflections of reality but are influenced by the brain’s predictions about sensory inputs. Seth outlines three core components of this view: the brain’s predictive capabilities, the integration of sensory signals, and the subjective nature of perceptual experience.

He uses various examples, such as visual illusions and the phenomenon of "The Dress," to demonstrate how context and personal biology shape perceptions. He highlights how perceptual experiences differ between individuals, reinforcing the idea that perception is more about internal constructions than objective reality.

Exploring the perception of colour, Seth explains that our experience of colour is not just a direct response to external properties but is also a product of the brain's inferences regarding context and lighting conditions. He argues against the notion that colours possess intrinsic qualities.

Seth addresses potential objections to the controlled hallucination perspective, reassuring that it does not deny the existence of an objective world but rather highlights how our perceptions are constructed. He concludes by affirming the role of predictions in shaping our experience and the complex ways in which the brain processes new information.

Odds

Thomas Bayes, an 18th-century philosopher, is renowned for his theorem on optimal inference under uncertainty, known as Bayesian reasoning. This method emphasises making informed guesses based on prior knowledge and new evidence, playing a significant role in understanding consciousness and perception.

Bayesian inference represents abductive reasoning, distinct from deductive (logic-based) and inductive (observation-based) reasoning. Abductive reasoning seeks the best explanation for uncertainties and incomplete information. Bayes’ rule updates beliefs with new data, facilitating the transition from prior to posterior probabilities. This process highlights how our brain forms perceptions based on sensory input. The brain operates Bayesian-like, constantly updating its best guesses as new sensory information is received. 

Priors reflect initial beliefs before new evidence arrives, while likelihoods assess how likely new evidence supports each hypothesis. The posterior emerges when priors and likelihoods combine, leading to an updated belief about observed data. In practical scenarios, Bayesian reasoning can lead to errors, like misdiagnosing diseases based on rare test results. It has applications across various fields, including medicine and science, where hypotheses are adjusted based on new evidence, deviating from traditional perspectives on scientific inquiry.

The Beholder’s Share

The chapter opens with a historical overview of key figures from early twentieth-century Vienna, including philosophers from the Vienna Circle (like Kurt Gödel and Ludwig Wittgenstein) and artists like Gustav Klimt and Sigmund Freud. The environment was one that intermingled science and art, leading to rich explorations of human experience. 

The 'beholder’s share' is a concept highlighting the active role of the observer in completing a work of art. This idea was popularised by Ernst Gombrich, emphasising that perception involves top-down processes influenced by expectations and prior knowledge. An example is given through Impressionist art, where the observer’s interpretation completes the visual experience, allowing for an imaginative engagement with the artwork. 

The chapter then discusses a series of experiments that illustrate how perceptual expectations shape conscious experience. One experiment involving continuous flash suppression demonstrated that expected images are perceived more quickly than unexpected ones. Follow-up studies affirm the idea of a ‘word superiority effect’, where letters are easier to identify when they form part of a recognisable word.

The text transitions to real-world experiences and continues with a personal anecdote about taking LSD, illustrating the brain's role in creating perceptual experiences. The author reflects on how drug-induced hallucinations underscore the brain's constructive role in perception. The chapter explores pareidolia, the phenomenon of seeing patterns such as faces in clouds, tying it to the brain's predictive capabilities.

An innovative 'hallucination machine' project is described, which uses VR technology to simulate hallucinatory experiences by projecting perceptual predictions onto sensory input, allowing for exploration of human perception in a controlled environment. The concept of 'objecthood' highlights how visual experiences are structured, such as the perception of solid objects versus abstract visual phenomena. The chapter explores theories like sensorimotor contingency theory, which accounts for perception through our interactions and actions in the physical world.

The concept of time as a critical aspect of perception is discussed, arguing that our experience of time is also a controlled hallucination. The author presents research findings supporting that time perception emerges from the rate of change in sensory inputs rather than an internal clock.

A new project on 'substitutional reality' aims to explore how perceptions can be manipulated to determine what makes experiences seem 'real'. This involves creating VR setups where participants are unaware they are experiencing pre-recorded footage. The chapter ultimately argues that perception is a form of controlled hallucination, shaped by the brain's best guesses based on sensory inputs and prior experiences. It invites readers to consider the implications of this perspective for understanding both conscious experience and the nature of reality itself. 

Delirium

In the summer of 2014, the author's mother entered a vegetative state due to an undiagnosed encephalopathy while hospitalised for bowel cancer. Four years later, in 2018, she experienced hospital-induced delirium after being admitted for severe bowel pain. This deterioration of her mental state marked a significant contrast to her previous condition. Delirium is defined as an acutely disturbed state of mind, characterised by restlessness, illusions, and incoherence. Unlike dementia, which is chronic and degenerative, delirium is typically temporary, although it can last for weeks. The experience of delirium can lead to a significant detachment from reality, manifesting in hallucinations and paranoia.

Upon visiting, the author found his mother confused, angry, and unable to trust others, believing she was part of a malicious experiment. The symptoms intensified, aided by factors such as unfamiliar surroundings and multiple stressors related to her medical condition. The hospital's sterile and overwhelming environment contributes significantly to the onset of delirium. Patients often have their mental health overlooked because the focus remains on treating physical ailments. Delirium can lead to serious long-term outcomes such as reduced cognitive ability, increased mortality risk, and potential future episodes of delirium and dementia.

In an effort to reorient his mother, the author brought personal items from her home, hoping they would provide comfort and familiarity. The nature of her delusions suggested a twisted logic related to her circumstances and the author's role as both a loved one and a medical professional. As her acute confusion diminished, the author's mother still exhibited changes in her temperament and cognition. The author grappled with the shift in her identity as he reflected on the concept of the 'self' and its fragility in the face of illness.

Through the challenging experience of aiding his mother, the author explored themes of identity, reality, and the nature of the self, raising questions about what constitutes one's sense of being amidst health crises. The interplay between consciousness and reality becomes particularly pronounced when dealing with conditions like delirium. Individuals in such states often experience a distorted perception of their surroundings, leading to a fragmented sense of self. This phenomenon challenges our understanding of what it means to be truly aware and raises profound questions about identity during health crises

Expect Yourself

The self is not simply the entity that perceives. It is a controlled hallucination, shaped by evolution to ensure survival. The various aspects of selfhood – personal identity, embodiment, and existential experiences – arise from our brain making the best guesses. A hypothetical scenario involving teletransportation raises questions about the nature of self and personal identity. If a replica of a person is made, does it possess consciousness? More importantly, what defines personal identity. Is it the continuity of consciousness or psychological continuity?

A thought experiment features a woman named Eva teletransported. If another version of her exists while the original remains, which Eva is the "real" one? This raises questions about personal identity, indicating both aspects could be real.

The self encompasses multiple dimensions:

1. Embodied Self: Feelings of ownership and connection to one's body.

2. Perspectival Self: A first-person viewpoint of perception.

3. Volitional Self: The experience of intention and agency.

4. Narrative Self: The self's continuity through memories and future projections.

5. Social Self: How we perceive ourselves within a social context.

These elements interact, but when the narrative and social aspects erode, personal identity can suffer dramatically.

Studies like the rubber hand illusion illustrate that our sense of ownership can be manipulated, exposing the malleability of our self-perception. Virtual reality experiments further reveal how our first-person perspective can shift, leading to profound experiences like body swapping.

Despite constant changes, humans experience a stable sense of self, leading to the intuition that we are unchanging. This continuity lends us a sense of control and awareness, shaped by our social interactions and awareness of how others perceive us. Selfhood is complex and constructed from perceptions, not a fixed essence. Our identity shifts over time, driven by experiences and relationships, suggesting that understanding ourselves is vital for navigating our existence in an ever-changing world.

Being a Beast Machine

Self-perception is fundamentally about physiological regulation and survival rather than discovering external realities. The conception of selfhood has evolved extensively since the medieval notion of the "Great Chain of Being," which placed humans between the divine and the animalistic. This framework was disrupted by René Descartes, who distinguished between mind and matter, resulting in the view that animals were mere "beast machines" without consciousness.

While Descartes emphasised the uniqueness of human rationality and suggested that animals lacked consciousness, La Mettrie expanded this idea, asserting that humans are also machines but endowed with consciousness. This sparked debates on the relationship between mind and life, with contemporary discussions still fluctuating between continuity and discontinuity.

Interoception — the perception of internal physiological states — plays a crucial role in our emotions and moods. Emotions are perceived not as fixed entities but as reactions to bodily states, supporting William James and Carl Lange's views that emotions arise from our perception of bodily changes. Appraisal theories further expanded this understanding by positing that emotions are dependent on cognitive evaluations of external contexts.

Interoceptive inference suggests that our emotions and moods arise from predictions about the internal state of our bodies. The brain continuously makes best guesses about physiological conditions, functioning similarly to how it interprets sensory data from the external environment. This framework supports the idea that consciousness does not arise in isolation but is deeply intertwined with our physiological states.

The insights from cybernetics emphasise that perception is not just for discovery but also for regulation. Emotions serve a control-oriented purpose, responding to essential physiological variables necessary for survival. This regulation influences how we perceive experiences of selfhood and regulate our internal condition to maintain physiological viability.

We experience ourselves as stable entities despite ongoing physiological changes. This perceived stability stems from the brain's prior expectations, which create a sense of self that remains relatively unchanging over time. This ties into the profound connection between our embodied existence and conscious experiences.

The chapter posits that our consciousness arises from interoceptive processes, emphasising that we are "conscious beast machines." This perspective challenges the traditional notion of a distinct, immaterial self and highlights the essential relationship between life and consciousness, reiterating that our experiences, emotions, and sense of self are deeply rooted in our bodily conditions and evolutionary needs for survival.

A Fish in Water

In September 2007, the author was eager to attend a seminar by Karl Friston but ultimately missed it. Friston's 'free energy principle' (FEP) is a significant concept in neuroscience, integrating ideas from various fields such as biology, physics, and machine learning. Despite its complexity and initial confusion, its elegance offers deep insights into life and consciousness.

For something to exist, there must be a boundary that separates it from everything else. Living systems actively maintain these boundaries over time, distinguishing them from non-living entities. This persistence of identity over time is crucial for the definition of life, as it allows organisms to resist entropy, a measure of disorder.

Living beings exist in a state of low entropy, meaning their internal states are less disordered than in death. According to the second law of thermodynamics, entropy in isolated systems increases over time. However, living systems manage to resist this law by maintaining a separation from their environment and seeking out states that keep them alive.

A key aspect of the FEP is that living organisms, like a fish in water, exist in states they statistically expect to be in. This involves actively seeking combinations of physiological states, such as body temperature, heart rate, or nutrient concentrations, that are compatible with survival.

The Role of Free Energy

Free energy approximates sensory entropy, which cannot be directly measured. However, organisms can measure free energy and, thus, minimise it to maintain the low-entropy states necessary for life. This connects free energy to sensory prediction error. Organisms maintain their existence by minimizing prediction errors through interactions with their environment.

According to the FEP, organisms model their world and their bodies, repeatedly revisiting the low-entropy states that define them. The act of minimising free energy means that organisms use models to predict sensory inputs, enabling them to maintain coherence between sensory signals and anticipated states. Living systems cannot isolate themselves from their environments without increasing sensory entropy. Therefore, staying alive requires engagement with evolving surroundings, necessitating movement and interaction to avoid stagnation.

While the FEP is not falsifiable in the experimental sense, it serves a philosophical purpose, addressing the conditions for existence. The framework can motivate more specific theories that allow for empirical testing, such as predictive processing. FEP enhances our understanding of consciousness by grounding theories in biological functions while also allowing exploration of predictive perceptions. Ongoing discussions between FEP and Integrated Information Theory (IIT) aim to bridge understandings of consciousness, suggesting new avenues for research and enhancing explanations of how mechanisms give rise to minds.

Degrees of Freedom

The existence, mechanics, and implications of free will have long been contentious topics in philosophy and neuroscience. The idea that one has the ability to control their actions feels intuitive, but the reality behind these experiences is complex. Key to understanding this experience is the radical sensation in making choices, suggesting that the self performs a causal role in actions.

Seth critiques the notion of “spooky” free will, which implies an intervention that allows conscious thought to override deterministic processes. He argues that real free will does not disrupt the causal flow of physical events but exists within a deterministic framework, exploring the implication of this understanding on the debate surrounding determinism and free will.

He discusses Benjamin Libet's famous experiments that suggested brain activity precedes conscious intention. This led to interpretations that undermine free will. However, later interpretations of Libet's findings point to emergent patterns of brain activity rather than direct causation, further complicating the relationship between intention and action.

Seth posits that experiences of volition should be viewed as self-related perceptions rather than indicators of an immaterial self causing actions. He presents three critical features that characterise these experiences: the feeling of doing what one wants, the sense of alternative possibilities, and the perception of actions originating from within.

The concept of degrees of freedom is introduced, referring to the various ways a system can respond to stimuli. Humans possess extensive degrees of freedom in their voluntary behaviours, controlled by a network of processes within the brain. These processes specify actions based on individual beliefs, goals, and environmental perceptions.

Experiences of volition provide feedback that helps individuals navigate their environment and learn from previous actions. Such experiences contribute to one’s self-identity and the ability to adapt future behaviour according to past outcomes.

The discussion inevitably leads to the implications of free will on moral responsibility. Situations wherein an individual's capacity to exercise free will is compromised raise questions about culpability in legal and ethical contexts. Seth emphasises the relevance of understanding the intertwining of neuroscience with moral accountability.

Ultimately the chapter argues against viewing free will as an illusion. While “spooky” free will does not hold water, the perceptual experiences tied to volition are real and essential to navigating the complexities of human behaviour, identity, and interaction with the world. The exploration into free will is not only a psychological and philosophical inquiry but also a foundational aspect of what it means to be human. The intricacies of consciousness explained reveal how individuals perceive their own agency and volition. By examining how we experience our intentions and responses, we can gain deeper insights into what drives our decisions and actions. This understanding has profound implications for our interpretations of free will and moral responsibility.

Beyond Human

From the ninth century until the mid-1700s, European ecclesiastical courts occasionally held animals legally accountable, resulting in bizarre outcomes such as the execution of pigs. Documented cases indicate that pigs were frequent offenders, with crimes ranging from eating children to consuming sacred items. The medieval view of animals suggested they might have conscious experiences, contrary to the Cartesian perspective that denied them rational minds. This historical perspective anticipates modern conversations surrounding animal consciousness and the potential of ‘personhood’ extending beyond humans.

Consciousness shouldn't be judged by language use or advanced cognition. Absent high-level cognitive abilities does not imply lack of consciousness. Researchers caution against anthropomorphism and anthropocentrism when assessing animal awareness and highlight that intelligence is not synonymous with consciousness. Generally, it’s believed that all mammals are conscious due to anatomical similarities in their brains. A set of seventeen properties were identified that could indicate conscious experience across mammalian species. Noteworthy similarities in brain activity patterns during sleep suggest shared conscious levels among mammals.

The ability of an animal to recognise itself in mirrors has been explored, with only a few species of great apes, dolphins, and elephants passing the mirror test. This raises questions about the mechanisms underlying self-awareness and cognition across different animal species. Time spent with octopuses revealed an intelligence vastly different from mammals, suggesting they may possess a unique consciousness. Their decentralised nervous systems challenge traditional interpretations of what it means to be conscious, showing behaviors that imply awareness, problem-solving, and adaptation.

Birds, particularly parrots and other intelligent species, show strong evidence of consciousness through complex behaviours. Their brain structure, while different from mammals, indicates possibilities for conscious experiences. The exploration beyond mammals calls into question how consciousness may manifest in many forms across species. It is important to recognise that consciousness may exist in forms unrecognisable to us and all life experiences variables of consciousness. Investigations into animal consciousness prompt reflections on our own experience and predicate ethical considerations regarding the treatment of animals.

Seth posits that consciousness can exist independently of intelligence. They can suffer without possessing high cognitive function. The implications of this carve a pathway toward considering artificial intelligence, raising questions about whether machines could achieve consciousness akin to living beings.

Machine Minds

In Prague, Rabbi Judah Loew created a golem named Josef from clay to protect his community. The story illustrates the dangers of creating sentient beings that can turn against their creators, as seen in various cultural references from Frankenstein to AI-themed movies. The advent of AI raises questions about machine consciousness and the implications of creating intelligent systems. Concerns include the potential for AI to displace jobs and disrupt society, along with existential fears about its capabilities.

People assume that consciousness will arise once machines reach a certain level of intelligence. This belief rests on two flawed assumptions: functionalism (the idea that consciousness is independent of the underlying material) and the link between intelligence and consciousness. Functionalism suggests that any system that processes information in the right way can be conscious, regardless of its material. However, this view is largely unproven, as consciousness may not be as straightforward as information processing alone.

Consciousness is not necessarily linked to intelligence. Current AI exhibits high capabilities in pattern recognition but lacks true consciousness. The relationship between consciousness and intelligence must be understood as separate and multidimensional. The possibility of designing machines that could be conscious involves understanding what consciousness requires, which varies with different theories of consciousness. Some propositions suggest consciousness may arise from integrated information processing, while others propose a necessity of being life-like.

Different theories, like Integrated Information Theory (IIT) and the Beast-Machine theory, examine the conditions under which consciousness might emerge in machines. The Beast-Machine theory suggests that consciousness is tied to biological drives and life processes, which current machines lack.

As technologies advance, ethical concerns arise regarding AI and machine consciousness. These discussions remain critical, especially if machines appear conscious but lack true awareness. Ethical considerations must include the potential psychological and societal impacts of interacting with seemingly conscious entities. There are fears of AI reaching a "Singularity" where it surpasses human intelligence. The potential merger of biological and technological consciousness raises further ethical dilemmas. Understanding and addressing the implications of creating conscious machines is paramount.

The allure of machine consciousness reflects a desire to transcend human limitations. However, maintaining a connection to our biological origins is essential as we develop new technologies. The exploration of consciousness should progress thoughtfully, respecting the potential impact on life and society.

Themes

Consciousness

"Consciousness is the ongoing controlled hallucination your brain generates to predict and make sense of sensory input." Anil Seth

Seth emphasises that perception is not a passive readout of the world but an active process of prediction and Bayesian inference: the brain continuously generates hypotheses about incoming sensory signals and updates them when prediction errors occur. This framework reframes what we experience as perception as the brain’s best-guess constructions rather than direct mirrors of external reality.

Seth describes the brain as a prediction machine whose models of the world and the body are continually refined. Conscious experience arises from this inferential process. Because these predictions incorporate both external (exteroceptive) and internal (interoceptive) signals, consciousness is shaped by bodily states as well as sensory input, linking perception tightly to physiology.

"Consciousness is a process, not a thing." 

Seth treats consciousness as graded and dynamic, varying in level and content according to the brain’s predictive dynamics and sensorimotor engagement. By focusing on mechanisms that minimise prediction error and maintain coherent models of body and world, his approach aims to bridge subjective phenomenology and neurocomputational explanations without treating qualia as metaphysical primitives..