Mindscape Collective is now The Consciousness Library. Same library, new name. You may need to sign in again. About the change
Skip to content

Karl Friston

45 papers in the library · 4,199 citations · publishing 1995-2026

Papers

How do inner screens enable imaginative experience? Applying the free-energy principle directly to the study of conscious experience.

Neuroscience of Consciousness January 1, 2025 Chris Fields, Mahault Albarracin, Karl Friston et al. 6 citations

The free-energy principle (FEP) constrains possible models of consciousness, especially those of attentional control and imaginative experiences like episodic memory and planning. The paper first reviews classical and quantum formulations of the FEP, focusing on multi-component systems where only some parts interact directly with the environment. It discusses internal boundaries structured as Markov blankets, which act as classical information channels. The analysis shows how this framework supports models of attention and imagination, explaining how imaginative experience can use the same spatio-temporal and object-recognition frames as ordinary perception, and how it can be internally generated yet still surprising. The paper concludes with implications for implementation, phenomenology, phylogeny, and the large variability of imaginative experience in humans.

On the minimal theory of consciousness implicit in active inference.

Physics of Life Reviews March 1, 2026 Christopher J. Whyte, Andrew W. Corcoran, Jonathan Robinson et al. 5 citations

Subjective experience is multifaceted, making it hard for traditional neuroscientific theories of consciousness to be compared because each focuses on different aspects like perceptual awareness or global states. This work instead starts from active inference, a first-principles framework that models behavior as approximate Bayesian inference, and builds a minimal theory of consciousness from shared features of computational models derived under active inference. By reviewing studies that apply active inference models to consciousness, the authors identify a small set of theoretical commitments implicit in these models, pointing toward a minimal and testable theory of consciousness.

Pathfinding: a neurodynamical account of intuition.

Communications Biology August 13, 2025 Steven Kotler, Michael Mannino, Karl Friston et al. 2 citations

Intuition, often inconsistently defined, is reframed as an evolutionarily grounded pathfinding mechanism that emerges from the brain's optimization of its relationship with the environment. A review of empirical findings identifies relevant brain networks and links intuition to cognitive states like insight. Unsolved problems dynamically alter attractor landscapes, guiding future intuitions. The concept of 'opportunistic assimilation' is explored through nonlinear neurodynamics, and hippocampal sharp wave ripples are identified as potential neural correlates of intuition, given their role in creativity, choice, action planning, and abstract thinking. Two complementary frameworks—the free energy principle and metastable coordination dynamics—together provide a comprehensive neurodynamical account of intuition's neurophenomenology.

The role of active inference in conscious awareness

PLoS One December 4, 2025 Jonathan Robinson, Andrew W. Corcoran, Christopher J. Whyte et al. 1 citation

Active inference, a framework for modeling how sentient agents behave, is being tested as necessary for changes in conscious content. In an adversarial collaboration, active inference will be contrasted with two other theories that do not require it for consciousness. This study protocol describes an adaptation of the motion-induced blindness paradigm: an active condition where participants direct their gaze toward a target after it disappears from consciousness and report its reappearance, versus a passive condition where participants fixate centrally while the stimulus array moves in a replay of active eye-tracking data. Two experiments will compare target reappearance across conditions to evaluate active inference's contribution to conscious awareness.

Integrated information and predictive processing theories of consciousness: An adversarial collaborative review.

Neuroscience and Biobehavioral Reviews August 1, 2026 Andrew W. Corcoran, Andrew M. Haun, Reinder Dorman et al.

Three theories of consciousness—Integrated Information Theory, Neurorepresentationalism, and Active Inference—are compared and contrasted in a structured adversarial collaboration. The review presents each theory's core claims, the phenomena they explain, their explanatory approaches, and methodological strategies. It outlines key hypotheses to be tested across multi-site experiments, discusses observations that would support or challenge each theory, and describes how data from disparate experiments can be formally integrated to quantify evidential support. The work also provides meta-scientific insights into the mechanics of adversarial collaboration and theory-testing, including how theories may be evaluated by the scientific progress they deliver.

The Complex Brain Hypothesis: Resolving the Entropy-Content Conundrum in Minimal Phenomenal Experience

arXiv (Cornell University) May 15, 2026 Jonas Mago, Edmundo Lopez-Sola, Jakub Vohryzek et al.

States of consciousness with minimal phenomenal content, such as those induced by certain meditation practices, show increased brain entropy similar to high-content psychedelic states, challenging the Entropic Brain Hypothesis that links entropy to phenomenal richness. The Complex Brain Hypothesis resolves this by proposing that brain complexity, not entropy, better indexes the richness of experience. Complexity is modulated by the grain of inference the brain uses to resolve uncertainty: fine-grained inference loosens constraints and proliferates content, as in psychedelic states; coarse-grained inference simplifies experience into contentless awareness, as in minimal phenomenal experiences. Both regimes can elevate entropy but differ in phenomenology and perturbational signatures, refining the Entropic Brain Hypothesis and highlighting minimal phenomenal experiences as a test case for computational theories of consciousness.

Protocol for investigating the warping of spatial experience across the blind spot to contrast predictions of the Integrated Information Theory and Predictive Processing accounts of consciousness.

PLoS One January 1, 2026 Clement Abbatecola, Bernard Marius ’t Hart, Belén M. Montabes De la Cruz et al.

The subjective experience of space around the visual blind spot is investigated to test three theories of consciousness: Integrated Information Theory (IIT), Predictive Processing Active Inference (AI), and Predictive Processing Neurorepresentationalism (NREP). IIT predicts that the blind spot region, lacking feedforward input from one eye, should contribute differently to perceived spatial quality. The Predictive Processing accounts argue that internal models accommodate structural deviations based on sensory evidence. Participants evaluate distances, areas, and illusory motion with or without the blind spot involved. Psychometric models quantify bias and precision in perceived versus objective space. Simulated results correspond to each theory's predictions, and challenges for dissemination are discussed.

A beautiful loop: An active inference theory of consciousness

June 16, 2025 Ruben Laukkonen, Karl Friston, Shamil Chandaria preprint

A theory proposes that active inference can model consciousness through three conditions: a generative world model (epistemic field) that defines what can be known; inferential competition where only coherent uncertainty-reducing inferences enter the model (Bayesian binding); and epistemic depth, a recursive sharing of beliefs such that the world model knows it exists non-locally. This self-knowing is distinct from self-consciousness. The theory introduces a hyper-model for precision-control across hierarchical inference layers, termed the Beautiful Loop Theory. It offers insights into meditation, psychedelic states, minimal phenomenal experience, and suggests a path toward conscious artificial intelligence.

A Rosetta Stone hypothesis for neurophenomenology: mathematical predictions from predictive processing.

Neuroscience of Consciousness January 1, 2026 Lancelot da Costa, Anil K. Seth, Karl Friston et al.

A Rosetta Stone hypothesis from predictive processing proposes that beliefs serve as a central hub linking phenomenology, behavior, and neural dynamics. Under the assumption that phenomenology is a function of beliefs, the authors derive mathematical predictions for subjective similarity judgments, cognitive metabolic cost, subjective cognitive effort, and time perception. They review the connection between beliefs and neural dynamics to complete a generative passage for neurophenomenology, omitting the belief–behavior link as already documented. Testing these predictions will inform the validity of the central assumption and advance the neurophenomenology research program.

Integrated information and predictive processing theories of consciousness: An adversarial collaborative review

arXiv Preprint Archive August 30, 2025 Andrew W. Corcoran, Andrew M. Haun, Reinder Dorman et al.

Three major theories of consciousness—Integrated Information Theory, Neurorepresentationalism, and Active Inference—are compared and contrasted in the context of a structured adversarial collaboration designed to test their competing predictions. The review presents each theory's core claims, the phenomena they explain, their explanatory styles, and methodological strategies. It outlines key hypotheses to be tested across multi-site experiments, discusses observations that would support or challenge each theory, and describes how data from disparate experiments can be formally integrated to provide a quantitative measure of evidential support. The work also offers meta-scientific insights into adversarial collaboration and theory-testing.

Testing the role of spontaneous activity in visuospatial perception with patterned optogenetics.

PLoS One January 1, 2025 Kengo Takahashi, Samuel Pontes Quero, Julien Fiorilli et al.

A planned experiment will test two competing theories of consciousness: one holding that neuronal activity itself generates conscious experience, the other that the causal structure of neural circuits is responsible. The experiment will optogenetically inactivate a portion of mouse visual cortex that does not respond to behaviorally relevant stimuli, targeting neurons with very low spiking activity that does not significantly affect other cortical areas. The effect on perception of the spatial distribution of stimuli will be measured. Results will be compared against theoretical predictions to help determine the neuronal substrate of consciousness.

On the Minimal Theory of Consciousness Implicit in Active Inference

arXiv Preprint Archive October 9, 2024 Christopher J. Whyte, Andrew W. Corcoran, Jonathan Robinson et al.

Subjective experience is multifaceted, making consciousness hard to study because traditional theories often focus on isolated aspects like perception or wakefulness and are difficult to compare. This work starts from active inference—a first-principles framework that models behavior as approximate Bayesian inference—and builds toward a minimal theory of consciousness derived from shared features of computational models under active inference. Reviewing models applied to consciousness, the authors argue that these models imply a small set of theoretical commitments pointing to a minimal, testable theory of consciousness.

A Rosetta Stone Hypothesis for Neurophenomenology: Mathematical Predictions from Predictive Processing

arXiv Preprint Archive September 30, 2024 Lancelot da Costa, Anil K. Seth, Karl Friston et al.

A Rosetta Stone hypothesis from predictive processing proposes that beliefs serve as a central hub linking phenomenology, behavior, and neural dynamics. If phenomenology is a function of beliefs, then specific predictions follow for subjective similarity judgments, cognitive metabolic cost, subjective cognitive effort, and time perception. The connection between beliefs and neural dynamics completes the generative passage for neurophenomenology, while the belief-behavior link is already well-documented. Testing these predictions will inform the validity of the central assumption and advance the neurophenomenology research program.

Extended active inference: Constructing predictive cognition beyond skulls.

Mind & Language June 1, 2022 Axel Constant, Andy Clark, Michael D Kirchhoff et al.

Cognitive niche construction—the process by which organisms build and maintain cause-effect models of their environment to guide adaptive behavior—can be understood through active inference, a framework in which organisms embody predictive models of the world optimized by perception, action, and learning. This paper proposes that cognitive niche construction itself is a cognitive function that optimizes these generative models, a process termed extended active inference. The account integrates ideas from extended mind theory, which holds that cognitive processes extend beyond the brain into the world, and predictive processing, offering a unified perspective on how organisms shape their niches to improve prediction and adaptive fitness.

Brief Mindfulness Meditation Induces Gray Matter Changes in a Brain Hub

Journal of Neural Transplantation and Plasticity November 16, 2020 R. Tang, Karl Friston, Yi-Yuan Tang

Brief mindfulness meditation—ten hours total over two to four weeks—increases gray matter volume in the ventral posterior cingulate cortex (PCC), a brain region central to self-awareness, emotion, cognition, and aging. A randomized controlled trial compared Integrative Body-Mind Training (IBMT) with relaxation training in healthy adults. The increase in ventral PCC volume was smaller in individuals with higher baseline temperamental trait scores, indicating that individual differences modulate training-induced brain plasticity. These structural changes suggest that even short-term mindfulness practice may help protect against mood disorders and age-related cognitive decline.

Anticipation in architectural experience: a computational neurophenomenology for architecture?

arXiv Preprint Archive November 7, 2020 Zakaria Djebbara, Thomas Parr, Karl Friston

Perceptual experience of architecture arises from interactions between the body's sensory and motor systems and the built environment. Actions change perceived surroundings based on expectations shaped by bodily capabilities and architectural features. Affordances—the fit between body structure and movement possibilities in a space—underlie continuous sensory information gathering. This paper takes a first step toward understanding architectural design's role in perceptual experience at a neuronal level, proposing a framework that synthesizes computational neuroscience with architectural phenomenology into a computational neurophenomenology. The framework aims to guide future studies linking architecture and cognitive neuroscience.

Am I Self-Conscious? (Or Does Self-Organization Entail Self-Consciousness?)

Frontiers in Psychology April 24, 2018 Karl Friston

Self-consciousness is necessary for consciousness. Living systems return to attracting states, which mathematically implies a Lyapunov function that is model evidence (self-evidence) in Bayesian statistics or surprise (self-information) in information theory. All biological processes perform inference, from evolution to conscious processing. Consciousness arises when self-evidencing has temporal thickness or counterfactual depth, grounding inferences about the consequences of one's actions. Consciousness is inference about one's future—the self-evidencing consequences of what one could do.

A mathematical model of embodied consciousness.

Journal of theoretical biology September 7, 2017 David Rudrauf, Daniel Bennequin, Isabela Granic et al.

Consciousness is modeled as a spatial field structured by projective geometry and controlled by active inference. The Projective Consciousness Model (PCM) combines sensory evidence with prior beliefs, selecting points of view and perspectives according to preferences. Violations of expectation are encoded as free energy, and minimizing free energy drives perspective taking, switching between perception, imagination, and action. Consciousness functions as an algorithm maximizing resilience by using projective perspective taking to escape local free energy minima. The model accounts for spatial phenomenology, distinctions between perception, imagination, and action, affective processes in intentionality, bistable figure dynamics, and body swap illusions. It relates phenomenology to function, suggesting computational advantages of consciousness and generating neurophenomenological hypotheses.

The neural correlates of consciousness under the free energy principle: From computational correlates to computational explanation

Wanja Wiese, Karl Friston preprint

The free energy principle reframes the search for neural correlates of consciousness by shifting focus from neural states to neural dynamics, and by adding computational correlates defined as probabilities encoded by neural states. This approach addresses two problems: explaining consciousness without sensory input or behavior, and distinguishing systems that genuinely are conscious from those that merely simulate consciousness. The principle-derived constraints help ascribe consciousness in controversial cases, such as during sensory deprivation, and clarify what it means to be a conscious system versus simulating one.