PLoS One
May 13, 2015
Melanie Boly, Shuntaro Sasai, Olivia Gosseries et al.
69 citations
A meaningful sequence of stimuli, like movie frames, triggers varied experiences, while meaningless stimuli, like TV noise, produce a uniform experience despite physical differences. The differentiation of cortical responses, measured by Lempel-Ziv complexity of functional MRI images, reflects the overall meaningfulness of a stimulus set rather than differences among stimuli. Brain activity patterns showed highest differentiation during a movie, intermediate for a temporally scrambled movie, and minimal for spatially scrambled TV noise, even though overall cortical activation was strong and widespread in all conditions. Meaningfulness also correlated with higher information integration among cortical regions. This approach can assess stimulus meaningfulness without identifying relevant features or neural locations.
Philosophy and the Mind Sciences
December 23, 2025
Tadahiro Taniguchi, Masafumi Oizumi, Noburo Saji et al.
1 citation
Language emergence and the relational structure of subjective experiences, termed qualia structure, may influence each other bidirectionally. The emergence of languages with distributional semantics, such as syntactic-semantic structures, is hypothesized to link to the coordination of internal representations shaped by experience, potentially facilitating more structured language through reciprocal influence. This mutual dependency is explored through theoretical frameworks like collective predictive coding. Computational studies show that neural network-based language models form systematically structured internal representations, and multimodal language models can share representations between language and perceptual information. Language emergence may serve not only as a communication tool but also as a mechanism for shared understanding of qualitative experiences. The paper outlines future constructive research directions in consciousness studies, linguistics, and cognitive science.
iScience
March 21, 2025
Genji Kawakita, Ariel Zeleznikow-Johnston, Ken Takeda et al.
A new method based on optimal transport can align the similarity structures of color experiences between people without assuming which colors correspond. After collecting subjective similarity judgments for 93 colors, the method correctly aligned the color qualia structures of color-neurotypical participants at the group level, but could not align the structures of color-blind participants with those of color-neurotypical participants. This provides quantitative evidence that color-neurotypical people's experience of "red" is relationally equivalent to other color-neurotypical people's "red", but not to color-blind people's "red". The method can be applied across sensory modalities.
Neuroscience of Consciousness
January 1, 2025
Elise G Rowe, Ken Takeda, Masafumi Oizumi et al.
Top-down attention affects not just whether we see something, but how we see it—the quality of conscious experience, or qualia. Using a dual-task paradigm, participants rated similarity of stimulus pairs in the periphery for rotated letters, bisected disks, and greyscale faces. Similarity ratings served as a proxy for qualia structures, analyzed with dimension reduction and optimal transport alignment. Attention withdrawal collapsed qualia structures for letters and disks, but alignment accuracy remained high for face qualia structures under both full and poor attention. This relational judgment approach extends previous binary categorization methods and offers a novel way to study qualia structures.
bioRxiv Preprint Server
July 12, 2021
Jun Kitazono, Yuma Aoki, Masafumi Oizumi
preprint
A method for hierarchically decomposing a brain network into cores based on the strength of bidirectional connections helps identify regions likely essential for consciousness. Applied to a whole-brain mouse connectome, cores with strong bidirectional connections included the isocortex, thalamus, and claustrum—areas thought to support consciousness—and excluded the cerebellum, which is not considered relevant. Simpler methods that ignore bidirectionality failed to show this correspondence. The findings suggest that analyzing bidirectional connectivity offers a novel way to relate brain network structure to consciousness.
Entropy (Basel, Switzerland)
March 6, 2018
Jun Kitazono, Ryota Kanai, Masafumi Oizumi
Integrated Information Theory (IIT) links the amount of integrated information (Φ) in the brain to the level of consciousness, proposing that Φ should be measured across the partition of a system where information loss from partitioning is minimized—the Minimum Information Partition (MIP). Exhaustively searching for the MIP is computationally infeasible for large systems. Previous work showed that if a measure of Φ is submodular, an optimization algorithm can find the MIP in polynomial time, but later versions of Φ are not submodular. This study empirically tested the algorithm on non-submodular Φ measures using simulated and real neural data, finding it identifies the MIP nearly perfectly, enabling practical Φ measurement in large systems.
arXiv Preprint Archive
December 19, 2017
Jun Kitazono, Ryota Kanai, Masafumi Oizumi
Integrated Information Theory (IIT) links consciousness to the amount of integrated information (Φ) in the brain, measured across a system's Minimum Information Partition (MIP). Finding the MIP is computationally expensive for large systems. Previous work showed that if Φ satisfies submodularity, an optimization algorithm can find the MIP in polynomial time, but later versions of Φ are not submodular. This study tests the algorithm on non-submodular Φ measures using simulated and real neural data. The algorithm identifies the MIP with near-perfect accuracy even for non-submodular measures, enabling practical measurement of Φ in large systems.
eNeuro
January 1, 2017
Andrew M. Haun, Masafumi Oizumi, Christopher K Kovach et al.
A pattern of integrated information—a measure inspired by integrated information theory—corresponds to what a person consciously sees, whereas broader information measures such as mutual information and entropy do not. Intracranial recordings from six neurosurgical patients showed that, in object-sensitive brain areas, the hierarchical causal structure of neural interactions matched the subjects' conscious percepts when they viewed faces or objects under conditions that dissociate perception from the physical stimulus (continuous flash suppression and backward masking). Unsupervised classification confirmed that integrated information patterns, but not other information measures, clustered according to the subjects' visual experiences. The findings suggest that locally integrated information plays a key role in the neural basis of conscious object perception.
PLoS Computational Biology
June 10, 2016
Masafumi Oizumi, Shun-ichi Amari, Toru Yanagawa et al.
A novel practical measure called Φ* quantifies integrated information in the brain, a property predicted by Integrated Information Theory (IIT) to reflect levels of consciousness. Earlier measures failed to satisfy theoretical lower and upper bounds: zero when no information is generated or when parts are independent, and the total information generated by the whole system. By applying mismatched decoding from information theory, Φ* meets these bounds. Under a Gaussian assumption, Φ* has an analytical expression applicable to experimental neural data. Φ* can serve as a measure of integrated information in consciousness research and as a tool for network analysis in biology.
Proceedings of the National Academy of Sciences of the United States of America
October 15, 2015
Masafumi Oizumi, Naotsugu Tsuchiya, S. Amari
A unified framework based on information geometry quantifies multiple causal influences among elements in a system. The proposed measure, geometrical integrated information, projects the system's probability distribution onto a constrained manifold to avoid overestimation and noncausal confounding that plague part-based approaches. It provides intuitive geometric interpretations harmonizing mutual information, transfer entropy, stochastic interaction, and integrated information, each characterized by how causal influences are disconnected. Inspired by consciousness studies of neural integration, the framework should have general utility for analyzing causal relationships in complex systems across physics and biology.
arXiv Preprint Archive
October 15, 2015
Masafumi Oizumi, Naotsugu Tsuchiya, Shun-ichi Amari
A unified theoretical framework based on information geometry is proposed for quantifying spatio-temporal interactions in stochastic dynamical systems. The degree of interaction is measured by the divergence between the actual probability distribution and a constrained distribution where the interactions of interest are removed. This approach provides novel geometric interpretations of mutual information, transfer entropy, and stochastic interaction, clarifying their relationships. Extending transfer entropy, the authors introduce a new measure of integrated information that quantifies causal interactions between parts of a system, capturing the extent to which the whole exceeds the sum of its parts. This measure is suggested as a potential biological indicator of consciousness levels.
arXiv Preprint Archive
May 17, 2015
Masafumi Oizumi, Shun-ichi Amari, Toru Yanagawa et al.
Integrated Information Theory (IIT) proposes that a brain's capacity to integrate information is essential for consciousness and that a measure called integrated information, Φ, should reflect levels of consciousness. Practical application has been hindered because existing measures fail to satisfy theoretical lower and upper bounds: zero when the system generates no information or consists of independent parts, and equal to the whole system's information when its parts generate no information independently. The authors derive a new practical measure, Φ*, using mismatched decoding from information theory. This measure satisfies the required bounds. They also provide an analytical expression for Φ* under a Gaussian assumption, making it applicable to experimental neural data.
PLoS Comput. Biol.
May 1, 2014
Masafumi Oizumi, L. Albantakis, G. Tononi
Integrated Information Theory (IIT) 3.0 proposes that consciousness is identical to a maximally irreducible conceptual structure (MICS) generated by a physical system. Starting from phenomenological axioms—information (each experience is specific), integration (each experience is unified), and exclusion (each experience has unique borders and grain)—IIT formalizes postulates for how mechanisms like neurons must be configured to produce experience. The theory defines intrinsic information as differences that make a difference within a system and integrated information as information of a whole irreducible to its parts. A MICS specifies the quality of an experience, and integrated information (ΦMax) its quantity. The theory entails that consciousness is always about the system itself, that inactive elements can generate experience, and that feed-forward systems can be unconscious zombies functionally equivalent to conscious complexes.
Naotsugu Tsuchiya, Masafumi Oizumi, Makiko Yamada et al.
preprint
The mystery of consciousness—how subjective experience arises from physical and chemical brain processes—is one of modern science's greatest challenges. Approaches that define humans and animals solely by observable behavior cannot scientifically address the conscious experiences of patients with hallucinations or delusions, nor those of non-verbal patients, infants, or animals. Recently, researchers confronting consciousness directly have advanced empirical studies of the brain-consciousness relationship. This article reviews several promising frameworks for consciousness research and introduces the authors' own strategy: clarifying the structure of conscious experience, the structure of information generated by the brain, and the relationships between these structures. The mathematical tool of category theory is briefly introduced as a means to achieve this, along with future prospects.