TLDR: The Modular Consciousness Theory (MCT) proposes that subjective consciousness arises from discrete ‘Integrated Informational States’ (IISs), each tagged with an ‘information-density vector’ that quantifies its subjective intensity and influences memory and behavior. This theory outlines a modular architecture (sensorimotor, subconscious, conscious modules) and an evolutionary path for consciousness, from basic reactive systems to advanced introspective ones. MCT provides a computational pipeline for how conscious experience is generated, distinguishing it from other theories, and offers testable predictions for both biological and artificial systems, as well as a framework for understanding mental disorders as specific modular dysfunctions.
A new theoretical framework, the Modular Consciousness Theory (MCT), offers a fresh perspective on how subjective experience arises from information processing in both biological and artificial minds. This theory, proposed by Michaël Gillon, aims to bridge the gap between neuroscience, cognitive science, and AI research by providing a biologically grounded and computationally explicit model of consciousness.
At its core, MCT posits that consciousness is not a continuous, ineffable flow, but rather a discrete sequence of ‘Integrated Informational States’ (IISs). Imagine these IISs as individual packets of integrated information. Each packet is tagged with a unique, multidimensional ‘information-density vector’ that quantifies its informational richness. When this vector is absent, the system remains unconscious. When present, its magnitude directly correlates with the intensity of subjective experience – the higher the information density, the more profoundly an informational packet influences memory, actions, and the overall continuity of experience.
The Modular Architecture of Consciousness
MCT describes cognition through a distributed architecture comprising specialized information-processing modules. These modules are defined by their functional roles, not necessarily by fixed brain regions. The theory categorizes them into three main types:
- Sensorimotor Modules: These govern perception, motor execution, and physiological regulation, operating outside conscious awareness but providing crucial inputs.
- Subconscious Modules: Responsible for reactive, associative, or automated outputs without subjective awareness, including memory storage, behavioral readiness, and decision-making.
- Conscious Modules: These directly contribute to constructing the IIS. They include modules for abstraction, narration, evaluation, and self-evaluation, forming the core content of conscious experience.
The theory outlines an evolutionary trajectory for consciousness, starting from simple unconscious architectures that support reflexive behavior and basic learning without IISs. It then progresses to a ‘minimal conscious architecture’ where IISs emerge through information filtering, narration, abstraction, and integration, enabling flexible learning and rudimentary planning. The ‘advanced conscious architecture,’ characteristic of humans, further incorporates self-evaluation, evaluation (for understanding others’ minds), and an ‘existential submodule’ for symbolic concepts, alongside a ‘linguistic loop’ for recursive thought and inner speech. This advanced stage supports introspection, moral reasoning, and complex social cognition.
The Conscious Cycle: From Input to Subjective Experience
Each conscious cycle begins with the ‘filter module’ selecting relevant signals from the subconscious stream based on their salience. These selected signals are then simultaneously broadcast to the conscious modules (abstraction, narration, evaluation, self-evaluation), which process them in parallel. Their co-modulated outputs are then integrated by the ‘integration module’ into a unified IIS. This IIS is immediately analyzed, and its informational density is measured along various dimensions (coherence, relevance, emotional salience), resulting in the multidimensional information-density vector. This vector is embedded within the IIS and dictates its priority for memory storage, its influence on behavioral readiness, and its role in decision-making.
This recurrent loop—from salience-based selection to parallel processing, integration, information-density tagging, and transmission—forms the stream of consciousness. The phenomenal continuity we experience arises from the rapid succession of these discrete, tagged informational states.
Distinguishing MCT from Other Theories
MCT builds upon and extends existing theories of consciousness. While Global Workspace Theory (GWT) emphasizes global information distribution, MCT specifies that broadcasting merely initiates the process, with consciousness emerging from the construction of a structured, information-density-tagged IIS. Integrated Information Theory (IIT) quantifies consciousness with Φ, but MCT reinterprets Φ as an indicator of the brain’s capacity to generate rich IISs, emphasizing that the *evaluation* of the IIS (the information-density vector) is what truly governs memory and behavior, not just integration alone. Similarly, MCT provides a detailed modular architecture that goes beyond the reflexive awareness of Higher-Order Theories (HOT) and integrates predictive coding mechanisms within its abstraction module, showing how these inferences become subjectively experienced and actionable.
Also Read:
- Integrating Causal Knowledge: Introducing the Causal Abstraction Network
- Solving AI Agent Forgetfulness: A Hybrid Memory System for Long-Running Low-Code Tasks
Implications for Artificial Minds and Clinical Understanding
A significant strength of MCT is its direct applicability to artificial intelligence. Its modular, computationally explicit design provides a blueprint for engineering synthetic systems capable of contextual modeling, coherent memory formation, and flexible action. An MCT-based AI could preferentially store highly relevant IISs, modify behavior based on internal salience, and generate linguistic outputs reflecting internal evaluation.
The theory also offers a powerful framework for understanding psychiatric and neurological disorders. Conditions like psychopathy, malignant covert narcissism, severe dissociative states, and bipolar disorder can be reinterpreted as specific failures in modular dynamics—such as deficient output generation, faulty integration, or aberrant modulation of salience. For instance, psychopathy might involve intact cognitive modules but dysfunctional evaluative circuits, leading to IISs without inhibitory moral weighting. Dissociation, on the other hand, could be a breakdown in generating stable, subjectively tagged IISs due to modular desynchronization.
In conclusion, MCT reframes consciousness not as an ineffable essence, but as an evolvable, quantifiable, and constructible feature of complex information processing. By proposing that the intensity of subjective experience correlates with the amplitude of an internal information-density vector, MCT offers a testable and implementation-ready framework for both biological and artificial systems. For more details, you can read the full research paper here.
