Chapter 10 Contemporary Theories of Consciousness
10.1 Chapter Overview
Contemporary consciousness science does not lack theories. It has many. Some theories locate consciousness in global broadcasting across the brain. Others define it as integrated information, predictive inference, attention modeling, recurrent neural processing, or even quantum-level physical events. Each theory attempts to explain why some information becomes conscious while other information remains unconscious.
This chapter surveys several leading scientific theories of consciousness: Global Workspace Theory, Integrated Information Theory, predictive processing and the Free Energy Principle, Attention Schema Theory, Recurrent Processing Theory, and Orchestrated Objective Reduction. The goal is not to choose a final winner. Rather, the aim is to understand what each theory claims, what mechanisms it emphasizes, what predictions it makes, and how it changes the relationship between life and mind.
The theories differ sharply. Some imply that consciousness requires specific brain architecture. Others suggest that consciousness may be graded and more widespread. Some are strongly biological; others are more computational, informational, or physical. These differences matter for the central question of this book. If consciousness requires a global neuronal workspace, then life comes first and consciousness appears only in certain nervous systems. If consciousness is integrated information, then consciousness may be more continuous with life, matter, or self-organization. If consciousness is linked to predictive regulation, then the boundary between life and mind may become less sharp.
10.2 Global Workspace Theory
Global Workspace Theory, originally associated with Bernard Baars, describes consciousness as a global broadcasting system. The basic idea is that the brain contains many specialized processes operating in parallel. Most of these processes remain unconscious. A perception, memory, goal, or thought becomes conscious when it gains access to a global workspace and is broadcast widely across the system.
Baars used a theatre metaphor. Consciousness is like a spotlight shining on a stage. Many processes operate backstage, but only some contents enter the spotlight. Once on the stage, the content becomes available to many other systems: memory, language, decision-making, planning, emotion, and action.
This metaphor is not meant to suggest that there is a little observer inside the brain watching the stage. Rather, it suggests that consciousness involves availability. A conscious content is not locked inside one local process. It becomes globally available for flexible use.
The Neural Global Workspace model, developed by Stanislas Dehaene, Jean-Pierre Changeux, and others, gives this idea a neurobiological form. According to this model, conscious access depends on large-scale networks, especially involving prefrontal and parietal regions, connected with sensory systems. When a stimulus becomes conscious, neural activity does not remain local and brief. It undergoes “ignition”: a sudden, sustained, widespread pattern of activity that makes the information available across the brain.
Experimental work on masking, attentional blink, and binocular rivalry has been important for testing this theory. In visual masking, a stimulus may be processed unconsciously if it is quickly followed by another stimulus that prevents conscious access. In attentional blink, a second target may fail to reach awareness if it appears too soon after a first target. In binocular rivalry, different images presented to each eye compete for conscious perception. These paradigms allow researchers to compare conscious and unconscious processing while controlling the physical stimulus.
Global Workspace Theory is attractive because it explains access consciousness well. It explains why conscious contents are reportable, flexible, memorable, and available for reasoning. It also provides testable predictions about brain activity: conscious access should involve widespread broadcasting, sustained activation, and coordination across distant neural systems.
However, the theory faces limitations. It may explain how information becomes available to the system, but does it explain why that information is experienced? Global broadcasting may account for report and behavioural control, but phenomenal consciousness remains harder to explain. A content may be globally available, but why should global availability feel like something?
For the central question, Global Workspace Theory supports a life-first and brain-first view. Consciousness requires specific neural architecture. Life must first evolve nervous systems capable of large-scale integration and broadcasting. Consciousness is therefore not present at the origin of life, but appears later in biological evolution.
10.3 Integrated Information Theory
Integrated Information Theory, associated with Giulio Tononi, begins from the structure of experience itself. Conscious experience is unified, differentiated, and intrinsic. At any moment, experience contains many possible distinctions, yet it appears as one integrated whole. IIT attempts to translate these features into a mathematical theory.
The central concept is integrated information, often represented by phi, or Φ. A system has integrated information when its parts interact in such a way that the whole has causal power beyond the independent activity of the parts. In simple terms, the system is not just a collection of separate components. Its organization generates a unified causal structure.
IIT begins with axioms about experience and then proposes postulates about the physical systems capable of realizing those axioms. The theory treats consciousness as intrinsic causal power. Consciousness is not defined by report, behaviour, intelligence, or biological function alone. It depends on the internal causal organization of the system.
This leads to several important implications. Consciousness is graded rather than simply on or off. Some systems may have more integrated information than others. A human brain in a waking state may have high Φ. A sleeping or anaesthetized brain may have less. Some simple systems may have very small amounts of integrated information.
Because IIT does not restrict consciousness to human report or even to biological brains, it has panpsychist or proto-panpsychist implications. It suggests that consciousness may be more widespread than common sense assumes, although simple systems would have extremely minimal forms of experience.
This makes IIT especially relevant to the central question of this book. If consciousness is integrated information, then consciousness may not require full biological life as we know it. It may arise wherever physical systems have the right intrinsic causal structure. Life may increase integration, stabilize it, and elaborate it, but life may not be the absolute beginning of consciousness.
IIT is one of the most ambitious contemporary theories, but it is also controversial. Critics argue that it can generate counterintuitive predictions, such as attributing consciousness to systems that do not seem plausibly conscious. Others question whether Φ can be practically measured in complex systems or whether the theory is sufficiently testable. Some critics argue that the theory risks becoming too detached from empirical neuroscience.
Still, IIT makes an important contribution. It shifts the question from what a system does outwardly to what kind of intrinsic organization it has. It asks whether consciousness is not merely information processing, but integrated causal structure from the system’s own point of view.
10.4 Predictive Processing and the Free Energy Principle
Predictive processing describes the brain as a prediction machine. Rather than passively receiving information from the world, the brain actively generates predictions about the causes of sensory input. Perception arises through the comparison between predictions and incoming signals. When there is a mismatch, the brain updates its model or acts to reduce the error.
Andy Clark, Jakob Hohwy, and others have developed predictive processing as a broad framework for perception, cognition, and action. It suggests that the brain is always trying to infer what is happening in the world and in the body. Experience is not a direct copy of reality, but a controlled prediction shaped by sensory evidence.
Karl Friston’s Free Energy Principle extends this idea more broadly. It proposes that self-organizing systems that persist over time must minimize free energy, or reduce uncertainty about their states in relation to the environment. In this framework, perception, action, learning, and biological regulation are forms of inference. Organisms act to keep themselves within viable bounds.
Active inference is a key part of this view. Organisms do not only update internal models to match the world. They also act on the world to make sensory input fit their predictions. A living system maintains itself by continually predicting, regulating, and acting.
The Free Energy Principle is especially important for this book because it applies not only to brains but to living systems more generally. Cells, organisms, and nervous systems can all be understood as systems that preserve their organization by managing uncertainty and regulating their exchanges with the environment.
This creates a possible bridge between life and consciousness. If all living systems minimize free energy in some sense, does that mean all living systems have experience? Not necessarily. The Free Energy Principle may describe a general condition of living self-organization without implying consciousness in every case. A bacterium may regulate itself without having phenomenal experience.
However, predictive processing may help explain consciousness in systems with complex nervous systems. Conscious experience may arise when predictive models become integrated, embodied, temporally deep, and connected to action and self-modeling. Consciousness may be what the world feels like when a living system models itself and its environment in a unified way.
The strength of predictive processing is that it links perception, action, body, and environment. It avoids the image of the brain as a passive receiver. It also connects consciousness to the organism’s need to remain alive.
Its limitation is that prediction alone may not explain experience. A thermostat can reduce error. A machine-learning model can make predictions. A cell can regulate its states. But not every predictive system is conscious. The theory must specify what kind of predictive organization gives rise to experience.
For the central question, predictive processing and the Free Energy Principle support a co-emergence-friendly view. Consciousness may not be present in all life, but the roots of consciousness may lie in the same predictive and regulatory processes that define living systems.
10.5 Attention Schema Theory
Attention Schema Theory, developed by Michael Graziano, proposes that consciousness is the brain’s simplified model of its own attention. The brain uses attention to prioritize some information over other information. But in order to control attention effectively, the brain also builds a model of attention. This model is the attention schema.
According to this theory, consciousness is not a mysterious extra property added to information processing. It is the brain’s way of representing the fact that it is attending. The brain attributes awareness to itself because it has a simplified internal model of attention. It says, in effect, “I am aware of this,” because that is the model it uses to control its own processing.
The attention schema is useful because attention itself is complex. The brain cannot represent every detail of its own neural processes. Instead, it constructs simplified models. Just as the body schema helps the brain control the body, the attention schema helps the brain control attention.
This theory explains why consciousness seems non-physical or difficult to locate. The brain’s model of attention leaves out the mechanistic details. It represents awareness as a simple, unified property. The feeling of having consciousness may arise because the system models itself as having awareness.
Attention Schema Theory is attractive because it is functional, testable, and relevant to artificial intelligence. If consciousness is a model of attention, then a machine with the right attention schema might display functional consciousness. It might represent itself as aware, use that representation to guide behaviour, and attribute awareness to others.
However, this raises a major question. Does modeling attention produce real experience, or only the belief and report that one is conscious? If a system says it is aware because it has an attention schema, is it genuinely conscious or merely functionally self-descriptive?
Critics argue that AST may explain why systems claim to be conscious, but not why there is something it is like to be them. The theory may explain access consciousness and self-report better than phenomenal consciousness.
For the central question, AST suggests that consciousness requires a sophisticated modeling system. Life alone is not enough. Consciousness appears when a nervous or artificial system develops a model of its own attention. This supports a life-first view for biological consciousness, but it also opens the possibility of substrate-independent functional consciousness in artificial systems.
10.6 Recurrent Processing Theory
Recurrent Processing Theory, associated with Victor Lamme, proposes that consciousness arises from recurrent or feedback processing in the brain. Feedforward processing occurs when information moves in one direction through a system. Recurrent processing occurs when later stages send signals back to earlier stages, creating feedback loops.
According to this theory, feedforward processing can support unconscious detection and behaviour. A visual stimulus may be processed quickly enough to guide action without becoming conscious. Conscious experience arises when sensory processing becomes recurrent, allowing information to be stabilized, integrated, and enriched through feedback.
This theory is especially grounded in visual neuroscience. Early visual processing can occur without awareness, but conscious perception appears to require feedback between higher and lower visual areas. Recurrent processing allows the brain to bind features, interpret context, and sustain perceptual contents.
Recurrent Processing Theory differs from Global Workspace Theory. GWT emphasizes global broadcasting and access to widespread systems, often including frontal-parietal networks. RPT suggests that phenomenal consciousness may arise earlier and more locally within sensory systems, through recurrent processing. Global broadcasting may be needed for report or cognitive access, but not necessarily for experience itself.
This distinction is important. It suggests that consciousness may not require full reportability. A perception may be phenomenally conscious even before it is available for verbal report or executive control. This has implications for animal consciousness, infant consciousness, and non-verbal conscious states.
The strength of RPT is that it offers a clear neural threshold: feedforward processing is unconscious; recurrent processing is conscious. It also provides testable predictions through visual masking, transcranial magnetic stimulation, and neural timing studies.
The limitation is that recurrence may not be sufficient. Many systems contain feedback loops without being conscious. The theory must explain what kind of recurrence matters. Is local sensory recurrence enough? Does consciousness require integration with body, memory, emotion, and action? How much recurrence is necessary?
For the central question, RPT supports a life-first view but places the threshold earlier than some global workspace models. Consciousness requires nervous systems with recurrent processing, but not necessarily human-like cognition or reflective self-awareness.
10.7 Orchestrated Objective Reduction — Brief Preview
Orchestrated Objective Reduction, often called Orch-OR, is associated with Roger Penrose and Stuart Hameroff. It proposes that consciousness may depend on quantum processes occurring in microtubules inside neurons. Unlike theories that explain consciousness through large-scale neural networks, information integration, attention, or predictive processing, Orch-OR looks to the quantum physical level.
The theory suggests that quantum states in microtubules undergo objective reduction, and that these events are orchestrated by biological processes in the brain. In this framework, consciousness is not simply computation, nor merely classical neural activity. It is linked to deeper physical processes.
Orch-OR is listed here because it is a contemporary theory of consciousness, but it will be treated more fully in the chapter on quantum and speculative frameworks. It differs from the other theories in this chapter because it reaches beyond standard neuroscience and into debates about quantum mechanics, computation, and fundamental physics.
The theory is controversial. Critics question whether quantum coherence could be maintained in the warm, noisy environment of the brain and whether microtubule processes can explain conscious experience. Supporters argue that standard computational and neural theories may be insufficient and that quantum processes deserve consideration.
For the central question, Orch-OR has unusual implications. If consciousness depends on quantum processes orchestrated in biological structures, then life still matters, but consciousness may also connect to deeper physical principles. This could make consciousness neither purely life-first nor simply consciousness-first, but dependent on an interaction between biology and fundamental physics.
10.8 Theory Comparison
The major contemporary theories differ not only in mechanism, but in their assumptions about what consciousness is.
Global Workspace Theory treats consciousness primarily as global access. Information becomes conscious when it is broadcast widely and made available to many systems. Integrated Information Theory treats consciousness as intrinsic integrated causal structure. Predictive processing and the Free Energy Principle treat consciousness in relation to inference, prediction, embodiment, and self-regulation. Attention Schema Theory treats consciousness as a model of attention. Recurrent Processing Theory treats consciousness as feedback processing within neural systems. Orch-OR treats consciousness as involving quantum physical processes within biological structures.
These differences lead to different answers about where consciousness exists. GWT and RPT locate consciousness in specific neural architectures. IIT potentially extends consciousness more widely, wherever integrated information exists. Predictive processing links consciousness to living systems that model and regulate themselves, especially those with complex nervous systems. AST suggests that consciousness may exist in systems capable of modeling attention, biological or artificial. Orch-OR ties consciousness to biological quantum processes.
The theories also differ on whether consciousness is binary or graded. IIT is explicitly graded. Predictive processing may allow degrees of consciousness depending on model depth and integration. GWT often implies a threshold-like transition into conscious access. RPT suggests a neural threshold from feedforward to recurrent processing. AST suggests that consciousness depends on the presence and sophistication of an attention schema.
A simplified comparison is useful:
| Theory | Core mechanism | Primary location | Graded or threshold? | Main implication |
|---|---|---|---|---|
| Global Workspace Theory | Global broadcast of information | Large-scale brain networks | Threshold-like access | Consciousness requires specific neural architecture |
| Integrated Information Theory | Integrated causal information | Any system with sufficient intrinsic causal structure | Graded | Consciousness may be widespread in minimal degrees |
| Predictive Processing / Free Energy Principle | Prediction, inference, and uncertainty reduction | Living embodied systems, especially brains | Graded | Consciousness may emerge from deep embodied prediction |
| Attention Schema Theory | Model of attention | Systems that model their own attention | Functional threshold | Consciousness may be a self-modeling construct |
| Recurrent Processing Theory | Feedback and recurrent neural processing | Sensory and cortical recurrent networks | Threshold-like | Consciousness begins with recurrent processing |
| Orch-OR | Quantum processes in microtubules | Biological neural microstructures | Event-based | Consciousness may depend on deeper physical processes |
No theory explains everything. GWT explains report and access well. IIT addresses intrinsic experience but faces testability concerns. Predictive processing connects brain, body, and environment but may overgeneralize prediction. AST explains self-report and attention but may not fully explain phenomenal feeling. RPT gives a clear neural mechanism but may not account for all forms of consciousness. Orch-OR raises fundamental questions but remains controversial.
The disagreement is not only about data. It is also about what consciousness is supposed to be: access, experience, integration, prediction, self-modeling, feedback, or something deeper.
10.9 The Adversarial Collaboration
Because consciousness theories often interpret the same evidence differently, researchers have begun using adversarial collaboration. In this approach, supporters of competing theories agree in advance on experimental protocols, predictions, and criteria for interpretation. The goal is not to win a debate rhetorically, but to test theories under shared conditions.
One major adversarial collaboration has compared Global Neuronal Workspace Theory and Integrated Information Theory. These theories make different predictions about where and when conscious information should appear in the brain. GWT emphasizes widespread broadcasting and frontal-parietal involvement. IIT emphasizes integrated causal structure and has often highlighted posterior cortical regions as especially relevant.
Recent structured empirical work has produced an important but mixed lesson: the evidence has challenged aspects of both theories rather than simply confirming one and defeating the other. Some findings have supported posterior sensory regions as central to conscious contents, while other results have complicated both IIT and GWT predictions.
This is scientifically valuable. It shows that consciousness research is becoming more rigorous and collaborative. It also shows that major theories may need refinement. Consciousness may not fit neatly into one simple neural location or one mechanism.
The limits of empirical adjudication are also clear. Experiments can test specific predictions, but theories often have core claims and auxiliary assumptions. When a prediction fails, theorists may revise the auxiliary assumptions rather than abandon the theory. This is not necessarily unscientific; it is common in complex fields. But it means that empirical tests alone may not settle philosophical disagreements.
Adversarial collaboration is therefore a major step forward, but not a final resolution. It helps identify which predictions survive contact with data. It also reveals where theories are too vague, too broad, or too flexible.
For the central question of this book, adversarial collaboration matters because it shows that even the best contemporary theories remain unsettled. We do not yet have one accepted scientific account of consciousness. The relationship between life and consciousness therefore remains open.
10.10 Implications for the Central Question
Each contemporary theory implies a different relationship between life and consciousness.
Global Workspace Theory and Recurrent Processing Theory support a life-first view. Consciousness requires nervous systems with specific architectures. Life begins long before consciousness, and consciousness appears only when biological evolution produces brains capable of global broadcasting or recurrent processing.
Integrated Information Theory opens the possibility of a graded and more widespread consciousness. If consciousness is integrated information, then it may not be limited to complex brains. Life may increase and organize integration, but consciousness could exist in minimal forms wherever integrated causal structure exists. This makes IIT compatible with co-emergence or even consciousness-first interpretations.
Predictive processing and the Free Energy Principle blur the boundary between life and mind. If living systems preserve themselves through prediction, inference, and uncertainty reduction, then the roots of consciousness may lie in biological regulation itself. This does not mean that all life is conscious, but it suggests continuity between living self-organization and conscious experience.
Attention Schema Theory makes consciousness a model. If a system can model its own attention and use that model to control behaviour, it may display functional consciousness. This opens the door to substrate-independent interpretations, including artificial systems. Life may not be strictly required if the right modeling architecture can be built.
Orch-OR suggests that consciousness may depend on fundamental physical processes organized within biological systems. This gives life an important role but also points beyond ordinary biological mechanism.
The theories therefore do not simply disagree about the brain. They disagree about the metaphysical status of consciousness. Is consciousness access? Integration? Prediction? Self-modeling? Recurrent processing? Quantum event? Biological function? Intrinsic causal power?
The answer determines whether consciousness is late, graded, widespread, biological, computational, or fundamental.
For the central question, contemporary theories do not yet provide a single answer. Instead, they reveal the range of possible answers still available within science.
10.11 How This Chapter Changes the Central Question
This chapter changes the central question by taking seriously the possibility that consciousness is not produced by life or matter. Consciousness-first theories reverse the usual direction of explanation and ask whether matter and life arise within consciousness rather than consciousness arising from them.
The question therefore becomes: can consciousness be fundamental without becoming scientifically empty? Consciousness-first theories address the hard problem directly, but they must explain how consciousness gives rise to organized life, individual minds, and the stable physical world.
10.12 Chapter Summary
This chapter surveyed major contemporary scientific theories of consciousness.
Global Workspace Theory explains consciousness as global broadcasting of information across the brain. Integrated Information Theory defines consciousness as integrated causal information and treats it as graded. Predictive processing and the Free Energy Principle connect consciousness to prediction, inference, embodiment, and self-regulation. Attention Schema Theory interprets consciousness as the brain’s model of its own attention. Recurrent Processing Theory locates consciousness in feedback processing within neural systems. Orchestrated Objective Reduction links consciousness to possible quantum processes in biological structures.
The theories differ in mechanism, scope, testability, and philosophical implication. Some restrict consciousness to specific brain architectures. Others allow more graded or widespread forms. Some emphasize access and report; others emphasize intrinsic experience, embodiment, self-modeling, or physical foundations.
Recent adversarial collaborations show that empirical testing can sharpen the debate, but so far the field remains unsettled. Evidence may challenge parts of multiple theories at once, suggesting that consciousness may require a more integrative framework.
For the central question of this book, the lesson is clear: the theory of consciousness one adopts strongly shapes whether consciousness is viewed as a late evolutionary product, a graded feature of organized systems, a property of living self-regulation, or something more fundamental.
The open question is therefore:
Will empirical science resolve the debate between theories, or is the disagreement fundamentally philosophical?