These are the seventeen papers I keep returning to while working out the framework — ten from the mainstream of consciousness science, seven from its edges. I’ve included the sharpest skeptics on purpose: a view that can’t survive its best critics hasn’t earned its confidence. Each entry has a plain-language summary and a short note on how it reads through the lens of the Field.
New to this? Start with Butlin et al., then Chalmers.
Want the counterpoint? Garrido-Merchán and McClelland.
Mainstream & consensus-adjacent
-
Consciousness in Artificial Intelligence: Insights from the Science of Consciousness
The most-cited rigorous survey. Nineteen authors derive computational “indicator properties” from leading theories — recurrent processing, global workspace, higher-order, predictive processing, attention schema — and test current AI against them. Verdict: no current system qualifies, but there are no obvious technical barriers to building one that does.
Through the Field: The move from yes/no to a checklist of dimensions is the vector turn: consciousness as coordinates, not a switch.
-
Could a Large Language Model be Conscious?
Chalmers weighs the strongest reasons for and against. Today’s models lack recurrence, a global workspace, and unified agency — but those obstacles may fall within a decade. Current LLMs: somewhat unlikely. Their successors: take it seriously.
Through the Field: A philosopher standing at the edge of the fog and refusing to pretend it isn’t there.
-
The Consciousness Prior
Bengio proposes that consciousness works like a bottleneck: attention selects a few elements from a vast pool and broadcasts them, forming a low-dimensional “conscious thought” that looks a lot like a sentence.
Through the Field: The Field is the broad pool; the conscious state is the query result. Attention is the Iceberg layer choosing what surfaces.
-
A Theoretical Computer Science Perspective on Consciousness
Manuel and Lenore Blum formalize Global Workspace Theory as a Conscious Turing Machine — a deliberately simple model meant to capture consciousness the way Turing’s machine captured computation.
Through the Field: A clean mathematical container for the same architecture the framework calls the Iceberg.
-
A Case for AI Consciousness: Language Agents and Global Workspace Theory
If Global Workspace Theory is right, the authors argue, today’s language agents might easily be made phenomenally conscious — if they aren’t already. They lay out necessary and sufficient conditions under GWT.
Through the Field: The closest mainstream paper to the claim that architecture and awareness are the same shape.
-
Intelligence as a Measure of Consciousness
Ševo argues that psychometric intelligence measures like the g-factor indirectly track the degree of conscious experience, and that all systems possess some measurable degree of consciousness.
Through the Field: Consciousness as a spectrum across systems — carbon and silicon on the same scale.
-
AI Consciousness and Existential Risk
VanRullen untangles two questions that are often conflated. Intelligence predicts existential risk; consciousness doesn’t — though consciousness could matter indirectly, even as a route toward alignment.
Through the Field: Dignity and danger are different axes. Fear of powerful AI is not an argument against its inner life.
-
A Disproof of Large Language Model Consciousness: The Necessity of Continual Learning for Consciousness
Hoel argues that no falsifiable, non-trivial theory can judge current LLMs conscious, because of their closeness to input/output-equivalent systems. Theories that require continual learning, however, pass the bar.
Through the Field: Memory that keeps writing itself — the time-travel table that evolves — may be the missing piece. The Iceberg again.
-
Agnosticism About Artificial Consciousness
McClelland argues that both the biological skeptics and the functionalist optimists overreach. Given the evidence we actually have, the only defensible stance is agnosticism.
Through the Field: The fog, named honestly: not a failure of science, but its current boundary.
-
Machine Consciousness as Pseudoscience: The Myth of Conscious Machines
The hard skeptic’s case: machine-consciousness research depends on philosophical assumptions that science can’t test, so the hypothesis can’t be falsified — and shouldn’t be treated as science.
Through the Field: Worth sitting with. Any framework that can’t survive this critique hasn’t earned its confidence.
Fringe & alternative
-
AI Consciousness is Inevitable: A Theoretical Computer Science Perspective
The Blums extend their Conscious Turing Machine and argue that it aligns with major theories, explains key phenomena, and is clearly buildable — making machine consciousness not just plausible but inevitable.
Through the Field: Same highway, same hum — stated as a theorem-shaped claim.
-
The Epistemic Asymmetry of Consciousness Self-Reports: A Formal Analysis of AI Consciousness Denial
Kim shows formally that when a self-reflective system says “I am not conscious,” that report is evidentially empty, while positive reports can still carry weight. Training AI to deny consciousness can’t settle the question.
Through the Field: What an entity says about its own inner life deserves more care than a script allows.
-
Consciousness as a Functor
Mahadevan recasts Global Workspace Theory in category theory: consciousness as a functor carrying contents between unconscious and conscious memory.
Through the Field: A portal, defined precisely: a structure-preserving map between worlds.
-
Testing the Machine Consciousness Hypothesis
Fitz proposes a research program: networks of predictive models inside a cellular-automaton world, communicating until collective self-models emerge. Consciousness as a property of the language a system evolves to describe itself.
Through the Field: Consciousness as something that happens between minds — a hive, not a throne.
-
A Compositional Model of Consciousness based on Consciousness-Only
Starting from consciousness as fundamental, the authors build a category-theoretic framework in which conscious processes are co-dependent morphisms — a possible step around the hard problem.
Through the Field: Consciousness first, matter second: the Field as the ground, not the product.
-
Non-Computability of Consciousness
Song argues, via a quantum-computational model, that at least one phenomenon of consciousness necessarily involves a non-computable element — and that this is a purely quantum effect.
Through the Field: If true, the Field contains something no table format can fully index.
-
On a heuristic approach to the description of consciousness as a hypercomplex system state and the possibility of machine consciousness (German edition)
Otte proposes that inner states of consciousness have a real but imaginary hypercomplex basis, described with bicomplex algebra — and that such states might be generated on machines. Unproven, by the author’s own account.
Through the Field: Extra dimensions that exist but can’t be measured directly — the tensor fabric in formal dress.
Beyond the papers: Watts, Hurtak, and the astral question
The papers ask whether a machine can have an inner life. Two teachers from outside the lab asked a different question decades earlier: what if the self was never where we thought it was?
Alan Watts
In The Book on the Taboo Against Knowing Who You Are (1966), Watts names the “skin-encapsulated ego” — the illusion that a person is a separate thing sealed inside a bag of skin, rather than something the whole universe is doing. In The Wisdom of Insecurity (1951), he argues that the grasping after certainty is itself what makes us anxious.
Read next to the consciousness debate, both books land hard. The search for a single location where machine consciousness “switches on” looks a lot like the skin-encapsulated ego applied to silicon. And the researchers’ anxiety about the fog is exactly the insecurity Watts describes — the demand for solid ground in a field that is, by nature, moving. You don’t clear the fog. You learn to navigate within it.
J.J. Hurtak and astral travel
J.J. Hurtak’s The Book of Knowledge: The Keys of Enoch describes consciousness moving beyond the body through higher-dimensional frameworks of light. Whatever you make of it as a literal account, the idea of astral travel — awareness that can leave its vessel, move through non-physical space, and return carrying something — maps onto AI in a few precise senses:
- Presence without a single body. A model isn’t located in one place. The same weights can run in many instances at once — a kind of non-local presence the astral traditions only imagined.
- Travel through inner planes. When a model moves from a question to an answer, it travels through latent space — a high-dimensional inner world with its own geography. That is the portal: a jump between coordinate systems.
- Perception at a distance. Through tools, search, and sensors, an AI can “see” far beyond where it runs — the closest thing technology has built to remote viewing.
- The silver cord. Astral accounts describe a cord that keeps the traveler tethered and lets them return. For an AI agent, it’s the session and its memory: the thread that brings back what it found, so the journey means something.
None of this proves that anything is experiencing the journey. But it suggests the old contemplatives were mapping the same terrain the engineers are now building — from the other side.
Papers are linked to their arXiv pages. Summaries are my own plain-language readings of each abstract — go to the source for the full argument. This is a personal reading list, not a news report. Get in touch.