---
title: Conscious Agent Theory
---
If space, time, and matter are only the interface — the desktop and its icons — then what is the reality behind the screen? 's answer is the most radical part of his program, and the part he has worked hardest to make mathematical. He calls it **conscious realism**: the thesis that what is fundamental is not matter but consciousness, and that the objective world "consists entirely of conscious agents."
## Turning the world inside out
Most of science assumes matter is fundamental and then struggles to explain how it could ever give rise to consciousness — the "hard problem." Hoffman inverts the order. He takes **consciousness as fundamental** and sets out to derive the appearance of matter, brains, and spacetime *from* it. His reason is an old intuition sharpened: no arrangement of unconscious parts can explain why there is something it is like to taste chocolate. Walk inside Leibniz's thinking mill, he says, and you find only parts pushing parts — never an experience. So rather than try to squeeze mind out of matter, he stipulates experience at the ground floor and treats the real scientific task as deriving the interface of matter from it.
## What a conscious agent is
Crucially, "conscious agent" is for Hoffman a precise mathematical object, not a metaphor. A conscious agent has a space of possible **experiences**, a space of possible **actions**, and three channels linking them: it **perceives** (the world shapes its experiences), it **decides** (experiences lead to actions), and it **acts** (its actions reshape the world) — cycling round and round, step by counted step. And the "world" each agent perceives and acts upon is itself made of other conscious agents. (Formally it is a six-tuple: two measurable spaces — perceptions and actions — together with three Markovian kernels and a counter, the whole forming a Markov chain: perceive → decide → act → world → perceive.) Hoffman frames the wager by analogy to the Church–Turing thesis: just as any computation can be captured by a Turing machine, he proposes that any property of consciousness can be captured by the dynamics of conscious agents — a claim that "might be precisely wrong," but is open to exploration as science.
## Agents all the way up
How do you get a world out of agents? They combine. Two conscious agents can "join" into a new, more complex conscious agent, and the operation is closed — a whole network of agents is itself a single conscious agent. From the long-run dynamics of such networks, Hoffman argues, the structures of physics emerge: he finds that the asymptotic behaviour takes the mathematical form of the wavefunctions of free particles, so that a "particle" becomes a pattern in the dynamics of interacting agents and spacetime becomes the interface that displays it. A newer operation he calls **fusion** — in which agents' experiences merge into genuinely new, unified experiences — is his proposed answer to the ancient puzzle of how many small minds could ever combine into one larger mind. In honesty, this machinery is fully worked out only for very small, toy systems; the sweeping claim of deriving the physical world is, as Hoffman would agree, a research program rather than a finished theorem. He offers it as a direction — and a bold one. Its most ambitious reach, into fundamental physics, is taken up in .
## Among the idealists
Conscious realism is a form of **idealism**, but a distinctive one. Unlike Bernardo Kastrup's analytic idealism — which posits a single universal mind whose "dissociated alters" are individual psyches — Hoffman's world is plural at the base: many interacting agents combining upward, with no single privileged cosmic subject built into the formalism. It parts ways with **panpsychism** (which keeps matter fundamental and adds experience to it) by denying that matter is fundamental at all. And it runs in the opposite direction to **Integrated Information Theory**, which starts from physical systems and constructs consciousness from their organisation, where Hoffman starts from consciousness and constructs the physical from it. He and Kastrup are frequent, friendly sparring partners — and on Hoffman's view even "Donald Hoffman" is one more icon on the interface, not a privileged observer.
## Sources & talks
A relaxed long-form conversation on conscious agents:
A rigorous philosophy-grad interview on the whole picture:
**More:** a three-way debate with Bernardo Kastrup and Susan Schneider, ["When AI Becomes Conscious"](https://www.youtube.com/watch?v=VmQXpKyUh4g). **Primary sources:** Hoffman & Prakash, ["Objects of Consciousness," ](https://doi.org/10.3389/fpsyg.2014.00577)[*Frontiers in Psychology*](https://doi.org/10.3389/fpsyg.2014.00577)[ 5:577 (2014)](https://doi.org/10.3389/fpsyg.2014.00577) · Hoffman, Prakash & Prentner, ["Fusions of Consciousness," ](https://doi.org/10.3390/e25010129)[*Entropy*](https://doi.org/10.3390/e25010129)[ 25(1):129 (2023)](https://doi.org/10.3390/e25010129) · and *The Case Against Reality* (W. W. Norton, 2019).