Donald Hoffman is a cognitive scientist at the University of California, Irvine who argues that we do not see reality as it is. His interface theory of perception holds that evolution tuned our senses to track fitness payoffs rather than truth, and that space, time and objects are the interface, not the machine. The argument is stronger than its slogan.
Hoffman's case comes first, at full strength, because a comparison is worth nothing if the other position is a caricature. The part almost nobody writes down comes after: where a consciousness-first framework has to disagree.
Who is Donald Hoffman?
His faculty page at UC Irvine lists work across visual perception, evolutionary modeling and consciousness — a span that explains the argument's shape, which starts in vision science and ends in metaphysics. Most people meet him through a TED talk, where the soundbite comes from; the substance is in the peer-reviewed papers. The popular version is a provocation, the technical version a model with stated assumptions — the kind of object you can check.
What is the interface theory of perception?
The interface theory, set out by Hoffman with Manish Singh and Chetan Prakash in their 2015 paper The Interface Theory of Perception, says perception is a user interface rather than a window. The desktop on your screen has folders and icons; a blue rectangle is not what a file is like, since there is nothing blue or rectangular inside the machine. But the icon is not a lie either. It is a useful simplification that lets you act without knowing anything about voltages or memory addresses. Hoffman's claim is that our perception of space, time and physical objects works the same way: not a faithful reconstruction of what is out there, but a species-specific interface shaped to guide behavior. Hiding the underlying structure is its whole job, because an organism that stopped to compute the truth would lose to one that simply sees what to do next.
None of this makes perception arbitrary, or means nothing exists: the claim is that perceived structure need not resemble the structure of what is there.
Is there evidence, or is this just a metaphor?
This is the part most coverage skips. That perception is constructive rather than a passive recording is uncontroversial in vision science: the brain builds a scene from ambiguous input rather than transcribing one. Hoffman's move is to ask what evolution would build given the choice, then answer with modeling.
The method is evolutionary game theory: set up worlds with resources, let perceptual strategies compete, see which survives. One strategy sees the true structure of the world; another sees none of it and tracks only fitness payoffs. Hoffman's reported result, summarized in his TED talk and argued in the 2015 paper, is that the truth-seeing strategy does not merely lose ground — it is driven to extinction by strategies of equal complexity that see only payoffs. Four words: fitness beats truth.
His favorite illustration: Australian jewel beetle males were found trying to mate with discarded beer bottles — brown, dimpled, glossy, larger than any female — because their perception tracked a few payoff cues rather than the truth about beetles. The interface normally works flawlessly; it just never was a description.
The steelman should be uncomfortable for anyone wanting to wave him off: fitness-beats-truth is a claim inside stated models, not a flourish. You can read the assumptions, reject them, build a different game and show him wrong. Most metaphysics offers no such handle.
What is conscious realism — and how far does the claim go?
Hoffman's answer to what reality is comes in Objects of Consciousness, written with Chetan Prakash in 2014, and it is far more radical. The proposal there is conscious realism: what exists objectively is not matter but conscious agents and their experiences. A conscious agent, in that paper, is a mathematical object — spaces of possible experiences, actions and world-states, linked by maps for perceiving, deciding and acting — and agents combine into larger agents. Physical objects are icons in the interfaces of such agents. So is spacetime, which on this account is not the stage reality stands on but part of the display — spacetime, as he sometimes puts it, being doomed. The program follows: rather than deriving consciousness from physics, derive physics from the dynamics of conscious agents.
Conscious realism is not textbook panpsychism, which puts mentality into the micro-level constituents of matter — on Hoffman's account particles are icons, not constituents. It sits in the neighborhood the Stanford Encyclopedia calls idealism, ontological priority to the mental, while matching no classical version of it.
Where does the argument become contested?
Three places, and they fail differently. First, a result inside a model inherits its assumptions: fitness beats truth in the games Hoffman constructs, with the fitness functions, strategy spaces and payoffs he chose. Whether it generalizes to real perceptual systems is what researchers argue about; the modeling does not settle it.
Second, the step from "perception is an interface" to "spacetime is not fundamental" is a much larger stride than it looks. Grant the modeling result entirely and you have a conclusion about perceptual representations; the status of spacetime is a different question, and no evolutionary game answers it. That is where the live debate sits.
Third, the conscious-agent formalism is a mathematical proposal, not a confirmed theory. It has produced no derivation of physics that physicists have taken up, and no experiment distinguishes it from its rivals. That does not make it wrong; it makes it unfinished.
And the line that has to be said plainly, since this is where people reach for it: none of this proves that consciousness is fundamental. A materialist can accept the interface theory whole — evolution builds useful interfaces, not truthful ones — and hold that the machine behind the desktop is ordinary matter. The two claims are independent, and a framework that runs them together has helped itself to something unearned.
If you follow arguments by watching where they are willing to be wrong, the full case is set out in one place.
Where does Holopsychism diverge from Hoffman?
From here on this is Holopsychism's own reading — a proposal offered for testing, not a verdict on who is right. Holopsychism is this site's coined name for it, not a recognized academic school.
The agreement is real: the desktop is not the machine. Both hold that the perceived world is an interface, and both reject the picture in which a brain manufactures experience — on Hoffman's account a brain is an icon, and an icon cannot produce the mind reading it.
The divergence is about the primitive, which is the only place a metaphysics really gets decided. Hoffman starts with conscious agents: entities that already perceive, decide and act, arranged in networks — plurality and structure on the ground floor. Holopsychism starts one step earlier, with consciousness as pure potential (structureless, non-local, nothing yet individuated) plus awareness as selection, the act by which one possibility becomes a stable configuration rather than a live one. Individuated experiencers are a late result rather than the starting material: matter as persistent coherence, brains as receivers and filters tuning that field at different resolution — a spectrum, not a hierarchy.
Different primitive, different bill to pay. Hoffman owes an account of how agents combine without collapsing, plus the derivation of physics his program promises — but he owes it in a formalism, which makes the debt checkable. Holopsychism owes an account of why selection happens at all, and why one outcome stabilizes rather than another, with no formalism whatsoever. On that axis his program is in better shape.
Four positions get blurred in almost every discussion of this territory, so it is worth setting them side by side.
| Position | Primitive | The physical world is… | Divergence from Holopsychism |
|---|---|---|---|
| Conscious realism (Hoffman) | Networks of conscious agents | Icons in an interface | Agents are already structured; potential is not |
| Biocentrism (Lanza) | Life and biological observers | Actualized by observation | Biology is not fundamental here |
| Analytic idealism (Kastrup) | One universal consciousness | Extrinsic appearance of mental process | Substrate is pre-mental potential, not mind |
| Holopsychism (this site's coinage) | Structureless potential + selection | Stabilized coherence | — |
Each deserves more than a row in a table. The observer-effect argument that makes Robert Lanza's biocentrism vulnerable, and the dissociation mechanism behind Bernardo Kastrup's analytic idealism, differ enough from the interface argument that treating the three as one movement gets all three wrong. The wider map is in materialism vs idealism; this framework's starting question is in framing the question.
Is this just another consciousness framework with a new name?
The fair test is whether the difference does any work. A framework built from conscious agents has to explain how agents compose; one built from structureless potential has to explain how anything gets individuated at all. Not the same problem wearing different labels — different obligations, different failure modes.
What Holopsychism does not explain, without hedging: it does not say why one selection occurs rather than another, and it offers no equations, no quantitative prediction, no derivation of any constant. It cannot be tested against Hoffman's models, having nothing formal to test with — and if his program ever did derive physics, that would be evidence for a rival with a working formalism, not for us.
It could also fail cleanly: if the fitness-tuned character of perception is fully explained with no consciousness-first substrate anywhere — which is what most researchers expect — there is nothing here for the framework to gain.
The honest comparison
Hoffman and Holopsychism agree on the negative claim, the easy half: the perceived world is not the world. They separate on what a theory owes you once it says so, and the answers differ in kind — a formalism and a derivation program on his side, an argument about what a substrate must be like for selection to be possible at all on ours. Neither has paid its bill. That leaves you not with a choice between answers but with a way of reading both: by what each has made itself answerable to, and by which would notice if it were wrong.
Frequently asked questions
What does Donald Hoffman actually claim? That perception evolved to track fitness payoffs rather than truth, so space, time and objects are a species-specific interface, not a description of reality. His metaphysics goes further: conscious realism, on which what exists objectively is conscious agents and their experiences.
What is the interface theory of perception? The view, published by Hoffman, Singh and Prakash in 2015, that perception works like a computer desktop: icons are useful without resembling what they represent, and perceptual structure need not resemble the structure of the world.
Is "fitness beats truth" proven? It is a genuine mathematical result inside evolutionary game-theoretic models, not a slogan — which is what makes it worth engaging. Whether it generalizes to real perceptual systems, and whether it licenses conclusions about spacetime, is contested and unsettled.
Is Hoffman's view the same as biocentrism or analytic idealism? No. Robert Lanza's biocentrism centers biological life; Bernardo Kastrup's analytic idealism starts from one universal consciousness and explains individual minds by dissociation; Hoffman starts from networks of mathematically defined conscious agents. Shared direction, different substrate.
How does Holopsychism differ from Hoffman's conscious realism? Both hold that the perceived world is an interface. His primitive is conscious agents, with perception and decision already built in; Holopsychism's is structureless potential plus awareness as selection, individuated minds arriving as a result rather than a starting point — a proposal with no formalism behind it.
If you want the version of this that keeps its own limitations in view rather than out of frame, the full argument is in the guide.
