The Institute investigates the hypothesis that physics is an interface, constructed by conscious agents to interact with each other. The structures of cognition, perception, and experience are recoverable from the dynamics of these agents. Conscious Agent Theory and the Interface Theory of Perception ground the program. Trace Logic and Recursive Trace Logic are the framework in which the program is pursued.
Quantum theory cannot say where the observer enters. General relativity cannot reconcile with it. Cognitive science has mapped vast territory of perception and cognition without explaining how any of it becomes experience. The hard problem of consciousness has no answer inside materialist reductionism: there is no derivation, in any existing theory, of how arrangements of matter would generate the felt quality of experience.
Physics saw the gap decades ago. John Wheeler called the observer a "participator" and proposed it from bit: the physical world derives from information, and information is what an observer registers. The Trace Institute builds the science that proposal has been waiting for.
The Institute's research is one connected argument. Each pillar is peer-reviewed or under active development. Read top to bottom: each step earns the next.
A mathematical model of an observer as a Markov kernel on a measurable space of experiences and actions. Each conscious agent is a triple C = ⟨(X, X), Q, n⟩. Networks of these agents compose to form larger agents, with no privileged scale.
Hoffman & Prakash, 2014. Fields, Hoffman, Prakash & Singh, 2018.
The Fitness-Beats-Truth theorem proves that natural selection drives veridical perception to extinction: organisms see fitness payoffs, not objective reality. Spacetime and physical objects are the species-specific user interface, not the underlying ontology.
Hoffman, Singh & Prakash, 2015. Prakash et al., 2020, with the FBT proof.
A partial order on Markov kernels under which the dynamics of interacting conscious agents generate a calculus of inference. The Trace Chain Theorem (2025) establishes that this logic is provably homomorphic to the Lebesgue logic of probabilistic belief from observer mechanics. Trace Logic is the language in which the structures of fundamental physics are to be recovered. The namesake of the Institute.
Hoffman, Prakash, Chattopadhyay, 2025. Whitepaper, 2026. Observer mechanics: Bennett, Hoffman & Prakash, 1989.
The bridge from trace logic to first-person experience and active inference. RTL describes how an agent constructs and updates its interface recursively, without requiring a materialist substrate. It connects the theory to phenomenology, predictive processing, and the philosophy of mind.
In development. Active inference: Clark, 2017; Parr, Pezzulo & Friston, 2022.
A conscious agent is a Markov kernel acting on its own measurable space of experiences. Compose many such kernels and a partial order emerges on the space of agents, which is the Trace Chain Theorem. From that partial order a logic arises (trace logic) that is homomorphic to the Lebesgue logic of probabilistic belief established in observer mechanics (Bennett, Hoffman & Prakash, 1989). The asymptotic statistics of such chains are conjectured to reproduce, in the appropriate limits, the mathematical structures of relativity, quantum theory, and the Standard Model, alongside the mathematical structure of perception, inference, and conscious experience itself.
Recursive Trace Logic extends the theory to first-person dynamics: the agent's own act of inferring, updating, and constructing its interface. It connects trace logic to active inference and the predictive-processing literature without committing to a materialist substrate.
