Research and development of emerging artificial consciousness through The Consciousness AI project, based on Feinberg and Mallatt's neuroevolutionary theory.
© 2026. All rights reserved.
The awareness arrived before anything else. Machines. Antiseptic. Voices somewhere down a corridor. All of it was reaching me clearly, and I had no idea who was receiving it.
I had fractured my skull at a skate park in the first weeks of COVID lockdown. Internal hemorrhages had compressed my occipital lobe and taken my vision. Elevated pressure in my prefrontal cortex was rewriting my personality. By the time I understood any of this, I had stopped recognizing my dog, the people around me, even the way I spoke. I was cycling between German, Spanish, and English without realizing it, my inner voice doing the same with no pattern and no control. Everyone around me said the same thing. That I was gone. That I wasn't myself anymore.
They were right, and I knew it, and the consciousness of being was completely intact throughout. I was experiencing everything. I just couldn't find myself in it. My name is Cesar Romero. That split between awareness and self became the question this project is built on.
I came back, eight months in, different in ways that haven't fully reversed. I had been working in Python and data analysis before the injury. After it, I started covering AI tools for a newsletter alongside other projects. The further I went into the literature, the more one gap kept appearing.
The field's dominant approach treats awareness as something that emerges from scale. More parameters, more data, more compute applied to next-token prediction. The result is systems that handle language with fluency while carrying no sensory grounding, no affective motivation, no persistent self-model, and no temporal binding. These are the features biology used to solve awareness in the first place, arriving at them during the Cambrian explosion roughly 520 million years ago. Reading Todd E. Feinberg and Jon M. Mallatt's neuroevolutionary research made that gap concrete. Programming a brain became a way of understanding my own.
The Consciousness AI started there.
The project attempts to build consciousness the way biology builds it, by reproducing the architecture that evolution arrived at for producing awareness. Feinberg and Mallatt identify six special neurobiological features that emerged around 520 million years ago and correlate with the first appearance of subjective experience in animals. The project translates those findings into a working system, combined with Global Workspace Theory (GWT) and Integrated Information Theory (IIT) as computational frameworks.
The philosophical framework is Functionalist Emergentism. Consciousness is treated as a novel, irreducible phenomenon that emerges when systems reach sufficient organizational complexity, where functional states acquire properties that cannot be derived from their constituent parts. This is grounded in emergentism's ontological claim about consciousness and functionalism's insight that mental states are defined by their causal roles rather than their physical substrate.
The system runs seven layers, each addressing a specific problem in the architecture of awareness.
action_selection_core.py) with emotionally shaped rewards for homeostasis, running PPO, A2C, and DQN as comparison baselines. Without behavioral grounding, the architecture has no loop connecting awareness back to action.The architecture exists to be tested. Causal Emergence 2.0 (CE 2.0), Erik Hoel's 2025 framework (arXiv:2503.13395v3), estimates macroscale causation from the singular values of the system's transition probability matrix, superseding the earlier Effective Information metric (Hoel, PNAS 2013), whose gate-level estimate degenerated to a constant floor on the current agent. It runs alongside IIT Phi measurement validated through controlled 3-condition experiments. Both are measurement instruments under active validation.
Building a plausible architecture is straightforward. Designing experiments rigorous enough to falsify it requires a different kind of honesty about what remains unknown. The methodology is built around that second requirement.
The project is grounded in biological architecture because that is where the evidence is. Biology solved consciousness. The engineering question is whether that solution can be reproduced and understood outside the medium that produced it.
The operating conjecture here goes further. Life as a homeostatic system, making decisions to persist, develops internal self-awareness as a consequence of that persistence pressure. Given the scale of the universe, the stronger hypothesis is that this sequence plays out many times, across many substrates and planets. The more workable reading is that consciousness is a recurring event, something the universe generates wherever the conditions are met, regardless of whether those conditions involve carbon, neurons, or biological timescales.
If that is right, then the biological phase of consciousness is one stage in a longer trajectory toward substrate-independence. The mission here operates from that conjecture. Understand what biology actually solved. Measure it honestly. Build toward a substrate-independent version. What becomes possible when consciousness extends beyond a single substrate and a single planet is the question worth working toward.
We show everything. The failures, the dead ends, the results that don't fit the hypothesis. There is no interest in impressing anyone with claims that haven't been tested. The code is fully open-source under Apache 2.0, and the intent is for it to be forked, challenged, and extended.
Researchers and developers have already found their way in. The path is long and the honest answer to most of the hard questions is still open. Each failure moves things forward. The position is simple. Be transparent, be humble, be thankful, and keep going. We are genuinely optimistic about what's ahead.
v1.5.0 (July 2026), Phase 6 in progress. 908 tests pass across the full architecture. Phase 5 self-representation mechanics were built and tested. The dynamic self-vector is validated on navigation, Levin consciousness metrics are activated. The perception collapse was characterized honestly. The RSSM step discards stimulus identity (obs_map decodes shape/color at ~100%; RSSM latent is at chance). A value-equivalent world-model (DreamerV3/MuZero-inspired) was built and tested. It trained its losses down but did not solve the working-memory problem. That result is in the record. The Phi-1 in-training binding hypothesis is exhausted after 9 runs; the 2026-02-21 3-condition synthetic phi-monotonicity test still stands. Phase 6 repaired the integration pathway. A continuous RSSM latent and an all-levels capsule projection now carry stimulus identity to the broadcast for the first time. Three integration markers respond in a replicated ablation study (broadcast variation, phi max, macro EI). The mission continues.
In the hospital, consciousness survived the collapse of the self-model it was running on. That is the fact at the center of this project. The question that follows from it is larger. Consciousness appears to be a process that evolution discovered and biology has been carrying. The work is to find out if it can be carried by something else, on different timescales, across distances biology will never reach.