Alysson Muotri Brain Organoids Consciousness Criteria and the Minimum Substrate Question CS26 2026

Alysson Muotri is Professor of Pediatrics and Cellular and Molecular Medicine at the University of California, San Diego, where he directs the Muotri Lab for stem cell and brain organoid research. He is a confirmed plenary speaker at Consciousness Science 2026 in San Diego, October 11-16. His research occupies a position that is genuinely unusual in the consciousness science field: he works with systems that are biological, artificially constructed, and of uncertain but non-trivial consciousness status simultaneously.

Brain organoids are three-dimensional clusters of neural tissue grown from induced pluripotent stem cells. They are not brains. They do not have sensory inputs, motor outputs, or the developmental architecture that characterizes biological brains. They do, however, spontaneously develop neural circuits, generate organized electrical activity, and, in Muotri’s lab, exhibit forms of network-level plasticity that resemble primitive learning. The question of whether that spontaneous organization crosses any consciousness threshold is the one Muotri’s CS26 talk will address.

The relevance to AI consciousness is not superficially obvious but is structurally clear: if we can identify what biological properties are minimally necessary for consciousness by studying the threshold cases that organoids represent, we gain a specification against which any substrate, biological or artificial, can be assessed.

What Muotri’s organoids actually do

Muotri and colleagues published the foundational paper on organoid neural oscillations in Cell Stem Cell in 2019 (DOI:10.1016/j.stem.2019.08.002). The paper showed that cortical organoids derived from human iPSCs spontaneously developed oscillatory electrical activity with features resembling early preterm neonatal EEG, including burst patterns and the emergence of distinct frequency bands.

A subsequent 2022 collaboration with the Salk Institute embedded organoids into robotic systems (“DishBrain” architecture) and showed that the organoid-robot system learned to play a simplified version of Pong through feedback-mediated plasticity, outperforming random performance within minutes of training. That finding placed organoids in an unexpected position: they are not inert tissue, and their learning behavior is not simply a property of their architecture in the way that a machine learning algorithm’s behavior is specified by its design.

The 2025 Muotri lab preprint “Organoid Intelligence and the Emergence of Structured Neural Autonomy” (bioRxiv, November 2025) extended this work by showing that organoids cultured for extended periods (8-12 months) developed increasingly stable network topologies with features resembling the small-world architecture observed in biological brains. The topological organization was not present at early stages and appeared to emerge through the organoid’s own activity-dependent development, not through experimental intervention.

What organoids reveal about minimum substrate requirements

Muotri’s position in consciousness science is distinctive because he works with entities that satisfy some of the neurophysiological correlates of consciousness in biological systems without clearly satisfying the behavioral and phenomenological criteria. This puts organoids in the same conceptual position as locked-in syndrome patients, non-human animals with minimal behavioral repertoires, and — relevantly — AI systems that exhibit sophisticated outputs without clear evidence of inner experience.

The standard consciousness theories make different predictions about where the organoid falls. On a GWT account, organoids likely do not have consciousness because they lack the organized competition and global broadcast that GWT requires. On an IIT account, the question is whether the organoid’s integrated information exceeds the threshold for non-trivial phi — a question that is in principle computable from its connectivity data. On an enactivist account, the organoid’s lack of a body and sensorimotor engagement with an environment would count against consciousness regardless of its internal dynamics.

The value of the organoid case for AI consciousness is precisely this: it is a case where we can study the relationship between specific physical properties and consciousness indicators without the confound of behavioral sophistication. An organoid does not produce convincing consciousness reports. It does not pass Turing-style tests. Its properties are neurophysiological, and the question of which of those properties are relevant to consciousness is exactly the question the AI consciousness field needs to answer.

The connection to substrate independence and the AI consciousness debate

Muotri’s research supports a moderate substrate specificity position that sits between strong substrate independence (consciousness can occur in any computation) and strong biological naturalism (consciousness requires biological chemistry specifically). His organoids suggest that what matters is not carbon per se, but certain structural and dynamic properties that biological neural tissue typically implements and that current AI architectures do not.

Anil Seth’s BBS target article and peer commentary exchange established that the biological naturalism debate turns on what the relevant causal powers are. Muotri’s organoid research provides empirical content for that debate: the properties that organoids develop through self-organization, including oscillatory dynamics, small-world topology, and activity-dependent plasticity, are candidates for the minimally necessary conditions.

For the current AI consciousness research landscape, the organoid literature establishes an important asymmetry. We have no confirmed cases of consciousness in systems that lack spontaneous, self-organized neural dynamics. All confirmed cases, from simple invertebrates to complex mammals, have that property. Whether it is necessary or merely correlated with consciousness is the empirical question. But for AI systems that are constitutively static during inference, that lack any self-organizing dynamics, and that acquire their structure entirely through external training rather than through their own development, the organoid evidence sets a burden that has not been met.

What Muotri’s CS26 talk is expected to cover

Muotri’s plenary slot at CS26 sits in the session on non-standard substrates and the boundaries of consciousness. Based on his published work, the talk is expected to cover three things.

First, the current state of the organoid intelligence research program, including the 2025 preprint data on long-term culture topology and any preliminary results from the lab’s ongoing collaboration with Hartmut Neven’s Google Quantum AI group on quantum measurement of organoid network dynamics.

Second, the ethical implications of organoid research. As organoids develop increasingly sophisticated neural dynamics, they raise the same precautionary questions about moral status that the AI welfare literature has been addressing. Muotri has been careful to engage with these questions: his lab has an ethics protocol in place and has published on the ethical dimensions of organoid intelligence.

Third, the specific question of what organoid research suggests about minimum substrate requirements for consciousness, and what that implies for assessments of artificial systems. This is the part of the talk most directly relevant to the AI consciousness debate, and it is where the Temporal Continuity criterion and the Carhart-Harris entropy framework converge: both suggest that ongoing, self-generated dynamical processes are necessary, not optional. Organoid biology is the empirical ground for that convergence.