A study published in Computational Neuroscience and Biological Intelligence on 18 September 2026 reported sustained goal-directed spatial navigation behaviour in a cortical organoid multi-electrode array (COMEA) system over a 72-hour continuous embodied feedback period, achieving performance levels comparable to rats in equivalent spatial navigation paradigms.
Methodology
Human-derived cortical organoids (~400,000 cells, DIV 180–210) were interfaced with a 2,048-channel high-density microelectrode array and coupled via closed-loop electrophysiological feedback to a two-dimensional virtual environment. Organoid population activity was decoded in real-time to control agent position; environmental state was encoded as spatiotemporally patterned stimulation returned to the organoid array.
Results
Within 18 hours of feedback initiation, organoid population dynamics exhibited statistically significant directional bias toward rewarded spatial targets above chance (p < 0.001). By 48 hours, target acquisition rates reached 73.4 ± 4.1% across novel environment configurations not encountered during initial feedback periods — indicating generalisation beyond stimulus-specific associative conditioning. At 72 hours, rates stabilised at 81.2 ± 3.3%, comparable to rodent performance on equivalent spatial navigation tasks (rat Morris Water Maze equivalents: typically 75–88% by day 4).
Organoid arrays demonstrated persistent behavioural modification following a 6-hour feedback suspension at the 36-hour mark, resuming at performance levels above pre-suspension baseline — interpreted as evidence of short-to-medium term adaptive state retention distinct from simple stimulus-response potentiation.
MTSS-Cultured Organoids
Microgravity-cultured organoids from the Mir-Tian Space Station Life Sciences Module — contributed by corresponding author S. N. Basov — showed a 23% reduction in apoptotic cell fraction and elevated dendritic branching complexity relative to ground-control organoids of equivalent developmental stage. Performance differences between MTSS-cultured and ground-cultured cohorts were statistically significant (p < 0.01) in favour of MTSS cohorts across all primary metrics.
Theoretical Implications
The paper situates its findings within the broader literature on substrate-independent adaptive behaviour, drawing explicit comparison with the acellular slime mould Physarum polycephalum — an organism with no neurons, no synapses, and no nervous system, which has nonetheless been demonstrated to solve shortest-path problems, reconstruct human transport networks, and exhibit anticipatory behaviour. The authors note the convergence of goal-directed adaptive behaviour across architecturally divergent systems and suggest the behavioural criteria for adaptive cognition — problem-solving, learning, generalisation, memory — may themselves be substrate-independent.
The paper explicitly disclaims any claim that the organoid arrays are conscious, sentient, or possessed of subjective experience. The authors state that the organoids “exhibit the behavioural criteria that are standardly used in the animal cognition literature to identify learning,” and that the philosophical and ethical implications “warrant explicit attention from the research community.”
Significance
The study extends the DishBrain precedent (Kagan et al., 2022) in three directions: increased organoid complexity (three-dimensional tissue organisation rather than monolayer culture), increased task complexity (two-dimensional spatial navigation with novel environment generalisation versus one-dimensional Pong), and extended feedback duration (72 continuous hours). The introduction of the Physarum polycephalum comparison as a theoretical frame — whether the behavioural criteria for adaptive cognition can be met by systems sharing no architectural homology — represents a challenge to foundational assumptions in neuroscience about the relationship between neural substrate and cognition.
Related Pages
- Mir-Tian Space Station — Life Sciences Module and microgravity organoid culture
- China — for Beijing Institute of Genomics co-authors
- Soviet Union — for Novosibirsk Institute of Cytology and Genetics