Dr. Michael Levin on Lex Fridman: How Cells Can Be Persuaded

Levin insists that claims about intelligence or consciousness should be tied to concrete interaction protocols and experiments (e.g., training, perturbations, barriers) instead of philosophical definitions, so we can be right or wrong about what tools work with a given system.

He proposes measuring minds by the scale of goals they can actively pursue across space and time—their ‘cognitive light cone’—and by how easily they can be reprogrammed or persuaded, which lets you place bacteria, dogs, humans, and future AIs on the same continuum.

Mechanisms like diffusing stress signals and electrically shared memories (via gap junctions) let cells extend concern beyond themselves and act as integrated agents, enabling development, regeneration, and also explaining how cancer can be seen as a collapse of that enlarged self.

Levin’s lab has shown non-neural systems can perform forms of learning (habituation, sensitization, associative conditioning) and goal-directed problem solving, implying that behavioral science tools can (and should) be applied well beyond brains.

By perturbing classic sorting algorithms, his group found behaviors like delayed gratification and spontaneous clustering of ‘like with like’ (algotypes) that were not explicitly coded, illustrating that even minimal deterministic systems can display nontrivial, behaviorist-recognizable patterns.