Toward Controlling Biology with Language: Offline Learning of Prompt-Conditioned Interventions for Cells, Organoids, and Biobots
Toward Controlling Biology with Language: Offline Learning of Prompt-Conditioned Interventions for Cells, Organoids, and Biobots
This research introduces a method to interface natural language with living matter by learning offline mappings between text prompts and electrical pulse interventions for xenobots. By re-tracking an archive of 155 recordings using two independent methods and replacing hand-made labels with a blind multimodal language model critic, the authors show that a prompt-to-intervention model can generalize to unseen instructions and bots. The approach establishes a calibrated interface for three controllable behaviors while explicitly identifying those it cannot control, without requiring new wet-lab experiments for each instruction.
The study demonstrates an offline learning framework that maps natural language instructions to electrical pulse durations for controlling xenobot behaviors, validated against a re-tracked archive of 155 recordings using both measured labels and blind language model scoring.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2610.02247v2 Announce Type: replace-cross Abstract: Natural language has become an interface to code, images and molecules, but not to living matter: a stimulus applied to cells has no fixed linguistic meaning, and every paired example of an instruction and its outcome costs a wet-lab experiment. We ask how far an existing archive of experiments can take this interface, using xenobots, motile multicellular cons
Open source S2
tructs made of frog cells, stimulated by a single electrical pulse of varying duration. We re-track 155 archived recordings with two independent methods, measure what the pulse does from each bot's own behaviour before and after it, and separate behaviours the pulse controls from behaviours it does not. Longer pulses stop bots and short pulses leave them moving, sometimes faster; turning direction and starting from r
Open source S3
est do not depend on duration, and restarting at most weakly. We then replace hand-made behaviour labels with a blind language critic: fresh sessions of a multimodal language model read plots of each bot and score how well its behaviour matches each of 176 instructions. The critic agrees with the measured labels whenever its plots contain the information the labels are defined on, and dose curves learned from its sco
Open source S4
res match those learned from the measurements. A prompt-to-intervention model trained only on critic scores maps unseen instructions, including negations, to the right end of the dose range on unseen bots, as well as the same model trained on measured labels. The archive therefore supports a calibrated, refusing language interface for the three behaviours a pulse can set, and identifies, rather than hides, the behavi
Open source S5
ours it cannot.
Open source S6
Source：arXiv Robotics — research abstracts · arxiv.org