First, the idea.
Biological computing uses living systems as part of an information-processing device. Neural cultures can respond to inputs and change their activity with feedback. Performance on a task is a measurable result; personal consciousness is a different question.
A way to picture it
A system can improve its score on a game without showing that it understands a life story. “Learning” must be tied to the particular task and control conditions.
What the researchers did.
DishBrain links human or mouse neuron cultures to a simplified game environment. The researchers compare structured feedback with control conditions.
Human and mouse neuron cultures altered performance in a simplified Pong-like feedback environment.
Where the conclusion stops.
The authors’ use of sentience is not a demonstration of human-like subjective experience.
Why this matters for the larger question.
Focus on the behavioural definition of learning and the controls. The title’s sentience wording should not be promoted as proof of a conscious person in a dish.
This interpretation is Death X’s editorial assessment. It is separate from the original authors’ findings and is not an endorsement by them.
How to read the evidence.
Look for task definitions, controls, repeatability, useful lifetime and the cost of the complete support system. Compare equivalent tasks before making energy-efficiency claims.
Before relying on the result, inspect the full methods, independent sample counts, comparison groups, uncertainty and available data. The explanation here does not claim that all of those details have been independently appraised.
Three terms, explained.
- Reservoir computing
- Using a system’s changing internal responses as features for a readout.
- Closed loop
- A system in which measured outputs affect later inputs.
- Benchmark
- A defined test used to compare performance.
Check your understanding
What can this source support? Human and mouse neuron cultures altered performance in a simplified Pong-like feedback environment.
What would go too far? The authors’ use of sentience is not a demonstration of human-like subjective experience.
The original record.
- Original title
- In vitro neurons learn and exhibit sentience when embodied in a simulated game-world
- Authors
- Brett J Kagan; Andy C Kitchen; Nhi T Tran; Forough Habibollahi; Moein Khajehnejad; Bradyn J Parker; Anjali Bhat; Ben Rollo; Adeel Razi; Karl J Friston
- Journal or provider
- Neuron
- Publication
- 2022-10-12
- Volume
- 110
- Issue
- 23
- Pages / article number
- 3952-3969.e8
- DOI
- 10.1016/j.neuron.2022.09.001
- PubMed ID
- 36228614
- PubMed Central ID
- PMC9747182
Original record checked: 2026-10-09. No retraction flag in returned Europe PMC record; not a full notice search.
AI-assisted editorial explanation of the linked publication or provider record. No independent scientific reviewer is appointed; full statistical appraisal is not claimed. Suggest a correction.