In a study published in 2022, their measured activity steered the paddle, and performance improved within game sessions. The experiment does not answer whether the cells experience anything.
Schematic drawing of a neuronal culture, not a microscope image.
Explanation
What can living cells do without the animal?
Outside a complete animal, muscle cells can drive small frames, groups of cells can move on their own, and neurons can be coupled to a technical system. Each of these experiments requires specific culture conditions.
What the experiments show
Groups of cells can produce directed movement. A neuronal culture can be connected through electrodes to a game simulation in a feedback loop.
What they do not show
Movement and better game performance do not demonstrate consciousness. The Pong study nevertheless uses the word “sentience” in its title; other researchers have criticised this choice of word.
Muscle cells as motors
In a 2014 study, Cvetkovic and colleagues combined a soft, printed frame with a strip of skeletal muscle tissue. Electrical pulses triggered contractions. Because the frame was asymmetric, it moved in one direction.
Two years later, Raman and colleagues made light-sensitive muscle constructs. There, pulses of blue light triggered the contraction.
Groups of cells that move on their own
For the xenobots by Kriegman and colleagues, possible body shapes were first explored on a computer. The researchers then assembled selected shapes from embryonic frog cells; skin and heart muscle cells together formed a moving group.
Gumuskaya and colleagues grew human airway cells into small groups. These groups move using their cilia.
How neurons were connected to Pong
In the DishBrain experiment by Kagan and colleagues, neurons sat on an array of electrodes. The position of the ball was translated into electrical stimulation. In return, the measured cell activity moved the paddle in the simulation.
Experiment diagram after Kagan et al. (2022), not measured data. The interactive version is part of the full explanation.
The authors report improved performance measures within game sessions. That is a measurement. Interpreting it as learning is a further step.
Why consciousness is a separate question
The title of the DishBrain study uses the word “sentience”. The authors use it in a narrow, formal sense: being “responsive to sensory impressions” through adaptive internal processes. In everyday language, the word means the capacity to feel. In a scholarly commentary, Balci and other researchers criticised this choice of term and the far-reaching conclusions drawn from it.
We therefore keep apart how the experiment was set up, what was measured, and how the authors interpret it.
What HP states According to HP, Instant Ink cartridges stop working at the end of the billing period in which the plan was cancelled. Whether they work depends on taking part in the service.
What does not follow This concerns this subscription model and is not a statement about every HP printer. HP names regular original HP cartridges as a way to keep printing. The source is HP’s own documentation.
What the study reports A US cohort analysis of 112,284 participants reports a hazard ratio of 2.45 for the highest intake group of sugar-sweetened beverages, compared with people who drank less than one serving a month (95% confidence interval 1.49 to 4.04).
What the number does not mean It is not a 245 per cent probability of developing the disease, and it does not show that the drinks are the cause. Earlier summaries do not show a consistently positive association for stomach cancer.
The key sources behind this page. Studies appear with their original title and the date we last checked them. All research sources are in the source register.
Every claim points to the work it comes from. The source list shows when we last checked it.
The limit right beside the finding
What a study does not show sits next to its finding. On this page, yellow always marks such a limit.
Manufacturer statements labelled as such
A company’s documents tell you what the company says. We label them that way and do not treat them as independent testing.
AI use disclosed
AI can help with drafts, structure, code, and visualisation. It replaces neither an original source nor editorial review. The narration in our videos is AI-generated.
