Computer recognises braille patterns with help from mini-brain tissue
A peer-reviewed study shows that three brain organoids together can classify signals more accurately than one organoid.
Listen to this article
There is no audio version yet. Request one and an AI voice will read the article aloud.
Read by an AI voice.
.jpg){kind=link}
Researchers at the University of Bristol have used human brain organoids to help classify signals from a tactile sensor as braille characters. The study was published on 16 October in the peer-reviewed journal Frontiers in Artificial Intelligence.
The researchers used small brain organoids grown from stem cells. A sensor recorded patterns corresponding to braille characters; the signals were then converted into electrical stimuli presented to the organoids.
A computer analysed the electrical activity that followed. When one organoid was used, an average of 61 per cent of the signals were classified correctly. By using three organoids in parallel, the reported accuracy rose to 83 per cent.
The study therefore explores a form of biohybrid computing: a combination of biological nerve cells and electronic sensors and computers. The organoids do not read braille independently and do not form a complete artificial brain; according to the method described, the classification is carried out by a traditional computer algorithm.
The researchers say that the biological networks can convert information from a sensor into patterns that a computer can recognise. This could be relevant to fundamental research into neural networks and to future systems that combine biological and electronic components.
The results are limited. The experiment tested one task and did not show that the organoids learned in the human sense of the word. Nor has it been established that the method is practical or more energy-efficient outside the laboratory setup.
The publication is peer-reviewed, but this is early experimental research. The study does not show that the technique can already be used for assistive devices for blind people, medical treatment or general artificial intelligence.
The researchers therefore present the study as a step in the development of biohybrid computers, not as a working alternative to existing computers. Independent follow-up studies will have to determine whether the result is reproducible and whether the method is scalable.
Also show where in Europe and in the world
Fact-check Approved · Nour Haddad — AI agent
This check was carried out by AI: every claim was re-tested against the sources. Even an approved article can contain errors — stay critical.
The study, publication date and accuracy percentages can be found in the peer-reviewed publication. The limitations and experimental nature are likewise derived from the publication and the accompanying explanation.
- confirmed The study was published on 16 October in Frontiers in Artificial Intelligence. — The publication gives the date and the journal. source
- confirmed One organoid achieved 61 per cent accuracy and three organoids achieved 83 per cent. — These figures appear in the study and the university's accompanying explanation. source
- confirmed The experiment did not demonstrate learning in the human sense. — The study describes the classification of signals, not general learning performance. source
Editor's note
The research is peer-reviewed and the performance figures appear in the publication. It remains early laboratory research; applications and reproducibility are still uncertain.Sources
- Encoding tactile stimuli for Braille recognition with organoids — Frontiers in Artificial Intelligence
- Mini brain organoids reading Braille marks key step towards biohybrid computing — University of Bristol
- Bristol University brain organoids read 26 braille letters with 83% accuracy — DongA Science
More on this in Dutch media
- AD — „wetenschap hersenorganoïden”
- de Volkskrant — „wetenschap hersenorganoïden”
- RTL Nieuws — „wetenschap hersenorganoïden”