Slime mould is clever stuff. It finds the quickest path between food and has even shown signs of having memory – despite not having a brain. Now we know what faces it might pull.
A human-like robot face has been hooked up so that its expressions are controlled by the electrical signals produced when yellow slime mould shies away from light, or moves eagerly towards food.
It is all part of an experiment aimed at figuring out why these animate puddles of neon goo are so surprisingly smart. Physarum polycephalum, is a common yellow slime mould which ranges in size from several hundred micrometres to more than one metre. It is an aggregation of hundreds or thousands of identical unicellular organisms that merge together into one huge “cell” containing all their nuclei. This makes it much more interesting than your average unicellular bacteria, says Ella Gale who studies unconventional computing at the University of the West of England in Bristol and at the Bristol Robotics Laboratory.
Slime mould’s remarkable problem-solving capabilities are well-documented and include finding the shortest path between different food sources. It also displays memory, in a similar way to a novel electrical component called a memristor, which has in turn been likened to the functionality of biological brains. “It’s one of the simplest organisms that can learn,” says Gale.
Gale placed slime mould on a forest of 64 micro electrodes, along with some oat flakes. As the mould moved across the electrodes towards the food, it produced electrical signals, which Gale converted into sound frequencies. “I got excited and wondered if you could hear cellular communication,” she says. The result was some strange noises, rather like an aeroplane taking off, says Gale.
Together with colleagues, Gale then split the sound data into chunks. They gave each one a positive or negative label depending on whether the slime mould was moving towards food or recoiling from light – plus an “arousal” score based on the volume of the chunk.
Using a popular psychological model, the team was then able to assign each sound chunk an emotion – anger would be negative, high arousal, for example, while joy might be positive, low arousal. Finally, the team then used an expressive, female Jules robot made by Hanson Robotics to re-enact the sequence of emotions while the soundtrack was played.
The installation was unveiled at the Living Machines conference in London on 1 August.
It is not the first time a slime mould has controlled a robot. In 2009, a team led by Soichiro Tsuda, then at the University of Southampton, UK, used a slime mould to control a faceless, six-legged, insect-like robot.
But the latest mould robot is a powerful reflection of Gale’s particular interest in P. polycephalum. “What is the connection between slime mould and the brain?” she asks. “You have two different systems made of different stuff, yet both can learn.”
Source: New Scientist