Robotic arms react to the neural activity of a few thousand rat neurons to produce this drawing.

The robot consists of three drawing arms equipped with three colored markers perched over a white canvas. The rat cells are monitored by 60 two-way electrodes that detect electrical activity (neural signals). A computer records and translates the signals into robotic movements. Cell firing determines which part of the canvas to draw on and which markers to use.

Steve Potter, a professor at Georgia Tech in Atlanta, believes this experiment will lead to cell-culture networks in labs resembling those in animal and human brains. Eventually, the robot's sensor data will be sent back through the network of cells as electronic signals. By closing the feedback loop, researchers hope the robot will learn something about itself and its environment. The Georgia team has already developed a robot (dubbed Hybrot) with movements controlled by a cultured neural network said to be capable of adaptive behavior and learning.