Skip to content

The Dishbrain

  • by

Although growing neurons in an artificial environment to observe their network formation is over two decades old, for the first time they have been observed to be trainable to play a “high-level” game; Pong!

Pong (being the very first popular computerized application in the early 1970’s), often serves as the standard merit-badge for new computing technologies. Researchers at Cortical Labs in Melbourne, Australia, have trained a self-organized natural neural-network to play Pong. Given that the neurons self-organize, eliminating tremendous leaps that electronic computers required in design (organization of transistors) can only help the progress of such computers far faster. If this is the 1970’s of the biological computer, it’s hard to imagine its specialized capabilities in a mere two decades. The work was mentioned on 2022-Oct-20 in Nature Vol-610 p.433.

The approach harnessed the property that neurons tend to repeat activity that yields a predictable environment. When the neurons responded in a way that corresponded with hitting the ball, they were stimulated in a location and at a frequency that was the same each time. If they missed the ball, the network was stimulated by the electrodes in random locations and at different frequencies. in other words, “predictability” (lowered entropy) was the incentive, and disorder/randomness (raised entropy) the “punishment”. Over time, the neurons learned to hit the ball to receive the patterned response rather than the random response.

The nerve-cell array called DishBrain at work. The colors mark different types of nerve cell and their components. Credit: Cortical Labs