Research

We design experiments to understand how living neural networks learn, adapt, remember, generalize, and solve computational tasks.

Dino

Can biological neurons make decisions in real time?

>90%

Task accuracy

Living neurons received one piece of information, the distance to an approaching obstacle, and controlled when the dinosaur jumped. No explicit feedback was provided during gameplay.

  • Decision making
  • Real time control
  • Feedforward
View experiment

Asteroid Dodger

Can biological neurons learn through feedback?

86.5%

Peak dodge rate

Living neurons controlled a spacecraft as asteroids approached from above. The network received feedback from successful dodges and collisions, allowing its behaviour to change as it interacted with the environment.

  • Learning
  • Closed loop
  • Reinforcement
  • Continuous control
  • Adaptation
View experiment

Approach

Understanding biological computation.

Our experiments are designed around measurable tasks with defined inputs, outputs, feedback, and controls.

We record how biological neural networks respond, how their activity changes during training, and how learned behaviour develops over time.

Where appropriate, experiments include controls, repeated trials, comparisons across cultures, and quantitative performance measurements.

The goal is not simply to demonstrate that neurons can perform tasks. It is to understand how biological networks compute, learn, and adapt.

AxoGrid

Build your own experiment.

The infrastructure behind our research is being made accessible through AxoGrid. Access living neural networks, BPU hardware, closed loop training infrastructure, and experimental tools through software.

Explore AxoGrid Collaborate with us