Visualised behaviour of dynamic neural networks

The dynamics of simple nonlinear systems can be dramatically beautiful. I visualised a large number of randomly-generated systems as an exploration.

In the process of investigating the relationship between recurrent and feedforward network architectures, I generated a large number of random two-neuron systems. I'd examined these networks before, but not visualised them on such a scale. The behaviour as they explored their phase space ranged from the simple…

…to the evocative…

…to the surprisingly optimistic…

…to the sublime.

These images show starkly the degree of complexity that can arise from even very small numbers of interacting neurons. Now imagine the intricate and convoluted dynamics of a single cubic millimetre of cortical tissue, with 9 kilometres of wiring joining its 2.4 million neurons…