RNN 2AFC

Trains a continuous time RNN on a 2AFC perceptual decision making task. It then plots a psychometric curve. It then performs PCA on the hidden state, logistic regression decoding of choices and finds the fixed points using fixed-point-finder. It also performs decoding over time during the evidence accumulation and decision phase of the task.

There are 2 notebooks:

1. Visualization - to see the task
2. Analysis - To train and analyse the network