Estimate the artificial receptive fields (aRFs) of each in silico neuron the models were trained on (Figure 7 and Methods 4.5.1). The aRFs are used to estimate the centre of the receptive field for subsequent analyses.
- predict_noise.py inference the movie-trained model on full-field white noise images and save the predicted responses in
npzfiles. - fit_aRFs.py fit 2D Gaussian to the weighted average response to the white noise images to estimate its aRF.