The neural code is robust to spike deletions

Multiple sources of noise can affect the encoding of information in the nervous system. We assessed the robustness of the code to spike deletions, simulating failures in information transmission due to spike propagation failures or the stochastic nature of neurotransmitter release. Here, we trained and tested the classifier after randomly deleting a fraction of the spikes (shown in the x-axis). Red=categorization, Blue=identification. Classifier parameters: MUA, n=256 sites, [100;300) ms time interval, bin size = 50 ms.