Assessing cross-contamination in spike-sorted electrophysiology data.

High-density silicon probes are widely used to record the activity of large populations of neurons while animals are engaged in complex behavior. In this approach, each electrode records spikes from many neurons, and "spike sorting" algorithms are used to group the spikes originating from each neuron together. This process is error-prone, however, and so the ability to assess spike sorting accuracy is essential for properly interpreting neural activity. The rate of inter-spike-interval (ISI) violations is commonly used to assess spike sorting accuracy, but the relationship between ISI violation rate and sorting accuracy is complex and poorly understood. Here, we describe this relationship in detail and provide guidelines for how to properly use ISI violation rate to assess spike sorting accuracy.