Diversify your workflow! - An inconvenient suggestion to analyze spike data from intracranial recordings

An important challenge of neuroscience research and future brain machine interfacing is the reliable assignment of spikes to putative neurons. By means of extracellular recordings, researchers try to match different types action potentials with their putative neuronal source and timing. Unfortunately, this procedure is by far not standardized and reliable, leading to many different suggestions and as many differing results. It appears that sharing of data is thus hampered by different processing pipelines in different labs, thus playing along the reproducibility crisis in neurosciences. To systematically shed light on this issue, we present preliminary results of several easy event detection schemes on one data set, meant to illustrate the inconsistencies arising from different processing pipelines already in its initial step. The results indicate that thresholding choices alter findings due to a lack of a ground truth for spike sorting. We suggest to increase reliability in findings by only accepting and further processing events accepted by more than one processing pipeline.