Comparing Surrogates to Evaluate Precisely Timed Higher-Order Spike Correlations

Abstract

The generation of surrogate data, i.e., the modification of data to destroy a certain feature, can be considered as the implementation of a null-hypothesis whenever an analytical approach is not feasible. Thus, surrogate data generation has been extensively used to assess the significance of spike correlations in parallel spike trains. In this context, one of the main challenges is to properly construct the desired null-hypothesis distribution and to avoid altering the single spike train statistics. A classical surrogate technique is uniform dithering (UD), which displaces spikes locally and uniformly distributed, to destroy temporal properties on a fine timescale while keeping them on a coarser one. Here, we compare UD against five similar surrogate techniques in the context of the detection of significant spatiotemporal spike patterns. We evaluate the surrogates for their performance, first on spike trains based on point process models with constant firing rate, and second on modeled nonstationary artificial data to assess the potential detection of false positive (FP) patterns in a more complex and realistic setting. We determine which statistical features of the spike trains are modified and to which extent. Moreover, we find that UD fails as an appropriate surrogate because it leads to a loss of spikes in the context of binning and clipping, and thus to a large number of FP patterns. The other surrogates achieve a better performance in detecting precisely timed higher-order correlations. Based on these insights, we analyze experimental data from the pre-/motor cortex of macaque monkeys during a reaching-and-grasping task.

Model of multiple synfire chains explains cortical spatio-temporal spike patterns

“…In order to keep the model as simple as possible, we did not include the significance testing performed in SPADE [51,54] either. Modeling it more closely could also improve the match of the pattern size distributions, however, it is almost impossible without fully dynamical simulations, since the rates of the neurons and thus the numbers of occurrences of the patterns are important factors required for the significance test.…”

“…This is achieved by applying a Frequent Itemset Mining algorithm [52,53]. The second step is the statistical evaluation of the significance of the patterns detected in the first step, under the null-hypothesis of mutual independence of spike trains given their firing rate (co-)modulations [54]. The third step is a conditional test performed on all significant patterns, in order to remove patterns arising from the overlap of true pattern spikes and chance spikes.…”

Third-order motifs are sufficient to fully and uniquely characterize spatiotemporal neural network activity

“…For future applications, when testing specific hypotheses using experimental data, a more nuanced surrogate (e.g. [37, 38, 39]) should be used: e.g. a stimulated neuron will not obey a simple Poisson process.…”

Detection and Removal of Hyper-synchronous Artifacts in Massively Parallel Spike Recordings

“…To estimate the complexity distribution of independent data, we use surrogate data, i.e., modified versions of the original data where spike times are intentionally altered. Stella et al (2022) compared a number of surrogate methods, and according to that, we employ here the 'time-shift' method (with a 30 ms shift width, (Pipa et al, 2008)), by which the spike trains are randomly shifted in time against each other. Time shifting destroys potential correlations between spike trains, while conserving many other features of the data such as the inter-spike interval distribution, the firing rate modulations and autocorrelation (Stella et al, 2022).…”