Accelerated Spike Resampling for Accurate Multiple Testing Controls

Harrison, Matthew T.

doi:10.1162/neco_a_00399

Cited by 4 publications

(8 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We hope that the observations highlighted here will encourage further development of computationally efficient procedures that can be rigorously understood or rigorously calibrated (Harrison, 2013; Jeck & Niebur, 2015). …”

Section: Discussionmentioning

confidence: 90%

Spike-Centered Jitter Can Mistake Temporal Structure

2017

View full text Add to dashboard Cite

Jitter-type spike resampling methods are routinely applied in neurophysiology for detecting temporal structure in spike trains (point processes). Several variations have been proposed. The concern has been raised, based on numerical experiments involving Poisson spike processes, that such procedures can be conservative. We study the issue and find it can be resolved by reemphasizing the distinction between spike-centered (basic) jitter and interval jitter. Focusing on spiking processes with no temporal structure, interval jitter generates an exact hypothesis test, guaranteeing valid conclusions. In contrast, such a guarantee is not available for spike-centered jitter. We construct explicit examples in which spike-centered jitter hallucinates temporal structure, in the sense of exaggerated false-positive rates. Finally, we illustrate numerically that Poisson approximations to jitter computations, while computationally efficient, can also result in inaccurate hypothesis tests. We highlight the value of classical statistical frameworks for guiding the design and interpretation of spike resampling methods.

show abstract

Section: Discussionmentioning

confidence: 90%

Spike-Centered Jitter Can Mistake Temporal Structure

2017

View full text Add to dashboard Cite

show abstract

“…As before α̂ Q ( X ) is a valid p-value for any n and has α̂ Q , n ( x ) → α ( x ) as n → ∞. For a discussion of α̂ Q , see Harrison (2012, 2013). …”

Section: Monte Carlo P-valuesmentioning

confidence: 82%

“…For some special choices of S and Ω the value of α ( x ) can be computed exactly (Amarasingham, 2004; Harrison, 2005, 2013). More commonly, the value of α ( x ) will need to be approximated.…”

Section: Testing For Fine-precision Spikingmentioning

confidence: 99%

“…

ℙ false(\overset{α}{true} false(bold-italicX; Y^{false(1 false)}, \dots, Y^{false(n false)} false) \leq ε false) \leq ε

for all ε ≥ 0; see Davison and Hinkley (1997); Harrison (2012, 2013). The Monte Carlo Appendix describes how to construct p-values for more complex Monte Carlo sampling schemes.…”

Section: Testing For Fine-precision Spikingmentioning

confidence: 99%

“…Of these, 1.346 × 10 4057 have 57 or more alignments, giving a p-value of α ( X ) = (1.346×10 4057 )/(4.842×10 4058 ) = 0.0278 from Equation (6). See Harrison (2013) for details about how these numbers were computed exactly. If we were trying to control the type I error rate at level 0.05, then we could safely reject the null hypothesis.…”

Section: Testing For Fine-precision Spikingmentioning

confidence: 99%

See 2 more Smart Citations

Spatiotemporal Conditional Inference and Hypothesis Tests for Neural Ensemble Spiking Precision

2015

Self Cite

View full text Add to dashboard Cite

The collective dynamics of neural ensembles create complex spike patterns with many spatial and temporal scales. Understanding the statistical structure of these patterns can help resolve fundamental questions about neural computation and neural dynamics. Spatio-temporal conditional inference (STCI) is introduced here as a semiparametric statistical framework for investigating the nature of precise spiking patterns from collections of neurons that is robust to arbitrarily complex and nonstationary coarse spiking dynamics. The main idea is to focus statistical modeling and inference, not on the full distribution of the data, but rather on families of conditional distributions of precise spiking given different types of coarse spiking. The framework is then used to develop families of hypothesis tests for probing the spatio-temporal precision of spiking patterns. Relationships among different conditional distributions are used to improve multiple hypothesis testing adjustments and to design novel Monte Carlo spike resampling algorithms. Of special note are algorithms that can locally jitter spike times while still preserving the instantaneous peri-stimulus time histogram (PSTH) or the instantaneous total spike count from a group of recorded neurons. The framework can also be used to test whether first-order maximum entropy models with possibly random and time-varying parameters can account for observed patterns of spiking. STCI provides a detailed example of the generic principle of conditional inference, which may be applicable in other areas of neurostatistical analysis.

show abstract

Closed form jitter methods for neuronal spike train analysis

Jeck

Niebur

2015

2015 49th Annual Conference on Information Sciences and Systems (CISS)

View full text Add to dashboard Cite

Interval jitter and spike resampling methods are used to analyze the time scale on which temporal correlations occur. They allow the computation of jitter corrected cross correlograms and the performance of an associated statistically robust hypothesis test to decide whether observed correlations at a given time scale are significant. Currently used Monte Carlo methods approximate the probability distribution of coincidences. They require generating N MC simulated spike trains of length T and calculating their correlation with another spike train up to lag τ max . This is computationally costly O(N MC × T × τ max ) and it introduces errors in estimating the p value. Instead, we propose to compute the distribution in closed form, with a complexity of O(C max log(C max )τ max ), where C max is the maximum possible number of coincidences. All results are then exact rather than approximate, and as a consequence, the p-values obtained are the theoretically best possible for the available data and test statistic. In addition, simulations with realistic parameters predict a speed increase over Monte Carlo methods of two orders of magnitude for hypothesis testing, and four orders of magnitude for computing the full jitter-corrected cross correlogram.

show abstract

Accelerated Spike Resampling for Accurate Multiple Testing Controls

Cited by 4 publications

References 14 publications

Spike-Centered Jitter Can Mistake Temporal Structure

Spike-Centered Jitter Can Mistake Temporal Structure

Spatiotemporal Conditional Inference and Hypothesis Tests for Neural Ensemble Spiking Precision

Closed form jitter methods for neuronal spike train analysis

Contact Info

Product

Resources

About