| name | antisymmetric-polyspectral-neural-interactions |
| description | Generalized framework of antisymmetric cross-polyspectral indices for identifying high-order neural interactions. Quantifies cross-frequency coupling while being intrinsically robust to volume conduction artifacts. Applicable to EEG/MEG analysis and personalized mTMS protocol design. Activation: antisymmetric polyspectral, cross-frequency coupling, high-order neural interactions, volume conduction robust, bispectral analysis, trispectral analysis, multi-frequency coupling, mTMS protocol. |
Antisymmetric Polyspectral Indices for High-Order Neural Interactions
A general family of antisymmetric cross-polyspectral indices that quantify harmonic dependencies between multiple frequency components while being intrinsically robust to instantaneous mixing (volume conduction).
Metadata
- Source: arXiv:2605.04636
- Authors: Alessio Basti, Rikkert Hindriks, Ruggero Freddi, Gian Luca Romani, Vittorio Pizzella, Guido Nolte, Laura Marzetti
- Published: 2026-05-06
- Categories: q-bio.NC, stat.ME
Core Methodology
Key Innovation
Conventional cross-frequency coupling metrics lack a robust framework to characterize genuine interactions among multiple time series where a frequency of interest arises from the combination of N components. This work introduces a general family of antisymmetric cross-polyspectral indices that:
- Quantify harmonic dependencies between multiple frequency components
- Are intrinsically robust to instantaneous mixing (volume conduction artifacts)
- Reveal higher-order dependencies that elude standard analytical approaches
Technical Framework
Cross-Frequency Coupling Problem
Given N source signals, their nonlinear combination produces interactions at frequencies:
- f_target = f_1 +/- f_2 +/- ... +/- f_N
- Volume conduction causes zero-lag artifacts that confound standard coupling metrics
Antisymmetric Polyspectral Indices
The framework derives indices based on the cross-polyspectrum:
- P(f_1, f_2, ..., f_{N-1}) = E[X(f_1) * X(f_2) * ... * X(f_{N-1}) * X*(f_1+...+f_{N-1})]
Key property: Antisymmetry ensures that contributions from instantaneous mixing cancel out:
- For purely instantaneous mixing: the antisymmetric component = 0
- For genuine nonlinear interactions: the antisymmetric component != 0
Implementation Steps
- Compute cross-polyspectrum of multi-channel recordings
- Extract antisymmetric component by appropriate index construction
- Test statistical significance against surrogate data
- Map identified interactions to brain network topology
Validation
- Simulation: Validated on simulated cubic nonlinearities with known ground truth
- Empirical EEG: Applied to real EEG recordings, revealing significant higher-order dependencies
- Robustness: Demonstrated intrinsic immunity to volume conduction artifacts
Implementation Guide
Prerequisites
- Multi-channel EEG/MEG time series data
- Spectral estimation tools (Welch method, multitaper)
- Statistical testing framework (surrogate data generation)
Step-by-Step
- Preprocessing: Filter and artifact-correct multi-channel neural time series
- Spectral estimation: Compute cross-spectra between channel pairs/triplets
- Polyspectrum computation: Estimate cross-polyspectrum at target frequency combinations
- Antisymmetric index extraction: Apply antisymmetric construction to isolate genuine interactions
- Statistical testing: Compare against phase-randomized surrogate data
- Network mapping: Map significant interactions to brain connectivity patterns
Applications
- Cross-frequency coupling analysis: Identify genuine phase-amplitude and phase-phase coupling in EEG/MEG
- Volume conduction robust analysis: Distinguish true neural interactions from field spread artifacts
- Personalized mTMS protocols: Enable selective monitoring and modulation of specific multi-frequency network interactions
- Higher-order brain connectivity: Go beyond pairwise connectivity to N-way interactions
- Epilepsy research: Detect pathological cross-frequency coupling patterns
Pitfalls
- Computationally expensive for large N (polyspectrum scales exponentially)
- Requires sufficient data length for reliable spectral estimation
- Statistical power decreases with higher-order interactions
- Interpretation requires careful consideration of frequency resolution and bandwidth
Related Skills
- higher-order-brain-networks
- hypergraph-functional-brain-network
- multi-view-o-information-brain-networks
- brain-higher-order-structures
- dcho-higher-order-brain-connectivity — complementary higher-order connectivity framework
- entropy-brain-connectivity-paths — information-theoretic connectivity analysis
- eeg-foundation-model-adapters