Aperiodic Exponent of Brain Field Potentials Is Dependent on the Frequency Range It Is Estimated

The aperiodic component of brain field potentials (EEG, LFP, intracortical recordings) is increasingly being recognized as an important topic in both basic and clinical neuroscience. Aperiodic activity is modeled as a power law of the power spectral density, with the aperiodic exponent proposed as a marker of the balance between excitatory and inhibitory activity. While an ideal power law would apply across frequencies, recent evidence suggests that low- and high-frequency ranges may not present the same aperiodic exponent. To test this, here we analyzed human resting-state intracortical recordings from 62 patients, estimating aperiodic parameters with two complementary estimation methods, Specparam and IRASA. We further validate these results using synthetic data. We systematically observed that the aperiodic exponent depends on its estimation frequency range: low frequencies displayed flatter spectra than high frequencies. This was consistent across estimation methods. The capacity of both methods to accurately estimate aperiodic exponents that vary across frequency ranges was demonstrated in silico. Our results show that the aperiodic exponent depends on its estimation frequency range, highlighting the need for caution when comparing exponents across studies and encouraging further research on the functional meaning of frequency-specific aperiodic estimates.