Disentangling neural and physiological contributions to fNIRS signals during active and passive auditory tasks

Functional near-infrared spectroscopy (fNIRS) is increasingly used to study speech processing because it balances spatial and temporal resolution, but fNIRS signals are susceptible to physiological interference, particularly during active-response tasks. This study examined how systemic-physiological signals influence task-evoked activation in an active word-monitoring task compared to a passive-listening paradigm. Three noise-correction models were applied: No-Correction, Short-Channel (SC) Correction addressing extra-cerebral signals, and combined Short-Channel + Physiology (SC + Physiology) Correction that also accounted for systemic physiology. Active-response trials showed negative amplitudes in the No-Correction model, consistent with scalp hemodynamic suppression. SC Correction shifted amplitudes to positive values and eliminated task differences in dorsolateral prefrontal cortex, secondary auditory cortex, and motor regions, suggesting that these task differences were at least partly attributable to superficial physiology. After adding systemic-physiology correction, the active-vs-passive difference in primary auditory cortex was no longer significant, however, the direct comparison between the SC Correction and SC + Physiology models did not reach significance. These findings indicate that active-response fNIRS paradigms are vulnerable to physiological interference and that global-mean-based short-channel correction substantially attenuates this interference and appears to reveal task-evoked activity that the uncorrected signal obscures.