Degraded neural coding of temporal fine structure with age predicts effortful listening in multi-talker environments

Middle age represents a critical window for early detection of neurophysiological decline. Hearing loss is increasingly recognized as both an early marker of neural degeneration and a modifiable risk factor for dementia. Yet many adults report difficulty understanding speech in noise despite normal audiograms, highlighting the limitations of current clinical tests that fail to capture the underlying physiology or effort required for real-world listening. Beyond hearing thresholds, speech comprehension in complex environments depends on precise neural encoding of temporal fine structure (TFS) cues that convey pitch and spatial information. Here, we use a noninvasive EEG-based measure of neural phase-locking (frequency modulation following responses or FMFRs) to quantify TFS encoding in young and middle-aged adults with normal hearing thresholds. Middle-aged listeners exhibited reduced FMFR amplitudes and shallower discriminability slopes, reflecting diminished neural synchrony despite preserved hearing thresholds. Using a multi-talker speech task we further found that pupil-indexed listening effort was significantly greater in middle-aged adults despite matched accuracy across groups. Further, increases in listening effort were predicted by decreases in TFS encoding. Together, these results reveal that degraded neural encoding of TFS underlies subclinical listening difficulties and increased cognitive load, establishing the FMFR as a sensitive biomarker of hidden auditory neural decline.