Early neural encoding of pitch drives cue weighting during speech perception.

Linguistic categories are conveyed in speech by several acoustic cues simultaneously, so listeners need to decide how to prioritize different potential sources of information. There are robust, replicable individual differences in how cues are weighted during speech perception, but the underlying mechanisms explaining this variability remain unclear. Here, we tested the hypothesis that the reliability of early auditory encoding of a dimension is linked to the weighting placed on that dimension during speech categorization. This hypothesis was tested in 60 first language speakers of Mandarin-learning English as a second language. Neural tracking of changes in the pitch contour of naturalistic speech was assessed using EEG, while speech cue weighting was behaviorally tested using word emphasis, lexical stress, and phrase boundary categorization tasks. We find that neural tracking of pitch is linked to pitch cue weighting during word emphasis and lexical stress perception. Specifically, higher pitch weighting is linked to increased tracking of pitch at early latencies within the neural response, from 15 to 55 ms. These results suggest that individuals' perceptual strategies are shaped by the reliability of encoding at early stages of the auditory system.