Probing the Underlying Mechanisms of Spectro-Temporal Modulation Discrimination

Speech perception relies critically on the processing of spectro-temporal modulations (STMs), yet the mechanisms supporting STM perception and their role in explaining individual variability remain incompletely understood. This study employed a reverse-correlation paradigm, previously used for STM detection, to investigate how listeners discriminate between upward- and downward-moving STMs of identical temporal and spectral rates. Normal-hearing participants completed the task under three surround conditions: (i) no surround sound, (ii) surround sound consisting of the upward-moving STM and (iii) surround sound consisting of the downward-moving STM. Each stimulus was perturbed with structured spectro-temporal noise, from which perceptual kernels were derived. The data showed that listeners primarily weighted upward-moving targets, showing limited evidence of simultaneous monitoring of both directions. Sensitivity was selectively reduced in the downward-moving surround condition, suggesting orientation-specific interference. Simulations with a modulation filterbank model under different configurations, as well as behavioural performance differences across different experimental conditions, further supported the conclusion that participants relied predominantly on upward-target templates, with phase sensitivity additionally playing a critical role in replicating behavioural patterns. Nonetheless, substantial variability was observed in individual perceptual kernels that could not be captured by the model in its normal-hearing configuration. Overall, the results provide new insights into the mechanisms of STM discrimination and demonstrate the utility of reverse-correlation approaches for evaluating auditory models and probing perceptual decision strategies.