This research axis investigates a fondational psychoacoustic ability of humans, namely, the perception of modulated sounds embedded in a background noise, primarily through auditory reverse-correlation experiments. Targets include amplitude-modulated signals, frequency-modulated signals, upward or downward ripples. These stimuli are critical for understanding the human auditory system as they constitue the basic building blocks of natural sounds, including speech. The empirical data collected from human participants is systematically compared with predictions from computational models of the human auditory system, e.g., to determine what are the mechanisms at play, estimate the temporal resolution of the human ear, or how our brain builds templates of a signal to be detected.

Our experiments revealed several forms of suboptimality in the auditory system’s processing of temporal modulations. First, participants exhibited limited sensitivity to envelope phase, yet some of them could still utilize phase information to some extent when the task demanded it. These interindividual differences could be relevant when trying to explain varibilitity in more complex listening tasks such as speech perception. Second, we found that the mental representations involved in these tasks include not only the target signal but also the background noise. However, these representations do not adapt dynamically to changes in noise level, highlighting a potential constraint in auditory processing.

Administrative details:

Collaborators: Christian Lorenzi, Emmanuel Ponsot

Fundings: Agence Nationale de la Recherche, projet fast-ACI (2021-2023)

Selected publications and presentations:

Scientific outreach

Varnet, L., Lorenzi, C. (2021). Probing AM detection in noise with reverse correlation. ARO 2021 (abstract, presentation, video)