Phillipp Hehrmann

PhD student

Gatsby Computational Neuroscience Unit, UCL

Alexandra House

17 Queen Square

London WC1N 3AR


Research Interests

Pitch Perception as Bayesian Inference (with Maneesh Sahani)

Pitch is a salient perceptual attribute of many environmental sounds. Not just a fundamental dimension of music, pitch carries important information in speech and provides strong cues for the parsing of complex auditory scenes. Despite more than a century of intense scientific investigation, the processes involved in transforming sound-evoked neural activity in the auditory nerve into our percept of pitch is still not well understood.

In my research, I try to relate the psychophysics of pitch perception to the framework of Bayesian probabilistic inference. From this perspective, pitch can be understood as the statistically optimal estimate of the periodicity of naturalistic sounds, indirectly observed through the corresponding evoked firing patterns in the auditory nerve. Our theory provides a computationally unified account of multiple facets of pitch perception that together have previously been interpreted as evidence for the existence of multiple pitch mechanisms, each operating on different physical cues.

Adaptive Coding of Auditory Spatial Cues (with David McAlpine)

Sensory neurons across many species and modalities adapt their responses to the longer-term characteristics of the stimulus. In many cases, these changes have been shown to improve the accuracy of the neural stimulus representation.

We investigate adaptive coding of interaural time delays, a major cue for horizontal sound localisation, in the guinea pig auditory midbrain. Specifically, I am trying to link experimentally observed changes in neural response tuning curves to the underlying neural circuit in the midbrain through the use of computational models.