Contextual modulation of auditory responses predicted by statistical inference

Brian Fischer (1) , Sophie Deneve (1)

1 LNC2 - Laboratoire de Neurosciences Cognitives & Computationnelles

Auditory neurons exhibit complex nonlinear and context-dependent responses that are difficult to capture by standard modeling techniques. Rather than using a bottom up approach and characterizing the responses of individual cells as a function of their input, we propose to use a top-down, normative approach to analyzing auditory processing. We develop a model of spiking neurons that perform probabilistic inference to estimate the state of auditory environment from sensory signals. The model predicts a form for the nonlinear, context-dependent modulation of inputs to central auditory neurons. We show that a simple model based on auditory inference can explain the presence of multiple aspects of contextual modulation in both frequency and time that are observed in the auditory system.

Mots clés: Others

Domaines: Neurosciences [q-bio.NC]; Neurosciences

Brian Fischer, Sophie Deneve. Contextual modulation of auditory responses predicted by statistical inference. Cinquième conférence plénière française de Neurosciences Computationnelles, "Neurocomp'10", Aug 2010, Lyon, France. ⟨hal-00553432⟩