The inferior colliculus (IC) is the main converging station in the auditory midbrain and important for processing of complex sounds. However, the functional mapping of natural complex sounds to its neural representation is not yet very well understood, and good modeling approaches would be useful. To evaluate prediction models, we use recordings from groups of neurons in the IC of guinea pigs which were acoustically presented a set of 11 conspecific vocalizations. The different vocalizations display various envelope types and spectral contents. Using cross-correlation, we compare the predicted and recorded temporal neural responses for two approaches. The first model is a modification of the biophysically detailed Meddis model, and the second one is a filtering approach around the neuron's preferred frequency. Surprisingly, we find that for responses to natural sounds from groups of neurons, the filtering approach yields better predictions than the biophysically detailed model. Thus, the collective, integrated response can be well described by a frequency-band selective representation.
展开▼
