Topic: natural sounds

Animals, including humans, learn to discriminate between the diverse sensory stimuli in their environments and to generalize their behaviors to new instances of those stimulus types. For example, rain can sound dramatically different as it falls on different surfaces and yet we have no problem classifying it as rain and discriminating it from speech. While we have some understanding of how generalization is done along a single stimulus dimension, such as sound frequency or light wavelength, natural stimuli are identifiable by a combination of dimensions. To understand perception, measuring the

Convolutional neural networks (CNNs) provide a generalizable architecture for encoding models that predict neural activity evoked by natural sounds with much greater accuracy than other established models. However, the complexity of CNNs makes it difficult to discern the computational properties that support their improved predictive power. We used a combination of local linear approximation and dimensionality reduction to measure the tuning subspace captured by a CNN. Single-unit data was recorded using microelectrode arrays from primary auditory cortex (A1) of awake, passively listening ferr

Neural responses throughout the auditory pathway show tuning for modulations in a time-frequency representation of sound, but how these spectrotemporal modulations are encoded in the human auditory cortex remains poorly understood. Classical, linear spectrotemporal receptive field (STRF) models are simple to understand and fit but have limited predictive power, particularly for complex natural sounds in non-primary regions of the human auditory cortex. We measured responses to a diverse set of natural sounds using spatiotemporally precise intracranial recordings from human neurosurgical patien

Normative neural network models provide a valuable tool for answering principled ‘why’ questions about neural structure, function, and organisation from an optimization perspective. However, typical approaches often compare models without controlling for differences in architecture, training dataset or model objective. As a result, it is often difficult to make strong conclusions relating variation in model performance to a given characteristic. Here we ask what factors tend to make models predict neural responses most effectively?To investigate this, we explored how hyperparameter choice impa