Cortical sensory neurons are commonly characterized using the receptive field, the linear dependence of their response on the stimulus. In primary auditory cortex neurons can be characterized by their spectrotemporal receptive fields, the spectral and temporal features of a sound that linearly drive a neuron. However, receptive fields do not capture the fact that the response of a cortical neuron results from the complex nonlinear network in which it is embedded. By fitting a nonlinear feedforward network model (a network receptive field) to cortical responses to natural sounds, we reveal that primary auditory cortical neurons are sensitive over a substantially larger spectrotemporal domain than is seen in their standard spectrotemporal receptive fields. Furthermore, the network receptive field, a parsimonious network consisting of 1–7 sub-receptive fields that interact nonlinearly, consistently better predicts neural responses to auditory stimuli than the standard receptive fields. The network receptive field reveals separate excitatory and inhibitory sub-fields with different nonlinear properties, and interaction of the sub-fields gives rise to important operations such as gain control and conjunctive feature detection. The conjunctive effects, where neurons respond only if several specific features are present together, enable increased selectivity for particular complex spectrotemporal structures, and may constitute an important stage in sound recognition. In conclusion, we demonstrate that fitting auditory cortical neural responses with feedforward network models expands on simple linear receptive field models in a manner that yields substantially improved predictive power and reveals key nonlinear aspects of cortical processing, while remaining easy to interpret in a physiological context.
Linear filter descriptions of sensory neurons have been with us since the 1970s, and have been enormously influential. But such models, and more recent nonlinear variants, are rather like modeling the entire network as a single neuron, failing to account for the neuron's response being a consequence of a network of many nonlinear units. Here we show how these limitations can be overcome by using recent advances in machine learning to fit “network receptive field models” to neural responses to natural sounds. Feedforward networks of 1–7 nonlinearly-interacting lower-order model neurons are required to model a cortical receptive field. Each lower order neuron is tuned to somewhat different stimulus features, arranged together in complex but interpretable structures, which cover a far wider range of sound frequencies and delays than current receptive field models indicate. The NRF models capture important nonlinear functional characteristics in auditory cortical neurons, including multiplicative gain control and conjunctive feature selectivity, where neurons respond when certain features are present together but not in isolation. This enables NRFs to predict the responses of auditory cortical neurons with considerably greater accuracy than conventional models.