Abstract

Rapid progress in technologies such as calcium imaging and electrophysiology has seen a dramatic increase in the size and extent of neural recordings. Even so, interpretation of this data often depends on manual operations and requires considerable knowledge about the nature of the representation. Decoding provides a means to infer the information content of such recordings but typically requires highly processed data and prior knowledge of the encoding scheme. Here, wedeveloped a deep-learning-framework able to decode sensory and behavioural variables directlyfrom wide-band neural data. The network requires little user input and generalizes across stimuli,behaviours, brain regions, and recording techniques. Once trained, it can be analysed to determineelements of the neural code that are informative about a given variable. We validated this approachusing data from rodent auditory cortex and hippocampus, identifying a novel representation ofhead direction encoded by putative CA1 interneurons.