论文部分内容阅读
Understanding how the auditory system recognizes auditory objects such as speech and animal vocalizations is one of the central goals of systems neuroscience.Auditory signals are inherently dynamic.Thus, auditory object recognition requires integration over time, typically in the range of hundreds of milliseconds to a few seconds for speech and animal vocalizations.It has been well established that sub-cortical neurons respond to sound features over much shorter time scales, usually tens of milliseconds.Therefore, the time integration for auditory object recognition must be accomplished in the cortical areas.We propose that cortical neurons connected into chain-like networks are essential infrastructure for auditory object recognition.Cortical neurons receive ascending inputs from the feature detecting sub-cortical neurons.The patterns of these feed-forward connections encode the specific auditory object to be recognized by the cortical neurons.The spiking activity of the cortical neurons, determined by the interaction between the sound-driven sub-cortical inputs and the intrinsic dynamics of the network, signals the recognition or rejection of the sound signal as the encoded auditory object.We will discuss how such encodings can be learned through auditory experience.We suggest that cortical networks are essential for auditory object recognition.