Neural Circuits: Models of Emergent Functions

Information about external world and internal actions is represented in the brain by the activity of large networks of heavily interconnected neurons, that is, neural circuits. To be efficient, this representation requires neuronal circuits to perform certain functions which cannot be performed by single neurons. A major factor that determines the emerging functional properties of neural circuits is the architecture and the distribution the connection strengths between the neurons in the network. Mathematically, this is reflected in the global transformation between the external inputs and the network states. The major network models are classified according to the architecture of the connections. In the feed-forward networks, information is flowing unidirectionally along the ordered sequence of distinct layers. These networks directly map their inputs into the outputs. In the recurrent networks, feed-back loops allow the autonomous dynamics of the network. The models of neural circuits were applied to a wide range of phenomena in the brain, such as the formation of feature maps, associative memory, representation of sensory inputs, and motor commands, etc.