Normative neurophysiology. (360G-Wellcome-095621_Z_11_Z)

Adaptive behaviour is brought about by the complex dynamical interactions of neurons and neural circuits in the nervous system. Much research has focussed on the biophysical elements of neuronal dynamics, such as action potential generation, synaptic transmission, and plasticity. However, in order to understand the link between neuronal phenomena and adaptive behaviour, a key question to be asked is how this plethora of basic biophysical mechanisms endows neurons and neural circuits with the cap acity to process information in highly efficient ways. I propose to study this key question systematically, starting from the central observation that neural circuits need to operate in the face of fundamental informational bottlenecks. My working hypothesis is that many aspects of neural circuit dynamics can be understood as adaptations towards overcoming these bottlenecks. This hypothesis is original and provocative because it implies that observations from widely different levels of neural organisation can be explained by a single underlying computational principle. Moreover, this research will require a unique combination of rigorous mathematical approaches for formalising relevant aspects of 'information processing' with an appreciation for the detailed biological constraints under which computations are carried out by the nervous system. We will study three, hierarchically nested, bottlenecks implied by 1) the conversion of analogue membrane potentials into digital spike tr ains, 2) the conversion of distributed patterns of neural activity into synaptic efficacies, and 3) the conversion of features of the environment into distributed patterns of neural activity. In turn, the specific questions we will address are how these bottlenecks are overcome by 1) excitable dendrites, and interactions between short- and long-term forms of synaptic plasticity and the plasticity of dendritic excitability, 2) the way long-term synaptic and structural plasticity is matched to the statistics and dynamics of neural activities, 3) controlled forms of variability in cortical responses.