16. Januar 2018 / Wissenschaftler
Scaling behaviour of neural structure derived from optimal wiring principles
Sprecher: Hermann Cuntz (FIAS & ESI)
Der Vortrag wird in Englisch gehalten.
Neural circuits in different species are required to perform similar computations on a wide range of circuit sizes. To ensure this, the number of neurons or the size of neurons can be increased. Alternatively, an entirely different connectivity can be in use. In order to study the scaling behaviour of neural circuits we use generative models that predict neural structure from simple principles such as optimal wiring. We find that responses to synaptic inputs are in principle independent of dendritic tree length or branching structure. The scaling behaviour of dendritic trees is well captured by connectivity requirements and optimal wiring constraints. Macroscopic features of neural circuits show more structure with increasing neuron numbers as a result of optimal wiring constraints. Our observations using simulations are consistent with the scaling behaviour in biological systems assuming more or less unchanged connectivity across species.