Dynamics of nonlinear dendritic computation in cortical pyramidal neurons and interneurons – DyndriteSymphony

The computational power of biological neurons in the mammalian cortex far surpasses that of their artificial counterparts due to their intricate dendritic structures and dynamic, active mechanisms. These structures give rise to a complex interplay between gradient and all-or-none mechanisms allowing neurons to perform sophisticated computations beyond simple integration. Despite accumulating evidence of complex dendritic behaviour, neurons, especially inhibitory interneurons, are still generally regarded as simple integrators. We propose an alternative view whereby the dendritic tree is considered as a recurrent network-like structure with intra-dendritic dynamics that govern the transformation of thousands of synaptic inputs into axonal output. To investigate this hypothesis, we propose to use advanced optical/optogenetic techniques in vitro and in vivo to evaluate the impact of evoked synaptic patterns on dendritic dynamics. Additionally, we propose to use detailed synthetic models reconstructed based on realistic neuron morphologies and biophysics to examine how neurons process these inputs efficiently to generate their output. We will develop a mathematical theory considering dendritic trees as dynamic network structures to provide insight into their possible spatiotemporal behaviours and computations. Finally, we will evaluate how the dendrite as a recurrent network interacts with the cortical microcircuit. This work will not only fundamentally advance our understanding of neuronal computation but also has far-reaching implications for artificial intelligence inspired by the details of biological neurons.