In the brain, information processing is defined by the properties of neurons and their interaction via structured synaptic connections. In the human cortex layer 2-3, recent studies have uncovered a greater pyramidal neuron diversity and divergent synaptic properties compared to rodents. However, to what extent these synaptic properties are related to cellular heterogeneity and how connectivity is structured in the human cortex is unknown.
In this talk, I will present unpublished data on 1214 electrophysiologically characterized pyramidal neurons and 1419 identified monosynaptic connections from 23 individuals. Using multi-neuron patch-clamp recordings, we observed functional subtypes in the human cortical layer 2-3 that exhibited substantial heterogeneity in their cellular and synaptic function within single individuals. We further identified multiple network principles that are in stark contrast to previous rodent studies, suggesting a directed and mostly acyclic graph topology of the human cortical microcircuit. Analysis and simulation of neural network models applying these wiring principles showed an increased dimensionality of network dynamics and improved performance in a speech recognition task. Taken together, our cellular and synaptic data attributes the human cortical microcircuit with the capacity for high-dimensional neural computation.