Leveraging heterogeneity for neural computation with fading memory in layer 2/3 cortical microcircuits

Computational studies addressing the dynamics and computational properties of biologically inspired spiking neurons and networks tend to assume (often for the sake of analytical tractability) a great degree of homogeneity in both neuronal and connectivity parameters. The biophysical reality, however, is radically different from a homogeneous system and multiple levels of complex heterogeneous properties co-exist and shape a local circuit's emergent collective dynamics and information processing properties. Despite their varying molecular, morphological and physiological features, cortical modules can be seen as variations on a common theme. The combined complexity of the circuit's heterogeneous building blocks can be used to provide a rich dynamical space, suitable for online information processing.In this study, we set out to systematically evaluate the role played by different sources of heterogeneity (structural, neuronal and synaptic) in the characteristics of population dynamics and the circuit's capacity for online stimulus processing with fading memory, using cortical layer 2/3 microcircuits as a core inspiration for the circuit specification. We cross-reference various sources of experimental data regarding the composition and patterning of these microcircuits, accounting for different phenomena of interest (e.g. neuron types and corresponding sub-threshold characteristics, conductance properties of different receptor types, circuit-level connectivity and activity statistics, etc.), across different cortical regions, assuming a certain degree of generalization is possible. The methods applied in this study to quantify the dynamics and generic processing properties, being system-independent, can provide a valuable set of tools for microcircuit benchmarking. As carefully curated and organized datasets become increasingly available, it will become possible in the near future to apply increasingly realistic constrains and comparatively study the properties of realistic microcircuits, built to model specific cortical regions and input-output relations