# Biological Basis of the Computational Model The provided code simulates an excitatory-inhibitory (EI) neuronal network using the NEST simulator to produce activity sequences. This type of network reflects certain structures and dynamics found in biological neural circuits, particularly in the cortex. Here, we focus on how the key components and design choices of the model correspond to biological principles. ## Neuron Types and Populations The model consists of excitatory and inhibitory neurons, represented by `popE` and `popI` respectively. This reflects the basic organization of cortical microcircuits where excitatory neurons (often pyramidal cells) form the majority, while inhibitory neurons (such as interneurons) modulate network activity. - **Excitatory Neurons:** Typically use the neurotransmitter glutamate which has a depolarizing effect on postsynaptic neurons, promoting action potential firing. - **Inhibitory Neurons:** Typically employ gamma-aminobutyric acid (GABA) which hyperpolarizes postsynaptic neurons, reducing the likelihood of firing. ## Connectivity Patterns ### Gaussian Connectivity The code uses Gaussian connectivity patterns (via `lcrn.lcrn_gauss_targets` function) to establish synapse connections, a common feature observed in cortical circuits. These patterns reflect local dense excitatory connections with decreasing probability and strength over distance: - **E->E, E->I Connections:** Excitatory neurons project to other excitatory and inhibitory neurons, potentially forming motifs similar to those seen in the cortex's connections. - **I->E, I->I Connections:** Inhibitory neurons project to both E and I populations, providing major feedback and feedforward inhibition within the network, serving to tune network excitability and sparseness of activity. ### Landscape and Asymmetric Connections The code includes a `landscape` feature, potentially reflecting heterogeneities or mapped functional specializations in biological tissue, introducing asymmetry in connectivity reflective of real neural circuit organizations. ## Noise and External Input The integration of noise generators (`noiseE`, `noiseI`) to simulate background synaptic input models real-world variability and stochastic nature of synaptic transmission due to both recurrent activity and noise-scale inputs, which are critical for meaningful dynamics in cortical circuits. ## Simulation Details and Dynamics ### Kernel and Simulation Configuration - **Temporal Resolution:** Setting the resolution reflects the temporal scales at which neuronal and synaptic processes occur. - **Neuronal Parameters:** The code sets parameters like synaptic weights and delays, representing the strength and speed of synaptic transmission characteristic of real synapses. ### Warming Up and Stability The model includes a warming-up phase, allowing the network to stabilize before measurements. This is biologically relevant as real neural circuits also require equilibrating inputs before onset of task-related activity or simulations. ## Synaptic Properties - The excitatory synapses typically show weights drawn from a positive distribution (e.g., `p.Ji`), while inhibitory synapses are negative (with an inversion factor, `p.g * -p.Ji`), reflecting the distinct synaptic potentials they mediate. ## Outcome: Activity Sequences Overall, the primary biological objective of this model is to study how sequences of population activity (perhaps akin to those observed in sequences during cognitive tasks or sensory processing) emerge from the connectivity and dynamics specified in these EI networks.