Despite large advances in neuroscientific research, understanding the spontaneous replay of patterns of activity and the emergence of diverse collective dynamics in the cortex from neuronal interactions is still a key challenge in neuroscience. This research elucidates how static structural connections may give rise to the emergence of a large repertoire of dynamic patterns in spontaneous brain activity on a large scale.We employ a modular neural model. Each module, consisting of an ensemble of leaky integrate and fire (LIF) neurons, represents a cortical region, to mimic neuronal cortical dynamics. During the learning process, connections among neurons are formed through spike-timing dependent plasticity, allowing the network to learn and propagate spatiotemporal patterns based on white-matter fiber density. In the spontaneous dynamics, both the order parameter, measuring the degree of overlap with the stored patterns, and its fluctuations are measured. We aim to clarify how static modular structural connections can generate a broad spectrum of dynamic behaviors, including bistable regimes with hysteretic first-order transitions, and a critical regime where transitions occur continuously and the diversity of spontaneously emerging patterns is maximized.The implementation of the modular neural network aids in understanding how the high flexibility and the structure-function relationship achieve optimal explanation when the system operates near a critical regime, with scale-free avalanches distribution. The empirical structure-function correlation (from experimental data) can be retrieved in silico when the model is set in a critical regime.