The dynamics of changes in the fractal dimension of the EEG in the process of solving cognitive problems on mathematical models of neural networks were simulated. Two models showed dynamics of changes in the total activation of neurons in the model similar to those observed in the experiments, consisting in a smooth decrease in the fractal dimension when approaching a successful solution to the problems. The results obtained may indicate the universal nature of this phenomenon, which reflects the systemic energy‑ informational transformations of the model in the process of effective learning.