Objectives: Creating a working model of memory by analyzing neural dynamics for implementation in artificial intelligence and further medical purposes. The project focused on the infrastructure in the neural circuitries that perform associative learning functions. System’s capacity and behavior were examined by adjusting parameters, representing biological properties. Methods & Results: A simulator (Mnemosyne) was constructed in MATLAB to imitate neural populations in action. Procedure starting from the translation of the exteroceptive input into an inner language, to constructing links to related nodes were simulated as many serial processing lines in parallel. Learning was modelled to be a diminution in entropy of the system, given against incoming input stimuli and the ability of defining a system even with a subset of its components. A chaotic itinerant recalling mechanism was proposed with a hypothetical unit of crosschecking named temporal checker. Attractor mechanisms, representing neuromodulators were used to modulate intrinsic properties in their environment. It was seen that by adjusting the initial parameters of the neural population, properties of the system such as learning speed, accuracy and robustness could be altered in an observable efficacy. Conclusion: A simulator with adaptable inner parameters can serve as a platform to optimize a neuromorphic artificial intelligence system. It can also be used as an artificial environment to develop theoretical cognitive neurobiology.