Acetylcholine-Mediated Neural Mass Model for Episodic Memory

Episodic memory is the cognitive skill for the spatiotemporal learning of sequential events. This inborn ability requires the participation of several brain regions, including the entorhinal cortex, the medial septum–ventral diagonal band of the Broca complex, and the hippocampus. Its correct functioning can be explained by means of encoding and retrieval, two separate processes occurring at distinct time intervals: episodic sensorial information is firstly encoded in the hippocampal neurons, then a portion of the previously memorized information, maintained in the working memory, can trigger the retrieval of the entire episode. Unfortunately, the whole mechanism is not completely understood yet, but new studies suggest that acetylcholine (ACh) could play an important role in the regulation of these two phases. In this thesis a neural mass model, based on the superimposition of theta and gamma rhythms, is proposed to simulate episodic memory based on septal cholinergic release, allowing for the simultaneous encoding and retrieval of temporally coded features. The model exploits different phases of the theta rhythm to avoid interference between the two mechanisms. Hebbian and anti-Hebbian learning methods are employed in the training phase to set the synaptic weights linking the computational areas together. Results, obtained through robustness and sensitivity analyses performed on the parameters defining acetylcholine behaviour, show good resistance to changes, fostering preliminary tests on clinical cases related to damages in cholinergic neural activity. Furthermore, this model was effectively used in simulations of low-ACh physiological states such as quiet waking and slow-wave sleep, demonstrating its reliability for both the theta and non-theta conditions.