The investigation of community structure in networks is a task of great importance in many disciplines, namely physics, sociology, biology and computer science where systems are often represented as graphs. One of the challenges is to find local communities from a local viewpoint in a graph without global information in order to reproduce the subjective hierarchical vision for each vertex. In this paper we present the improvement of an information dynamics algorithm in which the label propagation of nodes is based on the Markovian flow of information in the network under cognitive-inspired constraints \cite{Massaro2012}. In this framework we have introduced two more complex heuristics that allow the algorithm to detect the multi-resolution hierarchical community structure of networks from a source vertex or communities adopting fixed values of model's parameters. Experimental results show that the proposed methods are efficient and well-behaved in both real-world and synthetic networks.