This paper introduces a framework of the human deductive reasoning and its ACT-R based implementation called Human Reasoning Module. HRM provides a unified view of rule- and mental model-based strategies of deductive reasoning. The paper explores the role of bottom-up visual processes and implicit knowledge in mental model-based strategy. The validity of HRM is tested with cognitive models of two tasks involving casual deduction and reasoning on spatial relations. Via exploration through these models, it is shown how HRM combined with ACT-R's declarative memory give rise to a pragmatic reasoning.