The experimental work consisted in developing a behavioural protocol to evaluate both memory and executive function using navigation tasks (Starmaze and Y-maze) in order to test different mouse models related to Alzheimer's disease. As a first step, a Starmaze protocol and the creation of scores adapted to the evaluation of desired cognitive functions were developed. In a second step, three mouse models were tested with the protocol developed: a model of the familial form of Alzheimer's disease (APPPS1) at 14 months and a risk factor model of the sporadic form of Alzheimer's disease (APOE4) at 6 and 14 months by comparing them to control mice of the same age. The effect of normal aging on the cognitive functions assessed was also evaluated by comparing the navigation behaviours of control mice (C57BL/6) at 6 and 14 months. The results show that normal aging does not lead to cognitive deficit at 14 months, while 14-month-old APPPS1 mice have memory and executive deficits. ApoE4 mice do not show memory deficits at 6 and 14 months but have a similar flexibility deficit at 6 months as observed in elderly control mice, suggesting that the risk factor ApoE4 could be a model of early aging. My work has validated the use of navigation tests to jointly assess memory and executive function, opening up clinical applications for this type of test in Alzheimer's disease screening.