In France, over 1.5 million people suffer from mild cognitive impairment, often a precursor to Alzheimer's disease (AD), whose global prevalence could reach 153 million by 2050. With no curative treatment available, maintaining patients' autonomy at home is essential to mitigate the negative effects of institutionalization and reduce economic costs. Non-pharmacological approaches, such as cognitive training, have shown potential in stabilizing cognitive functions. Research suggests that motor networks and procedural memory remain relatively preserved in early AD, and bodily engagement during encoding enhances memory. For patients with motor impairments, motor imagery (MI) activates these networks without actual movement. Additionally, dynamic visual cues, like videos, reduce cognitive load compared to static images. This project aims to combine real action, MI, and dynamic visual supports to optimize memory. By validating 120 videos of daily activities, it will assess the impact of action on recall, evaluate MI's effectiveness for patients with mobility limitations, and confirm the superiority of videos over static images. The results could support the development of digital tools, such as serious games, to enhance patients' autonomy and address aging-related challenges.