Stroke is a leading cause of adult disability worldwide, commonly resulting in motor and cognitive impairments that significantly hinder daily activities, particularly dual tasks like talking while walking. Dual-task training is essential in stroke rehabilitation as it targets both motor and cognitive challenges, improving functional independence, community participation, and quality of life. Gait training is a key rehabilitation focus, though even after regaining walking ability, patients remain vulnerable to interference during multitasking. Motor learning which is critical in stroke recovery occurs in three phases: acquisition, consolidation, and retention, with different brain regions involved in each phase. Stroke-related brain damage affects these phases differently but may be mitigated by neural compensation. Two motor learning strategies-explicit (conscious, instruction-based) and implicit (automatic, unconscious) offer tailored approaches. While both methods can improve motor tasks, it’s still unclear which is more effective for dual-task situations. There is limited research comparing these approaches in real-life tasks as most studies have been done in non clinical tasks such as serial reaction time task. This study is aims to find out which learning strategy works best for improving dual-task performance of gait cognitive dual task after stroke, helping to guide better rehabilitation.

Methodology- This is a crossover experimental study. 18 participants who meet the selection criteria will participate in the study. Convenience sampling and then random allocation will be done. Pre training dual task measurement will be done for all participants. Each group will have nine participants. Group A will receive implicit training first, followed by explicit training after 48 hours, while Group B will follow the reverse order. All participants will undergo both types of training, with the sequence depending on their group assignment. Post training dual task measurement will be done.

Statistical analysis-Statistical analysis will be conducted at a 95% confidence level, with significance set at P < 0.05. Data normality will be assessed using the Shapiro-Wilk test; if normal, a paired t-test will be used, otherwise the Wilcoxon signed-rank test will be applied.