Abstract: Artificial light at night (ALAN), as an emerging pollutant, disrupts wild animals' nocturnal behaviours and physiological processes. Recent evidence indicates that ALAN can also impair diurnal cognition, especially in highly developed vertebrates. However, previous research has rendered mixed results across taxa and task types, with the parameters of the light source also scattered. That limits conclusion generalizability. Here we examined cognitive impacts of ALAN in housed Java Sparrows (Lonchura oryzivora), focusing on two questions: (1) whether ALAN uniformly impairs diverse cognitive traits and (2) how correlated colour temperatures (CCTs) modulate these effects. Sparrows were exposed to amber light (low CCT), neutral-white light (medium CCT), blue light (high CCT), or a no-light control. We then compared individual performance in three cognitive paradigms which were used to assess the animals' capacities of discrimination learning, reversal learning, and inhibitory control. Results showed no significant effects of ALAN on discrimination learning, but ALAN-exposed birds required fewer trials in reversal learning. Lower CCT (amber light) led to more failures in detour-reaching. These findings indicate that cognitive impacts of ALAN are not uniformly negative but depend on cognitive function and CCT. Our study highlights context-dependent effects of ALAN, providing insights for optimizing urban lighting policies to balance ecological and human needs.