Independent effects of valence and memorability in visual statistical learning.

Negative valence images benefit memory tasks as well as visual statistical learning (VSL) tasks. Another factor that strongly drives memory performance: image memorability—an inherent property of the image regarding its likelihood to be remembered across observers. Here, we examined the influence of image memorability and emotional valence on VSL and the potential role of memorability explaining the benefit of negative valence in VSL tasks. In three VSL experiments, participants viewed a stream of repeating triplets of images. Importantly, the triplets were constructed of varying images in their emotional valence and memorability scores. Memorability significantly influenced VSL performance, with high-memorability images enhancing learning compared to low-memorability images. The benefit of negative valence was also replicated but emerged only under low-memorability conditions. Similarly, when valence was held constant, the effect of memorability was evident only for neutral images. These findings indicate that memorability is a robust and consistent driver of VSL. However, its influence is not fixed, it can be modulated by emotional valence and cannot fully account for the benefit seen with negative emotion. Taken together, our study provides new insights into the role of image memorability in VSL and highlights the importance of considering its interaction with emotional valence in incidental visual memory.