I haven't a clue! Expectations based on repetitions and hints facilitate perceptual experience of ambiguous images

Abstract

In recent years, the role of top-down expectations on perception has been extensively researched within the framework of predictive coding. However, less attention has been given to the different sources of expectations, how they differ, and how they interact. In this article, we examined the effects of informative hints on perceptual experience and how these interact with repetition-based expectations to create a long-lasting effect. Over 7 experiments, we used verbal hints and multiple presentations of ambiguous 2-tone images. We found that vividness ratings increased from 1 presentation to the next, even after the object in the image had been identified. In addition, vividness ratings significantly increased when images were introduced with a hint, and this boost was greater for more detailed hints. However, the initial increase in vividness did not always carry over to the next presentation. When recognition of the image in the presentation was hard because of memory load, inconsistent presentation, or noise level of the image, the initial advantage in vividness was attenuated. This was most apparent when participants were primed with a grayscale version of the 2-tone image. A computational model based on evidence accumulation was able to recover these patterns of perceptual experience, suggesting that the effect of hints is short lived if it cannot be encoded in memory for future presentations. This notion highlights the different contributions of attention, memory, and their interactions on forming expectations for perception.