Do Memories of Inferred Visual Representations Guide Low-level Perception?
Wang, R. L.; Harrison, W. J.
Show abstract
Prior knowledge shapes how we interpret and adapt to a dynamic environment. A crucial aspect of this process is the lifelong development of structured object representations, enabling meaningful and survival-relevant interactions with the external world. In this study, we investigated the extent to which memories of complex objects influence perception by refining visual details. To test this, we had participants hold in working memory Mooney images, a stimulus class that requires top-down processing to perceive hidden image structure. While holding a Mooney image in memory, participants performed a detection task, in which they had to detect an edge feature that appeared at selected locations of the illusory contours of the image. Participants completed this task twice, once before they were shown the hidden Mooney structure, and then after learning the hidden structure. In a signal detection framework, we assessed whether learning the hidden content altered participants sensitivity and response bias in detecting edge-feature targets. We found that, while holding object memories enhanced Mooney image disambiguation, it did not refine visual sensitivity. This dissociation -- where categorical identification improves without corresponding perceptual refinement -- suggests that memories of complex objects improve the overall understanding of ambiguous information independent of refining relevant visual details. These findings have important implications for the theory of recurrent processing, which has traditionally emphasised perceptual refinement in top-down feedback. Our results highlight how prior knowledge improves the perceived clarity of degraded visual information without necessarily improving precision in local feature detection.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
- Objects, faces, and spaces: Organizational principles of visual object perception as evidenced by individual differences in behavior 96%
- Dynamic saccade context triggers more stable object-location binding 95%
- Expectation-Based Gist Facilitation: Rapid Scene Understanding and the Role of Top-Down Information 95%
Similar papers in this journal
- Feature distribution learning by passive exposure 97%
- Meaning maps and saliency models based on deep convolutional neural networks are insensitive to image meaning when predicting human fixations 96%
- Divergent effects of absolute evidence magnitude on decision accuracy and confidence in perceptual judgements 96%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.