Visual anagrams reveal high-level effects with 'identical' stimuli.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Visual Perception

Background:

• High-level properties (e.g., animacy, emotion, real-world size) are thought to be automatically encoded and organized by neural responses.
• A key challenge in perception research is distinguishing effects of high-level properties from those of lower-level visual features, as they often covary.

Purpose of the Study:

• To develop and apply a novel method to investigate the independent representation of high-level visual properties, specifically real-world size.
• To overcome the challenge of confounds between high-level properties and low-level visual features in perception research.

Main Methods:

• Introduced 'visual anagrams,' a diffusion-based technique creating images with radically different interpretations based on orientation.
• Generated visual anagrams depicting canonically large objects in one orientation and small objects in another.
• Employed classic experimental paradigms to assess the impact of real-world size under these manipulated conditions.

Main Results:

• Demonstrated that many effects attributed to real-world size persisted even when visual cues were systematically altered using visual anagrams.
• Found that not all size-related effects were independent of the manipulated visual features, indicating nuanced perception.

Conclusions:

• The visual anagram technique provides a powerful and broadly applicable tool for dissociating high-level and low-level feature processing in psychology and neuroscience.
• The findings suggest a robust, though not absolute, representation of real-world size in visual perception, independent of certain low-level visual confounds.