Visual memorability in the absence of semantic content

Area of Science:

• Cognitive Psychology
• Visual Perception
• Memory Studies

Background:

• Past research on image memorability focused on observer characteristics (age, emotion).
• Recent work introduced image-based memorability, the likelihood of an image being remembered across individuals.
• The role of semantic content versus visual features in image memorability remains debated.

Purpose of the Study:

• To investigate whether purely visual features, independent of semantic content, contribute to image memorability.
• To determine if visual properties alone can predict how likely an image is to be remembered.
• To differentiate between semantic and visual contributions to memory for images.

Main Methods:

• Phase scrambling was used to transform natural scene images, removing semantic content while preserving visual properties.
• Memorability of both intact and phase-scrambled images was tested using a recognition memory paradigm.
• Participants responded to repeated images in sequences, allowing for the assessment of short-term and longer-term memory.

Main Results:

• Reliable purely visual memorability was observed for both short-term memory (2-15 seconds) and longer-term memory (minutes).
• The memorability of phase-scrambled images could not be explained by the ability of observers to generate semantic labels for them.
• Visual features alone significantly predicted image memorability, irrespective of semantic interpretation.

Conclusions:

• Image memorability is a function of both semantic content and inherent visual characteristics.
• Perceptual features of an image play a crucial role in its memorability, separate from its meaning.
• This research highlights the importance of visual aesthetics and structure in memory formation and recall.