Exploring expression space: adaptation to orthogonal and anti-expressions

Area of Science:

• Cognitive Neuroscience
• Computational Vision

Background:

• Facial expression perception is crucial for social interaction.
• Previous models often treated expressions discretely, lacking a unified dimensional framework.
• Facial identity is known to be represented in a multidimensional
• face space

Purpose of the Study:

• To investigate if naturally occurring facial expressions are encoded within a multidimensional
• expression space
• analogous to facial identity representation.
• To compare the coding of naturalistic expressions with established models of facial identity.

Main Methods:

• Analysis of naturalistic facial expression sequences to extract optic flow variations.
• Application of principal component analysis (PCA) to define dimensions of an
• expression space
• .

Main Results:

• Adaptation to a specific expression selectively biased perception towards its
• anti-expression
• along the defined PCA axes.
• This perceptual bias was directional, supporting a multidimensional coding scheme.
• No significant bias was observed in orthogonal directions.

Conclusions:

• Naturally occurring facial expressions can be modeled within a multidimensional framework.
• This
• expression space
• is analogous to the
• face space
• used for identity.
• Suggests a unified representational system for facial features.