Wednesday, August 12, 2015

Categorical Perception of Fear and Anger Expressions in Whole, Masked and Composite Faces

Below:  Design of Experiment 1.
Illustration of a face morphed from the original fearful (outer left) to the original angry expression (outer right) in 9 intermediary steps, resulting in a total of 11 face morphs; a, whole face; b, upper half intact ('eyes' condition); c, lower half intact ('mouth' condition); due to copyright restrictions, the depicted example is an in-house generated averaged face based on 16 different identities and not depicting an actual person. This example is a representative illustration of the type of stimuli used but was not itself part of the present experiment.




Below:  Design of Experiment 2 (Composite Face).
Illustration of a face morphed from the original fearful (outer left) to the original angry expression (outer right) in 9 intermediary steps, resulting in a total of 11 face morphs; a, eye judgements with 100% angry lower face half; b, eye judgements with 100% fearful lower half; c, mouth judgements with 100% angry upper face half; d, mouth judgements with 100% fearful upper half; conditions a and b or conditions c and d were always presented in one block, to aid participants in focusing on one face half only; due to copyright restrictions, the depicted example is an in-house generated face which was not used in the present experiment.



Human observers are remarkably proficient at recognizing expressions of emotions and at readily grouping them into distinct categories. When morphing one facial expression into another, the linear changes in low-level features are insufficient to describe the changes in perception, which instead follow an s-shaped function. Important questions are, whether there are single diagnostic regions in the face that drive categorical perception for certain parings of emotion expressions, and how information in those regions interacts when presented together. We report results from two experiments with morphed fear-anger expressions, where (a) half of the face was masked or (b) composite faces made up of different expressions were presented. When isolated upper and lower halves of faces were shown, the eyes were found to be almost as diagnostic as the whole face, with the response function showing a steep category boundary. In contrast, the mouth allowed for a substantially lesser amount of accuracy and responses followed a much flatter psychometric function. When a composite face consisting of mismatched upper and lower halves was used and observers were instructed to exclusively judge either the expression of mouth or eyes, the to-be-ignored part always influenced perception of the target region. In line with experiment 1, the eye region exerted a much stronger influence on mouth judgements than vice versa. Again, categorical perception was significantly more pronounced for upper halves of faces. The present study shows that identification of fear and anger in morphed faces relies heavily on information from the upper half of the face, most likely the eye region. Categorical perception is possible when only the upper face half is present, but compromised when only the lower part is shown. Moreover, observers tend to integrate all available features of a face, even when trying to focus on only one part.

Read more at:   http://goo.gl/79BKBK HT @uniaktuell

No comments:

Post a Comment