Image and task
Which structures are actually available in the face or image, and what task is being performed? View, illumination, projection, quality, and temporal change determine what can be used.
Overview
How do visual judgements of faces arise, and under which conditions can they be studied systematically?
Age estimation, similarity assessment, and categorical comparison decisions are different tasks. What can be examined across them are effects of image conditions and observer variability.
Depending on the task, form, contour, skin structure, and other facial features may be relevant; viewpoint and image quality also constrain the information available.
Observation and observer variability
Which features do observers use, and how do their ratings differ? Structured ratings, repeated measurements, and comparisons between observers provide ways to examine these questions.
Judgement and confidence
How strongly do judgements vary, how confident are observers, and — for objectively scorable tasks — how well does that confidence correspond to actual accuracy?
Visual age perception
Age is often quickly apparent in faces, but rarely unambiguous. Wrinkles, furrows, hair features, contours, soft-tissue form, and skin texture may suggest age without producing a precise estimate.
Visible age can act as a social cue. Age judgements may draw on several visible features and can vary between observers in both estimate and confidence.
Studying visual age perception can show which cues are associated with observers’ judgements, how strongly estimates vary for individual faces, and how subjective confidence relates to accuracy.
Facial comparison and similarity
Similarity judgements can be influenced by perceived correspondences and differences as well as by the image conditions under which faces are shown. In categorical same/different decisions, the decision criterion introduces an additional component.
Uncertainty, consensus and overinterpretation
Uncertainty may be an appropriate response to limited or ambiguous information. Strong observer variability, low repeatability, or poorly calibrated confidence are therefore informative findings in their own right.