Abstract
A variety of methods have been used to study visual memory for images including recall tasks, recognition tasks, verbal description, and recently, drawing recall tasks. However, when it comes to recollection of faces, studies have largely failed to capture the precise details during free recall that allow for quantitative analysis. Here, we propose a new method utilizing an avatar creation tool to provide a numerical representation of face recollection. The interface contains a 3D character model, and sliders that range from -1 to 1 for each facial feature. Changes made to features such as nose width or prominence of laugh lines are updated in real-time, so participants can understand the effects of each slider on their representation of a face. Participants were first shown sample human faces that were constructed in the face tool for a period of time (20 seconds) and told to memorize as much as they could. After the face image was presented, they were then shown the character creation interface with instructions to recreate the face as best they could from memory. After this memory task, a perceptual task was conducted where participants had to recreate the faces again, this time with the image available as a reference. We identified key differences between face recall and perception, along with perception and the ground truth in participants’ abilities to reconstruct faces. Specifically, face reconstruction was improved when using a reference as opposed to relying on memory, demonstrating that this interface can be an effective method to capture representations of faces in perception and memory. Overall, the use of this interface allows for a new method to study the recollection of faces with precise quantification of specific features.