Encoding and recognition of faces involve different eye-movement dynamics

Abstract

Previous incidental evidence suggests that eye movement dynamics differ between the encoding and recognition of faces; however, restricted stimulus presentation time during recognition was a potential confound. The present study aims to distinguish the influences of experimental phase (encoding/recognition) and stimulus presentation time (short/long) on eye movements during facial perception. Participants completed an encoding ("study") phase and a subsequent recognition ("test") phase, with each phase divided into separate blocks of either one- or five-second stimulus presentation time conditions. For our participants, the second fixation was longer in duration during the test than the study phase, regardless of presentation time condition or whether a face had been previously studied. Also, relative fixation density was greater over the eye region during the study phase and greater over lower facial regions during the test phase. The fine-grained patterns of differential fixation density suggest, though, that phase interacted to some degree with the presentation time condition. Importantly, only the long study phase presentation time condition improved recognition, suggesting that facial feature information is integrated over many fixations during encoding, whereas a face identity representation formed through such integration can be rapidly and holistically activated with a small number of fixations during recognition. Our results challenge the scan path replay hypothesis, which has been a prominent hypothesis regarding general visual recognition. While a small number of prior research studies have attempted to challenge the scan path replay hypothesis, the novel theoretical contribution of our study is that it also provides a differential account of face encoding (slower and more integrative) vs. recognition (faster and more holistic). Insofar as face encoding and recognition may be a representative case of visual encoding and recognition more generally, our study may perhaps be a first step in uncovering more general visual perceptual mechanisms.

Meeting abstract presented at VSS 2017