Abstract
Our visual system is well equipped to rapidly inform us of whether or not an image depicts a possible viewpoint on a structurally coherent 3-D object. Adult observers can readily classify simple 2-D line drawings as depicting possible or impossible 3-D objects. We examined infants' capacity to detect critical structural variations that determine local relative depth of features and global 3-D object coherence.
Previous work has shown that young infants detect perceptual similarities and regularities in features and attributes of novel objects. We hypothesized, therefore, that infants can form a perceptually complete representation of coherent novel objects even when a critical junction is concealed.
We tested infants aged 4 to 9 months in a habituation paradigm using an ‘impossible cube’ with a small red occluder concealing the depth cues at the critical junction. The stimulus contained color, texture, and shadow-based depth cues in addition to contour junctions. After habituation infants were shown 6 test trials each consisting of the unoccluded possible or impossible cube. The impossible cube image was constructed by reversing the local interposition cue of two overlapping edges of the possible cube, causing it to be an ‘accidental view’ of any possible 3-D object.
The infants showed a novelty preference for the impossible cube (p < .001), providing evidence for discrimination between possible and impossible novel objects. This finding suggests that mechanisms for representing object coherence are available early in postnatal life. Further, whereas previous research suggests that static interposition and amodal completion cues (as well as other static, monocular depth cues) are not operative until about 6 months of age, the present study suggests that interposition without motion of the occluded object relative to the occluder may be a robust depth cue at 4 months of age.
We would like to thank all the families and their babies for volunteering to participate in our studies.