Abstract

Our ability to recognise objects is impaired in cluttered scenes, even for objects that are clearly visible in isolation. This process, known as crowding, is most evident in peripheral vision where it disrupts the recognition of a range of visual stimuli (including orientation, motion, colour, and faces) across large spatial regions. Crowding also becomes elevated in foveal/central vision in visual disorders ranging from amblyopia to dyslexia.

Is there a common mechanism for these diverse effects? I will argue that ‘pooling’ models, which depict crowding as an unwanted combination of target and flanker elements, can indeed provide this. In peripheral vision, crowded errors are not random, but rather reflect the combined appearance of the target object and surrounding flankers. We observe a similar pattern of errors in the foveal crowding that affects children with amblyopia, which pooling models simulate closely. Similarly, although it has been argued that crowding disrupts face recognition in a manner unlike that of simpler objects, we find that simple pooling processes provide a more parsimonious account. Finally, I will demonstrate that crowding can independently disrupt judgements of colour and motion for the same target object, suggesting the existence of multiple instances of crowding. Although this finding challenges ‘higher level’ crowding approaches, it is easily explained by pooling models.