Abstract
Detecting and categorizing particular entities in the environment are important visual tasks that humans have had to solve over evolutionary time. Here we investigate whether characteristics of ecologically-significant entities play a particular role during the development of visual categorization. To do this, the current project examined the effects of developing visual abilities, visual properties and ecological significance on categorization. Our stimuli were monochromatic photographs of structure-like assemblies and surfaces taken from three categories: vegetation, non-living natural elements, and artifacts. A set of computational and rated visual properties were assessed for these stimuli. We conducted two empirical studies: (a) two card sorting tasks with 76 preschool children (age: 4.1-6.1 years) and 72 adults (age: 18-50 years) which assessed classification and similarity judgments, and (b) a gaze-contingent eye-tracking search task with 39 infants (8-month-olds) in which target-structure patches were placed on a discrepant background structure, and the impact of properties and categories on infants' search performance was investigated. We found that category membership and visual properties impacted the performance of all participant groups. Sensitivity to the respective categories varied between tasks and over the age groups. For example, artifact images hindered infants' visual search but were classified best by adults, whereas sensitivity to vegetation was equally high for all age groups during visual search or similarity sorting. In children and infants, rated depth predicted task performance stronger than shape-related properties. Moreover, children and infants were sensitive to variations in the complexity of low-level visual statistics. These results suggest that classification of visual structures, and attention to particular visual properties is affected by the functional or ecological significance these categories and properties may have for each of the respective age groups. Based on this, the project highlights the importance of further developmental research on visual categorization with naturalistic, structure-like stimuli.