At one level, this talk is about the development of visual object recognition and specifically abstract structural descriptions of object shape --the kind of description under which, for example, all varieties of chairs might be seen as the "same shape." Despite the important of object recognition to an understanding of human cognition and despite considerable efforts in the study of adult object recognition and machine vision, there is no agreed upon account and extraordinarily little data (or theory) on how the object recognition system develops.

      This talk will present new evidence of significant, and dramatic changes in visual object recognition between the ages of 18 and 24 months, and age period also known for marked and, I will argue, related changes in learning names for categories of things and for acting on things in ways that may inform visual processing of object shape.

      At another level, this talk is fundamentally about the nature of developmental process --of how cognition is made. Cognitive development is far more complex and dynamic than our usual debates allow. Developmental change is multi- causal nature, with weird loops of causes which are also consequences and consequences that are causes, with considerable and nontrivial causal spread. The interdependencies among developmental changes in visual object recognition, action, and object naming learning present a compelling example.

References: http://www.iub.edu/~cogdev/labwork/currentpapers.html
(Look under the head of "shape bias and related")