Investigating the development of visual object perception from noisy images (360G-Wellcome-202627_Z_16_Z)

As we perceive the world, our brain continuously makes educated guesses about what we will see next. These inferences allow us to distinguish objects in our field of vision without having to examine every detail. This is known as the Bayesian model of visual perception. Recent evidence suggests that object recognition tasks for which such inferences are likely to be crucial may develop well into adolescence. There is anecdotal evidence that children find tasks such as identifying objects in poor lighting conditions or when borders are unclear (Bova, 2007) very difficult, even when they know exactly what the object they are looking for looks like (Yoon, 2007). This suggests that the robust object perception described in Bayesian models of the adult system takes surprisingly long to develop. This piece of research will test this hypothesis by (a) measuring children’s ability to recognize objects in a distorted ("noisy") image, and (b) testing how expectations about the objects (e.g. what it will look like) can improve perception. This aims to further our understanding of how the brain learns to use existing knowledge to interpret new sensory information, and make better inferences about the world.