S.P. Arun, Centre for Neuroscience, Indian Institute of Science Bangalore, India, has given a talk on Zoom on March 7th.
http://www.cns.iisc.ac.in/home/people/sp-arun/
Short abstract:
A fundamental question in object vision is whether we can understand the whole object in terms of its parts, features or properties. While this enterprise seems plausible at least when we consider objects like a car or person, there are many objects that seem to transcend their parts. Some objects like circles and squares don't seem to have identifiable parts, while other objects have properties like symmetry that is not contained in either part alone. Finally, we have the experience of instantly recognizing visual words, suggesting that we recognize whole words at once rather than by reading individual letters. So can we explain whole objects using their parts or not? I will describe our recent work in which we have related whole object representations to their parts using quantitative models, and which challenge widely held intuitions about holistic processing.