Paul Linton

Centre for Applied Vision Research, City, University of London
  •  10
    Deep neural networks are not a single hypothesis but a language for expressing computational hypotheses
    with Tal Golan, JohnMark Taylor, Heiko Schütt, Benjamin Peters, Rowan P. Sommers, Katja Seeliger, Adrien Doerig, Talia Konkle, Marcel van Gerven, Konrad Kording, Blake Richards, Tim C. Kietzmann, Grace W. Lindsay, and Nikolaus Kriegeskorte
    Behavioral and Brain Sciences 46. 2023.
    An ideal vision model accounts for behavior and neurophysiology in both naturalistic conditions and designed lab experiments. Unlike psychological theories, artificial neural networks (ANNs) actually perform visual tasks and generate testable predictions for arbitrary inputs. These advantages enable ANNs to engage the entire spectrum of the evidence. Failures of particular models drive progress in a vibrant ANN research program of human vision.
  •  4
    I use the distinction between perception and cognition to advance a novel account of visual shape and visual scale. 1. INTRODUCTION (Preface & Chapter 1) - I outline the distinction between perception and cognition, and give an overview of how cognitive influences were rejected, and later reincorporated back into, 3D vision. 2. VISUAL SHAPE (Chapter 2 & Chapter 3) - Traditional theories argue that 3D vision integrates binocular disparity, perspective, and shading into a single coherent percept. …Read more