PhD Laurent Caplette | Characterizing mental representations using deep image synthesis and behavior

Guest Lecture

  • Date: Oct 14, 2022
  • Time: 03:00 PM - 04:00 PM (Local Time Germany)
  • Speaker: PhD Laurent Caplette
  • Turk-Browne Lab, Department of Psychology, Yale University, New Haven, USA
  • Location: MPI for Human Cognitive and Brain Sciences
  • Room: Zoom Meeting
  • Host: CBS CoCoNUT
Accessing mental representations of visual concepts is a longstanding goal of cognitive psychology. Although progress has been made in uncovering some low-level representations, there is currently no general framework for investigating representations of high-level visual concepts. In a first set of experiments, we synthesized images made of random complex visual features using a deep neural network (DNN) and we asked observers to indicate what they perceived in them. This allowed us to uncover a general mapping between visual and semantic information, and reconstruct mental representations of hundreds of visual concepts. In a second set of experiments, we extended this method to investigate the level of abstraction of these representations in different contexts. To do so, we synthesized images using more or less abstract features from distinct DNN layers and we compared how well they explained human responses across different tasks. These methods are promising tools in the quest to better characterize internal visual representations.
Go to Editor View