Translation Tolerance in Vision

Ryan Blything, Ivan Vankov, Casimir J H Ludwig, Jeffrey S Bowers

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

Abstract

A fundamental challenge in object recognition is to recognize an image when it is projected across different retinal locations, an ability known as translation tolerance. Although the human visual system can overcome this challenge, the mechanisms responsible remain largely unexplained. The ‘trained-tolerance’ approach holds that an object must be experienced across different retinal locations to achieve translation tolerance. Previous studies have supported this approach by showing that the visual system struggles to generalize recognition of novel objects to translations as small as 2° of visual angle. The present paper outlines a series of eyetracking studies that show novel objects can be recognized at translations as far as 18° from the trained retinal location, challenging the standard account of translation tolerance in neuroscience and psychology.
Original languageEnglish
Title of host publicationProceedings of the Cognitive Science Society
Pages1416
Number of pages5
Publication statusPublished - 2019

Structured keywords

  • Cognitive Science
  • Language

Fingerprint Dive into the research topics of 'Translation Tolerance in Vision'. Together they form a unique fingerprint.

Cite this