September 2015
Volume 15, Issue 12
Vision Sciences Society Annual Meeting Abstract  |   September 2015
Neural tuning changes underlying visual shape learning
Author Affiliations
  • Sach Sokol
    Zanvyl Krieger Mind/Brain Institute, Johns Hopkins University
  • Charles Connor
    Zanvyl Krieger Mind/Brain Institute, Johns Hopkins University
Journal of Vision September 2015, Vol.15, 373. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sach Sokol, Charles Connor; Neural tuning changes underlying visual shape learning. Journal of Vision 2015;15(12):373. doi:

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Learning to discriminate new shapes ultimately depends on processing changes at the level of individual neurons and neural populations. We sought to quantify and model the precise neural tuning changes that underlie shape learning. We trained a macaque monkey to discriminate 8 categories of letter-like stimuli. These stimuli were configurations of medial axis components (line segments, curves, junctions) drawn from a common set of 8 components. This enforced learning of global shape, since stimuli sharing components were confusable on a local level. In addition, the 8 stimuli constituted only a subset of the potential component configurations, making it possible to test for processing of specific configurations. After training we characterized neural tuning functions in anterior inferotemporal cortex (AIT), the final stage in the object pathway of monkey visual cortex. We used a genetic algorithm to guide large-scale sampling (500–1500 initially random, progressively evolving stimuli) in the medial axis shape domain. Our previous study showed that this method can constrain quantitative, predictive models of AIT tuning for medial axis structure (Hung et al., Neuron, 2012). Here, we hypothesized that a substantial number of AIT neurons would show structural tuning for medial axis components and component configurations unique to the learned stimuli. This hypothesis was confirmed in our current sample of AIT neurons. Many neurons were tuned for components common to multiple stimuli. Many neurons were tuned for configurations diagnostic for specific stimuli. We did not observe tuning for potential component configurations outside the learned stimulus set. This is the first direct observation of structural tuning changes that could explain visual shape learning.

Meeting abstract presented at VSS 2015


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.