September 2021
Volume 21, Issue 9
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2021
Implementing TVA as a Bayesian classifier in a foraging task
Author Affiliations
  • Sofia Tkhan Tin Le
    HSE University
  • W.Joseph MacInnes
    HSE University
  • Árni Kristjánsson
    HSE University
Journal of Vision September 2021, Vol.21, 2363. doi:https://doi.org/10.1167/jov.21.9.2363
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sofia Tkhan Tin Le, W.Joseph MacInnes, Árni Kristjánsson; Implementing TVA as a Bayesian classifier in a foraging task. Journal of Vision 2021;21(9):2363. https://doi.org/10.1167/jov.21.9.2363.

      Download citation file:


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

      ×
  • Supplements
Abstract

Foraging involves a natural search for many targets. Kristjánsson and colleagues (2014) found that changes in human foraging strategies depend on the complexity of the target and distractor relations. We aimed to understand the mechanisms underlying these differences. Bundesen’s (1990) TVA model involves an attempt at explicating the attentional mechanisms needed for selection, dividing the selection process into filtering and pigeonholing of the sensory input from the visual system and parameters from the executive system. We will combine these ideas in a generative model to simulate human foraging during a computerized task. First, we created a series of classifiers to determine which scene properties were important for predicting human behavior. Data from Kristjánsson et al. (2014) was used as input for a Naïve Bayesian model and an augmented Naïve Bayesian classifier. Switching rates between target types were the initial predictive parameter. EM log-likelihood and the strength of influence between parameters revealed high accuracy for both the Naïve Bayesian network and the augmented Naïve Bayesian network for conjunction foraging, and both models showed a connection between switching and previously selected target. We also looked at the strength of learned connections with network variables to see if they matched cognitive aspects of search behavior, and confirmed that switching appears regardless of the type of target stimulus. There was also a strong connection between switching and target type. This result relates to previous foraging findings in humans, which indicate that changes in search strategies depend on the complexity of objects in the visual field. The obtained Bayesian networks confirmed the general finding that increasing the complexity of the target changes foraging behavior minimizing switches in favor of exhaustive foraging of one category.

×
×

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.

×