Modeling visual exploration in rhesus macaques with bottom-up salience and oculomotor statistics

Seth D. König, Elizabeth A. Buffalo

Research output: Contribution to journalArticlepeer-review

5 Scopus citations


There is a growing interest in studying biological systems in natural settings, in which experimental stimuli are less artificial and behavior is less controlled. In primate vision research, free viewing of complex images has elucidated novel neural responses, and free viewing in humans has helped discover attentional and behavioral impairments in patients with neurological disorders. In order to fully interpret data collected from free viewing of complex scenes, it is critical to better understand what aspects of the stimuli guide viewing behavior. To this end, we have developed a novel viewing behavior model called a Biased Correlated Random Walk (BCRW) to describe free viewing behavior during the exploration of complex scenes in monkeys. The BCRW can predict fixation locations better than bottom-up salience. Additionally, we show that the BCRW can be used to test hypotheses regarding specific attentional mechanisms. For example, we used the BCRW to examine the source of the central bias in fixation locations. Our analyses suggest that the central bias may be caused by a natural tendency to reorient the eyes toward the center of the stimulus, rather than a photographer's bias to center salient items in a scene. Taken together these data suggest that the BCRW can be used to further our understanding of viewing behavior and attention, and could be useful in optimizing stimulus and task design.

Original languageEnglish (US)
Article number23
JournalFrontiers in Integrative Neuroscience
Issue numberJUNE2016
StatePublished - Jun 30 2016
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2016 König and Buffalo.


  • Attention
  • Bottom-up
  • Free viewing
  • Natural behavior
  • Random walk
  • Salience


Dive into the research topics of 'Modeling visual exploration in rhesus macaques with bottom-up salience and oculomotor statistics'. Together they form a unique fingerprint.

Cite this