TAM: Explaining off-object fixations and central fixation tendencies as effects of population averaging during search

Abstract

Understanding how patterns are selected for both recognition and action, in the form of an eye movement, is essential to understanding the mechanisms of visual search. It is argued that selecting a pattern for fixation is time consuming—requiring the pruning of a population of possible saccade vectors to isolate the specific movement to the potential target. To support this position, two experiments are reported showing evidence for off-object fixations, where fixations land between objects rather than directly on objects, and central fixations, where initial saccades land near the centre of scenes. Both behaviours were modelled successfully using TAM (Target Acquisition Model; Zelinsky, 2008). TAM interprets these behaviours as expressions of population averaging occurring at different times during saccade target selection. A large population early during search results in the averaging of the entire scene and a central fixation; a smaller population later during search results in averaging between groups of objects and off-object fixations.

Keywords:

• Centre of gravity fixations
• Global effect
• Saccade target selection
• Selection for action

Acknowledgements

I would like to thank the current and previous members of the eye movements and visual cognition (Eye Cog) lab for many invaluable discussions. This work was supported by NIH grant R01-MH063748.

Notes

¹Of course, search judgements are also usually indicated by a manual buttonpress; this action occurs after the search task has been completed (Zelinsky & Sheinberg, 1997) and is therefore contingent on the target's recognition.

²Why an object must be selected to be recognized is a question beyond the scope of this paper, but one plausible speculation is that selection is needed to constrain the routing of feature information to higher visual areas responsible for recognition (Olshausen, Anderson, & van Essen, 1993).

³Although a saliency map or an activation map as a whole might be considered a sort of population code (Zelinsky, 2008), the time needed to find the peak on these maps is not thought to increase with the number of active points (the size of the population).

⁴Although sequences of eye movements can certainly be preprogrammed (e.g., Zingale & Kowler, 1987), a form of motor prioritization, such preprogramming probably characterizes only a small percentage of eye movement behaviour in the real world (with reading being a common exception to this rule; see Schad & Engbert, this issue 2012).

⁵Estimates of motor noise in the saccadic eye movement system vary with saccade amplitude (e.g., Abrams, Meyer, & Kornblum, 1989). The mean amplitude of saccades landing more than 1.0° from an object in Experiment 1 was 2.6°. Assuming conservatively that motor noise is 10% of saccade amplitude, this produces a mean noise estimate of only 0.26°, roughly one quarter of the minimum gaze-to-object distance used to define off-object fixations in this experiment.

⁶Each change in the threshold used to prune points from the target map is assumed by TAM to take one unit of time. However, the current version of TAM does not specify the mapping between this unit and saccade latencies; it may be the case that each change in the threshold consumes a constant amount of time, but more likely this relationship is nonlinear. Until this mapping is specified in future work, TAM will lack the ability to make the sorts of detailed predictions of fixation duration found in many current models of reading and scene perception (e.g., Nuthmann & Henderson, this issue 2012).