Elsevier

Cognition

Volume 123, Issue 1, April 2012, Pages 185-189
Cognition

Brief article
The emergence of frequency effects in eye movements

https://doi.org/10.1016/j.cognition.2011.12.011Get rights and content

Abstract

A visual search experiment employed strings of Landolt Cs to examine how the gap size of and frequency of exposure to distractor strings affected eye movements. Increases in gap size were associated with shorter first-fixation durations, gaze durations, and total times, as well as fewer fixations. Importantly, both the number and duration of fixations decreased with repeated exposures. The findings provide evidence for the role of cognition in guiding eye-movements, and a potential explanation for word-frequency effects observed in reading.

Introduction

The extent to which oculomotor and cognitive factors influence eye movements in reading is widely debated (Starr & Rayner, 2001). Oculomotor theories contend that the timing and location of fixations are predominantly determined by visual acuity and oculomotor constraints (e.g., McDonald et al., 2005, O’Regan, 1990, Yang and McConkie, 2001). Cognitive (processing) theories (e.g., Just and Carpenter, 1980, Reichle et al., 1998, Reilly and Radach, 2006) argue that cognitive processes like word identification largely drive eye movements, as evidenced by word-frequency effects, or the finding that higher frequency words receive shorter fixations and are skipped more often than lower frequency words (e.g., Inhoff & Rayner, 1986).

To determine cognition’s involvement in eye-movement guidance during reading, eye movements during reading have been compared to eye movements during tasks presumed to minimally engage cognition, such as searching for a target word in a text or performing z-string “reading” (Rayner and Fischer, 1996, Rayner and Raney, 1996, Vitu et al., 1995). In these tasks, the word-frequency effects that are ubiquitous in reading are absent. Assuming that word-frequency effects arise because more exposures make a word’s representation in memory easier to access, as posited by several episodic models of printed word identification (e.g., Ans et al., 1998, Reichle and Perfetti, 2003), this may be evidence that eye movements are guided by the demands of accessing lexical representations from memory during reading, but not during tasks that do not require full lexical retrieval.

The current experiment investigates how manipulating the frequency of orthographic patterns in a visual-search task affects eye movements. Demonstrating frequency effects in a non-reading task would provide strong evidence that cognition rapidly influences eye movements and address the etiology of word-frequency effects in reading. Because associations between word frequency and fixation durations are inherently correlational, some have argued that the causal nature of this relationship remains conjectural (cf., Kliegl et al., 2006, Rayner et al., 2007). The current experiment directly tests the hypothesis that frequency effects arise from frequency of exposure by manipulating the latter to determine if and how eye-movement measures are affected.

We used a paradigm from Williams and Pollatsek (2007; see also Corbic, Glover, & Radach, 2007), in which participants scanned lines of Landolt-C clusters, or circles with a missing segment of varying size and orientation, to locate clusters containing an O, i.e., targets. In Williams and Pollatsek (2007), the number of pixels in the missing segments, or gap size, was held constant within a given cluster, but manipulated between clusters. Their results indicated that fixation durations on the non-target or distractor clusters were related to the gap size of the fixated cluster, but not to the gap sizes of neighboring clusters. Furthermore, gaze durations on clusters during scanning were nearly equivalent to reaction times for target present-absent judgments on the same clusters in isolation. Because these results parallel the finding that gaze durations on words in reading correlate with the identification times of those same words displayed in isolation (Schilling, Rayner, & Chumbley, 1998), Williams and Pollatsek argued that eye movements in their task were also driven by identification processes, and thus cognition. However, given that gap-size manipulations directly affect CO discriminability, gap-size effects are likely to be influenced by perceptual fluency. Even more compelling evidence for cognitive effects would be provided by a demonstration that the accessibility of individual clusters’ memory representations affects eye movements.

Our study extends Williams and Pollatsek (2007) by manipulating the number of times a cluster appeared in the experiment. We predict an inverse relationship between the number of encounters and fixation durations on the distractor clusters. This finding would suggest that the accessibility of a cluster’s representation in memory (as determined by frequency of exposure) influences how long that cluster must be processed and thus how long the eyes stay on it. This would provide stronger evidence of cognition influencing eye movements and would exemplify how the mind-eye link may develop in tasks like reading.

Section snippets

Participants

Sixteen University of Pittsburgh undergraduates with normal or corrected vision received partial course credit in an introductory psychology course for their participation.

Procedure

Each trial began with participants viewing a dot on the leftmost side of the screen. Participants initiated the trial with a button press, which displayed the stimuli for 12 s. Participants were instructed to scan a horizontal row of clusters from left to right and to identify any targets, i.e., a cluster containing a letter O

Behavioral results

Participants’ mean accuracy on the search task was greater than 90%. The average hit rate was 87%, the false alarm rate less than 1.5%, and there was only one failure to respond within 12 s.

Eye-movement results

Analyses focus exclusively on distractor clusters because only distractor frequency was manipulated. The following eye-movement measures were examined on a given cluster: (a) first-fixation duration, or the duration of the initial fixation during first-pass scanning; (b) gaze duration, or the sum of all

Discussion

This study investigated the role of cognition in eye-movement behavior in a visual search task. Distractor clusters that were encountered more often had shorter gaze durations and total-viewing times, and received fewer fixations. These findings parallel word-frequency effects in reading (Inhoff & Rayner, 1986) and have several important theoretical implications. First, the effect of cluster frequency indicates that participants did not solely engage in letter-by-letter discrimination, but

Acknowledgements

This research was supported by NIH Grant HD053639 to the second and last authors. We thank Dr. Natasha Tokowicz for helpful comments on earlier drafts of the manuscript. We thank Steve Walenchok for his help in data collection.

References (32)

  • A.W. Inhoff et al.

    Parafoveal word processing during eye fixations in reading: Effects of word frequency

    Perception and Psychophysics

    (1986)
  • M.A. Just et al.

    A theory of reading: From eye fixations to comprehension

    Psychological Review

    (1980)
  • J.K. Kaakinen et al.

    Task effects on eye movements during reading

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (2010)
  • R. Kliegl et al.

    Tracking the mind during reading: The influence of past, present, and future words on fixation durations

    Journal of Experimental Psychology: General

    (2006)
  • J.L. McClelland et al.

    An interactive activation model of context effects in letter perception: Part 1. An account of basic findings

    Psychological Review

    (1981)
  • S.A. McDonald et al.

    An anatomically constrained, stochastic model of eye movement control in reading

    Psychological Review

    (2005)
  • View full text