A rapid effect of stimulus quality on the durations of individual fixations during reading

Abstract

We developed a variant of the single fixation replacement paradigm (Yang & McConkie, 2001, 2004) in order to examine the effect of stimulus quality on fixation duration during reading. Subjects' eye movements were monitored while they read passages of text for comprehension. During critical fixations, equal changes to the luminance of the background produced either an increase (Up-Contrast) or a decrease (Down-Contrast) of the contrast of the text. The durations of critical fixations were found to be lengthened in the Down-Contrast but not the Up-Contrast condition. Ex-Gaussian modelling of the distributions of fixation durations showed that the reduction in stimulus quality lengthened the majority of fixations, and a survival analysis estimated the onset of this effect to be approximately 141 ms following fixation onset. Because the stimulus quality of the text during critical fixations could not be predicted or parafoveally previewed prior to foveation, the present effect can be attributed to an immediate effect of stimulus quality on fixation duration.

Keywords:

• Eye movements
• Reading
• Fixation duration
• Stimulus quality
• Stimulus contrast

This research was supported by the Atkinson Fund to Keith Rayner and by an NSERC grant to Eyal Reingold.

This research was supported by the Atkinson Fund to Keith Rayner and by an NSERC grant to Eyal Reingold.

Notes

¹ Because it was unlikely that the white text and black text produced identical effective text contrast under the Up-Contrast and Down-Contrast conditions, we conducted a post-hoc (3 × 2) repeated measures ANOVA on fixation durations to test for the interaction between Display Change condition and Text Contrast Polarity. This revealed a significant interaction, F(2, 40) = 30.41, p < .001; the Down-Contrast manipulation produced a slightly larger effect on fixation duration for black text (mean = 273 ms) compared to white text (mean = 251 ms), and the Up-Contrast condition produced a slightly larger effect for white text (mean = 216 ms) than black text (mean = 208 ms).

² The appearance of the Down-Contrast distribution relative to the No-Change condition suggests a dip in saccadic activity between 100 and 150 ms. This is reminiscent of a saccadic inhibition profile (Reingold & Stampe, 2000, 2002, 2004), and it is possible that this feature might in part be caused by saccadic inhibition in response to the display change that occurred during the prior saccade. This is unlikely, however, because a different pattern is present in the Up-Contrast condition, which contains equivalent luminance changes during the prior saccade. Moreover, we re-computed the histograms while only considering fixations for which the display change had been completed 15 ms prior to fixation onset as opposed to the 7 ms requirement used to generate Figure 2. This did not change the qualitative appearance of the “dip” in saccadic activity seen in the Down-Contrast condition, confirming that it is not tied to the display change but rather the difficulty in text processing imposed by that condition.

³ Following our post-hoc analysis of mean fixation duration, we also estimated the divergence point between Down-Contrast and Up-Contrast conditions separately for the black text and white text conditions. The divergence points were found to be 140 ms and 161 ms, respectively. This is consistent with the finding of a stronger effect on mean fixation duration for the Down-Contrast manipulation for black text compared to white text.