The influence of predictability, visual contrast, and preview validity on eye movements and N400 amplitude: co-registration evidence that the N400 reflects late processes

ABSTRACT

Two eye movement/EEG co-registration experiments investigated effects of predictability, visual contrast, and parafoveal preview in normal reading. Replicating previous studies, in Experiment 1 contrast and predictability additively influenced fixation durations, and in Experiment 2 invalid preview eliminated the predictability effect on early eye movement measures. In both experiments, predictability influenced the amplitude of the N400 component of the fixation-related potential. In Experiment 1, visual contrast did not influence the N400, and in Experiment 2, the effect of predictability on the N400 was larger with invalid preview, in opposition to the eye movement pattern. The N400 may reflect a late process of accessing conceptual representations while the duration of the eyes’ fixation on a word is sensitive to the difficulty of perceptual encoding and early stages of word recognition. The effects of predictability on both fixation duration and the N400 suggest an influence of this variable at two distinct processing stages.

KEYWORDS:

• Eye movements
• N400
• predictability
• word recognition
• reading

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Eye movement and EEG data for both experiments, as well as analysis scripts, are available at: https://osf.io/6nr9t/?view_only=61eb1851bba4454f822563e0f5c7bb92.

Notes

1 The full set of electrodes was: E31, E36, E37, E42, E52, E53, E54, E55, E61, E62, E78, E79, E80, E86, E87, E92, E93, E104 and E129.

2 Results from the models using log-transformed reading times for the pre-target word were different in two respects: In the log first fixation model, the interaction effect was significant; in the log gaze duration model the effect of target word predictability was not significant.

3 The interaction between preview validity and predictability was significant with log go-past time as the dependent measure. All other patterns remained the same with log transformed reading time measures.

4 Models using log reading times resulted in the same statistical patterns.

Additional information

Funding

This work was funded by a grant from the NSF (BCS 1732008).