# Gaze patterns during visual mental imagery reflect part-based generation

## TL;DR

Eye movements during mental imagery resemble part-by-part viewing, suggesting images are generated in parts.

## Contribution

Shows gaze patterns during imagery reflect part-based generation, independent of prior perception.

## Key findings

- Fixation scanpaths during imagery mirror part-based viewing (GCW condition).
- Refixation patterns during imagery align with part-based perception.
- Gaze patterns during imagery remain part-based regardless of initial encoding method.

## Abstract

Eye movements during visual mental imagery resemble those made during prior perception. Across two experiments, we investigated whether eye movements during imagery reflect a part-by-part generation of mental images, by comparing gaze patterns during mental imagery to those during part-based viewing (using a gaze-contingent window, GCW) and to those during holistic viewing (using an artificial scotoma, AS). In Experiment 1, participants freely encoded and imagined pictures before reinspecting them either part-by-part (GCW condition), or holistically (AS condition). The results show that fixation scanpaths (MultiMatch) and refixation patterns (recurrence quantification analysis) during mental imagery largely mirror those during GCW viewing. In Experiment 2, we examined whether this effect depends on prior perceptual encoding. Pictures were initially encoded either freely, with the AS, or with the GCW, and subsequently imagined. The results show that regardless of how the pictures were initially encoded, gaze patterns during mental imagery systematically resembled part-based perception. The current study provides direct evidence that eye movements during mental imagery reflect a part-by-part generation process of the imagined content, independent of prior perceptual encoding.