Disentangle guidance carryover from task familiarity

Determine whether the post-guidance timing gains produced by Co-Annotator’s gaze-aligned AOI guidance and combined AOI–VLM guidance result from genuine perceptual reorientation rather than task familiarity or practice effects.

Background

Co-Annotator evaluates whether gaze-aligned area-of-interest heatmaps improve ophthalmology residents’ OCT-reading efficiency. In US2 and US3, residents read a pre-guidance control block, a guidance block, and a post-guidance control block in a fixed order. Reading time decreased after guidance, which the authors interpret as evidence that residents internalized more efficient visual-search strategies.

Because the post-guidance block always followed the guidance block, however, participants may also have become faster through repeated exposure to the task or images. The authors explicitly state that the timing carryover cannot be fully separated from familiarity and identify a crossover design as a way to obtain stronger causal evidence.

References

Ordering effects on diagnostic accuracy were nonsignificant, so the accuracy results are not attributable to practice; the timing carryover, by contrast, cannot be fully disentangled from familiarity under this design, which we treat as suggestive rather than confirmatory.

— Co-Annotator: Expert-Distilled ViT and VLM for Visual and Documentation Guidance in Age-Related Macular Degeneration  (2608.30352 - Li et al., 31 Aug 2026) in Section 3, User Study 3, subsection “Diagnostic Accuracy and Efficiency”; Section 4, “Limitations and Future Directions”