AOI Geometry Is an Analytical Assumption

AOI boundaries encode a substantive decision about where a meaningful visual region begins and ends. Even with a fixed stimulus, that decision can be uncertain.

When to use this perspective

Treat geometry as a sensitivity dimension when fixations cluster near borders, AOIs are small relative to spatial error, neighboring regions nearly touch, or more than one boundary coding is scientifically defensible.

When not to use perturbation

Do not use AOI perturbation to repair poor calibration, search for significance, or replace a preregistered primary AOI after seeing the result. The nominal geometry remains the reference; perturbations test dependence on it.

Boundary ambiguity

eyeprocesspy does not silently choose the first AOI when regions overlap. The default assignment is __ambiguous__, making geometry conflicts visible in downstream summaries.

Practical design rule

Define a perturbation envelope from the display, acquisition quality, and stimulus design. A small set such as ±0.25°, ±0.50°, and ±1.00° is easier to defend than a large result-driven search grid.

Continue to AOI Perturbation Sensitivity Analysis.