Why attention is not the lever these studies point to
The Ebbinghaus illusion is the classic size illusion in which a central circle looks larger or smaller depending on the size of the circles surrounding it. One study asked whether the way people look at such displays—gaze-mediated attention—could account for differences in how the illusion affects perception and visually guided action. Its conclusion was that the differences are better explained by the different neurocomputational demands of the tasks involved, and it explicitly set gaze-mediated attention aside as an explanation [1]. In plain terms, the visual system is being asked to do computationally different jobs in different tasks, and that difference—not where attention is directed—drives the result.
This matters for the reader's question because it means adaptation effects in size illusions are not simply a matter of attention being allocated more or less strongly. If attention were the modulating factor, you would expect gaze behavior to track the changes; instead, the study treats gaze-mediated attention as a factor that does not account for the pattern [1].
How the visual system recalibrates to surrounding size
A second body of work approaches size illusions through adaptation-level theory, which holds that perceived size is judged relative to a reference point built up from the surrounding context. Under this account, when the contextual sizes change, the reference point shifts, and the same object is perceived differently—an assimilation effect in which perceived size is pulled toward the context [2]. This is a recalibration of the comparison standard rather than a change in how attention is deployed.
The theory also treats the weight given to different contextual elements as variable, which helps explain why size-contrast illusions can shift as the surrounding context changes [2]. The mechanism being described is about how context sets the benchmark for size judgments, not about attention selecting or amplifying one part of the display.
What this means if you expected attention to be the answer
Put together, the evidence points away from attention as the modulator of adaptation-induced changes in visual size illusions. The Ebbinghaus work attributes the differences to neurocomputational requirements rather than gaze-mediated attention [1], and the adaptation-level account explains size changes as shifts in a context-derived reference point with variable weighting [2]. Neither line of evidence assigns attention a modulating role.
One caveat belongs right next to these claims: both sources are available only at the abstract level, so the conclusions rest on the stated summaries rather than on the full experimental detail. Within that scope, the answer to the question is that attention is not supported as the factor that modulates these adaptation effects; the evidence instead points to task demands and contextual recalibration.
Sources used in this answer
Looking at the Ebbinghaus illusion: Differences in neurocomputational requirements, not gaze-mediated attention, explain a classic perception-action …
Differences in neurocomputational requirements, not gaze-mediated attention, explain the pattern of effects observed with the Ebbinghaus size illusion in perception and visually guided action.
Assimilation predicted by adaptation-level theory with variable weights
Adaptation-level theory accounts for size-contrast illusions through a context-derived reference point with variable weights, without addressing attention as a modulator.