In natural behavior, humans make trade-offs between sampling information from the visual environment and relying on memory. As is often the case, observers favor visual sampling when its cost is low (e.g., a sampling only takes a few saccades); but when the sampling cost is artificially increased (e.g., by imposing longer waiting times), they favor visual working memory over visual sampling. Studies investigating this ubiquitous trade-off have neglected the stability in the real world, where repetitive patterns may engage a different memory system: long-term memory (LTM). Two competing hypotheses were derived from previous studies: when stable environments allow the formation of LTM (and sampling costs are relatively low), observers may either (a) continue to rely on visual sampling, or (b) choose to favor memory over visual sampling. We provide evidence for the latter hypothesis, based on a copying task in which participants reproduced an example display containing several colored polygons that either changed or remained stable over consecutive trials. Experiment 1 showed that, when the example display was repeated, the sampling frequency and durations decayed exponentially, eventually disappearing entirely. In Experiments 2 and 3, we replicated the reduction in sampling behavior when only half of the example polygons were repeated. Moreover, participants' recall of repeated items on a surprise memory test predicted this reduction in sampling behavior, confirming the involvement of LTM. Our findings demonstrate that repetitive patterns in stable visual environments make memory use preferable over visual sampling by reducing memory cost through the recruitment of LTM. (PsycInfo Database Record (c) 2026 APA, all rights reserved).
Xu et al. (Mon,) studied this question.