Spatial contextual cueing refers to visual search performance's being improved when invariant associations between target locations and distractor spatial configurations are learned incidentally. Using the instance theory of automatization and the reverse hierarchy theory of visual perceptual learning, this study explores the acquisition of visual specificity in spatial contextual cueing. Two experiments in which detailed visual features were irrelevant for distinguishing between spatial contexts found that spatial contextual cueing was visually generic in difficult trials when the trials were not preceded by easy trials (Experiment 1) but that spatial contextual cueing progressed to visual specificity when difficult trials were preceded by easy trials (Experiment 2). These findings support reverse hierarchy theory, which predicts that even when detailed visual features are irrelevant for distinguishing between spatial contexts, spatial contextual cueing can progress to visual specificity if the stimuli remain constant, the task is difficult, and difficult trials are preceded by easy trials. However, these findings are inconsistent with instance theory, which predicts that when detailed visual features are irrelevant for distinguishing between spatial contexts, spatial contextual cueing will not progress to visual specificity. This study concludes that the acquisition of visual specificity in spatial contextual cueing is more plausibly hierarchical, rather than instance-based.