Repeated Scenes Speed Visual Search by Guiding Attention, Not by Speeding Decisions

Every day, in countless small ways, the human brain exploits the statistics of the visual world. When you walk into your own kitchen, you find the coffee machine faster than you would in a stranger’s, because the layout of the scene is familiar and your attention is quietly steered toward likely locations. In the laboratory, this phenomenon is known as contextual cueing: when observers repeatedly search through the same spatial arrangement of distractors, they detect the target faster than in novel arrangements, even though they typically cannot consciously report having seen the layouts before. Since Marvin Chun and Yuhong Jiang first described the effect in 1998, it has become one of the standard paradigms for studying how implicit statistical learning shapes visual attention. For decades, the dominant interpretation has been that repeated contexts speed up search by improving attentional guidance, helping the visual system home in on the target more efficiently.

Yet an alternative idea has lingered in the literature. Perhaps some of the benefit of a repeated context arises after the target has already been found. Once attention lands on the target, the brain still has work to do: it must identify the target, select an appropriate response, and execute it. If familiarity with the search display also lubricates these post-target stages, then contextual cueing would reflect a broader processing advantage rather than a purely attentional one. Some earlier studies, including work combining electroencephalography with search tasks, hinted that response-related processes might be modulated by context repetition, keeping this hypothesis alive. A new study by Feifei Zhao and Markus Conci of Ludwig-Maximilians-Universität München, published in Attention, Perception, & Psychophysics, puts this idea to a systematic and unusually clean test, and the verdict is negative: repeated contexts do not appear to facilitate response-related processing at all.

The logic of the new experiments is elegant in its simplicity. If contextual cueing speeds up response selection, then the size of the contextual cueing effect should depend on how demanding the response is. When the decision after finding the target is easy, response selection consumes little time, leaving little room for a contextual boost to manifest. When the decision is hard, with more possible responses to choose among, response selection takes longer, and any genuine facilitation from the repeated context should have more opportunity to express itself as a larger cueing effect. In other words, increasing response difficulty should amplify contextual cueing if, and only if, the learned context genuinely helps with the response stage. Zhao and Conci manipulated exactly this variable while holding attentional guidance constant, creating a situation in which the two candidate mechanisms, guidance and response selection, could be cleanly dissociated.

In the first experiment, participants performed a standard contextual cueing task in which they searched for a target among distractors. Some displays repeated across the session, forming the learned, or old, contexts, while others were generated anew on each trial, forming the novel, or new, contexts. The critical manipulation concerned the response: on some blocks, participants had to choose between two possible responses, whereas on other blocks the same search displays required a choice among four possible responses. The four-alternative condition makes the decision stage substantially more difficult, because more candidate responses must be considered and discriminated before a keypress can be issued. Crucially, the search displays themselves, and therefore the attentional guidance they support, were identical across these conditions, so any difference in how contextual cueing behaved could be attributed to the response side of the task.

The results were unambiguous. Participants showed a reliable contextual cueing effect, responding faster in repeated than in novel displays, confirming that the paradigm was working as intended. But the size of that effect did not grow when the response became harder. Reaction time benefits from repeated contexts were statistically indistinguishable between the two-alternative and four-alternative conditions. If repeated contexts facilitated response selection, the harder decision should have produced a larger cueing benefit, and it did not. The authors complemented their null results with Bayesian analyses, which quantify the evidence in favor of the null hypothesis rather than merely failing to reject it, strengthening the conclusion that the absence of an interaction reflects a genuine absence of response facilitation rather than a lack of statistical power.

One might object that the manipulation was too coarse, or that response facilitation requires the response itself to be predictable from the context. Zhao and Conci anticipated this concern in two further experiments. In the second and third experiments, they introduced response cues that were themselves learnable: the repeated search contexts were paired consistently with particular responses, so that the spatial layout did not merely guide attention but also carried predictive information about what the correct response would be. Under these conditions, a learner sensitive to response regularities could, in principle, prepare or select the response in advance, producing a substantial speedup on repeated-display trials. This is the strongest possible version of the response-facilitation hypothesis, because it hands the participant exactly the kind of consistent, repeating response regularity that statistical learning mechanisms are known to pick up.

Even here, the effect refused to appear. Adding learnable response cues did not enhance contextual cueing, whether participants faced two or four response alternatives. The cueing effect remained what it had been in the first experiment, a benefit attributable to attentional guidance, with no additional boost from the predictable response mapping. This pattern is theoretically important because it rules out a family of explanations in which the contextual cueing effect is a composite of attentional and response-related components. Across three experiments, the data converge on a single conclusion: the benefit of a repeated visual context is confined to the stage of locating the target, and it does not extend to the stages of deciding what to do about it once the target has been found.

The findings fit within a broader theoretical framework in which visual search is decomposed into separable stages, from preattentive guidance through attentional selection to target identification and response selection. Models such as Guided Search treat these stages as distinct, and the new results suggest that implicit context memory feeds into only the earliest of them. Consistent with this, electrophysiological work by the same research group and colleagues has shown that repeated contexts modulate early, attention-related brain potentials, such as the N2pc component that indexes the allocation of visuospatial attention, while leaving later, response-related potentials largely untouched. The behavioral evidence now aligns with the neural evidence: contextual learning is a memory system for where to look, not a memory system for what to press.

Why does this matter beyond the laboratory? Contextual cueing is often cited as a model of how expertise and experience make everyday visual tasks effortless, from radiologists scanning medical images to drivers navigating familiar streets. If the benefit of scene familiarity were partly a response-level phenomenon, training programs might profitably emphasize consistent response mappings alongside consistent visual layouts. The new results suggest otherwise: the dividend of a familiar environment is paid almost entirely in faster target localization. For human factors research, this implies that designing environments with stable, predictable spatial layouts should speed up the finding of relevant objects, but pairing those layouts with consistent actions will not, by itself, buy additional time savings. Familiarity helps us find things quickly; what we do once we have found them remains our own, slower, deliberate business.

The study also exemplifies the growing emphasis on transparency in cognitive psychology. All raw trial-level data, subject-level means, and analysis scripts have been made publicly available through the Open Science Framework, allowing any reader to reanalyze the evidence and test alternative models of the reaction time distributions. Although the design and analysis plans were not preregistered, the multi-experiment structure, the Bayesian support for the null results, and the open data collectively make a strong case. After nearly three decades of research, the field can now say with considerable confidence what contextual cueing is, and, just as importantly, what it is not: a powerful form of implicit memory that guides the eyes and the mind’s attention to where targets hide, and nothing more.

Subject of Research: Whether contextual cueing in visual search facilitates response-related processing in addition to attentional guidance

Article Title: No evidence for facilitated response-related processing in contextual cueing

Article References: Zhao, F., & Conci, M. (2026). No evidence for facilitated response-related processing in contextual cueing. Attention, Perception, & Psychophysics, 88(7), Article 199. https://doi.org/10.3758/s13414-026-03349-x

Image Credits: AI Generated

DOI: 10.3758/s13414-026-03349-x

Keywords: contextual cueing, visual search, attentional guidance, response selection, statistical learning, implicit memory, reaction time, Bayesian analysis, visual attention, cognitive psychology, evidence, facilitated