Mechanisms of priming of pop-out: Stored representations or feature-gain modulations?

Hyunkyu Lee, Michael C. Mozer, Shaun P. VeceraAttention 2009 · DOI

Abstract

Priming of Pop-out (PoP) is the phenomenon where repeating a target's feature, such as color, facilitates visual search. This paper investigates whether PoP relies on Visual Short-Term Memory (VSTM) or results from low-level feature-gain modulations. Through dual-task experiments that loaded VSTM to capacity, the authors found that PoP persists independently of memory load. These results suggest that PoP is an automatic, bottom-up process where recently attended features receive a sensory gain boost, facilitating subsequent search.

Introduction

Visual search is often facilitated by recent experience. When a target's defining feature (e.g., color or shape) is repeated from the previous trial, reaction times (RT) typically decrease—a phenomenon known as Priming of Pop-out (PoP). Historically, two competing theories have attempted to explain this effect. The top-down VSTM account posits that the brain stores a representation of the previous target in a high-level memory buffer to guide attention. Conversely, the bottom-up feature-gain account suggests that successful selection of a feature "turns up the volume" or increases the gain on specific feature maps, making subsequent matches pop out more readily. This study aims to distinguish between these mechanisms by testing PoP under conditions of high VSTM load.

Methods

The researchers conducted four behavioral experiments with human participants to isolate the mechanisms of PoP.

Experimental Design

In Experiment 1, a baseline for PoP was established using Landolt C stimuli (rings with a small gap). Participants were tasked with identifying the gap location on a uniquely colored target within a search array.

Example of the visual search display featuring Landolt C stimuli.

Experiment 2 introduced a dual-task paradigm to "stress test" the VSTM account. Participants were required to hold four distinct colors in VSTM—filling it to capacity—while simultaneously performing the pop-out search. If PoP relies on VSTM, filling this buffer should eliminate the priming effect. Experiment 3 served as a control to ensure the memory load effectively occupied VSTM. Finally, Experiment 4 explored cross-task priming by interleaving a non-search shape-discrimination task with the visual search, testing if simply attending to a color in a different context would induce priming.

Results

The primary finding across experiments was the persistence of PoP regardless of VSTM load. In Experiment 2, PoP remained robust, providing a 25–40 ms benefit even when the VSTM buffer was fully occupied. Statistical analysis (ANOVA) confirmed a significant main effect of target color repetition ($p < .001$), while the interaction between VSTM load and priming was non-significant ($F < 1$), indicating that memory load did not interfere with the priming mechanism.

Reaction time (RT) benefits for repeated target features (Same) vs. changed features (Different) across trial positions.

As shown in the figure above, the priming effect was most pronounced for the immediate subsequent trial ($n-1$) and gradually decayed over approximately 3 to 5 trials. Furthermore, Experiment 4 demonstrated that priming is task-independent; attending to a colored shape for a non-search task significantly improved RTs in a subsequent search for that same color ($p < .01$).

Discussion

The results strongly support the feature-gain modulation account of PoP. Because priming persisted under full VSTM load, it is unlikely that PoP relies on the same high-level memory architecture used for conscious short-term storage. Instead, the data aligns with the Guided Search model, suggesting that attending to a feature automatically boosts the gain of the corresponding feature map. This increased gain enhances the signal in the brain's master saliency map on subsequent trials, facilitating "pop-out" search.

Implications and Limitations

The automaticity of PoP is further evidenced by the fact that participants could not suppress the priming cost even when color sequences were predictable. While this study clarifies the functional architecture of PoP, the specific neural locus—whether in early visual areas like V1/V4 or attentional centers like the Superior Colliculus—remains to be identified. Future research should also investigate whether these VSTM-independent mechanisms extend to other visual features like size or orientation.

Conclusion

Priming of Pop-out is a low-level, automatic sensory adjustment rather than a high-level memory process. By demonstrating that PoP is independent of VSTM, this study provides critical evidence for feature-gain modulation as a fundamental mechanism of visual attention.

Summary updated

BibTeX
@inproceedings{lee2009mechanisms,
  author = {Hyunkyu Lee and Michael C. Mozer and Shaun P. Vecera},
  title = {{Mechanisms of priming of pop-out: Stored representations or feature-gain modulations?}},
  booktitle = {Attention},
  year = {2009},
  doi = {10.3758/app.71.5.1059},
}
Figure