Increasing Signal, Reducing Noise: Contrasting Neural Mechanisms of Attention in Visual Search.

Abstract

When invariant target-distractor arrays are presented repeatedly during visual search, participants respond faster on repeated versus novel configuration trials. This effect reflects attentional guidance through long-term memory (LTM) templates-a phenomenon termed contextual cueing. Subsequently, relocating the target within the same distractor layout abolishes any contextual cueing effects, and relearning the new target position is much harder than the initial learning-likely due to consistent attentional misguidance toward the initial (learnt) target position. Here, we studied how the different processes involved in contextual cueing and relearning affect the variability of neural responses in human participants as measured with EEG. Attention has previously been shown to reduce trial-by-trial variability in EEG [Arazi, A., Yeshurun, Y., & Dinstein, I. Neural variability is quenched by attention. Journal of Neuroscience, 39, 5975-5985, 2019], indicating that, in addition to increasing the neural response to an attended stimulus, attention may reduce the noise within the neural response itself. While repeated versus novel contexts did not modulate the trial-by-trial variability during initial learning, significant lateralized variability reductions were observed for repeated but not novel context trials in the relocations phase. This contrasts with how contextual cueing affected lateralized ERPs in past research. Zinchenko and colleagues [Zinchenko, A., Conci, M., Töllner, T., Müller, H. J., & Geyer, T. Automatic guidance (and misguidance) of visuospatial attention by acquired scene memory: Evidence from an N1pc polarity reversal. Psychological Science, 31, 1531-1543, 2020] found that lateralized ERPs signal correct and incorrect (i.e., misguided) attentional selection of target positions learned earlier. This phenomenon was observed during both the learning and relocation phases. Thus, variability and lateralized ERPs may represent different facets of attention, where variability becomes evident specifically under high attentional demand conditions, such as when participants must override the misguidance caused by LTM templates.