Boosting the signal: Expectation-driven gain modulation of preparatory spatial attention

Abstract The visual system can flexibly adjust attentional deployment to match task demands, but whether observers can proactively modulate the spatial scope of attention based on expectations about upcoming search difficulty remains unclear. According to the zoom lens model, attention can narrow or broaden its spatial extent, with narrower focus enhancing processing efficiency, a mechanism that would benefit target discrimination in crowded displays. We tested whether observers adjust attentional scope when expecting sparse versus dense search arrays by combining spatial cueing with block-wise manipulations of display density expectations. Participants performed a visual search task in which endogenous cues predicted target location, while blocks predominantly contained either sparse (1 target, 3 distractors) or dense (1 target, 7 distractors) displays. We applied inverted encoding models to broadband EEG data to reconstruct spatial channel tuning functions, enabling precise characterization of both the locus and breadth of attentional deployment. Behaviorally, expecting difficult searches selectively improved accuracy at cued locations without costs elsewhere. Consistent with this selective benefit, neural measurements revealed that expectancy enhanced the amplitude of spatially selective responses at the attended location but did not alter tuning width. These findings demonstrate that expectations about search difficulty modulate attention through gain-based signal enhancement rather than adjustments to spatial scope, revealing that preparatory attentional control operates via amplitude modulation within a stable spatial focus. This mechanism complements reactive attentional adjustments and provides an efficient means for the visual system to optimize processing under predictable task demands. Competing Interest Statement The authors have declared no competing interest. Footnotes Conflict of Interest: The authors declare no competing financial interests This revised version of the manuscript incorporates several changes made in response Elife peer review, improving the methodological rigor, statistical reporting, and interpretational transparency of the work. The most substantive change concerns the statistical reporting of the behavioral analyses. In the original version, fixed effects from the generalized linear mixed-effects models (GLMMs) were reported using Type III Wald F statistics with large denominator degrees of freedom, which reflected trial-level observations rather than participant-level uncertainty. In the revised version, we now report fixed effects as regression coefficients (β), standard errors, z values, and associated p values, consistent with standard practice for binomial GLMMs implemented in lme4. The random-effects structure, which includes by-participant random intercepts and random slopes for all within-subject factors, is now described explicitly in the Methods. Additionally, pairwise comparisons previously reported in the Figure 2 caption using conventional t-tests have been replaced with GLMM-based contrasts derived from estimated marginal means, ensuring consistency across all statistical reporting. These changes do not affect the overall pattern of results or the substantive conclusions of the study. We have also expanded the Discussion to address two interpretational concerns raised during peer review. First, we now explicitly acknowledge that the absence of a tuning width difference in the neural data does not definitively rule out subtle or temporally limited changes in attentional scope that may not be fully captured by the spatial resolution of our approach. While the findings remain most consistent with a gain-based account, we adopt a more cautious interpretation of this null result. Second, we address the alternative explanation that the amplitude enhancement observed in expect-hard blocks could reflect a general change in alertness or motivational state rather than spatially specific preparatory attention. We highlight two aspects of the data that argue against this account: the behavioral benefit of expectation was selective to the cued location with no corresponding effects elsewhere, and multivariate decoding of expectancy condition remained at chance throughout the cue-target interval, suggesting that the two conditions did not produce globally distinct neural states. Several additional clarifications have been made throughout the manuscript, including specifying the broadband nature of the EEG analysis in the Abstract, clarifying the interpretation of the Serences et al. (2004) finding in the Introduction, adding context about the IEM approach in the Results, correcting an ambiguous reference to stimulus onset, and updating Figure 2 to correct an erroneous significance annotation and improve visual clarity.