Introduction: Recent research has shown that attentional control can be modulatedat the item-specific level, meaning that different control strategies can be applied toindividual stimuli based on their relative frequencies within a given set. However,an alternative account suggests these findings may be confounded by stimulusresponsecontingency learning. Contingency is the proportion of trials where adistractor appears with its corresponding target; higher contingency enables fasterassociative response preparation, producing behavioral effects that mimic itemspecificcontrol. While neuroimaging studies provide some evidence for anteriorcingulate cortex (ACC) and dorsolateral prefrontal cortex (dLPFC) involvement duringcontrolled processes, differentiating item-specific control from contingency learningat the neural level remains challenging. In particular, no prior study has examinedhow temporal factors influence prefrontal engagement during these two processes.Methods: This study utilized item-specific proportion congruency (ISPC)manipulation via the flanker task. Hemodynamic activity in the prefrontal cortex wasmeasured using functional near-infrared spectroscopy (fNIRS) alongside reactiontime and error rates. The role of contingency learning was examined by manipulatingstimulus onset asynchrony (SOA). A 2 (congruency) × 2 (proportion congruency) ×3 (SOA: -250, 0, +250 ms) within-subject design with repeated-measures ANOVAwas used. To control for multiple comparisons across the 16 channels and 7 ANOVAeffects, a hierarchical false discovery rate (FDR) procedure was applied.Results: The ISPC effect was significant and modulated by SOA, being significant at-250 and 0 ms but not at +250 ms. In fNIRS data, the SOA main effect was significantin eight channels (3,9–14,16), and the ISPC effect was significant in four channels(1, 10, 11, and 12), after hierarchical FDR correction. The three-way interaction wasnot significant in any channel. Contingency analysis revealed a main effect, withthe largest contingency effect at -250 ms. The fNIRS contingency main effect wasalso significant in channels 1, 10, 11, and 12.Conclusion: While behavioral results mostly support contingency learning, fNIRSdata revealed ISPC-related activation in prefrontal channels with sensitivity to BA46 (lateral, channel 1) and the BA 10/BA 11 overlap zone (medial, channels 10–12),with channel 11 emerging as a convergence point for both SOA and ISPC effects.This spatial dissociation is consistent with the possible involvement of both controland learning processes operating in parallel within distinct prefrontal subregions.Keywords: Attention, dorsolateral prefrontal cortex, hemodynamics, learning,orbitofrontal cortex, spectroscopy