Spatial Numerical Associations (SNAs) refer to the compatibility between space and numerical magnitude. They are commonly operationalized as faster left-hand responses to small numbers and faster right-hand responses to large numbers (R/SNA). By contrast, the compatibility between stimulus spatial position and numerical magnitude (S/SNA) has received little attention. We disentangled R/SNA and S/SNA contributions for peripheral target numbers, with their magnitude being either task-irrelevant or task-relevant, while also manipulating the spatial position-response hand compatibility (S/R). Across two Posner exogenous cueing experiments, we orthogonally varied target spatial position (left/right), cue validity (valid/neutral/invalid), and the S/R-compatibility (S/R21 compatible/incompatible). Experiment 1 employed a magnitude classification task, and Experiment 2 a localization task. We applied a chronometric framework to decompose response latencies into three components: reflexive attention (cue-driven), voluntary attention (target-driven), and response planning. In Experiment 1, SNAs were driven by the response planning component and were progressively inhibited by voluntary attention, decreasing from the valid to the invalid cue condition. In Experiment 2 SNAs emerged in the invalid cues condition only, suggesting that they were driven by the voluntary attention component. Notably, S/R exhibited a striking reversal: in Experiment 1, it was driven by voluntary attention, whereas in Experiment 2, it hinged on response planning and was actually suppressed by voluntary attention. These results align with a relative-dependence framework, suggesting that compatibility effects are guided by either voluntary attention or response planning, depending on whether the critical dimension underlying the effect is explicit or implicit.
