Introduction Approximate numerical comparison is often influenced by various non-numerical sensory cues, yet whether they act via uniform inhibition (inhibitory control theory) or cue-weighted integration (sensory integration theory) remains debated. Methods To clarify this theoretical controversy, the present study tested a cue-specific, temporally staged account by orthogonally manipulating numerosity with a holistic, highweight cue (convex hull) and a basic, lower-weight cue (average dot size) while recording fronto-central ERPs (P2, N450). Twenty-five adults performed a rapid dot array comparison under four congruency conditions. Results Behavior showed clear convex-hull dominance: accuracy was high whenever convex hull aligned with numerosity and dropped when it conflicted, regardless of dot-size consistency; response times were unchanged. ERPs revealed a two-stage dynamic: the early P2 selectively tracked dot-size congruency (larger for dot-size–congruent trials), consistent with automatic integration of basic features, whereas the later N450 scaled with conflict structure and cue weight (fully congruent dot-size–congruent fully incongruent convex-hull–congruent/dot-size–incongruent), with no latency differences. Discussion These converging results support a time-resolved, weightsensitive mechanism in which basic features bias integration early and holistic configurations dominate later choice and recruit control when misaligned. The account reconciles sensory-integration and inhibitory-control views and motivates further tests of how cue–cue and cue–number conflicts shape numerical decisions.
Xiao et al. (2025) studied this question.