Contrasting signatures of visual attention emerge from the same underlying neural dynamics

Abstract Behavior is a window into cognition, yet behavioral effects rarely map one-to-one onto mechanisms: different mechanisms can produce the same effect, and the same mechanism can produce opposing effects across contexts. Visual attention provides a test case. We introduce the Shaped Priority Account, a neural-dynamical model in which salience and target guidance are nonlinearly transformed, unfold on different timescales, and compete within a recurrent selection field. With one fixed parameterization, the model reproduced 34 conditions from 12 experiments spanning seemingly distinct phenomena including triple-conjunction search, task-irrelevant feature effects, attentional capture, resistance to capture, and distractor-present benefits. These results show that apparently distinct behavioral signatures can emerge from the same neural dynamics rather than from distinct mechanisms. Behavioral effects are therefore not mechanistic fingerprints: before invoking distinct processes for each effect, researchers must pursue a benchmark, interrogating whether a neural process account can unify such effects under a single dynamical theory.