When making decisions humans often violate the principles of rational choice theory. Recent experiments, involving rapid experiential decisions, uncovered a mechanism that is responsible for various rationality violations. According to this selective gating mechanism, incoming value samples are accumulated across time, but prior to their accumulation they are weighted in proportion to their momentary rank-order. Here, using a data-driven approach, I present a dynamic extension of this mechanism, which involves potentially asymmetric inhibition between the inputs. As a result, and contrary to the previous selective gating implementation, the vigour of gating is modulated by the difference between two value samples (a distance effect) as well as by the absolute magnitude of the samples (a magnitude effect). This extension offers a superior explanation to existing and new data; and links high-level decision phenomena with computational principles previously described in theories of selective attention and visual search.