Performance is impaired when a distracting stimulus is incongruent with the target stimulus (e.g., “green” printed in red). This congruency effect is decreased when the proportion of incongruent trials is increased, termed the proportion congruent effect. This effect is typically interpreted in terms of the adaptation of attention in response to conflict. In contrast, the contingency account argues that the effect is driven by the learning of predictive relationships between words and responses. In a recent report, Abrahamse, Duthoo, Notebaert, and Risko (2013) demonstrated larger changes in the magnitude of the proportion congruent effect when switching from a mostly congruent list to a mostly incongruent list, relative to the reverse order. They argued that this asymmetric list shifting effect fits only with the conflict adaptation perspective. However, the current paper presents reanalyses of this data and an adaptation of the Parallel Episodic Processing model that together demonstrate how the contingency account can explain these findings equally well when considering the generally accepted notion that performance improves with practice. The contingency account may still be the most parsimonious view.