List-level transfer effects in temporal learning: Further complications for the list-level proportion congruent effect

Abstract

Congruency effects are larger when most trials are congruent relative to incongruent. According to the conflict adaptation account, this proportion congruent effect is due to the decreased attention to words when most of the trials are conflicting. This paper extends on previous work arguing that list-level (contingency-unbiased) proportion congruent effects might be explainable by temporal learning biases. That is, congruency effects are larger in an easier task (i.e., mostly congruent) due to the faster pace of the task. Two non-conflict analogues of the proportion congruent effect are presented, one with a contrast manipulation and another with a contingency manipulation. Critically, both experiments control for potential item-specific temporal learning biases by intermixing biased context and unbiased transfer items. Results show a proportion congruent-like interaction for both item types, supporting the notion of task-wide temporal learning as an explanation for list-level proportion congruency effects. Distributional analyses lend further credence to the temporal learning account by showing that proportion congruent and proportion congruent-like effects are localised in the fastest and intermediate responses.

Keywords:

• Conflict adaptation
• Contingency learning
• Contrast
• Delta plot
• Distributional analyses
• Proportion congruent
• Temporal learning
• Transfer

I would like to thank Keith A. Hutchison for the suggested addition of distributional analyses.

I would like to thank Keith A. Hutchison for the suggested addition of distributional analyses.

Notes

¹ Note that the transfer items did have an item-specific manipulation. However, the manipulation for transfer items was identical in the mostly congruent and mostly incongruent lists. Thus, any item-specific biases were equated.

² The default payment for participants in the department increased partway through the experiment.

³ Of course, monitor settings will affect how a given RGB value is displayed. Although a light reader was not available for more precise measurement, monitors were set to fairly standard settings, which made high-contrast stimuli easily perceivable and low-contrast stimuli perceivable but not immediately. Any variance across testing computers did not affect the general direction of effects reported.