Lettuce entertain you: Assessing Sandwich Builder as a measure of auditory short-term memory.

Abstract

Numerous language models propose a critical role of auditory short-term memory in spoken language processing, making reliable estimations of individual listeners' memory capacities essential in linguistic research. Although a variety of cognitive tasks exist for assessing both visual and auditory short-term capacities in a laboratory environment (e.g., digit span), these types of tasks are typically not engaging for the participant. In the current study, we assess the validity and reliability of a novel "gamified" short-term memory assessment tool called Sandwich Builder. In Sandwich Builder, the participant completes 12 trials of a sandwich-building game. In each trial, a drive-thru customer's sandwich order is presented auditorily, including a list of 1 to 10 sandwich ingredients (e.g., salami, cheddar cheese, mayo, and olives). The participant needs to remember these sandwich ingredients and then build the customer's sandwich by selecting the necessary ingredients in the correct order. We evaluate Sandwich Builder as an estimate of auditory short-term memory, finding significant correlations with other classic short-term memory measures. Additionally, we measure affect (i.e., mood), fatigue, and motivation following Sandwich Builder as compared to classic tasks: Participants leave Sandwich Builder in a better mood, less fatigued, and more motivated than they leave other memory tasks. Sandwich Builder shows strong convergent, discriminant, and predictive validity, and moderate-to-good test-retest reliability. Finally, we examine the practicality of Sandwich Builder as a measure of individual differences in speech perception research, providing expected power curves for two common paradigms. Altogether, we robustly demonstrate the utility of Sandwich Builder, which is openly available online.