Derivation and Demonstration of a New Metric for Multitasking Performance

Abstract

Objective We proposed and demonstrate a theory-driven, quantitative, individual-level estimate of the degree to which cognitive processes are degraded or enhanced when multiple tasks are simultaneously completed. Background To evaluate multitasking, we used a performance-based cognitive model to predict efficient performance. The model controls for single-task performance at the individual level and does not depend on parametric assumptions, such as normality, which do not apply to many performance evaluations. Methods Twenty participants attempted to maintain their isolated task performance in combination for three dual-task and one triple-task scenarios. We utilized a computational model of multiple resource theory to form hypotheses for how performance in each environment would compare, relative to the other multitask contexts. We assessed if and to what extent multitask performance diverged from the model of efficient multitasking in each combination of tasks across multiple sessions. Results Across the two sessions, we found variable individual task performances but consistent patterns of multitask efficiency such that deficits were evident in all task combinations. All participants exhibited decrements in performing the triple-task condition. Conclusions We demonstrate a modeling framework that characterizes multitasking efficiency with a single score. Because it controls for single-task differences and makes no parametric assumptions, the measure enables researchers and system designers to directly compare efficiency across various individuals and complex situations. Application Multitask efficiency scores offer practical implications for the design of adaptive automation and training regimes. Furthermore, a system may be tailored for individuals or suggest task combinations that support productivity and minimize performance costs.