Losing Track of Time: Computational Cognitive Modeling of Prospective Timing Under Sequential Multitasking

Abstract

While extensive research has examined how interruptions affect task performance, the impact on our time perception remains unexplored. Yet, accurately estimating elapsed time can be critical in domains related to safety, useful in workplaces, and essential for intentional technology use. This thesis investigates whether task interruptions shorten prospective duration judgments through a systematic integration of sequential multitasking and time perception research. We approached this question through three methods: theoretical analysis connecting interruptions to temporal processing via executive resources, a controlled laboratory experiment, and a prototype computational cognitive model. Twenty-six participants completed simple typing tasks that were either interrupted by or performed sequentially with N-back tasks of varying complexity, then provided verbal time estimates of the trial's duration. The computational model, based on ACT-R principles, formalized how cognitive operations supporting task switching compete with operations that sample and encode temporal duration. Results confirmed that interruptions significantly impaired task performance—slower typing, more errors, longer resumption lags—replicating established findings. However, effects on time perception were minimal, with only a marginal trend toward decreased timing accuracy but no systematic underestimation. The computational model successfully produced theoretically predicted patterns of temporal underestimation under interruptions, but these diverged from human behavior. While the weak empirical effects may stem from insufficient cognitive load manipulation or measurement limitations, they also suggest that prospective timing may be more resilient to interruption-based disruption than theory predicts. This work demonstrates the value of theory-driven computational modeling for exploring underexplored cognitive phenomena and informs the design of intelligent systems that support human temporal awareness in multitasking environments.

Code and materials available at: github.com/emfrg/multitasking-n-time-perception-study