Does it deliver what it promises? Evaluation of cognitive distraction caused by speech-based interfaces with detection response and box task

Alexandra Loew, Yannick Forster, Frederik Naujoks, Bianca Biebl, Andreas Keinath, Klaus Bengler

Research output: Contribution to journalArticlepeer-review

Abstract

Speech is considered a promising modality for human-machine interaction while driving, especially in reducing visual and manual distraction. However, speech-based user interfaces themselves have shown to increase cognitive distraction. There remains a lack of standardized and unambiguous methods for measuring the impact of speech-based assistants on cognitive distraction while driving. This work aims to investigate whether the combination of the box task and the detection response task (DRT) is a suitable method for assessing the cognitive distraction caused by speech-based assistants. For this purpose, participants (N = 39) engaged in artificial (n-back tasks) and natural speech-based secondary tasks (interaction with Android's Google Assistant and Apple's Siri) differing in predefined levels of cognitive workload while performing the box task and the DRT. The results showed that DRT performance differed between the 0-back and 1-back task but not between the different cognitive workload levels of the speech-based assistants. No clear effects emerged for the box task parameters. Thus, the combination of the box task and DRT is well-suited for measuring cognitive distraction caused by artificial secondary tasks but not by natural interactions with speech-based assistants.

Original languageEnglish
Pages (from-to)1-16
Number of pages16
JournalTransportation Research Part F: Traffic Psychology and Behaviour
Volume91
DOIs
StatePublished - Nov 2022

Keywords

  • Cognitive workload
  • Driver distraction
  • Evaluation methods
  • Speech-based assistants

Fingerprint

Dive into the research topics of 'Does it deliver what it promises? Evaluation of cognitive distraction caused by speech-based interfaces with detection response and box task'. Together they form a unique fingerprint.

Cite this