Only an unreliable warning system is a good system – but which errors help?
• The combined effect of missing critical events und false alarms was examined for a pedestrian warning system. • A system with no missings, but 33 % false alarms did not induce a cry-wolf effect but induced possibly dangerous brakings. • The system without false alarms but 33 % misses led to slower and missing reactions in critical situations. • The systems with 70 % hits and 33 % false alarms could counteract this effect. • However, for the drivers, this system is annoying and does not seem very effective. Warning systems have two error types: misses (fail to warn during critical events), being a problem when overtrust develops, and false alarms, causing system mistrust (cry-wolf effect). Mistrust can reduce harm from misses but may also prompt undue reactions which may create new hazards. Prior studies suggest systems with up to 30 % misses and false alarms remain beneficial, but seldom assess both errors jointly. We conducted a driving-simulator study with 47 participants across three reliability conditions: (1) 100 % detection with 33 % false alarms; (2) 70 % detection with no false alarms; (3) 70 % detection with 33% false alarms. Participants encountered 100 pedestrian scenarios: standing or approaching pedestrians that stopped (uncritical event; 90 %) or crossed (critical event; 10 %). In condition 1, 33 % false alarms did not impair reactions to critical events but induced non-critical braking. In condition 2, misses led to minimal missed reactions but significantly delayed braking when warnings were absent. In condition 3, drivers reacted faster to correct warnings than in condition 1 and did not brake more slowly during undetected events, indicating that experiencing both error types reduces overtrust while preserving responsiveness. However, condition 3 was subjectively rated as least useful and most annoying. These results highlight a trade-off between objective effectiveness and user acceptance, suggesting that balanced system reliability can enhance driver response, but at the expense of user satisfaction.
Authors
- Mark Vollrath (ORCID: https://orcid.org/0000-0001-8508-8934)
- Julie Morawietz
Institutions
- Technische Universität Braunschweig (DE)
Publication Details
- Journal
- Transportation Research Part F Traffic Psychology and Behaviour
- Published
- 2025-11-20
- DOI
- https://doi.org/10.1016/j.trf.2025.103446
- Primary Topic
- Human-Automation Interaction and Safety
- Type
- article
- Field-Weighted Citation Impact
- 0.00