Time Pressure Increases Automation Reliance in a Face Matching Task

Area of Science:

• Cognitive Psychology
• Human-Computer Interaction
• Biometrics

Background:

• Automated facial recognition (AFR) systems are widely used for identity verification.
• Human operators often rely on AFR systems as decision aids.
• Previous research indicates time pressure negatively impacts human face matching accuracy.

Purpose of the Study:

• To investigate how time pressure affects human reliance on AFR system decisions.
• To determine if increased time pressure leads to greater automation reliance.
• To assess the impact of AFR assistance on human accuracy under varying time constraints.

Main Methods:

• 129 participants completed a face matching task under high (2s), medium (5s), and low (10s) time pressure.
• Participants made unaided decisions, then reviewed an AFR system's decision (92.3% accurate), and submitted an aided decision.
• Stimuli disappeared after the allocated time, simulating real-world operational constraints.

Main Results:

• Human accuracy improved with AFR assistance across all time pressure conditions.
• The greatest accuracy improvements were observed under higher time pressure.
• Aided human accuracy remained lower than the AFR system's standalone accuracy.
• Participants struggled to override incorrect AFR system decisions, leading to decreased accuracy.

Conclusions:

• Increased time pressure enhances human reliance on automated facial recognition systems.
• While AFR assistance improves human performance, it does not fully compensate for time pressure effects.
• Effective human oversight of AFR systems requires strategies to mitigate over-reliance and improve error correction capabilities.