Cognitive Bias Affects Perception and Decision-Making in Simulated Facial Recognition Searches

Area of Science:

• Cognitive Psychology
• Forensic Science
• Biometrics

Background:

• Cognitive biases can lead to errors in forensic pattern analysis.
• Facial recognition technology (FRT) is increasingly used in criminal investigations.
• The influence of cognitive biases on FRT accuracy is not fully understood.

Purpose of the Study:

• To investigate whether cognitive biases, specifically contextual and automation biases, impede accurate perpetrator identification using FRT.
• To determine if extraneous information or confidence scores influence FRT-based identification decisions.

Main Methods:

• 149 participants completed two simulated FRT tasks.
• Participants compared a perpetrator's probe image against three candidate faces.
• Candidate faces were randomly paired with guilt-suggestive biographical information or biometric confidence scores.

Main Results:

• Participants were more likely to identify a candidate as the perpetrator when paired with guilt-suggestive information or a high confidence score.
• These biasing details were randomly assigned, indicating a susceptibility to contextual and automation biases.
• Misidentifications were most frequent when candidates were associated with guilt-suggestive information.

Conclusions:

• Cognitive biases, including contextual and automation biases, significantly impact the accuracy of facial recognition technology in identifying criminal perpetrators.
• The study underscores the need for implementing procedural safeguards to mitigate the effects of cognitive bias in FRT applications within criminal justice.
• Ensuring the unbiased application of FRT is crucial for maintaining the integrity of criminal investigations and preventing wrongful identifications.