Forensic evidence feels objective, but much of it involves human judgement — and human judgement can be nudged by expectation. Recognising cognitive bias is one of the most important developments in modern forensic science.
The main biases
- Contextual bias — irrelevant case information (that a suspect confessed, say, or is already in custody) shapes how an examiner interprets the evidence.
- Confirmation bias — once an examiner forms an expectation, they unconsciously weight the evidence toward confirming it.
- Expectation and reference bias — starting a comparison from the "answer" (a database candidate, a named suspect) rather than from the evidence itself.
The case that made it real
The Brandon Mayfield fingerprint misidentification is the textbook example: an initial (wrong) candidate shaped every subsequent judgement, and examiners explained away real differences to fit it. It showed that bias can defeat even careful, experienced experts.
What's being done
The response is procedural, not a matter of "trying harder":
- Linear, blind approaches — analyse the crime-scene mark and record its features before seeing the suspect's, so the reference can't steer the analysis (sometimes called sequential unmasking).
- Blind verification — a second examiner who doesn't know the first's conclusion.
- Context management — keeping domain-irrelevant information away from the examiner.
- Honest reporting of uncertainty and error rates.
Why it matters
Acknowledging bias isn't an admission that forensic science is unreliable — it's what makes it more reliable. The disciplines that have confronted bias head-on, and built procedures to contain it, are exactly the ones a court can trust most.