Case File 070 · Investigation and Psychology

How Confirmation Bias Can Affect an Investigation

How early beliefs can narrow an investigation into tunnel vision, what experiments show about its effects, and which safeguards are proposed.

By Murder Book Editorial TeamPublished 16 September 2026Updated 16 September 2026

Confirmation bias affects an investigation when an early belief about what happened, or who did it, starts to steer how evidence is gathered and interpreted. Instead of testing that belief, investigators may look mainly for information that supports it, read ambiguous evidence as confirmation, and give less weight to evidence that points elsewhere. In criminal cases this pattern is often called tunnel vision. As summarised by psychologists Saul Kassin, Itiel Dror and Jeff Kukucka, legal scholars Keith Findley and Michael Scott describe it as a rigid focus on one suspect that leads investigators to favour incriminating evidence while overlooking evidence that could clear them. It is not usually deliberate misconduct. Research shows it arises from ordinary thinking processes that affect everyone, including experts. It can shape which questions are asked in an interview, how a suspect's reactions are read, and even how forensic evidence is judged.

Key takeaways

Confirmation bias is the tendency to seek, interpret and remember information in ways that fit what we already believe.
In criminal cases it can become tunnel vision: a narrow focus on one suspect or theory that filters out contradicting evidence.
In an experiment, interrogators who expected guilt asked more guilt-presumptive questions and applied more pressure, especially to innocent suspects.
Contextual information, such as knowing a suspect confessed, can change how people judge forensic evidence.
Suggested safeguards include blind testing, independent review, and actively considering alternative explanations.

What is confirmation bias in an investigation?

Kassin, Dror and Kukucka use the term forensic confirmation bias for "the class of effects through which an individual's preexisting beliefs, expectations, motives, and situational context influence the collection, perception, and interpretation of evidence during the course of a criminal case." Their 2013 article in the Journal of Applied Research in Memory and Cognition draws on decades of psychology research on first impressions, expectancy effects and observer effects.

A 2018 report for Canada's Federal/Provincial/Territorial Heads of Prosecutions, Innocence at Stake, published by the Public Prosecution Service of Canada, defines tunnel vision in similar terms: a tendency of participants in the system, such as police or prosecutors, "to focus on a particular theory of a case and to dismiss or undervalue evidence which contradicts that theory."

Why does tunnel vision happen?

The Canadian report stresses that tunnel vision grows out of normal human thinking rather than bad character. It points to three sources:

Confirmation bias: unconsciously preferring information that supports a conclusion already reached.
Hindsight bias: believing, after the fact, that an outcome was more predictable than it really was.
Mental shortcuts: the heuristics people rely on when they are under time pressure and heavy workloads.

It adds that institutional pressures can reinforce these tendencies, including pressure to secure convictions, public scrutiny and high caseloads.

How can bias enter at each stage?

Bias does not need to appear all at once. It can build as a case moves forward.

Stage	How bias can enter	Research example
Choosing a suspect	An early impression, such as nervous behaviour, marks someone as the likely culprit	People judge lies and truths correctly only about 54% of the time (Bond and DePaulo)
Questioning	Expectations shape which questions are asked and how hard a suspect is pressed	Kassin, Goldstein and Savitsky (2003)
Evaluating evidence	Knowing about a confession or other context changes how ambiguous evidence is judged	Kukucka and Kassin (2014)
Reviewing the case	Evidence that fits is recorded and emphasised; evidence that does not is explained away	Described in the Innocence at Stake report

Can an interrogator's expectations change a suspect's behaviour?

A 2003 experiment by Saul Kassin, Christine Goldstein and Kenneth Savitsky tested this directly. Fifty-two participants, some guilty and some innocent of a mock theft, were questioned by 52 interrogators. The interrogators had been led to believe that most of the suspects were either guilty or innocent.

Interrogators who expected guilt chose more guilt-presumptive questions, used more interrogation techniques, were more likely to judge the suspect guilty, and put more pressure on the suspect to confess. This was especially true when they were questioning suspects who were actually innocent.

In a second phase, neutral observers listened to the recordings. They perceived suspects in the guilty-expectation condition as more defensive, and as somewhat more guilty. The authors describe this as behavioural confirmation: the expectation influences the interrogator's behaviour, which influences the suspect's behaviour, which then appears to confirm the original expectation.

This is one reason behaviour in an interview is weak evidence. A defensive reaction can be a response to the way someone is being questioned. See why there is no single reliable sign of deception for more on this.

Can context affect forensic evidence?

Many people assume forensic analysis is immune to this problem. Research suggests otherwise. Kassin, Dror and Kukucka open their 2013 article with a high-profile fingerprint misidentification in which an official review listed confirmation bias as a contributing factor. They then review studies showing that judgements in several pattern-matching disciplines, such as fingerprints, can be influenced by irrelevant contextual information.

In a 2014 experiment, Jeff Kukucka and Saul Kassin gave participants a case summary and asked them to compare handwriting samples. Participants who read that the defendant had confessed were more likely to wrongly conclude that the samples were written by the same person, and more likely to judge the defendant guilty, than those who were not told about a confession.

This matters because confessions are not always reliable, as covered in why innocent people sometimes confess. If a confession then shapes how other evidence is read, a single error can appear to be supported by several independent pieces of evidence.

What safeguards are suggested?

Researchers and prosecutors have proposed several ways to reduce the risk. None of them removes bias completely.

Blind procedures. Kassin, Dror and Kukucka recommend shielding forensic examiners from information they do not need, such as whether a suspect has confessed. They also suggest that verification should be blind where possible, so the second examiner does not know the first examiner's conclusion.

Evidence line-ups. Borrowing from eyewitness line-ups, they propose that an examiner compare crime scene evidence with several samples, only one of which comes from the suspect, instead of a single suspect sample.

Independent review and challenge. The Innocence at Stake report lists measures such as second opinions and case reviews, formal "devil's advocate" roles in prosecution offices, training on tunnel vision, and structured case management for police.

Considering alternatives. A 2026 study in PLOS ONE by Lenz and colleagues tested brief interventions with 293 current and future police officers in Germany. Participants were asked to focus either on a suspect or on reconstructing scenarios from the evidence, and either to confirm or to try to falsify their assumptions. The short interventions did not change how guilty officers judged an innocent suspect to be. They did change the next investigative steps officers proposed: an evidence focus led to more proposals for objective investigation, and a falsification focus encouraged more critical testing of assumptions.

Why does recording eliminated suspects matter?

A disciplined case file can act as a check on tunnel vision. The Los Angeles Police Department's murder book format, described in What Is a Murder Book?, includes a section for eliminated suspects. Documenting who was considered and why they were ruled out makes an investigation's reasoning visible to others. Comparing accounts carefully, rather than fitting them to a theory, is part of the same discipline; see how investigators find inconsistencies in a story.

How does this connect to Murder Book?

Murder Book: The Art of Interrogation is an upcoming narrative interrogation game from Gungrounds in Zagreb, built around unreliable stories, behavioural changes and contradictions. Development screenshots show an evidence board that links statements to each other, and a case select screen whose results include "not guilty".

Tunnel vision is a trap for players as much as for investigators. Once you decide a character is lying, every pause can start to look like proof. The research suggests a better habit: treat a behaviour as one clue among many, ask what else could explain it, and let a contradiction raise a question before it settles an answer. Murder Book is fiction, but that kind of uncertainty is what makes reading a suspect interesting. See how Murder Book turns interrogation into gameplay for more.

Murder Book is in development for PC (Steam) and mobile, with release currently planned for the end of Q2 2027.