Case File 286 · Investigation Psychology

How Do Investigators Test Competing Hypotheses?

By Murder Book Editorial TeamPublished by Gungrounds, the studio making Murder BookPublished 3 October 2026Updated 3 October 2026

Investigators test competing hypotheses by evaluating multiple possible explanations for the evidence side by side, rather than picking one theory and looking for proof. The most widely cited framework for this is the Analysis of Competing Hypotheses (ACH), developed by CIA analyst Richards Heuer in the 1970s. The method asks investigators to list all plausible explanations, lay out the evidence, and assess which hypothesis is most consistent with the full body of facts while paying special attention to evidence that rules hypotheses out. The goal is to reduce the influence of confirmation bias and avoid the tunnel vision that research has linked to wrongful convictions.

Key takeaways

Testing competing hypotheses means evaluating several explanations at once rather than building a case for a single theory.
The Analysis of Competing Hypotheses (ACH), created by Richards Heuer, is a structured method that uses a matrix to compare evidence against each hypothesis.
The method emphasises disconfirming evidence: the strongest test of a hypothesis is finding facts that are inconsistent with it.
Research by O'Brien (2009) found that asking investigators to generate reasons why their initial suspect might be innocent significantly reduced confirmation bias.
Maintaining competing hypotheses is a safeguard against investigative tunnel vision, which Findley and Scott (2006) identified as a central factor in wrongful convictions.

What is the Analysis of Competing Hypotheses?

The Analysis of Competing Hypotheses (ACH) is a structured analytical technique developed by Richards J. Heuer Jr. during his career at the Central Intelligence Agency. Heuer published the method in his book Psychology of Intelligence Analysis (1999), which remains a foundational text in the intelligence community and has influenced investigative practice more broadly.

The core idea is straightforward. Instead of starting with a preferred hypothesis and looking for evidence to support it, the analyst identifies all plausible hypotheses, then systematically evaluates the evidence against each one. The method uses a matrix: hypotheses across the top, evidence items down the side, and each cell marked as consistent, inconsistent or irrelevant.

The critical step is focusing on inconsistency. Heuer argued that the hypothesis with the most inconsistent evidence should be rejected, rather than selecting the hypothesis with the most supporting evidence. This distinction matters because supporting evidence is often ambiguous, fitting multiple hypotheses equally well. Evidence that could support more than one theory is common in complex investigations, which is precisely why disconfirmation is a stronger test.

How does this apply to criminal investigation?

Criminal investigations face the same cognitive challenges that Heuer identified in intelligence analysis. Investigators form an initial impression of a case early, often based on limited evidence, and that impression shapes how they interpret everything that follows. O'Brien (2009), publishing in Psychology, Public Policy, and Law, demonstrated this experimentally.

In the study, participants reviewed a mock police file that contained weak circumstantial evidence pointing to one suspect. Partway through, a second, stronger suspect emerged. Participants who had been asked to form an initial hypothesis tended to stick with their first suspect even after encountering the stronger evidence. In a second experiment, participants who were asked to generate reasons why their initial suspect might be innocent showed significantly less confirmation bias.

This finding mirrors the ACH principle: the antidote to premature commitment is not simply having more evidence, but actively looking for reasons why the current theory might be wrong. For a deeper look at why this is psychologically difficult, see Why Is It Hard to Abandon an Investigation Theory?.

What are the steps for testing competing hypotheses?

Heuer's original ACH framework involves several steps, adapted here for an investigative context:

List all reasonable hypotheses. Include explanations that seem unlikely but are not impossible. The point is to avoid narrowing the field prematurely.

Gather the significant evidence. This includes physical evidence, witness statements, forensic results, alibis and any other relevant information.

Build a matrix. Place each hypothesis across the top and each piece of evidence down the side. For each cell, ask: is this evidence consistent with, inconsistent with, or irrelevant to this hypothesis?

Focus on inconsistencies. Look for evidence that rules out one or more hypotheses. A single strong inconsistency is more diagnostic than several pieces of supporting evidence.

Reassess the surviving hypotheses. After removing the hypotheses with the most inconsistent evidence, examine what remains. If multiple hypotheses survive, identify what additional evidence would distinguish between them.

Monitor for new evidence. As the investigation continues, new facts may change the matrix. A hypothesis that was previously consistent with all evidence may become inconsistent with a new finding.

Question assumptions. Some cells in the matrix may rely on assumptions rather than established facts. Identifying and testing those assumptions can reveal weaknesses in the analysis.

Why is disconfirming evidence more useful than confirming evidence?

This is a central insight of hypothesis testing in both science and investigation. Confirming evidence is often ambiguous: a suspect's fingerprint at a scene supports the hypothesis that they committed the crime, but it also supports the hypothesis that they visited the location for an innocent reason. The fingerprint is consistent with multiple explanations.

Disconfirming evidence is more decisive. If a suspect's DNA is absent from a crime scene where the perpetrator would certainly have left biological traces, that is inconsistent with the hypothesis that the suspect was the perpetrator. It does not prove innocence, but it weakens the hypothesis in a way that no amount of confirming evidence can match.

Heuer emphasised this asymmetry because of how confirmation bias works. People naturally seek and notice information that supports their existing beliefs. Deliberately looking for inconsistencies is an unnatural cognitive act, which is why it must be structured into the process.

What goes wrong when investigators do not test competing hypotheses?

Findley and Scott (2006), in their analysis of wrongful conviction cases, found that tunnel vision was a recurring theme. Investigators settled on a suspect early, interpreted ambiguous evidence as confirming guilt, and failed to pursue alternative leads. Once the theory of the case was set, the investigation became an exercise in building support rather than testing possibilities.

Rossmo (2009) described the same pattern in Criminal Investigative Failures, identifying three categories of error: cognitive biases such as tunnel vision, organisational pressures such as groupthink, and probability errors in evaluating forensic evidence. In each category, the common thread was a failure to maintain and test alternative explanations.

The consequences are serious. When an investigation commits to the wrong theory, the real perpetrator may remain free, evidence may be lost as time passes, and innocent people may be convicted. Considering alternative explanations is not a procedural formality; it is a safeguard against these outcomes.

How does this connect to Murder Book?

Murder Book: The Art of Interrogation is an upcoming narrative interrogation game from Gungrounds in which players must weigh competing explanations for what they hear and see. Development screenshots show an evidence board linking statements to physical evidence, and a technique menu that includes options such as Confrontation and Baiting alongside Build Trust. These tools suggest a game built around testing accounts against evidence rather than simply extracting confessions.

The branching case map shown in development materials, with multiple possible endings including a "not guilty" result, reinforces the idea that the evidence may genuinely support more than one conclusion. Players who commit to one theory without testing it against alternatives risk reaching the wrong outcome.

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