How Confirmation Bias, Tunnel Vision And Institutional Momentum Can Turn An Investigation Into A Conclusion
Executive Summary
An investigation begins with a hypothesis. That is unavoidable.
The danger begins when the hypothesis stops being tested and starts controlling the meaning of everything that follows.
Part Four of My Morocco Observer’s Equal Treatment? investigation examines how confirmation bias, tunnel vision and institutional momentum can cause professionals acting in good faith to become increasingly certain while alternative explanations receive less attention.
The central question is not whether investigators make mistakes. It is whether institutions are designed to discover those mistakes before suspicion becomes punishment.
A file is opened.
The first evidence appears persuasive.
A witness identifies a person.
A telephone number is linked to a name.
A payment seems difficult to explain.
An anonymous source provides information.
A digital message appears to support the theory.
Investigators form a hypothesis.
At first, it is only one possible explanation.
Then the investigation begins to move.
More records are requested.
More people are interviewed.
More data are collected.
Each new fact is interpreted through the existing theory.
A contradiction becomes deception.
Silence becomes suspicious.
A missing record becomes concealment.
An international contact becomes an association.
A denial becomes predictable.
Cooperation becomes an attempt to manage appearances.
The investigation may still be described as open-minded.
But the central question has quietly changed.
It is no longer:
What happened?
It has become:
How did this person do it?
That change can occur without dishonesty.
No investigator needs to fabricate evidence.
No prosecutor needs deliberately to suppress the truth.
No judge needs to approach the case with prejudice.
Professionals can act conscientiously and still become trapped inside a theory that increasingly determines what they see.
This is why tunnel vision is so dangerous.
It does not always feel like error from inside the institution.
It can feel like progress.
A Hypothesis Is Necessary
Investigations cannot begin without direction.
When police receive a report, they must decide which explanation deserves attention.
When a bank encounters an unusual transaction, it must determine whether further questions are necessary.
When a regulator identifies irregularities, it must decide whether they reflect error, negligence or deliberate misconduct.
When prosecutors review a file, they must assess whether the available evidence supports a charge.
A hypothesis helps professionals organise information.
It identifies which records should be obtained, which witnesses should be interviewed and which events may be connected.
Without hypotheses, investigations would become limitless.
Every possibility would remain open.
Every person could be examined.
Every record might appear relevant.
That would be neither practical nor fair.
The problem is therefore not that investigators form theories.
The problem begins when a theory becomes resistant to meaningful challenge.
A responsible hypothesis creates questions.
An entrenched hypothesis begins supplying answers.
The First Story Has an Advantage
The first coherent explanation of an event has unusual psychological power.
Once information has been understood through a particular narrative, later facts are rarely encountered in isolation.
They are compared with what the decision-maker already believes.
This is anchoring.
The original theory becomes a reference point.
Suppose investigators are told that a business owner is laundering money.
A large transfer may then appear suspicious.
An overseas company may appear deliberately complex.
A cash payment may look concealed.
A discrepancy in an invoice may appear intentional.
Each observation can support the original theory.
But imagine that the first information had been different.
Suppose investigators were initially told that the same person operated an international family business with irregular payment cycles and weak administration.
The transfer might then appear commercial.
The foreign company might look ordinary.
The cash payment might reflect the nature of the sector.
The invoice discrepancy might appear careless rather than criminal.
The facts have not changed.
The frame has.
This does not mean that one explanation is necessarily correct.
It means that the first story can shape how every later fact is interpreted.
Confirmation Bias Is Not Corruption

Confirmation bias is sometimes discussed as though it were evidence of bad faith.
It is not.
It is a normal feature of human reasoning.
People tend to notice information supporting an existing belief, search more actively for confirmation and interpret ambiguous facts in ways consistent with what they already expect.
Contradictory information may receive less attention or be explained away.
Belief perseverance can then preserve a conclusion even after some of its foundations have weakened.
These effects are not limited to police officers or prosecutors.
They affect doctors diagnosing patients.
Scientists testing theories.
Auditors examining accounts.
Journalists investigating allegations.
Judges assessing evidence.
Compliance professionals reviewing customers.
The professional nature of the work does not remove cognitive bias.
Expertise can reduce some errors.
It can also make someone more skilled at defending an established interpretation.
The proper response is therefore not to demand that professionals become psychologically perfect.
It is to build procedures that make doubt operational.
When the Question Changes
At the beginning of an investigation, several explanations may remain possible.
A payment may be legitimate or criminal.
A witness may be accurate or mistaken.
A person may use a telephone or merely have access to it.
A message may be literal, coded, exaggerated or written by somebody else.
An inconsistency may reflect dishonesty, confusion or poor memory.
An association may indicate participation, family connection, professional contact or coincidence.
A fair investigation tests these alternatives.
Tunnel vision begins when the institution stops asking which explanation best fits the evidence and starts asking how the evidence can be fitted into the preferred explanation.
The distinction is subtle.
An investigator may sincerely believe they are following the facts.
But the facts selected for attention may already have been shaped by the theory.
A fact supporting the hypothesis is developed.
A fact weakening it remains in unused material.
A witness who confirms the theory is interviewed again.
A witness who contradicts it is considered unreliable.
A new lead is pursued because it might strengthen the case.
An alternative lead is rejected because it appears less promising.
Each individual decision may be rational.
Together, they can make one scenario increasingly visible while competing scenarios gradually disappear.
The Evidence That Never Enters the Story
Criminal investigations and regulatory inquiries generate far more material than eventually appears in a case.
Digital devices can contain thousands of messages.
Financial investigations may involve years of transactions.
Witnesses may provide conflicting accounts.
Surveillance can produce long periods in which nothing relevant occurs.
Investigators must decide what matters.
That selection is unavoidable.
But unused material may contain the evidence most capable of challenging the central theory.
A message that changes the meaning of another message.
A location record inconsistent with the alleged event.
A witness statement naming somebody else.
A failed identification.
A transaction following an ordinary pattern.
An expert limitation that weakens the certainty of a technical conclusion.
The problem is not always that material is deliberately concealed.
Sometimes its significance is simply not recognised.
A professional who already accepts the central theory may genuinely regard a contradictory item as irrelevant.
The same cognitive frame that created the risk can therefore influence the decision about what the defence, regulator or reviewing authority needs to see.
The Digital Evidence Paradox

Modern technology records more human activity than ever before.
Messages.
Locations.
Searches.
Payments.
Photographs.
Contacts.
Call records.
Vehicle movements.
Cloud data.
Security-camera footage.
This can make investigations more accurate.
A digital record may expose a false account.
It may connect participants, reconstruct movements or establish a chronology that memory cannot provide.
But volume creates a new problem.
The more data investigators possess, the more opportunities they have to find material supporting almost any theory.
A single sentence may appear decisive when removed from a conversation containing thousands of messages.
A location point can appear incriminating without establishing who carried the device.
Contact between two people can prove communication without proving its purpose.
A nickname can be attributed to a person through several indicators without any individual indicator being conclusive.
A repeated pattern may be significant.
It may also emerge because investigators searched the material with one hypothesis already in mind.
Digital evidence often feels objective because a machine preserved it.
But machines preserve data, not meaning.
Human beings still decide:
Who used the device?
What did the language mean?
Which messages matter?
Which dates define the relevant period?
What alternative explanations are realistic?
What missing information would change the conclusion?
Technology can reduce uncertainty about what was recorded.
It cannot automatically resolve uncertainty about interpretation.
When Intelligence Becomes Evidence
Intelligence serves a different purpose from courtroom proof.
It may identify threats, relationships or investigative directions before information can be fully verified.
An anonymous source may provide an important lead.
A confidential report may identify a possible participant.
A database connection may show that two people were associated with the same address, vehicle or telephone.
Such information can legitimately guide an investigation.
But intelligence frequently contains uncertainty.
The source may be mistaken.
The information may be incomplete.
The motive behind it may be unknown.
The connection may be indirect.
The record may be outdated.
The danger arises when the origin and limitations of intelligence disappear as it moves through a file.
“Unverified information” becomes “intelligence”.
“Intelligence” becomes “known association”.
“Known association” becomes part of the factual narrative.
What began as a reason to investigate can eventually be presented as part of the evidence produced by that investigation.
The label becomes stronger.
The underlying information does not.
Institutional Momentum

An investigation is not only an intellectual process.
It is an organisation at work.
Time is allocated.
Specialists are assigned.
Surveillance may be authorised.
Property may be seized.
Senior officials may be briefed.
Prosecutors may approve charges.
Public statements may be made.
Once an institution has invested heavily in one direction, reversal becomes costly.
Not only financially.
Psychologically and professionally.
Changing course may require people to accept that earlier judgments were mistaken.
It may require an institution to release property, discontinue proceedings, correct a public statement or acknowledge that intrusive powers were used without sufficient foundation.
The longer the process continues, the more the investigation itself begins to resemble evidence.
The person has been investigated for two years.
Surely there must be something.
Several departments have examined the file.
Surely the theory cannot be baseless.
Senior officials approved the measures.
Surely the original information was strong.
But duration is not evidence.
Institutional agreement is not independent confirmation when every participant relies on the same underlying assumptions.
The number of people accepting a theory says little when they are all reading versions of the same file.
The Cost of Turning Back
Institutions often describe correction as a sign of strength.
In principle, it is.
In practice, admitting error can trigger consequences.
Compensation claims.
Disciplinary questions.
Political criticism.
Reputational harm.
Damage to public confidence.
Reviews of similar cases.
The possibility that one mistake was not isolated.
These consequences can create an incentive to defend an earlier decision for longer than the evidence justifies.
Again, no conspiracy is required.
An institution may sincerely believe that defending its earlier actions protects public confidence.
But confidence built on refusal to reconsider is fragile.
The legitimacy of a justice system does not depend on never making mistakes.
No system can promise that.
It depends on whether the system possesses the intellectual and procedural capacity to identify mistakes before they become irreversible.
The Self-Sealing Theory
A healthy hypothesis can be disproved.
A self-sealing theory cannot.
Every possible response becomes evidence supporting the same conclusion.
If a person refuses to answer, the silence appears suspicious.
If the person answers in detail, the explanation appears rehearsed.
If documents are provided, they may be described as having been created to justify the activity.
If documents are missing, their absence becomes concealment.
If the person has contact with an alleged participant, that becomes association.
If the person denies contact, the denial becomes deception.
If the person behaves calmly, they appear controlled.
If they appear distressed, they look nervous because they know more.
A theory that interprets every outcome as confirmation has stopped functioning as an empirical theory.
It has become institutionally unfalsifiable.
That is the moment an investigation becomes most dangerous, because the person is no longer able to demonstrate innocence when the system has lost any agreed understanding of what innocence would look like.
The person is not merely struggling to disprove an allegation.
They are struggling against a framework in which every attempt at disproof can be absorbed as further confirmation.
Alternative Explanations Must Be Real, Not Ceremonial
Investigative files often state that alternative scenarios were considered.
The important question is how.
Were investigators merely able to imagine another explanation?
Or did they actively test it?
Did they seek information capable of disproving the central theory?
Did somebody independent review the assumptions?
Were contradictory witnesses examined seriously?
Was an alternative digital attribution tested?
Were expert conclusions presented with their actual limitations?
Could investigators identify the evidence that would cause them to abandon their preferred explanation?
An alternative hypothesis does not deserve equal resources merely because it is imaginable.
Some scenarios are remote.
Some are contradicted by strong evidence.
Some are raised only to create confusion.
But dismissing a reasonable alternative should require reasons connected to evidence.
It should not be rejected merely because the investigation has already travelled too far in another direction.
The Duty to Pursue Reasonable Lines of Inquiry
A fair system does not require investigators to explore every theoretical possibility.
It does require them to pursue reasonable lines of inquiry, including those capable of pointing away from the suspect.
Modern disclosure rules in England and Wales provide one explicit example.
They require continuing review when new evidence or unused material emerges, when the defence identifies relevant issues or when further reasonable searches are requested.
They also recognise that digital investigations require careful search strategies, record-keeping and repeated consideration of what may assist the defence.
These safeguards matter far beyond one legal system.
The underlying principle is universal.
An investigation should not be evaluated only by the strength of the evidence it assembled.
It should also be evaluated by the seriousness with which it searched for evidence capable of proving itself wrong.
Disclosure Is Not a Technical Afterthought
Disclosure can sound administrative.
A schedule.
A document category.
A digital search.
A procedural deadline.
But disclosure is one of the principal mechanisms through which institutional certainty can be tested.
The defence does not need access only to the material the prosecution considers persuasive.
It may need material that changes context, supports an alternative explanation, challenges reliability or reveals that an investigative assumption was weaker than presented.
European fair-trial principles recognise that the ability to know and challenge the evidence is central to adversarial proceedings and equality of arms.
Restrictions can be justified in limited circumstances.
Witnesses may require protection.
Investigative methods may need to remain confidential.
National security may be involved.
But the institution constructing the allegation cannot always be the sole and final judge of which contradictory information remains hidden.
Independent judicial control and effective counterbalancing safeguards are essential.
The purpose is not to overwhelm the defence with every irrelevant file.
Nor is it to expose confidential techniques unnecessarily.
It is to prevent the institution that built the allegation from maintaining exclusive control over the information capable of dismantling it.
When the Defence Is Treated as an Obstacle
A defence lawyer’s role is not to obstruct the truth.
It is to test whether the state has established its allegations lawfully and reliably.
That may involve challenging witnesses.
Requesting unused material.
Questioning digital attribution.
Identifying investigative gaps.
Offering an alternative chronology.
Examining the reliability of experts.
Arguing that intelligence has been mistaken for evidence.
These challenges can feel frustrating to investigators already convinced of the case.
A request for additional inquiry may appear tactical.
A demand for disclosure may appear excessive.
An alternative theory may appear manufactured.
Sometimes those assessments will be correct.
Defence strategies can be speculative.
Requests can be disproportionate.
Proceedings can be delayed unnecessarily.
But a system that treats challenge itself as evidence of bad faith removes one of its most important protections against tunnel vision.
The defence does not need to be correct for the challenge to be valuable.
It forces the institution to explain why it believes it is correct.
The Judge Is Not Outside the System
Courts provide independent review.
They hear the prosecution and the defence.
They test legal thresholds.
They can exclude evidence, reject arguments and acquit where proof is insufficient.
But judges encounter a case after it has already been shaped.
The investigation determines which people were interviewed.
The prosecution determines which charges were brought.
The file determines much of what reaches the courtroom.
The parties determine which issues are contested.
A judge can critically examine the case presented.
They cannot automatically reconstruct every investigative path that was never pursued.
Judges are also human.
They can be influenced by anchoring, framing, repetition and belief perseverance.
That is not an argument against judicial independence.
It is an argument for procedures that expose judges to the strongest available challenge, rather than only the most coherent institutional narrative.
An impressive dossier is not necessarily a complete dossier.
A consistent narrative is not necessarily a correct one.
A large quantity of evidence is not necessarily a large number of independent confirmations.
The Schiedam Park Murder: When the Theory Took Control

The Netherlands does not need to treat tunnel vision as an abstract psychological possibility.
It has already seen what can happen when an investigative theory becomes stronger than the evidence challenging it.
In the Schiedam park murder case, Kees B. was convicted of killing and sexually assaulting a ten-year-old girl and attempting to kill an eleven-year-old boy.
He had been near the scene and called the emergency services after the surviving child asked him for help.
He initially confessed, but later withdrew those confessions.
According to the Dutch judiciary’s own public account, police and prosecutors subsequently worked towards his conviction through tunnel vision.
Evidence pointing towards him was embraced.
Exculpatory information was not taken seriously or was explained away.
Approximately four years later, the actual perpetrator confessed.
The significance of the case extends beyond one wrongful conviction.
It demonstrates how an early suspect, a confession and a coherent theory can begin to control the interpretation of everything that follows.
The problem was not simply that one piece of evidence proved unreliable.
The investigative system had lost sufficient capacity to treat contradictory information as a reason to reconsider its central theory.
The case led to an official evaluation and recommendations intended to reduce the risk of future miscarriages of justice.
But its lasting lesson is more fundamental:
A system can follow procedures, assemble a substantial file and reach judicial certainty while still being wrong about the person at the centre of the case.
What Investigators and Prosecutors Are Entitled to Say
Fair journalism must acknowledge the strongest institutional response.
Serious investigations are difficult.
Criminal organisations conceal evidence, intimidate witnesses, use intermediaries and communicate through coded or encrypted systems.
Fraud can involve complex companies, false documentation and transactions across multiple countries.
Investigators frequently work under pressure.
They must make decisions before every fact is known.
Digital evidence can be vast.
Alternative scenarios can be invented strategically.
Confidential sources and methods cannot always be disclosed.
Victims and the public have a legitimate interest in effective enforcement.
Prosecutors should not abandon a well-supported case merely because the accused provides another explanation.
All of this is true.
Tunnel vision should not become a slogan used to dismiss every strong investigation.
The existence of an investigative theory does not make the investigation biased.
Consistency across evidence may genuinely confirm a hypothesis.
Multiple records may provide truly independent support.
A denial may be demonstrably false.
An alternative scenario may be incompatible with objective facts.
The demand is not permanent uncertainty.
It is disciplined certainty.
What Institutions Must Be Able to Demonstrate
An institution confident in its conclusion should be able to answer difficult questions.
What was the original hypothesis?
What information created it?
Which parts were independently verified?
Which records came from the same source?
What evidence contradicted the theory?
How was that evidence assessed?
Which alternative explanations were pursued?
Why were others rejected?
What information would have caused investigators to change direction?
Were experts shown only selected material or the broader context?
Could investigators distinguish between evidence of association and evidence of participation?
Was digital material interpreted in context?
Were the limitations of attribution recorded?
Was unused material preserved and reviewed?
Could the defence meaningfully challenge confidential information?
Did anybody outside the core team test the theory?
Were major decisions reconsidered when the evidence changed?
These questions do not weaken investigations.
They distinguish investigations designed to establish the truth from investigations designed to prove an initial belief.
The Value of the Contrarian

Institutions often assign people to challenge decisions.
A second reader.
An independent prosecutor.
A peer-review panel.
A devil’s advocate.
A red team.
A judge authorising intrusive measures.
These mechanisms work only when disagreement is genuinely permitted.
A reviewer who receives the same summary, inherits the same assumptions and faces pressure to approve quickly may add procedure without adding scrutiny.
Effective challenge requires access to the foundation of the theory.
It requires institutional permission to say:
This association is weaker than it appears.
These records are not independent.
This witness changed their account.
This message may have another meaning.
This expert conclusion contains uncertainty.
This alternative has not been tested.
The case may remain strong after those challenges.
If so, it becomes stronger for having survived them.
The Warning From Wrongful Convictions
European legal systems contain mechanisms for reviewing possible miscarriages of justice because final judgments can be wrong.
Witnesses can be mistaken.
Confessions can be unreliable.
Experts can overstate conclusions.
Investigators can overlook information.
Prosecutors can fail to disclose relevant material.
Courts can accept a coherent but incorrect narrative.
The existence of review procedures is not proof that the justice system generally fails.
It reflects the opposite principle.
A serious legal system understands that its decisions require routes for correction precisely because its institutions are powerful and human.
The Schiedam case was not the only European warning.
But it provides a lesson that every institution should understand.
Error rarely announces itself as error.
From inside the process, it can resemble evidence accumulating, professionals agreeing and procedures being followed.
That is why confidence alone can never be the final safeguard.
When the System Stops Questioning Itself
The greatest danger does not arise when an investigator forms a hypothesis.
It arises when the institution forgets that it remains a hypothesis.
A suspicion becomes a working theory.
The theory shapes the investigation.
The investigation creates a dossier.
The dossier gives the theory authority.
Senior approval gives it legitimacy.
Time gives it permanence.
The prosecution gives it legal form.
Eventually, questioning the conclusion begins to look less like caution and more like disloyalty to the work already completed.
That is when the system stops questioning itself.
The institution may still hold meetings.
It may still record decisions.
It may still follow procedures.
But procedure alone cannot prevent tunnel vision when every stage inherits the same untested assumption.
The essential safeguard is not endless doubt.
It is the continuing existence of a meaningful path from certainty back to uncertainty.
A path through which contrary evidence can enter.
A path through which an investigator can change direction without humiliation.
A path through which a prosecutor can reconsider without appearing weak.
A path through which a court can see what the investigation failed to pursue.
A path through which an institution can say:
We believed this.
We investigated it.
The evidence did not ultimately support it.
We were wrong.
That statement does not weaken the rule of law.
The inability to make it does.
Because the strength of an institution is not measured only by how confidently it reaches a conclusion.
It is measured by whether it remains capable of questioning that conclusion after everybody inside the system has begun to believe it.
This is Part Four of MMO’s “Equal Treatment?” investigation.
Sources & Methodology
This investigation draws on official judicial research, European human-rights standards, public disclosure guidance and institutional reviews concerning confirmation bias, belief perseverance, tunnel vision, unused material and fair-trial safeguards.
The principal sources include:
– Research Memoranda 2025: Judicial Decision-Making, published through the Dutch judiciary. The research examines confirmation bias, belief perseverance and other cognitive mechanisms capable of influencing professional judgment. (Rechtspraak)
– The Dutch judiciary’s public account of the Schiedam park murder and tunnel vision. It describes how incriminating information was embraced while exculpatory material was not taken seriously or was explained away, before the actual perpetrator later confessed. (Rechtspraak)
– The Dutch judiciary’s broader information concerning miscarriages of justice, including the Schiedam park murder, the Putten murder case and the case of Lucia de Berk. (Rechtspraak)
– Attorney General’s Guidelines on



