Post cover image

July 30, 2026

What is Trust?

What kind of relation comes into existence when one actor trusts another?

By Gael Van Weyenbergh

13 min read

We use the word trust everywhere.

We trust people, institutions, information and technology. Governments seek to restore trust. Companies promise to build it. Researchers measure it. A fall in trust is treated as a warning about the health of a relationship, an organisation or an entire society.

Yet these uses do not always describe the same thing.

A bridge may be reliable. An institution may exercise power fairly. A friend may keep a confidence. All three give us reasons to proceed, but the nature of our reliance differs in each case.

This ambiguity is a reason to ask more precisely what comes into existence when we trust.

The standard definition

One of the most influential definitions in organisational research comes from Roger Mayer, James Davis and F. David Schoorman. They describe trust as a willingness to become vulnerable to another party on the expectation that the other will perform an important action, even when they cannot be monitored or controlled.

A major cross-disciplinary synthesis by Denise Rousseau, Sim Sitkin, Ronald Burt and Colin Camerer later arrived at a similar formulation:

Trust is a psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behavior of another.

These definitions established a strong common ground. Trust involves vulnerability. It arises under uncertainty and interdependence. It depends on expectations about what another actor will do. It is not identical to cooperation, which can be produced by coercion or calculation, nor to the act of taking a risk.

This is an important achievement. It removes trust from the comfortable but imprecise language of virtue. It tells us what must be present for trust to arise.

But it describes trust mainly from one side.

The trustor is willing to become vulnerable. The other party appears as the object of an expectation. We learn that a risk has been accepted, but less about what has been placed at risk, what kind of power the trusted party receives, or what changes between them.

The definition brings us to the threshold of trust.

It does not yet take us across.

When another's freedom matters to us

Imagine giving a neighbour the key to your home while you are away.

You may know them well. They may have helped you before. You may have every reason to expect that they will water the plants, lock the door and use the key only as agreed.

By handing it over, however, you do more than accept a risk.

You allow another person's judgement to enter a protected part of your life. They will decide when to enter, what to notice and how to respond if something unexpected happens. The home remains yours. The freedom to act is temporarily theirs.

Something you value now depends on what they choose to do.

This suggests a fuller definition:

Trust is the relationship that arises when we allow something we value to depend on another's freedom to act, expecting them to take our vulnerability into account.

Or, more simply:

We trust when we allow another's freedom to matter to our future.

This definition retains uncertainty, vulnerability and positive expectation. But it places them inside a relation.

Something is valued, so loss is possible. Another actor has room to act, so the outcome cannot be guaranteed. The trustor nevertheless allows that freedom to affect their future. In doing so, they expect the trusted party not merely to behave predictably, but to keep what has been entrusted within their field of concern.

That final element is essential.

I do not need to expect that you will always place my interests above your own. I do expect that my vulnerability will count in how you decide.

Trust is an unguarded boundary

Trust is an unguarded boundary.

The boundary separates what we keep under our own authority from what another person is permitted to affect. It may concern a home, a confidence, a decision, a body, a piece of work, a child or part of a shared future.

Unguarded does not mean unprotected.

We may set conditions, keep a spare key, agree on a purpose, verify competence or preserve a route of recourse. These safeguards can make trust more warranted and limit the harm of a breach. They define the opening. They do not act in the other person's place.

At the boundary itself, some freedom remains.

If every action were prescribed, observed and enforced, the relationship would no longer depend on trust in the same way. The other party might comply, but they would exercise little or no judgement on our behalf. Complete control would not perfect trust. It would change the nature of the relation.

Trust therefore does not require the absence of safeguards. It begins where safeguards leave another actor with discretion over something that matters to us.

The boundary is unguarded not because we have abandoned judgement, but because judgement cannot remove the other's freedom.

Uncertainty is not only missing information

Trust is often treated as a response to insufficient knowledge. If only we had more information, better credentials or a longer record, uncertainty would fall and trust would become easier.

Up to a point, this is true.

Experience gives us reasons. A professional qualification can establish competence. Repeated conduct can reveal care, honesty or reliability. Evidence matters because trust without grounds easily becomes credulity.

But the uncertainty at the centre of trust is not only an information gap.

It also arises because the other remains an agent.

No record of the past can fully determine how someone will respond to a situation that has not yet occurred. Even a person we know intimately may surprise us, fail under pressure, revise their priorities or exercise sound judgement in a way we did not foresee.

Information can orient trust. It cannot close the future without removing the freedom that makes trust necessary.

Guido Möllering describes trust as involving a moment of suspension: reasons and routines carry us towards trust, but cannot eliminate the remaining uncertainty. Something still has to be crossed.

Trust is how a relationship carries its past into a future it cannot guarantee.

More than a positive expectation

We form confident expectations without trusting.

We may expect an adversary to exploit a weakness, a train to follow its timetable or a supplier to honour a contract because the penalty for breach is severe. The prediction may be well founded. It does not follow that trust is present.

The difference lies in what we expect to move the other party.

Annette Baier placed goodwill at the centre of trust and described it as accepted vulnerability to another's possible ill will. Karen Jones later emphasised optimism that another person's goodwill and competence will extend to the domain in which we depend on them. Russell Hardin proposed that the trusted party has reason to take the trustor's interest seriously because it has become part of their own interest in maintaining the relationship.

These accounts differ, but they point towards a shared distinction. Trust is not simply the expectation of a favourable outcome. It is the expectation that the fact of our dependence will have some bearing on how the other acts.

We can call this regard.

Regard does not require affection, self-sacrifice or perfect alignment. A surgeon need not love a patient. A public official need not share a citizen's preferences. A business partner may properly defend interests that conflict with our own.

What matters is that the vulnerability entrusted to them is not treated as irrelevant.

They understand what has entered their hands. They allow that fact to count.

Trust changes both parties

The usual image of trust focuses on the person who takes the risk.

But trust also changes the position of the person trusted.

They receive discretion over something another person values. That discretion may be narrow or extensive, temporary or enduring. It may be accepted explicitly, as when someone agrees to keep a confidence, or arise through a role, as when a doctor, teacher or public official becomes responsible for people who depend on their judgement.

With that discretion comes a form of answerability.

This is why trust can be betrayed rather than merely disappointed. If a machine fails, we ask what caused the failure. If a friend deliberately reveals a confidence, we ask how they could do that to us. The outcome matters, but so does the way they used the freedom we allowed them.

Betrayal is not simply an incorrect prediction. It is a failure to give due regard to what was entrusted.

This normative claim has limits. Trusting someone does not allow us to invent obligations they never accepted, to demand unlimited loyalty or to make them responsible for every hope we attached to the relationship. The object, purpose and conditions of trust still matter.

But where the trust is legitimate and understood, the trusted party becomes answerable for how they hold another person's vulnerability.

Trust is therefore neither a private feeling nor a unilateral gamble.

It is a relation that redistributes freedom, exposure and responsibility.

The grammar of trust

We often say, "I trust her," as though trust were a general verdict on a person.

Usually, the sentence is incomplete.

A more accurate grammar is:

A trusts B with X, for Y, under conditions C.

I may trust a friend with a confidence but not with money. I may trust a colleague's scientific judgement but not their ability to manage a deadline. I may trust an institution to keep an accurate record without trusting it to decide how that record should be used.

This does not make trust weak or inconsistent.

It makes trust precise.

The object tells us what is entrusted. The purpose tells us why. The conditions establish the limits within which the other person's discretion is accepted. Change any of these and the judgement may change.

This grammar also explains why trust is not transitive.

If I trust you and you trust someone else, I do not automatically trust that person. Your judgement may give me a reason to meet them, listen to them or begin with a limited commitment. But the new relationship must establish its own history.

An introduction can travel.

Trust begins again at the new boundary.

Reliance, confidence and trust

Several neighbouring ideas become clearer once trust is understood in this way.

  • Dependence is a condition: an outcome is affected by someone or something beyond us.
  • Reliance is an action: we proceed while depending on it.
  • Confidence is an expectation: we believe a person, institution or system will perform in a certain way.
  • Trust adds a relation: something we value is exposed to another actor's freedom, and we expect them to take that exposure into account.
  • Trustworthiness belongs to the other side of the relation: it is the capacity and disposition to exercise entrusted freedom with competence, integrity and regard.

These distinctions are especially useful when we speak of technology.

A bridge does not decide what to do with our vulnerability. We normally have confidence in its design, maintenance and reliability. We may, however, trust the engineers, operators and institutions responsible for it because they exercise judgement, hold duties and can answer for how their decisions affect us.

Institutions can also become objects of trust when they organise human agency across roles, rules and time. We do not need to know each person inside an institution. We need grounds to expect that its power will be exercised within a legitimate purpose, that duties will survive changes of personnel and that failures can be challenged and repaired.

Keeping these meanings distinct is not semantic housekeeping.

It tells us where responsibility lies.

Trust through time

Trust has a temporal structure.

It develops as actions acquire meaning through repetition. Someone arrives when promised, handles a small responsibility with care, admits an error before being confronted or tells an inconvenient truth. None of these acts guarantees the next one. Together, they form grounds for allowing more of the future to depend on the relationship.

This is why trust is usually slow to build.

It is also why a single breach can do more than add one negative event to an otherwise positive record. Betrayal may change the meaning of what came before. Generosity begins to look strategic. Reassurance sounds manipulative. The relationship's history is not erased, but it is reread.

Repair therefore requires more than renewed reliability.

The person who breached trust must recognise what was entrusted, how their freedom was used and why the other person's vulnerability did not count as it should have. Explanation, acknowledgement and changed conduct may create grounds for a new judgement. They cannot compel it.

Trust cannot be restored on behalf of the person who was harmed.

The boundary is theirs to reopen.

The strange economy of trust

Trust has an unusual economy.

There is no business case for trust; yet there is no business without it.

Keeping a confidence, disclosing a difficulty or refusing to exploit an ambiguity can carry an immediate cost. In any single decision, betrayal may offer the greater short-term return. Trust becomes economically intelligible only across time, through the relationship such decisions preserve.

No contract can prescribe every response to uncertainty. No rule can determine how every ambiguity will be resolved. Business — and much of social life — depends on freedoms that cannot be fully supervised: whether problems are revealed early, knowledge is shared honestly, disagreements are handled fairly and another person's vulnerability is treated with care.

Where no such freedom can safely remain, every opening must be replaced by verification, surveillance, collateral or control. Eventually, the cost of protection begins to consume the value of the relationship.

The benefits of trust compound because each vulnerability carefully held enlarges what the relationship can bear. A small responsibility handled well creates grounds for a larger one. Candour met with care permits greater candour. Reliability reduces the attention each party must spend defending against the other.

Breach follows a different path.

It does not merely cancel one successful interaction. It can change the meaning of what came before and narrow what becomes possible afterwards. Generosity begins to look strategic. Reassurance is heard differently. The past remains, but the relationship can no longer carry it into the future in the same way.

This is the cost asymmetry of trust:

Its benefits compound. Its failures cascade.

The compounding is gradual because trust grows through repeated acts whose meaning reinforces what came before. The cascade can be sudden because a single breach may travel backwards through the history of the relationship and forwards into its possible future. Trust is not an account balance in which ten faithful acts necessarily cancel one betrayal.

Social embeddedness changes this economy.

A relationship held within a family, profession, community or institution does not exist only between two isolated individuals. It has a social audience, even when that audience remains silent. Others may know something of its history. Their judgement matters. The trustee acts not only in relation to the person who trusted them, but within a field of peer exposure and future consequence.

No explicit threat is required. The possibility of disappointing the group, losing its regard or weakening one's place within it can become part of the trustee's reasons for acting. Embeddedness thus operates before any breach: it creates a psychological pressure to hold another person's vulnerability carefully.

It also matters afterwards. Shared context can help others distinguish error from deception. Memory and witnesses can make a breach harder to deny. Continuing relationships may provide routes to acknowledgement, answerability and repair.

In this sense, embeddedness can make trustworthy conduct self-reinforcing. Acting with care preserves not only one bond, but the wider field of relationships in which that bond exists.

But the same force is morally ambivalent. Peer pressure can discourage betrayal; it can also demand silence. A group can expose misconduct or protect the person responsible for it. Shared history can support repair or be used to prevent withdrawal. Dense relationships can carry care, but also collusion, patronage and exclusion.

Embeddedness therefore does not make trust inherently antifragile. Betrayal does not automatically strengthen it. What can become more resilient is a social fabric in which peer exposure encourages care before a breach, shared context supports judgement after it, and repair remains possible without requiring the vulnerable party to reopen the boundary on someone else's terms.

Trust changes both parties — and, over time, changes what their relationship is capable of holding.

Trust is not always good

Trust is often praised as though its presence were inherently desirable.

It is not.

Trust can be misplaced. It can protect collusion, favouritism and abuse. Loyalty within a group can make its members more willing to deceive outsiders. A trusted authority can use confidence accumulated in one domain to claim obedience in another. Intimate relationships can turn vulnerability into leverage.

Distrust may be perceptive rather than cynical.

This becomes especially important where dependence is imposed. A child, patient, employee, citizen or platform user may have little practical choice but to rely on people and institutions they did not select. Calling that reliance trust can conceal the unequal distribution of freedom and risk.

Interdependence makes reliance unavoidable.

It does not make trust deserved.

When an institution says "trust us" without making its purpose, power and obligations visible, it asks the exposed party to solve a problem of trustworthiness through attitude. When a person is expected to become more open while the other side remains unaccountable, vulnerability is being moralised rather than protected.

The relevant questions are not only whether trust exists, but:

  • Who is exposed?
  • Who receives discretion?
  • What has been entrusted?
  • On what grounds?
  • Can the exposure be limited or refused?
  • What happens if the trust is breached?
  • Who has the power to define repair?

Trust becomes worthy not because it is trust, but because the relation, its purpose and the distribution of vulnerability can be justified.

Learning to see vulnerability

The standard definition directs our attention to the willingness to become vulnerable.

An extended view must also direct attention to the vulnerability we receive from others.

For the person who trusts, this means neither celebrating openness nor treating all uncertainty as danger. It means recognising what is at stake, judging the grounds for trust, setting the boundary and remaining able to revise that judgement.

For the person trusted, the attitude is different. It means noticing that access is not ownership, discretion is not permission to ignore consequence and another person's dependence is not an opportunity to extract advantage. Being trusted places part of someone else's future within the reach of our choices.

For institutions, the task is broader still. They must identify vulnerability even when it is dispersed, silent or involuntary. They must avoid demanding personal trust where evidence, limits, duties and recourse should be provided. And they must ensure that those who bear the greatest exposure are not those with the least capacity to contest it.

This double awareness changes the meaning of a culture of trust.

Trust is no longer imagined as a social quantity that should simply rise. The aspiration becomes a society more literate in vulnerability: better able to see where lives are open to one another, to judge which openings are warranted, to protect those who cannot choose their exposure and to accept responsibility for what others place in our hands.

Such a society would not confuse caution with hostility or openness with virtue. It would make the grounds for trust visible, its boundaries negotiable, its withdrawal possible and its breaches open to repair.

To trust well is to open ourselves without abandoning judgement.

To be trustworthy is to understand that another person's future has entered our field of action.

Trust is an unguarded boundary. Its measure is not how widely we open it, but how consciously we cross it — and how carefully we hold what has crossed towards us.

→ The full framework: Trust as Infrastructure

Selected references