Some of the most consequential words in English are very short.
Yes.
No.
I agree.
I understand.
I accept.
You may.
Each can look like consent.
They are not automatically the same thing.
A person can understand without agreeing.
They can agree to one thing without agreeing to another.
They can click a button because there is no realistic alternative.
They can stay silent because they did not notice the question.
They can agree today and withdraw tomorrow.
This is why consent is not simply a polite version of permission.
Consent is English used to make another person’s choice explicit enough that everyone can tell what was agreed, by whom, for what purpose, under what information, and for how long.
That is the communication problem this article owns.
Quick Read
One-sentence answer: consent works when a person with the relevant ability to decide is given understandable and sufficiently specific information, has a genuine choice, takes an affirmative action that communicates agreement, and remains able to withdraw or change that agreement where the governing context permits.
Current UK Information Commissioner’s Office guidance describes valid consent for data protection as freely given, specific, informed and unambiguous, signified by a statement or clear affirmative action. It also requires organisations relying on consent to keep evidence of what people agreed to, make requests clear, and make withdrawal possible. Current Health Research Authority guidance likewise emphasises autonomy, clarity, accessibility, proportionate information and evidencing consent in research.
These are domain-specific rules, not a universal law of every consent situation. But they reveal a broadly useful communication architecture:
- decision-maker: who is being asked to consent;
- choice: whether there is a genuine option to say yes or no;
- information: what the person needs to understand before deciding;
- scope: exactly what activity or purpose the consent covers;
- affirmative signal: what action expresses agreement;
- record: how the agreement is evidenced;
- duration: how long the consent remains relevant;
- withdrawal: how a person can change the decision;
- change control: what happens when the purpose or circumstances change.
The Consent Is Not the Permit
A permit is granted by an authority to a holder.
Consent is given by the person whose agreement matters.
Permit:
The authority permits the company to conduct the approved activity.
Consent:
The participant agrees to the specified use of their information.
The direction of authority is different.
permit = institution grants bounded authority
consent = affected person communicates agreement
This keeps Consent separate from How English Works | The Permit.
The Consent Is Not the Licence
A licence creates an ongoing permission relationship with an authority.
Consent creates an agreement relationship with a person.
A licensed organisation may still need consent for particular activities.
Likewise, a person’s consent does not replace whatever licence or legal authority the organisation itself may need.
institutional authority and personal agreement are separate layers
The Consent Is Not the Application
An application says:
Please decide whether I may receive this status or permission.
Consent says:
I agree to this specified activity or use.
The application asks an authority to act.
The consent records a person’s choice.
One process can contain both.
For example, an application may include a consent section for checking information with another source.
That section should not be mistaken for the application itself.
The Consent Is Not the Contract
This boundary is often blurred.
A contract may require agreement between parties and may contain consent clauses.
But not every contractual term is based on consent in the same legal or ethical sense used in privacy, research or healthcare.
ICO guidance is explicit that consent is only one possible lawful basis for using personal information and should not be presented as a choice if the organisation would proceed regardless.
asking for consent when there is no real choice creates misleading language
So:
agreement to a contract ≠ consent to every possible related activity
Consent Starts with a Real Decision-Maker
Before asking:
Do you consent?
the system needs to know:
Who has the relevant authority and ability to decide?
Depending on context, that may be:
- the individual directly affected;
- a parent or guardian;
- an authorised representative;
- another legally recognised decision-maker.
The rules differ across domains and jurisdictions.
The communication principle is stable:
agreement from the wrong person cannot be repaired by perfect wording
Information Must Come Before the Choice
Weak consent:
I agree.
Agree to what?
Strong consent architecture places relevant information before the affirmative decision.
That information may need to explain:
- what will happen;
- why;
- who is involved;
- what information or activity is covered;
- important risks or consequences;
- important alternatives;
- how long the consent matters;
- how withdrawal works.
Current HRA guidance emphasises that information should be clear, understandable, accessible and proportionate rather than simply long.
more words do not automatically create more informed consent
Information Overload Can Defeat Informed Choice
A 48-page legal notice may contain every technical detail.
A reader may understand almost none of it.
Consent writing therefore faces a difficult optimisation problem:
enough information to make the choice meaningful
without burying the choice inside information nobody can practically use
HRA guidance explicitly encourages proportionate, user-friendly information rather than treating a template as the goal.
This is a broader How English Works principle:
information is useful only if the receiver can convert it into a decision
Specificity Prevents Consent from Spreading
Weak:
I consent to use of my information.
Use by whom?
For what purpose?
For how long?
Shared with whom?
Stronger:
I agree that Organisation A may use my email address to send the monthly programme update described above.
The second statement has scope.
ICO guidance requires consent to be specific and to identify the relevant purposes.
specific consent puts a fence around the yes
Bundled Consent Hides Several Decisions Inside One Button
Example:
By clicking Agree, you consent to receive service messages, marketing emails, partner promotions, research invitations and data sharing.
That is one visible button carrying several conceptually different choices.
Where choices are separable, bundling them can make it impossible to know what the person actually wanted.
one click can be too small a container for five independent decisions
A better design may separate:
- required service communication;
- optional marketing;
- optional research participation;
- optional third-party sharing.
The exact legal requirement depends on context. The information-design lesson is that independent decisions should remain independently legible.
Silence Is Ambiguous
A person receives a message.
They do nothing.
What does that mean?
- They agree?
- They disagree?
- They did not see it?
- They forgot?
- They did not understand?
- They assumed no action was needed?
This is why modern consent systems often require affirmative action.
ICO and HRA guidance both reject silence, inactivity or pre-selected choices as sufficient signals in the contexts they govern.
silence is an empty state, not automatically a yes
This connects directly to How English Works | The Empty State.
Pre-Ticked Boxes Convert Inaction into Apparent Agreement
Box:
☑ Send me promotional messages
The user clicks Continue without noticing.
Did the user choose marketing?
Or did the interface choose marketing and ask the user not to object?
The difference matters.
opt-in asks the person to create the yes
preselection asks the person to discover and remove a yes created for them
That is a fundamentally different interaction design.
Affirmative Action Should Match the Decision
Possible signals include:
- ticking an unchecked box;
- signing a consent form;
- selecting “I agree”;
- giving an oral statement where the context permits;
- another clear action whose meaning is understood.
The signal should be observable enough to distinguish agreement from accidental navigation.
consent needs a behaviour that can reasonably be interpreted as this specific yes
A Signature Is Evidence of a Decision, Not the Whole Decision
Signed form:
Signature: Maya Tan
The signature may provide evidence.
It does not prove by itself that:
- the information was understood;
- the choice was voluntary;
- the scope was specific;
- the signer had capacity;
- the document was the correct version.
signature can evidence consent
signature does not manufacture valid consent out of a broken process
Consent Is a Process Before It Is a Record
This is especially clear in research guidance.
HRA describes consent as more than procedural formality and emphasises the conditions that allow people to make voluntary and informed decisions.
The record comes after or alongside the process.
information → questions → understanding → voluntary choice → affirmative signal → record
When systems focus only on collecting the final checkbox, they preserve the receipt and lose the decision.
Voluntariness Means the “No” Must Be Real
A consent request says:
You are free to decline.
But the person believes refusal will cause an unrelated penalty.
The sentence and the real choice architecture now disagree.
ICO guidance says consent is not appropriate when people do not have genuine choice and control.
voluntary language cannot compensate for coercive structure
The real question is not merely:
Was “voluntary” written?
It is:
Could the person realistically say no?
Power Imbalance Makes Consent Harder to Interpret
Consider:
- employer and employee;
- school and student;
- doctor and patient;
- government and citizen;
- platform and dependent user.
These relationships can contain unequal power.
That does not automatically make consent impossible in every context.
It does make the voluntariness question more serious.
the stronger the power imbalance, the less safe it is to infer freedom from the word “agree” alone
Consent Records Need the Version of the Information Shown
Record:
Maya consented on 4 September 2026.
Consent to which wording?
If the information notice changed on 1 September, the date alone may not be enough.
A stronger record may preserve:
- who consented;
- when;
- how;
- which purpose;
- which information version;
- which choices were selected.
ICO guidance specifically emphasises keeping evidence of who consented, when, how and what they were told.
consent receipt should let the future reconstruct the decision environment
The Record Should Not Contain More Personal Data Than Necessary
Wintour V1.0 adds an important production principle here.
Evidence of consent should be sufficient.
It should not become an excuse to collect unrelated personal information.
The record needs enough to prove the consent state.
It does not automatically need:
- extra identity documents;
- unrelated demographic details;
- full conversation transcripts;
- sensitive material irrelevant to the decision.
evidence should be proportionate to the claim it proves
Withdrawal Is a New State, Not Proof the Original Consent Was Invalid
A person agrees on Monday.
They withdraw on Friday.
Both statements can be true:
Consent was valid on Monday.
Consent is no longer active after Friday.
Withdrawal changes the current state.
It does not necessarily rewrite the past.
consent history and current consent status are different records
Withdrawal Should Be Understandable and Practical
Consent button:
One click.
Withdrawal process:
Print a form, mail it internationally, wait thirty days.
The two sides are badly imbalanced.
ICO guidance requires withdrawal to be possible and easy in its context.
The general interaction principle is:
do not make yes frictionless and no deliberately exhausting
Withdrawal Needs a Clear Effective Point
Person clicks:
Withdraw consent.
When does the state change?
- immediately;
- when received;
- after verification;
- after processing by the system?
The governing rules determine the answer.
The communication design should make the timing legible.
state transition needs a timestamp
Withdrawal Does Not Always Undo What Already Happened
Suppose information was lawfully used before withdrawal.
Does withdrawal erase every prior act?
Not necessarily.
The answer depends on context and law.
Good consent language should not promise retroactive deletion, destruction or reversal unless that is actually true.
stop future authorised action and undo past completed action are different operations
Changed Purpose Can Require a New Consent Decision
Original request:
May we use your email to send course updates?
Later proposal:
May we share your email with partner companies for advertising?
That is not obviously the same purpose.
Specificity means consent should not stretch silently across materially different uses.
new purpose can require a new yes
Current ICO lawful-basis guidance also emphasises reconsidering the lawful basis when purposes change.
Purpose Drift Is Consent Cannibalisation
This is an especially useful Wintour-style analogy.
A consent begins with one clear job.
Over time, new activities are quietly attached.
- service communication;
- marketing;
- profiling;
- third-party sharing;
- research.
The original consent becomes overloaded.
one consent should not be forced to own every later purpose
The same canonical discipline used in publishing applies to consent design: one decision should have a clear owner and boundary.
Consent Can Expire Through Change Even Without a Fixed Date
Some consent has an explicit term.
Some does not.
Even without a formal expiry date, old consent can become stale when:
- the purpose changes;
- the organisation changes;
- the relationship changes;
- the information given is no longer accurate;
- the person’s circumstances materially change;
- the governing rules require refresh.
consent freshness depends on whether the original decision environment still exists
Refresh Consent Only When Refresh Has a Real Job
Sending repeated consent requests can create fatigue.
People begin clicking without reading.
A refresh should have a reason:
- changed purpose;
- new data type;
- new participant group;
- new risk;
- changed terms;
- stale or uncertain record.
Repeated ceremony without changed substance can make consent less meaningful rather than more.
Granular Consent Lets People Say Yes to One Thing and No to Another
Example:
- ☑ monthly programme email;
- ☐ partner marketing;
- ☑ anonymous research survey invitation;
- ☐ personalised advertising.
The person’s preference becomes a vector rather than one global bit.
granularity preserves several independent choices instead of compressing them into one yes/no field
Too Much Granularity Can Also Be Bad Design
Imagine 117 consent toggles.
Every sub-purpose has a separate checkbox.
The person cannot build a coherent mental model.
Granularity should match meaningful decision boundaries.
split choices where the person may reasonably want different answers
do not split until the interface becomes a maze
Consent Language Should Be Written for the Actual Audience
Consent request:
The undersigned hereby irrevocably acknowledges and consents to all foregoing processing activities notwithstanding any antecedent limitation…
The words may look formal.
The decision may be less informed.
HRA guidance emphasises meaningful, understandable and accessible communication.
The plain-language principle is:
formal consequence does not require unnecessarily difficult prose
Children and Capacity Need Their Own Rules
This area cannot be safely collapsed into one universal sentence.
Different jurisdictions and contexts use different rules for children, young people and adults who cannot make particular decisions for themselves.
Current HRA guidance, for example, distinguishes multiple research contexts and specifically warns that the applicable rules depend on the type of study and location.
Therefore:
do not copy an adult consent model into a child or capacity-sensitive context and assume the legal problem is solved
This article deliberately stops at the communication architecture. Domain-specific legal and professional guidance must govern the actual process.
Consent and Assent Are Not Automatically the Same
In some child or capacity-sensitive contexts, systems distinguish formal consent from assent: a person may be involved in and positively agree with participation even where another legally recognised decision-maker provides the formal consent.
The exact terminology and effect vary.
The communication lesson is:
respecting a person’s expressed wishes can matter even when formal decision authority sits elsewhere
Refusal Needs a Clean State
A consent system should not store only:
consent = false
That value may mean:
- refused;
- not yet asked;
- withdrawn;
- expired;
- not applicable.
These states have different histories.
no is not one state
This is a Schema problem as much as a language problem.
Consent Receipts Need State History
Useful history:
1 June — consent granted for Purpose A
12 July — Purpose B offered, declined
4 September — Purpose A withdrawn
Now future systems can reconstruct what authority existed at each point.
One current boolean cannot explain the past.
consent is temporal data
Consent Cannot Be Assumed from Relationship Alone
A loyal customer is not automatically consenting to every new use.
An employee is not automatically consenting to every optional monitoring activity.
A student is not automatically consenting to every publicity use.
A patient is not automatically consenting to every research use.
Context changes the rules.
The broader principle remains:
relationship can create expectations; it does not automatically create unlimited consent
Consent Is Not Proof of Understanding
A person clicks:
I understand and agree.
The click proves an action occurred.
It does not prove deep comprehension.
High-stakes systems may need additional safeguards such as:
- plain-language explanation;
- opportunity for questions;
- teach-back or comprehension checks where appropriate;
- time to decide;
- accessible formats.
Whether these are legally or professionally required depends on the context.
evidence of agreement and evidence of understanding are related but distinct claims
Consent Is Not a Shield Against Bad Practice
An organisation should not reason:
They consented, therefore anything that follows is acceptable.
Consent does not automatically erase:
- safety duties;
- fairness obligations;
- professional standards;
- data minimisation;
- consumer protection;
- ethical limits.
The precise obligations depend on the domain.
The communication principle is universal enough to teach:
consent is one control, not a moral or legal blank cheque
Primary School: Consent as a Clear Choice
Use a simple classroom example:
May we display your drawing on the classroom wall?
Give two clear choices:
- Yes, display it.
- No, keep it in my folder.
Then change the purpose:
May we post the same drawing on the school website?
Ask:
Does the first yes automatically answer the second question?
Children quickly understand scope.
Lower Secondary: Find the Broken Consent
Give students:
By continuing to use this service, you agree to all current and future uses of your information by us and our partners.
Ask them to identify problems:
- scope too broad;
- future purposes unknown;
- partners undefined;
- continuing may not be affirmative choice;
- no withdrawal information.
This teaches precision, power and interface design at once.
Upper Secondary: Audit a Consent System
- Who is being asked?
- Do they have a real choice?
- What information is provided?
- Is the purpose specific?
- Are independent purposes separated?
- What affirmative action records agreement?
- Which version of the information was shown?
- How is withdrawal performed?
- What changes require a fresh decision?
- What states distinguish refusal, withdrawal, expiry and not-yet-asked?
Ten Failure Modes of Consent English
- Silent consent. Inactivity is treated as agreement without a clear affirmative action.
- Bundled consent. Several independent purposes are hidden behind one global yes.
- Purpose vagueness. “Use your information” provides no meaningful boundary.
- Information overload. Formal completeness makes practical understanding harder.
- Power blindness. A nominal yes is treated as freely given despite strong pressure or dependency.
- Receipt without process. The system keeps the checkbox but cannot show what information or choice architecture produced it.
- Version amnesia. Nobody knows which consent wording the person saw.
- Withdrawal friction. Saying yes is easy while saying no later is deliberately difficult.
- Purpose drift. New activities are quietly attached to an old consent.
- Consent absolutism. Agreement is treated as permission to ignore other legal, ethical or professional controls.
How to Write a Better Consent Process
- Identify the person whose choice matters.
- Confirm the relevant decision-making authority and capacity rules for the domain.
- Explain the activity and purpose in usable language.
- Separate materially different purposes.
- Make refusal a real option where consent is the basis of the activity.
- Require a clear affirmative signal.
- Record who agreed, when, how and to which information version.
- Make withdrawal understandable and proportionate.
- Define what changes require a fresh consent decision.
- Keep consent history without collecting unnecessary personal data.
Then perform the stranger test:
If a careful outsider reviewed this consent six months later, could they reconstruct what the person was told, what choices were available, what they affirmatively agreed to, and whether that agreement is still active now?
The Deeper Idea: Consent Turns Agency into a Recorded Boundary
Consent is not mainly a checkbox.
It is a boundary-setting act.
A person says:
This, yes.
That, no.
For this purpose.
Under these conditions.
Until I change my decision or the context changes enough that a new decision is needed.
The language gives agency a durable shape.
information → voluntary choice → specific yes → recorded boundary → withdrawal or refresh
This is why Consent belongs to How English Works.
It is English used to ensure that another person’s agreement does not become larger, longer or more convenient for the organisation than the decision the person actually made.
Reader Checklist
- Who is consenting?
- What exactly are they agreeing to?
- Do they have a realistic choice?
- What information was provided?
- Was the action affirmative?
- Are separate purposes unbundled?
- What version did they see?
- How is consent recorded?
- How can it be withdrawn?
- What change would require a new decision?
Related eduKateSG Reading
- How English Works | The Application
- How English Works | The Permit
- How English Works | The Licence
- How English Works | The Identifier
- How English Works | The Schema
- How English Works | The Empty State
Research and Further Reading
- Information Commissioner’s Office — What Is Valid Consent?
- Information Commissioner’s Office — When Is Consent Appropriate?
- Health Research Authority — Informing Participants and Seeking Consent
- Health Research Authority — Proportionate Approach to Seeking and Evidencing Informed Consent
Consent rules differ across privacy, research, healthcare, education, employment, contracts and other settings, and they vary by jurisdiction. Some current UK consent guidance is under review following 2026 legal changes. This article explains communication architecture and is not legal, medical or regulatory advice; use the current rules that govern the actual context.
Final idea: consent is good when the recorded yes is no larger than the choice the person actually understood and made.