A task is rarely only the action other people can see.

Someone answering an email may appear to be typing. At the same time, they may be deciding what needs saying, searching for missing information, remembering what must happen next, anticipating the recipient’s reaction and judging themselves for not replying sooner.

The visible task is one task.

The cognitive task may be several tasks running at once.

This article proposes a way of separating those demands into three interacting layers:

• operational load — doing the task;
• instrumental load — managing how the task will be done;
• evaluative load — processing what doing, refusing or failing at the task means.

This is a provisional clinical framework, not an established cognitive test or a claim that attention literally consists of three separate systems. Its purpose is practical: to explain why two apparently identical tasks can have radically different nervous-system costs.

The central question is not simply:

• How hard is the task?

It is:

• Where is the load?

How the idea developed

The starting observation was that attention seemed to become diluted as a task acquired additional layers.

At the first layer, attention could be largely operational:

• Put the key in the lock.
• Cut the vegetables.
• Write the sentence.
• Swim the next length.

At the second layer, the person also had to hold the architecture of the activity in mind:

• What comes next?
• What equipment do I need?
• Where did I put it?
• What must happen before I can begin?
• What must I remember afterwards?

The person was no longer simply performing. They were managing the performance.

At the third layer, more questions could enter:

• Why am I doing this?
• Do I want to?
• What happens if I do not?
• What will somebody think?
• What does struggling with this say about me?

Now the person might be simultaneously doing, managing and judging.

Calling these progressively diluted “levels of attention” was tempting, but too crude. Higher-order meaning does not necessarily weaken attention. Interest, purpose and chosen value can stabilise attention and sustain considerable effort.

The model therefore shifted from three levels of attention to three layers of task load.

Higher layers can organise attention towards the operation. They can also remain active as additional work competing with it. The difference depends less on whether a task has meaning and more on whether that meaning is settled, chosen and able to recede — or unresolved, threatening and continually demanding supervision.
 

Existing theories: relatives rather than replicas

The framework did not appear in a theoretical vacuum.

Activity Theory distinguishes operations, shaped by immediate conditions; actions, directed towards conscious goals; and broader activities, organised by motives. That is a striking predecessor for the movement from concrete execution to goal and meaning.

Action Identification Theory, developed by Robin Vallacher and Daniel Wegner, similarly proposes that the same behaviour can be represented at different levels: lower-level descriptions explain how it is performed, while higher-level descriptions explain why it is performed or what effect it has. When an action becomes difficult, a more concrete representation can provide better guidance.

These theories show that human activity can be represented at more than one level. The Three-Layer Task Load framework asks a slightly different question:

What happens to usable capacity when several layers must remain actively processed while the person is trying to act?

Research on prospective memory provides some support for the instrumental part of this question. Holding an intention to do something later while performing an ongoing task can create measurable costs in the ongoing task. Researchers debate the precise mechanisms — including active monitoring and strategic slowing — so this should not be reduced to a single finite “bucket” of attention. Still, remembering what else must happen can affect what is happening now.

The three-layer framework is therefore a synthesis and clinical hypothesis. The component processes have substantial prior art; the particular arrangement and its proposed application have not, to my knowledge, been validated as one model.
 

Layer one: operational load — doing

Operational load is the immediate demand of interacting with the task:

• Read this line.
• Move this hand.
• Choose this word.
• Listen to this sound.
• Turn here.

The load may be perceptual, sensory, cognitive, linguistic or motor. A task can be operationally expensive because the instructions are difficult to understand, the movement requires precision, the room is painfully noisy, the person is fatigued, or language has become less accessible under stress.

With familiarity, some operations become more automatic. What once required deliberate sequencing can become fluid enough that the person no longer needs to represent each component consciously.

At one end of successful operational absorption sits flow. Research describes flow as deep involvement in an activity, often involving intense concentration and a merging of action and awareness.

Operational attention is not the same as flow. Brushing teeth or entering numbers may be fluent without being absorbing or rewarding. Flow is better understood here as a particular state of operational dominance: attention is embedded in the activity while competing management and self-conscious evaluation recede.

This fits particularly well beside my article on monotropism and the costs of switching attention. Deep focus may make an activity vivid, sustaining and regulating; it may also make interruption and transition disproportionately costly.

Flow may therefore depend partly on what is present — interest, sufficient skill, workable challenge — and partly on what becomes quiet.
 

Layer two: instrumental load — managing the doing

Instrumental load is the work required to create and maintain the conditions for action. It may include:

• initiating;
• sequencing steps;
• remembering future actions;
• finding tools or information;
• organising materials;
• making decisions;
• shifting between activities;
• monitoring progress;
• correcting errors;
• knowing when the task is finished;
• preserving the goal while completing its parts.

Consider making tea.

For someone in their own kitchen, with enough capacity and a familiar routine, much of the sequence may run automatically.

For someone in an unfamiliar kitchen — or someone depleted, interrupted or unable to access the usual sequence — the same task may require repeated reconstruction:

• Where are the mugs?
• Did I switch the kettle on?
• Where is the tea?
• I opened the fridge for something. What was it?

The visible action has not changed. Its instrumental overhead has.

This is particularly relevant to executive-function differences. Describing somebody as unable to perform a “simple task” may completely miss where the difficulty lies.

The operation may be simple.

Having to construct the runway every time before taking off may not be.

My article on neurodivergent routine explores practical ways to reduce this layer: visible sequences, event-based anchors, cognitive bookmarks, fewer decision points, pre-positioned materials, body doubling and versions of a task that fit different levels of capacity.

External support is not cheating. It moves some of the task architecture out of working memory and into the environment.
 

Layer three: evaluative load — positioning myself towards the doing

Evaluative load concerns the task’s significance and the person’s relationship to it:

• Why am I doing this?
• Do I want the outcome?
• Does it matter to me?
• Am I allowed to refuse?
• What happens if I stop?
• Who expects this from me?
• What does my performance say about me?

This layer is not inherently harmful. Evaluation can support action or compete with it.

The distinction is better described as autonomous versus controlled motivation, rather than intrinsic versus extrinsic motivation. Self-Determination Theory shows why. A task does not have to be enjoyable for the motivation to be autonomous. I may dislike doing my accounts while freely endorsing the financial independence they support. Conversely, a pressure can be thoroughly internalised and still feel controlling.
 

Autonomous evaluation: value that points and then quietens

Autonomous evaluation may sound like:

• I want to understand this.
• This matters to me.
• I choose the consequence.
• I do not enjoy the process, but I value what it achieves.

Here, value can function as a priority signal:

This matters. Put resources here.

Once attention has been organised, the value may no longer need to remain in the foreground. “I want to” can disappear into the doing.

That is one possible route towards flow: value directs attention into the task and then becomes quiet enough to let the task occupy it.
 

Controlled evaluation: motivation that keeps watching

Controlled evaluation may sound like:

• I have to.
• I should.
• Somebody will be disappointed.
• I will look incompetent.
• A decent person would have done this already.

Self-Determination Theory calls one relevant form introjected regulation: pressure has been taken inside without being fully accepted as one’s own. Action may then be regulated through guilt, shame, approval or contingent self-worth.

The person may be extremely motivated. That does not mean all of their attention is available for the task.

Some of it may be occupied by monitoring compliance, anticipating consequences or defending the self from judgement.

The useful distinction is:

• “I want to” can disappear into the doing. “I must” often has to keep checking that I am doing.

Performative self-monitoring

One form of controlled evaluation can be called performative self-monitoring.

“Performative” does not mean fake or insincere. It means that attention has partly moved from participating in the activity to observing oneself as the performer of it.

Instead of only asking:

• What am I doing?

the person may also be processing:

• How am I doing?
• How do I look while doing it?
• Am I doing it in the acceptable way?
• Do I look competent, caring, productive or normal?
• What will they conclude about me?

The audience may be present, anticipated or so thoroughly internalised that nobody else needs to be in the room.

The person becomes both performer and observer.

This term is most useful for audience-shaped monitoring; it should not be used as an umbrella for every form of controlled motivation. Some pressure is better understood as moralised duty or direct fear of consequences.

Research does not validate “performative self-monitoring” as a distinct clinical construct. However, its proposed mechanism is compatible with evidence that anxiety can reduce the efficiency of goal-directed attentional control and that social-evaluative threat can disrupt demanding cognitive performance. Performance may sometimes be preserved through increased effort, but at a greater internal cost.

Operational attention does the task.

Performative self-monitoring watches the person doing it.

For neurodivergent people, this can overlap with masking: monitoring voice, face, posture, timing, eye contact, productivity and apparent ease while also trying to complete the actual activity. It can also be fuelled by rejection sensitivity, especially when ambiguity creates a strong expectation that other people have noticed a failure before the person knows what rule they broke.
 

Moralised self-monitoring

A related process is moralised self-monitoring.

Here a practical difficulty becomes evidence about personal goodness:

• I forgot to reply.

becomes:

• I should have replied.

then:

• Keeping somebody waiting is inconsiderate.

and finally:

• I am inconsiderate.

A missed action has travelled upwards until it becomes a character verdict.

The operation may require five minutes. The moral trial surrounding it may consume hours.

This is not an argument against responsibility. A late reply may affect another person. Repair may be needed. The distinction is between evaluating an action and its impact and turning difficulty into global evidence of badness.

Moral judgement often presents itself as motivation:

• If I stop criticising myself, I will do even less.

But surveillance can consume the capacity needed for action. The resulting loop may look like this:

• importance → monitoring → reduced usable capacity → delayed or disrupted action → increased threat → more monitoring

Shame then claims to be the engine while repeatedly flooding it.
 

One form, three different costs

Consider the instruction: send the form.

Operationally, this might involve attaching a document and pressing a button.

Instrumentally, it might mean:

• Find the form → locate the reference number → scan the evidence → remember where the file was saved → check the address → send it.

Evaluatively, it might mean:

• I should have done this last week → they will think I am irresponsible → perhaps I have ruined the opportunity → why can’t I manage basic administration? → I absolutely must do it today.

The nervous system is not merely sending a form.

It is sending a form while carrying prospective memory, executive sequencing, uncertainty, anticipated rejection, temporal consequences and a trial concerning personal adequacy.

Calling this a lack of motivation tells us very little.
 

A dynamic model, not a staircase

The layers should not be imagined as a staircase in which each higher step automatically produces worse attention.

• Operational processing performs the task.
• Instrumental processing enables and coordinates it, but creates overhead when substantial conscious management remains necessary.
• Evaluative processing establishes significance and priority. When autonomous and settled, it may strengthen engagement. When conflicted, threatening, performative or moralised, it may become another concurrent task.

The model’s central hypothesis is:

• Task cost is determined not simply by how many layers are present, but by how much unresolved processing each layer requires.

“I care about finishing this” may require almost no continuing attention.

“Do I want to do this, am I allowed not to, what happens if I refuse, will they be angry, does refusing make me selfish?” may occupy enormous capacity.

The problem is not meaning.

It is meaning that has to remain computationally live.
 

Clinical application: ask where the task becomes expensive

The model changes the opening question from “Why can’t you do this?” to:

• At which layer does this task become expensive?
   

When operational load is high

The activity itself may exceed current sensory, linguistic, motor, cognitive or attentional capacity.

Useful responses might include:

• reducing sensory input;
• changing the pace or method;
• offering clearer information;
• practising the skill without an audience;
• reducing pain or fatigue where possible;
• allowing recovery rather than escalating demands.

During autistic burnout, previously accessible operations may become far more costly. Loss of capacity should not be treated as proof that the person has stopped caring.
 

When instrumental load is high

The person may be able to perform the operation once its infrastructure is supported.

Possibilities include:

• writing down the sequence;
• pre-positioning materials;
• reducing choices;
• separating preparation from completion;
• using reminders that name the next action, not only the deadline;
• creating a clear definition of “finished”;
• using body doubling where company feels safe rather than supervisory;
• leaving a cognitive bookmark for re-entry.

The aim is not to become better at carrying an invisible administrative department in one’s head. It is to make the department smaller, clearer or external.
 

When evaluative load is high

More planners and reminders may do very little. The useful questions may be:

• Whose requirement is this?
• Do I choose the outcome, even if I dislike the process?
• What consequence am I anticipating?
• What does not doing it mean about me?
• Who am I performing competence for?
• What rejection am I trying to prevent?
• What moral judgement has become attached to this task?
• Would I still choose this action without that judgement?

The aim is not to persuade the person that the task does not matter. It may matter considerably.

The work is to distinguish value from coercion.

Someone may discover:

• I genuinely want this outcome.

Autonomous value can then be returned to the task.

Or they may discover:

• I do not choose this. I am frightened of what will happen if I refuse.

That is not primarily a productivity problem. It may concern consent, relational power, boundaries, threat or material consequences. My article on assertiveness as owned aggression explores the capacity to recognise and act from one’s own “no” without converting it into harm.
 

Neurodivergent relevance

The framework may be particularly useful when the visible simplicity of a task bears little relationship to its nervous-system cost.

• sensory or motor demands can enlarge operational load;
• executive-function differences and transition costs can enlarge instrumental load;
• monotropism can make switching away from an active attention tunnel costly;
• repeated correction, exclusion and punishment for difference can enlarge evaluative load;
• masking can add performative self-monitoring to almost any public task;
• rejection sensitivity can keep an actual or imagined audience cognitively present;
• anxiety, pain, sleep loss and burnout can reduce available capacity across all three layers.

This does not mean every difficulty is caused by neurodivergence, or that understanding the load removes responsibility for impact. It means that moral explanation should not be the first or only explanation.

A person can care deeply and still be unable to access an action.

Indeed, excessive controlled motivation may be part of what blocks access:

• I must get this right.

can create more interference than:

• I would like to do this, and here is the smallest available next step.

Care and surveillance are not the same cognitive state.
 

A provisional definition

The Three-Layer Task Load framework describes task engagement as involving three interacting forms of processing:

Operational load is the immediate attentional, sensory, cognitive, linguistic or motor demand of performing an activity.

Instrumental load is the executive demand of initiating, organising, sequencing and maintaining the conditions necessary for performance.

Evaluative load is the processing associated with the significance, consequences, value and self-relevance of performing, refusing or failing to perform the task.

Evaluative processing may be autonomous, when personally endorsed value directs and stabilises engagement, or controlled, when action is regulated through pressure, anticipated consequences, rejection, shame, guilt, imposed duty or contingent self-worth.

Controlled evaluation may include performative self-monitoring, in which attention is allocated to oneself as an actual or imagined object of evaluation, and moralised self-monitoring, in which task performance becomes evidence about personal goodness, worth or adequacy.

The framework predicts that two objectively similar tasks may carry substantially different costs because their operational, instrumental and evaluative overheads differ.

Its practical question is not:

• Why can’t I do this simple thing?

It is:

• Where is the load, and what would reduce it without reducing my agency?

Good engagement does not always require adding more motivation. Sometimes it requires skill, sensory accommodation or external structure.

And sometimes it requires removing the layer that keeps watching, judging and threatening the person who is trying to act.