When Your Bandwidth Is Gone: Navigating Burnout by Severity (Part 5 of 8)

Part 5: Profound Depletion

When Functioning Becomes Increasingly Uneven—and Support Has to Come Toward the Person

In Part 4, I talked about Significant Depletion, which is the point where it isn’t only everyday demands that have become expensive. Even the things that are supposed to help can start requiring more capacity than we realize.

Using the planner takes executive functioning. Requesting support takes organization and follow-through. Going to therapy takes time, emotional capacity, transitions, and often some ability to process afterward. Even asking for help requires recognizing what you need, figuring out whom to ask, finding the words, initiating the conversation, and managing whatever happens next.

Within the Bandwidth Model, I call this Implementation Cost.

Part 5 takes us further down that continuum, into what I call Profound Depletion. And this is where I think one of the most confusing things about burnout becomes especially important: a profoundly depleted person may not look profoundly impaired in every area of their life.

In fact, they may look surprisingly capable in very narrow moments.

Someone may be able to have an incredibly sophisticated conversation and then struggle to make themselves lunch. They may understand exactly what is happening but be unable to organize a shower. They may write something thoughtful but be unable to answer a routine message. They may spend hours deeply engaged in one compelling activity and then find that transitioning into brushing their teeth feels almost impossible.

From the outside, this can look contradictory.

It makes much more sense when we stop assuming that human functioning is uniform.

And this is where I want to bring the idea of the spiky profile more explicitly into the Bandwidth Model.

What I Mean by a Spiky Profile

The idea of a spiky cognitive profile did not originate with the Bandwidth Model.

There is empirical literature describing uneven cognitive profiles in neurodevelopmental populations. Wilson’s (2024) meta-analysis of WAIS-IV and WISC-V performance found that autistic participants, at the group level, showed a relatively “spiky” profile, with stronger verbal and nonverbal reasoning relative to processing speed and, to a lesser extent, working memory. ADHD participants showed a less distinctive Wechsler profile, although working-memory performance was somewhat reduced on average.

That does not mean every autistic person has that particular profile. And it does not mean a cognitive profile can diagnose autism or ADHD.

But it illustrates something important:

Different areas of functioning can be very different within the same person.

Adult ADHD research similarly shows considerable heterogeneity in executive and other cognitive functioning rather than one universal pattern of impairment (Boonstra et al., 2005).

So uneven neurocognitive functioning is not a Bandwidth claim.

What I’m proposing within the Bandwidth Model is something more specific:

As accessible capacity narrows, pre-existing differences in the cost and accessibility of different functions may become harder to compensate for. As a result, the person’s functional profile may become even more visibly uneven.

That specific proposition still requires empirical testing.

In other words:

Architecture predates load; depletion may reveal it.

That is a Bandwidth Model hypothesis, not an established scientific finding.

Profound Depletion Doesn’t Mean Everything Stops Working

I think this distinction matters because we often imagine severe depletion as though someone gradually turns down one global neurological dimmer switch.

At full capacity, everything works.

At half capacity, everything works half as well.

At profound depletion, everything barely works.

But human functioning does not appear to operate that neatly.

Different activities recruit different combinations of attention, working memory, initiation, switching, processing speed, language, motor planning, sensory processing, emotional regulation, motivation, learned automaticity, and environmental support.

That means one capacity can remain surprisingly intact while another becomes much less reliable.

Research on autistic burnout gives us some evidence adjacent to this idea. Raymaker et al. (2020) described autistic adults reporting chronic exhaustion, reduced tolerance to stimulus, and loss of previously available skills. More recent work has continued trying to characterize and measure autistic burnout (Arnold et al., 2023), and Ali et al.’s (2025) systematic review describes difficulties involving memory, communication, executive functioning, sensory and social tolerance, and access to previously available abilities across the emerging literature.

None of that proves the Bandwidth Model’s Profound Depletion level.

And it does not prove that burnout universally exaggerates a pre-existing spiky profile.

But it does make it difficult to defend the assumption that severe depletion must produce uniform impairment across every domain.

“But You Can Do This—So Why Can’t You Do That?”

This may be one of the most painful consequences of spiky functioning.

If someone can hold a thoughtful conversation, why can’t they make food?

If they can research something for three hours, why can’t they make the appointment?

If they can explain exactly what is happening, why can’t they just implement the solution?

If they can spend hours absorbed in one interest, why can’t they take a shower?

But those tasks are not interchangeable.

A person may understand exactly what medical appointment needs to happen and why.

Actually arranging it is a different task.

Now we are asking for initiation, decision-making, finding the phone number or portal, switching away from whatever they are doing, tolerating waiting, answering questions, recording the appointment, remembering it later, preparing to leave or log in, and perhaps navigating transportation.

Their reasoning may remain strong while the executive bridge between knowing and doing has become much less reliable.

That is why two of the distinctions I keep returning to in the Bandwidth Model are:

Insight is not access.

And:

Task complexity is not neurological cost.

Something can look incredibly simple from the outside while requiring exactly the combination of functions that is currently most expensive for that person.

The 2e Profile Can Make This Even Harder to See

This becomes particularly interesting with twice-exceptionality because the top of the person’s profile may remain exceptionally high.

Someone can retain sophisticated verbal reasoning, pattern recognition, creativity, specialized knowledge, or analytical thinking while struggling substantially with initiation, switching, working memory, processing speed, sensory tolerance, communication, or activities of daily living.

That can create an enormous perceptual problem.

We see the highest point on the person’s profile and use it to estimate everything else.

You’re obviously capable.

Yes.

They may be extraordinarily capable.

But capability in one domain doesn’t establish equivalent access in another.

There is some adjacent evidence here from research on autistic compensation. Livingston, Shah, and Happé (2019), for example, examined compensatory strategies in autistic adults whose observable behavior could conceal underlying cognitive difficulties or differences.

What we do not have is a validated “2e burnout profile” showing that gifted cognition predictably remains intact while other capacities collapse.

So I don’t want to claim that.

The Bandwidth hypothesis is narrower:

Exceptional strengths may sometimes allow someone to compensate for areas of greater cost while enough functional margin remains. If that margin becomes smaller, the underlying unevenness may become harder to compensate for.

That is a theoretical proposition, not an established 2e burnout mechanism.

And the Profile Isn’t Static

There’s another complication: the same person’s profile can look different depending on the conditions.

Someone may function much better in a quiet, familiar environment after adequate sleep with one clear task in front of them.

The same person may function very differently after poor sleep, multiple transitions, sensory overload, conflict, pain, or prolonged decision-making.

Their intelligence did not disappear.

The operating conditions changed.

This is where the Bandwidth Model’s concept of Context Fit becomes important.

The same function may cost very different amounts under different conditions.

So instead of:

“Can this person do this?”

we ask:

“Under what conditions can this person reliably access this function, for how long, and at what cost?”

That is a more useful question.

Profound Depletion Can Therefore Look Surprisingly Capable

This is why someone can be articulate, insightful, funny, engaged, and cognitively sharp in a particular conversation—and still be profoundly struggling with daily life.

They may have marshaled the capacities necessary for that particular moment.

What you saw does not necessarily tell you what happened before it, what it required to produce, or what happens afterward.

They may still struggle with preparing food, showering, changing clothes, responding to messages, taking medications consistently, tolerating noise, making decisions, completing paperwork, leaving the house, or moving from one activity into another.

The observable moment does not necessarily reveal the cost.

This leads to another Bandwidth distinction:

Peak access is not representative access.

The fact that I can observe someone functioning at their highest point does not tell me how reliably that capacity is available across ordinary life.

And it does not tell me whether using it is sustainable.

When the Behavioral Menu Gets Smaller

There can also be a point where depletion affects flexibility itself.

We do have established neuroscience showing that significant stress can interfere with aspects of prefrontal cortical functioning involved in higher-order cognitive control. Arnsten (2009), for example, reviewed mechanisms through which stress can impair prefrontal functioning.

That research does not prove the Bandwidth Model’s idea of a narrowing “behavioral menu.”

That is my functional interpretation.

What I’m proposing is that under substantial constraint, some adaptive responses may become harder to access quickly and reliably.

A person who would normally be able to pause, switch strategies, find different words, tolerate ambiguity, inhibit an immediate response, or organize alternatives may find some of those options harder to access under significant load.

That does not mean:

“I had no control.”

And it does not eliminate accountability.

It means something more nuanced:

Behavioral flexibility may be reduced under constraint.

The menu has fewer readily accessible options.

When Help Is Available but Not Accessible

This may be the central problem at Profound Depletion.

Someone may technically have support.

They may have family.

Friends.

A therapist.

A physician.

Insurance.

Community services.

People who genuinely want to help.

But accessing all of that still requires something from them.

They have to call.

Schedule.

Explain.

Decide.

Complete forms.

Remember appointments.

Coordinate transportation.

Figure out what to ask for.

Manage multiple people.

And the functions required to coordinate support may be some of the very functions that are least accessible.

That is why:

Support availability is not support accessibility.

At Significant Depletion, implementing support becomes expensive.

At Profound Depletion, the person may sometimes need another person to carry part of that Implementation Cost.

Support Doesn’t Have to Mean Taking Over

Supporting someone whose functioning is severely constrained does not mean deciding that they are globally incapable and taking control of their life.

A spiky profile gives us a better alternative.

Find where access still exists and preserve agency there. Then scaffold the gaps.

Maybe the person knows exactly what they want but needs help executing the logistics.

Maybe they can complete a task once someone helps identify where to start.

Maybe writing is accessible while speaking is not.

Maybe familiar routines remain accessible while novel tasks are not.

Maybe they can make a phone call if someone sits beside them.

Support can become precise rather than paternalistic.

Instead of:

“Let me know if you need anything.”

maybe:

“I’m bringing over some food. Would Tuesday or Wednesday be better?”

Instead of:

“You really need to make that appointment.”

maybe:

“Would it help if I sat with you while you called?”

Instead of:

“You need to figure out all these forms.”

maybe:

“Let’s look at which one actually has to be done first.”

Instead of asking someone to explain a complicated situation from scratch:

“If you tell me the main thing you need your provider to know, I can help you organize the message.”

The goal is not to remove autonomy.

It is to reduce the number of inaccessible steps standing between the person and the support they need.

Sometimes We Need to Stop Adding Strategies

This is also the point where I think we need to be very careful about optimization.

When somebody is struggling, our instinct is often to give them more strategies.

Try this planner.

Use this app.

Start this routine.

Do this exercise.

Schedule this appointment.

Practice this coping skill.

But every strategy has an Implementation Cost.

At Profound Depletion, the better question may be:

What can this nervous system stop carrying?

Maybe household standards temporarily change.

Maybe meals become extremely simple.

Maybe unnecessary appointments are reduced.

Maybe optional social contact decreases.

Maybe there are fewer transitions.

Fewer decisions.

Less sensory input.

Less expectation for immediate emotional processing.

More predictable routines.

More practical assistance.

That does not mean withdrawing from life forever.

And it does not mean avoidance is always therapeutic.

It means recognizing the difference between avoidance that perpetuates a problem and temporary demand reduction when current demands exceed sustainable access.

Autistic burnout research gives us some adjacent support here. Raymaker et al. (2020) identified chronic life stress and mismatch between expectations and abilities without adequate support as important themes. Ali et al. (2025) similarly identified sensory and social overwhelm, camouflaging, everyday demands, stigma, and insufficient accommodation across the literature. Mantzalas et al. (2022) proposed a conceptual model emphasizing interactions among individual, social, and environmental risk and protective factors.

Those findings support taking environment, demands, and support seriously.

They do not validate the Bandwidth Model’s particular severity continuum.

Recovery May Be Spiky Too

If functioning becomes uneven during depletion, why would we assume everything returns at exactly the same rate during recovery?

It may not.

Someone may regain interest before endurance.

They may be able to read again while still struggling with initiation.

They may tolerate conversation better while sensory capacity remains narrow.

They may manage familiar routines while novel demands still feel impossible.

Those specific sequences are examples, not established recovery stages.

Within the Bandwidth Model, the hypothesis is simply that recovery may be multidimensional and uneven because functioning itself is multidimensional and uneven.

And that is why the first good day can be misleading.

Someone wakes up and thinks:

I’m back.

So they answer every message, clean everything, run errands, tackle the backlog, make plans, and use virtually all of the newly available margin.

Then the next day they are flattened again.

Within the Bandwidth Model, I call that pattern Bandwidth Overdraft when present performance uses temporarily available capacity in ways that may compromise subsequent functioning or recovery.

That is a metaphorical and theoretical construct.

There is no scientifically established neurological bank account from which people literally borrow bandwidth.

The underlying point is simpler:

A temporary improvement in performance does not prove that recovery is complete or sustainable.

And None of This Means We Assume It’s Burnout

This becomes increasingly important as impairment gets more severe.

If someone experiences profound fatigue, cognitive changes, withdrawal, sensory changes, loss of self-care, difficulty eating, difficulty managing medication, or a substantial decline in basic functioning, we should not simply label it burnout and stop asking questions.

Depression, anxiety, trauma-related conditions, sleep disorders, medication effects, substance use, chronic pain, anemia, thyroid disease, nutritional deficiencies, hormonal changes, migraine, infection, autoimmune or inflammatory conditions, neurological disorders, and other medical conditions can produce overlapping symptoms.

Autistic burnout itself also overlaps phenomenologically with depression and other forms of distress, even though autistic people describe aspects they experience as distinct (Ali et al., 2025; Raymaker et al., 2020).

So the Bandwidth Model should organize questions, not replace assessment.

That boundary becomes more important as someone’s functioning declines.

The Moral Problem With Spiky Functioning

There is one final reason I think the spiky profile belongs so centrally in this part of the series.

We moralize inconsistency.

If someone never could do something, we may understand that they need help.

But when we have seen them do it before—or watched them do something we perceive as harder—we start assigning motives.

You can do it when you want to.

You’re not trying.

You’re being lazy.

You’re avoiding responsibility.

But the fact that a person can access a function under one set of conditions does not establish that the same function is equally accessible under another.

And the fact that one difficult function remains available does not prove that every supposedly easier function should also be available.

That does not mean every inconsistency is neurological.

It means inconsistency alone is not enough evidence to tell us why something is happening.

So instead of asking:

“If you can do this, why can’t you do that?”

I think the more useful question is:

“What is different about the conditions in which this function is accessible?”

That preserves scientific uncertainty.

It preserves accountability.

And it preserves the person’s dignity.

What the Science Says—and What the Bandwidth Model Adds

I want to keep this distinction explicit throughout the series.

Research supports substantial variability in human cognitive functioning. Autistic people can show uneven or “spiky” cognitive profiles at the group level, although those profiles are neither universal nor diagnostic (Wilson, 2024). Adults with ADHD show heterogeneous patterns of cognitive functioning rather than one uniform deficit profile (Boonstra et al., 2005). Autistic people may use compensation that makes underlying difficulties less visible (Livingston et al., 2019). And the emerging autistic burnout literature describes profound exhaustion, reduced tolerance, executive and cognitive difficulties, and reduced access to previously available abilities in at least some autistic people (Raymaker et al., 2020; Arnold et al., 2023; Ali et al., 2025).

Research also supports the broader principle that stress can interfere with aspects of higher-order cognitive functioning (Arnsten, 2009).

The Bandwidth Model takes those established and emerging findings and proposes a larger functional framework around them.

It proposes that underlying ability and currently accessible capacity are not identical.

It proposes that pre-existing spikiness may become more visible when the resources used to compensate for it become constrained.

It proposes that peak performance should not be treated as representative or sustainable access.

It proposes that Implementation Cost can become disproportionately important when accessible capacity is low.

It proposes that Context Fit changes the cost of accessing different functions.

And it proposes that recovery itself may be uneven.

Those are Bandwidth Model hypotheses.

They are informed by existing science.

They are not established scientific facts simply because the adjacent literature makes them plausible.

At Profound Depletion, Help Has to Become Easier to Reach

Ultimately, that is where Part 5 lands.

Profound Depletion does not necessarily mean the whole person stops functioning.

Sometimes that is precisely why we miss it.

The person may still think.

Still talk.

Still laugh.

Still create.

Still become deeply engaged in something meaningful.

Still show flashes of remarkable ability.

But those preserved strengths may represent only the highest points on an increasingly uneven functional landscape.

So I do not want to ask only:

“What can this person do?”

I want to ask:

“What can this person reliably access across ordinary daily life, under what conditions, for how long, and at what cost?”

Then support can be designed around the actual profile.

Preserve agency where access remains.

Scaffold where it does not.

Reduce unnecessary Implementation Cost.

And stop requiring recovery to depend entirely on the very capacities that depletion has made hardest to access.

Because:

Capacity is not character.

Potential is not availability.

Ability is not access.

Peak access is not representative access.

Task complexity is not neurological cost.

Support availability is not support accessibility.

And a spiky profile is not evidence of moral inconsistency.

At Profound Depletion, perhaps the most important question becomes:

What does this person still have to carry alone that no longer needs to be carried alone?

Part 6 will move into Early Recovery, because once capacity starts coming back, the next challenge is not simply getting someone functioning again.

It is protecting returning capacity long enough for it to become sustainable.

Darcy Stephens, LPCC

References:

Ali, D., Bougoure, M., Cooper, B., Quinton, A. M. G., Tan, D., Brett, J., Mandy, W., Maybery, M., Magiati, I., & Happé, F. (2025). Burnout as experienced by autistic people: A systematic review. Clinical Psychology Review, 122, 102669. https://doi.org/10.1016/j.cpr.2025.102669

Arnold, S. R. C., Higgins, J. M., Weise, J., Desai, A., Pellicano, E., & Trollor, J. N. (2023). Towards the measurement of autistic burnout. Autism, 27(7), 1933–1948. https://doi.org/10.1177/13623613221147401

Arnsten, A. F. T. (2009). Stress signalling pathways that impair prefrontal cortex structure and function. Nature Reviews Neuroscience, 10(6), 410–422. https://doi.org/10.1038/nrn2648

Boonstra, A. M., Oosterlaan, J., Sergeant, J. A., & Buitelaar, J. K. (2005). Executive functioning in adult ADHD: A meta-analytic review. Psychological Medicine, 35(8), 1097–1108. https://doi.org/10.1017/S003329170500499X

Livingston, L. A., Shah, P., & Happé, F. (2019). Compensatory strategies below the behavioural surface in autism: A qualitative study. The Lancet Psychiatry, 6(9), 766–777. https://doi.org/10.1016/S2215-0366(19)30224-X

Mantzalas, J., Richdale, A. L., & Dissanayake, C. (2022). A conceptual model of risk and protective factors for autistic burnout. Autism Research, 15(6), 976–987. https://doi.org/10.1002/aur.2722

Raymaker, D. M., Teo, A. R., Steckler, N. A., Lentz, B., Scharer, M., Delos Santos, A., Kapp, S. K., Hunter, M., Joyce, A., & Nicolaidis, C. (2020). “Having all of your internal resources exhausted beyond measure and being left with no clean-up crew”: Defining autistic burnout. Autism in Adulthood, 2(2), 132–143. https://doi.org/10.1089/aut.2019.0079

Wilson, A. C. (2024). Cognitive profile in autism and ADHD: A meta-analysis of performance on the WAIS-IV and WISC-V. Archives of Clinical Neuropsychology, 39(4), 498–515. https://doi.org/10.1093/arclin/acad073