CBT for autistic adults is often the first therapy offered and the first therapy that leaves them feeling worse. Cognitive behavioral therapy is the most widely available talk therapy in the world, and for a great many people it works. This post is for autistic and ADHD adults who have sat through a course of it, done the worksheets, and walked out wondering what was wrong with them. It looks at where the unadapted version tends to misfire, what adapted CBT changes, and what to ask before starting again.
A client I will call Rania came to me after two rounds of CBT with two different therapists. She is in her forties, identified as autistic at thirty-eight, and she described both courses the same way: polite, structured, and quietly humiliating. She had filled in thought records for months. Every week she wrote down the thought, the evidence for it, the evidence against it, and a balanced alternative. Every week she arrived with the same conclusion, which was that the evidence for the thought was pretty good. Her second therapist told her she was being resistant. By the time she found me, Rania had stopped believing she was someone therapy could help.
In this blog post, I want to take that mismatch apart carefully, because the conclusion most people draw from it is the wrong one. I will walk through what CBT is actually asking a person to do, then through three specific places where the unadapted version tends to break down: thoughts that were accurate rather than distorted, emotion labelling that assumes access most people never question, and homework that assumes executive function already spent. Then I will look at what adapted CBT changes, and what you can reasonably ask a therapist before committing to another course of it.
3 Key Takeaways
CBT is not the problem. The unadapted version of it, delivered without any accommodation for how a neurodivergent person processes thought, feeling, and demand, is where most of the trouble starts.
Three specific mechanics tend to misfire: treating an accurate social reading as a cognitive distortion, asking for an emotion label when locating the emotion is itself the difficulty, and assigning homework that costs more executive function than the person has left.
Adapted CBT is a real thing with real support behind it, and the adaptations are mostly unglamorous: more concrete language, more visual structure, more time, and less reliance on identifying internal states on demand.
What CBT Is Actually Asking You to Do
Underneath the worksheets, CBT makes a fairly specific request. Notice a thought. Identify the feeling attached to it. Examine whether the thought is accurate. Then act differently on the basis of what you found. The model assumes distress is often maintained by inaccurate or unhelpful thinking, and that changing the thinking changes the feeling and the behavior downstream of it.
For a great many people that request is reasonable and the results are good. None of what follows is an argument that CBT does not work. What follows is an argument about a specific set of assumptions the model quietly makes about the person in the chair.
It assumes you can catch a thought as it happens. It assumes you can name the feeling attached to it with reasonable confidence. It assumes your read of a social situation is more likely to be skewed than accurate. And it assumes that between sessions you have enough planning and initiation capacity left over to do structured written work at home.
Each of those is doing more work than it looks like it is doing. For autistic and ADHD adults, each one can be the exact place the process comes apart.
When a “Cognitive Distortion” Was an Accurate Reading
This is the one that does the most damage, and it is the one Rania hit.
A central move in CBT is identifying cognitive distortions: mind reading, catastrophizing, personalization, all-or-nothing thinking. The therapeutic value comes from noticing that the thought overshot the evidence, then adjusting.
But consider the thought they were uncomfortable around me and I do not know why. For a neurotypical client with social anxiety, that thought often does overshoot. For an autistic adult, it very often does not. Milton (2012) described the double empathy problem: the breakdown in understanding between autistic and non-autistic people is mutual rather than one-sided, with both parties working from different social frames and both misreading the other. The implication matters here. An autistic person who senses that an interaction went badly is frequently sensing something real. What they could not do was explain the mechanism.
So when a therapist labels that perception a distortion and asks for the evidence against it, the client is being asked to argue themselves out of an accurate observation. Some comply, and lose a little more trust in their own perception each time. Some push back, and get described as resistant.
Why this happens
It is usually not a therapist being careless. It is a manual written for a general population meeting a client whose social experience does not match the population the manual assumed. The distortion categories were built around thinking that outruns the evidence. They were not built to distinguish that from thinking that reads the evidence correctly and cannot show its work.
What this can look like
Being told a specific worry is catastrophizing when the thing you were worried about has happened to you repeatedly.
Being asked for evidence against a read of a social situation, and quietly deciding not to mention that you have been right about this before.
Being coached toward a balanced thought that is measurably less accurate than the one you came in with.
Leaving sessions with the sense that being believed depends on being less certain.
What helps
A therapist working well with this separates two questions that unadapted CBT tends to collapse into one: is the perception accurate, and is holding it this way useful right now. Those have different answers surprisingly often. You are probably reading that correctly, and it is costing you a great deal to carry it alone is a sentence that keeps the client’s perception intact and still opens somewhere to go.
Naming the Feeling Assumes You Can Find It
The second step of the standard sequence asks for an emotion label. Most worksheets have a column for it, sometimes with an intensity rating out of a hundred.
For a substantial share of autistic adults, that column is the hardest part of the page. Kinnaird, Stewart and Tchanturia (2019) found in their meta-analysis that 49.93% of autistic participants met the threshold for alexithymia on the Toronto Alexithymia Scale, compared with 4.89% of the neurotypical comparison participants in those studies. The authors’ own framing is that alexithymia is common in autism but not universal, and that figure describes those samples on that instrument rather than the autistic population as a whole. It is still a large enough share that a method built around fluent emotion labelling will meet it constantly.
What that looks like in a session is a person who can describe the situation in precise detail and then go quiet when asked what it was like. The detail was never a dodge. The second question is simply a harder question. They are not avoiding the feeling — they are looking for it and not finding a name that fits.
The body signal underneath
Garfinkel and colleagues (2016) found that autistic participants showed lower objective accuracy in detecting internal bodily signals alongside higher subjective confidence in their ability to do so, and it was the size of that gap, rather than accuracy alone, that predicted anxiety. That is more specific and more interesting than “poor body awareness.” It suggests a system where the internal signal and the read of the internal signal are out of register with each other.
Ask that system for a number out of a hundred and you will get one, because the client wants to cooperate. The number will not mean much. And the work built on top of it will be built on sand.
What helps
Describe the body instead of the emotion. Jaw, chest, hands, temperature, the urge to leave. Physical description is often available when emotional vocabulary is not.
Work from situation and behavior rather than from feeling. What happened, what you did next, what it cost.
Use a written list or a wheel to select from rather than generating a word cold. Recognition is a different task from recall, and a much cheaper one.
Allow “I don’t know” to be a complete answer the session continues past, rather than a gap to be filled in.
Homework Assumes Executive Function You Have Already Spent
Between-session work is not a peripheral feature of CBT. It is a substantial part of where the change is supposed to happen.
It also requires a specific cluster of abilities: remembering the task exists, initiating it without an external prompt, sustaining attention through it, and doing all of that on a day that has already asked a great deal. In my practice, the initiation step is the one that quietly defeats people. Not the difficulty of the task, not the willingness — the moment of starting, unprompted, at the end of a day already spent.
Emotional load compounds it. When a task has become the subject of a repeated conversation about whether you are trying, it costs more to approach than it did the first week. And when the homework is not done, the session that follows frequently becomes a conversation about motivation. For a client who has spent thirty years being told they are not trying hard enough, that conversation lands as confirmation rather than curiosity.
What helps
Do the work in session rather than between sessions, at least at first.
Shrink the task until it is genuinely small, then shrink it again. One thought record a week is not a modest ask for everyone.
Build the external prompt into the plan rather than assuming it. A phone alarm is not a failure of independence.
Treat undone homework as information about the design of the task, not about the person’s commitment. If it did not fit, the fit is the thing to change.
What Adapted CBT Actually Changes
None of this is an argument for abandoning CBT, and I want to be careful here, because the version of this conversation that circulates online often lands there.
Spain, Sin, Chalder, Murphy and Happé (2015) reviewed cognitive behaviour therapy for autistic adults with co-occurring psychiatric difficulties and found reasonable support for it, particularly where the delivery had been adapted. The adaptations that show up consistently in this work are mostly practical rather than theoretical: more concrete and less abstract language, greater use of written and visual material, more structure and predictability in the session itself, more time given to each stage, involvement of a partner or family member where the client wants it, and less reliance on the client identifying and reporting internal emotional states on demand.
Read that list again and notice what it is not. It is not a different theory of change. It is the same model, delivered in a way that does not quietly require the client to be neurotypical in order to participate.
That is the honest version of this topic. Unadapted CBT frequently misfires with neurodivergent adults. Adapted CBT is a legitimate option with support behind it. The gap between those two sentences is where a lot of people have been hurt, and closing it is a therapist’s job rather than a client’s.
What to Ask Before You Start Again
If you are considering another course of CBT, or considering it for the first time, a few questions tend to reveal quickly how a therapist thinks. You are not interviewing them to catch them out. You are finding out whether the fit is likely to work before you spend three months discovering it.
How do you adapt CBT for autistic or ADHD clients? A therapist who does this work will have a specific answer. A blank pause is itself an answer.
What happens if I think a thought you have called a distortion is actually accurate? Listen for whether disagreement is treated as data or as resistance.
I have trouble naming emotions. How do we work with that? This one sorts quickly.
What do you do when homework does not get done? You are listening for curiosity about the task rather than about your motivation.
Have you worked with late-identified autistic or ADHD adults before? Not a disqualifier if the answer is no, but worth knowing.
You are allowed to ask these in a first session. You are allowed to end after a first session. A therapist worth working with will not be offended by any of it.
Where This Leaves You
If two rounds of therapy left you more confused about yourself than when you started, there is a reading of that available which is not I am difficult to help.
A method can be well evidenced and still be a poor fit for a particular nervous system as delivered. Both of those are true at once, and holding them together is more accurate than either alone. You are not obliged to conclude that therapy does not work for you. You are also not obliged to try the same delivery a third time and hope for a different result.
What Rania needed was not a different theory. It was someone who would not ask her to argue herself out of accurate perceptions, who would let “I don’t know what I’m feeling” be a real answer instead of an obstacle, and who would notice when a between-session task was costing more than it returned. Most of what changed for her was structural rather than profound.
Nothing about you is broken. The room was set up for a different person, and nobody told you that.
If you would like to talk through whether a different fit might work, I offer a conversation with no agenda attached. Book a FREE 20-minute “Clarity & Connection” Zoom session
I’m Barbara (Blaze) Lazarony, MA, LMFT #151788. I work with neurodivergent adults and neurodiverse couples.
Client stories referenced in this post are composites, with identifying details changed to protect privacy.
Frequently Asked Questions
Is CBT bad for autistic or ADHD people?
No. The evidence base for CBT is substantial, and reviews of adapted CBT for autistic adults report reasonable support for it. The difficulty is with the unadapted delivery, which makes assumptions about thought-catching, emotion labelling and between-session capacity that do not hold for everyone. A therapist who adapts is offering something meaningfully different from one who does not, even though both would describe their work as CBT.
How do I know whether my thought was a distortion or an accurate reading?
Often you cannot know for certain in the moment, and that uncertainty is real rather than a failure. What helps is a therapist willing to hold both possibilities open instead of assuming the thought overshot. If you have a track record of being right about a particular kind of social read, that track record is evidence and belongs in the conversation.
My therapist says I am resistant. Am I?
Resistance is a real phenomenon, and it is also a label that gets applied when a method is not fitting and nobody has noticed. If the specific thing you are being called resistant about is disagreeing with a distortion label, struggling to name emotions, or not completing homework, it is worth raising the fit question directly before accepting the description.
What should I try instead if CBT has not worked twice?
There is no single correct answer, and it depends on what you are working on. Adapted CBT with a therapist who genuinely does the adapting is often the shortest path, since the model itself may not have been the problem. Other approaches place less weight on identifying internal states or on between-session written work. It is worth discussing openly with a clinician rather than deciding alone.
Does this apply to ADHD as well as autism?
Some of it applies to both and some of it splits. The homework and executive function difficulty is most pronounced in ADHD. The emotion labelling difficulty is most researched in autism, though not exclusive to it. The distortion problem shows up in both, for somewhat different reasons. AuDHD adults frequently run into all three.
References
Garfinkel, S. N., Tiley, C., O’Keeffe, S., Harrison, N. A., Seth, A. K., & Critchley, H. D. (2016). Discrepancies between dimensions of interoception in autism: Implications for emotion and anxiety. Biological Psychology, 114, 117–126. https://doi.org/10.1016/j.biopsycho.2015.12.003
Kinnaird, E., Stewart, C., & Tchanturia, K. (2019). Investigating alexithymia in autism: A systematic review and meta-analysis. European Psychiatry, 55, 80–89. https://doi.org/10.1016/j.eurpsy.2018.09.004
Milton, D. E. M. (2012). On the ontological status of autism: The ‘double empathy problem’. Disability & Society, 27(6), 883–887. https://doi.org/10.1080/09687599.2012.710008
Spain, D., Sin, J., Chalder, T., Murphy, D., & Happé, F. (2015). Cognitive behaviour therapy for adults with autism spectrum disorders and psychiatric co-morbidity: A review. Research in Autism Spectrum Disorders, 9, 151–162. https://doi.org/10.1016/j.rasd.2014.10.019
