Does CBT help autistic children handle big feelings and anger?

What 9 sources say about talk-and-practice programs for big feelings, anger and outbursts — the evidence, the caveats, and what it means for your family.

Research synthesisUpdated 2026-10-019 sources includedProtocol v4.5 (Amdt. v4.6)
CBT for emotion regulation Not enough research yet — one small trial could be read

A care note before you read. These programs teach children ways to notice and handle big feelings like anger. They are talk-and-practice programs, often with a parent taking part. If your child has meltdowns that put anyone at risk, please talk with your child’s doctor or psychologist.

Podcast·Two-voice deep dive

Listen to the discussion

0:00
Chapters & transcript

This is an AI-generated discussion of this article, transcribed automatically. The written summary below is the reviewed source — if the two differ, go by the summary.

0:00

What is this discussion about?

Host A Imagine you're trying to stop a runaway train.

Host B Okay.

Host A But the only tool you actually have in your hands is, like, a therapy manual.

Host B Right. That's not going to do much.

Host A No, exactly. And for families navigating life with an autistic child, the sudden intensity of a meltdown can, you know, feel exactly like that.

Host B Oh, absolutely.

Host A We're talking about those moments that completely shift the gravity in the room. I mean, the sudden anger, the intense outburst, the overwhelm. It's something so many families navigate every single day. But when they go looking for clear scientific answers on how to help their child manage those huge feelings, they often just, well, they hit a wall of confusion.

Host B Yeah, a massive wall.

Host A So today, we are diving into the source material to look at exactly that. We're doing a deep dive into cognitive behavioral therapy, or CBT, specifically tailored for emotion regulation in autistic children and adolescents.

Host B And that distinction of emotion regulation is really vital here. We are not talking about therapy for general anxiety today. We're not looking at social skills training either. We're looking specifically at programs that were designed from the ground up to target anger and emotional outbursts. We want to see what the evidence actually says.

Host A Yeah, and to do that, we are drawing on a very specific set of sources for this deep dive. We have an internal audit document and a findings report from a demo grade practice run by an organization called Spectrum Connect.

Host B Right.

Host A But before we go one step further, we need to make something crystal clear for you. This deep dive is strictly a summary of research findings.

Host B Exactly.

Host A It is not medical advice, and absolutely nothing we discuss here today is a recommendation for any individual child.

Host B Yeah, our goal is simply to look at what the data says, how the science was built, and, I mean, just as importantly, what the data completely fails to answer.

1:46

What is CBT for emotion regulation, and which programs use it?

Host A Okay, let's unpack this, because before we look at the results of these studies, we have to understand what kind of therapy we are even talking about. Because, you know, CBT is a huge umbrella term. If you type CBT for autism into a search engine, you're going to get a million different things.

Host B Which is exactly why the audit document we're looking at is so specific. It explicitly routes out CBT meant for anxiety or social skills because those are handled separately. Right. The intervention we are defining today is manualized CBT targeted strictly at emotion regulation.

Host A And when you say manualized, what does that actually look like?

Host B So when we say manualized, we mean there's a specific step-by-step guidebook that the therapist follows. Week one covers this specific concept. Week two covers the next and so on. The primary goals are very focused. Learning to calm down, handling anger, and having fewer outbursts or meltdowns.

Host A And the sources actually name the specific programs that fall into this category.

Host B Yeah, they do.

Host A We're talking about programs with names like Secret Agent Society, Operation Regulation, which is often abbreviated as SAS:OR. Right. There is also one called Exploring Feelings, another called STAMP, which stands for Stress and Anger Management Program, and one called Regulating Together.

Host B Mm-hmm.

Host A But how do these actually work? Like, what is the mechanism here? Because telling a child who's having a full blown meltdown to just calm down obviously doesn't work.

Host B It definitely doesn't. And that's why these aren't just, you know, open-ended conversations where a therapist sits there and asks, how did that make you feel? Right, right. These are highly targeted talk and practice programs. They teach a child how to recognize the physiological signs of anger before that runaway train leaves the station.

Host A Wow, okay.

Host B For example, in a program like Secret Agent Society, they use a spy theme to make the concepts really concrete for the kids.

Host A A spy theme?

Host B Yeah. A child might learn to detect their body's physical clues like clenched fists or a racing heart or a tight chest as warning signals.

Host A Okay, that makes sense.

Host B And once they identify the clue, they practice specific structured coping mechanisms to de-escalate their own nervous system.

Host A That's fascinating. And the sources give us a great example of the structure using that secret agent society operation regulation program. The typical intensity for that specific program involves 10 weekly one-hour individual visits. But here is the detail that really stands out to me. The caregiver is a co-participant in these sessions. They aren't just sitting in the waiting room scrolling on their phone or answering emails. They are in the room actively participating.

Host B And think about the mechanism of change there. The intervention isn't treating the child in a vacuum. Right. By having the parent in the room, the therapy is actively creating a shared language between the child and the caregiver. The parent is literally learning the exact same spy-themed coping strategies and emotional vocabulary.

Host A I mean, if the parent and child are building the shared vocabulary in the clinic, it sounds incredibly promising.

Host B It does.

Host A It means when a meltdown is brewing in the middle of like a busy grocery store, the parent knows the exact secret agent prompts to use.

4:52

How much research is there?

Host A But knowing how to spot a feeling in a calm therapy room is very different from stopping a meltdown in the real world. So do we have hard data showing this actually works? How did researchers actually prove these talk and practice programs are effective?

Host B Well, to understand the evidence, the audit document forces us to look at the volume and the type of studies.

Host A Okay.

Host B The sources note mentions of case reports, single group studies, and randomized controlled trials. But what the researchers could actually verify when auditing a specific program like SAS:OR is much narrower. How so? For instance, the developer's website, which is the lab at York University, lists six peer-reviewed research papers for their program.

Host A Wait, really? The sources show they list six published papers?

Host B Yep, six.

Host A Because if you're a parent looking for help and you see six peer-reviewed papers, you assume this is a massive, highly researched franchise.

Host B No, of course you do.

Host A But looking at the audit, all six of those papers stem from one single trial of 68 kids.

Host B Yeah.

Host A It's like seeing six different movie posters and thinking it's this massive cinematic universe, but it's literally just the exact same movie marketed six different ways.

Host B It's the perfect way to look at it. And it really highlights a major issue in how evidence is presented to families versus how it is evaluated by an auditor.

Host A Right.

Host B Those six papers might analyze different secondary angles of the data. You know, maybe one looks at parent stress, another looks at the kid's self-reporting. But from an evidence standpoint, it counts as exactly one test from one single non-independent lab.

Host A Wow.

6:26

Who was in the one trial we could read?

Host A Okay, so who was actually in this one movie? Who were the 68 kids in that specific trial, the Weiss 2018 trial?

Host B The demographics are highly specific. The trial studied children with an average age of about 9.75 years old, ranging from 8 to 12.

Host A Okay.

Host B They all had a formal autism diagnosis. And crucially, they all had what the study defined as, quote, at least average language skills.

Host A Interesting. And how did they actually set up the test?

Host B They split the 68 kids into two groups. Well, 69 were recruited, but 68 were randomized. Got it. 35 children received the active treatment, so the 10 weeks of the CBT program. The remaining 33 children were placed on a wait list.

Host A Okay, so therapy versus wait list.

Host B Exactly. And the researchers measured emotion regulation as the primary target, but they also measured externalizing behaviors.

Host A Let's define that real quick for everyone. Externalizing behaviors are essentially outward actions, right? Like the shouting, the throwing things, the visible aggression.

Host B Precisely. They measure those outward expressions of anger, and they included a follow-up assessment to see if any positive changes lasted after the 10 weeks of therapy ended. Right. But this is where the audit hits a significant roadblock. The Weiss 2018 trial is the one we have readable data for. However, the audit document notes that other trials for other programs absolutely do exist.

Host A Oh, they do?

Host B Yeah. There's a trial for exploring feelings from 2007, and there are pilot studies for regulating together.

Host A But the digital scrapers running this audit couldn't actually access them, right?

Host B Exactly.

Host A Yeah.

Host B They hit strict publisher paywalls and server blockers from major academic publishers, you know, places like Wiley, Guilford, Springer, and BMC. They were getting HTTP 403 and 429 errors.

Host A Oh, man. It's like trying to map an ocean, but we are only legally allowed to look at one single puddle.

Host B That's exactly what it is.

Host A We know the ocean of data is out there, but we're completely blocked from seeing the results of those other emotion regulation programs, which means we really have to narrow our focus down to the puddle we can actually see.

Host B Unfortunately, yes.

8:39

Did the program help children handle big feelings?

Host A So what happened to the kids who took the secret agent society operation regulation program in that Weiss trial?

Host B In that specific 2018 trial, the findings were positive.

Host A Oh, good.

Host B Yeah. The children who went through the program improved significantly more than the children sitting on the wait list.

Host A Okay.

Host B They showed improvements on measures of emotion regulation and those externalizing behaviors. And the sources note that these gains were maintained at that later follow-up check.

Host A Okay, that's really promising.

9:06

What does the big 2024 review of talk therapies show?

Host A And the sources also bring in some broader context, though. There is a 2024 meta-analysis by a researcher named Rosenau.

Host B Yes, the Rosenau meta-analysis is a massive review. It looked at 29 randomized controlled trials of psychotherapy for autistic youth, covering a total of 1,464 participants.

Host A Wow, that's a lot bigger.

Host B Huge difference. And that review found a small overall effect size of 0.38 across all therapies and a medium effect size of 0.59 specifically for externalizing behaviors.

Host A Hold on, though, because if that meta-analysis pulls all psychotherapies together, I mean, everything from treating anxiety to building social skills using dozens of different methods, how can we possibly credit that medium effect score to these specific anger management programs?

Host B We can't.

Host A We can't?

Host B Nope. You've spotted the exact trap the audit warns about. You cannot take a class-level benefit, a score that blends dozens of different therapies, populations, and goals, and magically transfer it to validate one specific manualized program.

Host A Ah, okay.

Host B The meta-analysis provides excellent background context, showing that talk therapies, as a broad category, can have an effect. But it does not prove that this specific emotion regulation program works better than anything else.

Host A That makes total sense.

10:30

Are there side effects?

Host A What about the downside, though? Did the studies mention any adverse effects, dropouts, or harms? Like, did the therapy make anyone's meltdowns worse?

Host B This is a glaring omission in the capture data. The sources clearly state that they found zero information regarding adverse events, dropouts, or deterioration.

Host A Wait, really? Nothing at all?

Host B Nothing. We simply do not have the data in this summary on whether anyone experienced negative side effects or if the demands of the therapy caused, like, increased stress for the child or the family.

Host A Wow. So we have one trial that looks like a win on paper, but we have no data on harms. We can't borrow the good grades from the massive meta-analysis. And we only have 68 kids from one lab.

11:14

Why is the evidence rated very low?

Host A So how does the scientific community and how would a clinician actually evaluate the strength of this evidence?

Host B This is where we examine the limitations stated in the sources. And honestly, the foundation here is quite fragile. Okay. First, we have the waitlist design. Because the comparison group was simply waiting, they weren't receiving a different active treatment.

Host A But why is a waitlist such a weak comparison? Doesn't it show that the kids who got therapy did better than the kids who got nothing?

Host B It does, but it introduces the active ingredient problem.

Host A The active ingredient problem?

Host B Yeah. We don't know if the structured CBT manual was the active ingredient that helped the child, or if the kids just got better because they were spending an hour a week receiving dedicated, undivided attention from a highly trained professional.

Host A Oh, I see.

Host B Right. Like, if the waitlist kids had been spending an hour a week just playing board games with a therapist, would they have improved just as much? The study design simply can't answer that.

Host A Wow, that's a huge point. And second, we are relying on a very small sample size coming from a single lab at York University.

Host B Exactly.

Host A The developers of the program are the ones testing the program. There's no independent replication in the data we could read.

Host B Nope.

Host A But wait, getting back to the mechanics of the therapy, earlier we said the parents attended every single session?

Host B Yes, they're in the room.

Host A They are co-participants learning the spy language. If those same parents who just spent 10 weeks of their lives sitting in therapy are the ones filling out the outcome rating forms at the end, isn't that a massive conflict?

Host B It is a critical flaw, and it is precisely what the audit flags. Let's look at the psychology of the parent rater bias here. If a parent sacrifices 10 weeks driving to a clinic, participating in an hour-long session, and practicing the tools at home, they have a massive psychological investment in the therapy working.

Host A Oh, absolutely.

Host B It's basic cognitive dissonance. When they're handed a survey at the end asking if their child's behavior improved, they are highly motivated, even unconsciously, to see progress. They want their hard work to mean something.

Host A They're looking for the wins.

Host B Exactly. Without blinded assessors, meaning independent experts or teachers who do not know which kid got the therapy and which kid was on the wait list, the risk of bias jumps to high.

Host A High risk of bias.

Host B Yeah. Wow. The parents' perception of their child's improvement is fundamentally entangled with their own involvement in the intervention.

Host A So putting on a clinician's lens, how do they weigh all of these factors, the positive results against the weightless flaws and the parent bias?

Host B A clinician looks at this and sees a positive signal. There is a plausible mechanism. And the initial data hints that it might be helpful. Right. But they recognize the foundation of that evidence is rated as very low certainty or basically insufficient. Insufficient. To strengthen their confidence, a clinician wants to see these exact programs tested by independent labs, teams completely unaffiliated with the developers.

Host A Makes sense.

Host B They want much larger sample sizes. And crucially, they want to see blinded raters evaluating the outcomes to remove that parent rater bias. The clinician doesn't dismiss the program, but they understand the science is in its infancy.

14:30

Does it work for teens, younger children, or children who speak few words?

Host A Which brings us to the gaps in the map. We know what we know based on this one trial, but what does the evidence not answer? What are the unknowns?

Host B The most significant gaps are in representation.

Host A Okay.

Host B The children in the Weiss trial were all 8 to 12 years old, and they all had at least average language skills.

Host A Which leaves out enormous portions of the autistic community.

Host B Huge portions.

Host A We have absolutely no idea from these sources if these emotion regulation programs work for teenagers dealing with, you know, the complexities of puberty and high school stress.

Host B Right.

Host A We have no idea if they work for young children under eight. And we have absolutely zero evidence here regarding minimally speaking youth or youth with intellectual disabilities. They were completely unrepresented in the captured data.

Host B Yeah, and we should frame these not as reasons to avoid the therapy, but as massive blinking gaps in the science.

Host A Definitely.

Host B We don't know if the therapy doesn't work for them because researchers just haven't included them in the trials we could read. And the other glaring unknown brings us back to our ocean and puddle metaphor.

Host A The paywalls blocking the other studies.

Host B Yep. The audit explicitly notes that anger-specific tests from 2007 and newer protocols for programs like Regulating Together do exist.

Host A They're out there.

Host B They are. There might be vast amounts of long-term data, larger samples, and answers to these very representation gaps sitting out there right now. But because they remain locked behind publisher paywalls or were blocked by server errors during this run, those questions remain completely unanswered in this summary.

Host A That is so frustrating.

16:06

What are the key takeaways?

Host A So what does this all mean for you? Let's recap the crucial points you should take away from this deep dive today. We have five key facts.

Host B Okay.

Host A Number one, emotion regulation CBT specifically targets anger and meltdowns. It uses manualized talk and practice sessions, teaching kids to recognize physiological clues, and often requires a parent to be present as a co-participant.

Host B Number two, one fair test of 68 children showed that kids in the program got better at handling big feelings compared to those who were just placed on a wait list.

Host A Right. Number three, broader reviews do show small to medium positive effects for psychotherapy in general for autistic youth, though we cannot apply that broad score directly to this specific program.

Host B And number four is the major limitation to remember. The current readable evidence for these specific emotion regulation programs is incredibly thin. It relies on a single, small lab test where parent participation might have significantly biased the outcome ratings.

Host A Which brings us to number five, the open question. There is currently a complete lack of evidence in this audit regarding how these programs affect teens, minimally speaking kids, or younger children.

Host B That's right. As we wrap up, we have a very important message that we need to deliver exactly as intended. Decisions about any individual child belong in a conversation with that child's pediatrician or care team. This deep dive cannot make that call for you.

Host A Absolutely.

Host B And research on this topic continues. Spectrum Connect will update its coverage as new evidence appears and as those digital roadblocks are bypassed.

Host A Spectrum Connect translates autism research into plain language for parents and checks findings for accuracy against misinformation. If you are a credentialed expert, an MD, PhD, PsyD, or DO, and would like to help validate our findings, you can email research@spectrumconnect.us.

Host B I will leave you with this final thought to mull over. In an age of endless information, consider the irony we uncovered today. Sometimes the biggest barrier to understanding a scientific treatment isn't the science itself, but the digital paywalls and gatekeepers that prevent us from actually reading the results. Who really gets to access the science that shapes our care?

Key Takeaways

One fair test found the program helped. A 2018 trial from Canada tested Secret Agent Society: Operation Regulation in 68 children, average age about 10. Children in the program improved more than children who were waiting on measures of emotion regulation, and the gains were still there at a later check-up. We could not read how big the gains were.

One test is not enough to be sure. It was small, it came from one research lab, and families knew whether their child was in the program or waiting. Other tests of these programs exist, but we could read the results of only one. This status does not mean the programs fail. It means we cannot yet say how well they work.

A large 2024 review is not proof for any one program. It pooled 29 trials of many kinds of talk therapy for autistic youth (1,464 young people) and found a small overall effect (0.38), and 0.59 for behavior problems. It mixes many therapies and goals, so we did not count it as proof for any single program.

We found no information about side effects. No source we could read reported on harms or on children getting worse, so we cannot say these programs are safe or unsafe. That is a gap in the research, not a warning sign.

If you are thinking about a program, it is reasonable to ask who runs it, what training they have, what age and language level it was tested with, how progress will be checked, and what the parent is asked to do.

What this means for you

Cognitive behavior therapy (CBT) teaches people to notice their thoughts, feelings and body signals, and to try new ways of coping. Some CBT programs for autistic children aim mainly at emotion regulation: calming down, handling anger, and having fewer outbursts or meltdowns. Programs with this aim include Secret Agent Society: Operation Regulation (from a lab at York University in Canada), Exploring Feelings, the Stress and Anger Management Program (STAMP), and Regulating Together. This page is only about feelings, anger and outbursts. Worry and anxiety, and social skills (see PEERS), are covered in other reports and are not counted here.

The one program test we could read found that children handled their feelings better than children who were waiting. But several other tests exist, and our review could open the results of only one of them, so “not enough research yet” partly reflects what we could read, not only what research exists. In that test the comparison group was children waiting for the program, not children getting a different kind of help. We also do not know who filled in the rating forms. That matters, because a parent who takes part in every session may rate things differently from an outside observer.

One small fair test found children improved at handling feelings, but one test is not enough to be sure.

Who was studied. In the test we could read, children aged 8 to 12 with a formal autism diagnosis and at least average language skills took part in 10 weekly one-hour visits, with a parent at every visit, in Canada. We do not know how well these programs work for younger children, teenagers, or children who speak few or no words.

Where the studies landed

Not enough research yet

1 Points toward it helping

8 Results not read, or not specific to these programs

Tap the cover to open it. Most covers are records we found but could not read the results of. They are shown so you can see what exists, not because we judged them.
One studyReview of many studies

Tap any tile to read it. Most tiles are records we found but could not read the results of. They are shown so you can see what exists, not because we judged them.

Points toward it helping1
One randomized trial from Canada (2018, 68 children) found children who took part in Secret Agent Society: Operation Regulation improved more than waiting children on emotion regulation and some other mental-health measures, and the gains held at a later check-up. We read only the summary of this trial, not the full paper, so we could not see the size of the gains. The program’s lab lists six research papers, but they come from the same group of children, so they count as one test.
Results not read, or not specific to these programs8
Five records point to program tests: an anger-management test for children with Asperger syndrome (2007), STAMP, two papers on Regulating Together (one a telehealth pilot), and a plan for a new Regulating Together trial. We found them but could not read their results, and the new trial has no results yet. Three are reviews: the 2024 review of many talk therapies (a small overall effect, but not specific to any one program), a 2026 review of CBT for emotion regulation, and a review comparing emotion regulation approaches, both of which we could not open. None of these tells us the programs do or do not work.

Spectrum Connect reviews published research on interventions parents are exploring for their autistic children — so you can see where the evidence actually stands. No agenda, no selling, no cherry-picking. Just the studies, our method, and what it means for you.

See the research behind this For anyone who wants to check our work · 9 sources
01

Where we looked

We did not search the research databases directly. This review used two web searches and ten attempts to open pages, and five of those pages opened. PubMed and similar pages blocked automated access, and several publisher pages refused to open. A full search would cover PubMed, Embase, APA PsycINFO, the Cochrane Library, ERIC, Epistemonikos, PROSPERO, and trial registers (ClinicalTrials.gov, ISRCTN and ANZCTR). Below is the PubMed search built for that purpose. You can run it yourself, but our findings do not rest on its results.

("Emotional Regulation"[MeSH] OR "emotion regulation"[tiab] OR "emotion dysregulation"[tiab] OR anger[tiab] OR irritab*[tiab] OR outburst*[tiab] OR meltdown*[tiab] OR tantrum*[tiab]) AND ("Cognitive Behavioral Therapy"[MeSH] OR "cognitive behavio*"[tiab] OR CBT[tiab] OR "Secret Agent Society"[tiab] OR "Exploring Feelings"[tiab] OR "Regulating Together"[tiab] OR "Stress and Anger Management"[tiab]) AND ("Autism Spectrum Disorder"[MeSH] OR autis*[tiab] OR asperger*[tiab]) AND (child*[tiab] OR adolescen*[tiab] OR youth[tiab]) Run on PubMed →

Per-database record counts are not available, because the database searches were not run. No protocol for this review was registered in advance.

02

What we did with what we found

23found (18 web results and 5 leads handed in as titles)
−12duplicates, write-ups of studies already listed, and off-topic items
11kept after a title-level look
−2a program page and a trial registry record, counted with the trials they describe
9sources included (6 trial records and 3 reviews)

This is not a standard screening flow. Items were judged by title and were not screened in duplicate. No full trial papers were read. We read a summary of one trial, a publisher page for one review, and the program’s own web page.

03

What the strongest evidence says

Handling big feelings (emotion regulation)

One small fair test from Canada found children in the program improved more than waiting children, and the gains held at a later check-up. The size of the gains was not read.

How sure
Very low
Anger and outbursts

A fair test of an anger-management program exists (2007), but we could not read its results. The 2018 trial’s other mental-health measures are not clearly about anger.

How sure
Very low
Side effects and safety

No source we could read reported on side effects or on children getting worse.

How sure
No data
Talk therapies in general for autistic youth

A 2024 review of 29 trials found a small overall effect (0.38), and 0.59 for behavior problems. It mixes many therapies and goals, so it is not counted for these programs.

How sure
Low (all therapies)
Ray Kawai · Protocol v4.5 BCAT · Open record
Spectrum Connect is not a medical provider, and nothing here is medical advice. This page shows where the research stands and how we got there. It is not a recommendation, and it is not a substitute for your child’s doctor or care team. What you do with it is yours to decide, together with them.

What would you like to do?

Want to talk to a parent who’s been there?

We match you with families whose kid is like yours. Free, takes two minutes.

Join Free
Heard of something we haven’t covered?Suggest an article

If another parent, a clinic or a video mentioned a therapy you can’t find here, tell us. We’ll read the research and review it next.

Suggest an article
Think we got something wrong?Report a factual error

We publish the whole record so it can be checked, and that only counts if we act on what you find. If a number looks wrong, a study is missing or has been retracted, or we’ve read a finding in a way the evidence doesn’t support, tell us.

You don’t need a research background to file one. “This doesn’t match what our doctor told us” is a useful report. Every one reaches a person: we reply within seven days, and within thirty we have either corrected the page or told you when we will. Substantive reports send the affected steps back through the protocol and need fresh sign-off before anything here changes.

Report a factual error
Are you a clinician or researcher?Review this article

If you have read this and the evidence holds up, you can sign off on it. We send you a personal review link where you confirm your credentials, and our team checks every review before anything is published. Once approved, your name and a link to your profile appear on this article.

Ask to review this