The therapeutic relationship with autistic adults: reflections on an encounter between neurotypes

In short

This text is for psychologists and is about the therapeutic relationship with autistic adults. There is little research on this topic, and the text explains why. The existing studies show that autistic people ask for therapists who understand autism, speak clearly and ask for feedback regularly. The text also discusses the difficulties that therapists themselves experience.

On the one hand, we say that not everything that comes into the net is fish. But it is also true that fishing together with a net is not just a matter of casting it and waiting for the fish. Of course, we are not really talking about fishing, but about the therapeutic relationship, and more specifically about the therapeutic relationship with autistic adults.

Over the years, I have observed that psychologists who train in the field of autism, and specifically in adult autism, are becoming increasingly able to suspect that the person in front of them may be someone with this diagnosis, or better able to carry out a diagnostic assessment of this condition, for example. However, when it comes to the therapeutic relationship, many of the difficulties seem to persist. Silences that last an endless amount of time in sessions, the challenges felt around insight, social reciprocity, etc., can make psychological support for autistic adults a challenge for many of us.

The therapeutic relationship is probably the most robust common factor we know of in psychotherapy. However, almost everything we know about it was learned with non-autistic clients. To a large extent, it was also described by non-autistic therapists.

When an autistic adult walks into the consulting room, two different ways of perceiving, communicating and relating meet. This text starts from a simple question: what happens to the therapeutic alliance when this encounter is between different neurotypes?

The reflection has three threads. The first is what research already tells us about the alliance with autistic adults. The second is why it tells us so little. The third is the challenges that psychologists and psychotherapists feel, often in silence, when working with this population.

The expression "autistic person" is used here, as preferred by most of the autistic community (Kenny et al., 2016), and ASD (autism spectrum disorder) when referring to the diagnosis. In clinical practice, the rule is a different and simpler one: ask the person.

What we know about the therapeutic alliance

The alliance explains a modest, but consistent part of psychotherapy outcomes. In a meta-analysis of 295 studies, with more than 30,000 patients, the association between alliance and outcome was r = .28 (Flückiger et al., 2018). The value held across theoretical models, measures, rating perspectives and countries.

Bordin (1979) proposed a definition that is still the most widely used. The alliance rests on three elements: agreement on goals, agreement on tasks and a bond of trust. The definition is transtheoretical, which makes it useful for thinking about any approach.

More recent research has shifted the focus from the alliance as a state to the alliance as a process. Ruptures are frequent and are not, in themselves, a bad sign. What is associated with better outcomes is the ability to repair them: r = .29 in a meta-analysis of 11 studies (Eubanks, Muran & Safran, 2018).

We also know that therapists differ in their ability to build alliances, and that this difference weighs more than that of clients (Del Re et al., 2012). This idea is central to the rest of the text. When the alliance with an autistic person fails, the useful question is not only "what is the client lacking?". It is also "what is failing in me and between us?".

A word of caution is in order. The relationship between alliance and outcome is correlational. The alliance improves when the client improves, and it is not always clear which comes first.

The alliance with autistic adults: what changes

An alliance with autistic people is possible and, by all indications, is also associated with outcomes. For a long time the opposite idea was taken as self-evident, but the few data available do not confirm it. Kerns et al. (2018) found, in autistic young people in CBT for anxiety, alliances comparable to those of non-autistic young people. Brewe et al. (2021) observed that the alliance fluctuates over the course of the intervention, dipping in the middle and rising again. Early alliance predicted less dysphoria at the end. In both cases, however, we are mostly talking about adolescents and young adults.

A mismatch that takes two

The concept of double empathy (Milton, 2012) changes the clinical question. Communication difficulties between autistic and non-autistic people are not a one-sided deficit. They are a mismatch between styles, and the responsibility is shared.

In the therapy room this has concrete consequences. Silence may be read as resistance, averted gaze as disinterest, a literal answer as defensiveness. The therapist misreads, just as the client may misread. The difference is that the therapist usually has more power to define what counts as "normal" in the relationship.

The same conditions, expressed more explicitly

In Hume's (2022) study, with 17 autistic adults and 3 therapists, the need for a different relationship did not emerge. What emerged was the need for the same Rogerian conditions, expressed more intensely and more explicitly: congruence, empathy and unconditional positive regard.

Many participants described a heightened sensitivity to the therapist's emotional climate. "Performed" empathy, artificial rapport techniques or disguised irritation were detected and damaged trust. Positive regard gained strength when it was stated directly and shown in practice: remembering an interest, preparing the session, adjusting the room.

What autistic people ask for

Pappagianopoulos et al. (2025) interviewed 19 autistic adults about their experiences in therapy. The most frequent themes were a therapist who understands autism and a flexible, collaborative approach. These were followed by validation, space to talk and, to a lesser extent, adaptations in communication, structure and the sensory environment.

To a large extent, this is general good practice. What changes is that it needs to be informed about autism: camouflaging, a history of invalidation, sensory processing, a different pace.

Camouflaging and silent ruptures

Camouflaging is perhaps the greatest hidden risk. A person may seem cooperative, punctual and satisfied while making an enormous effort to seem so. The apparent alliance may then be exactly the opposite of the real alliance.

Withdrawal ruptures, which are already hard to detect with any client, become almost invisible here. That is why asking for concrete and regular feedback, as suggested by the IACP/AsIAm guidelines (2025), is no longer an extra. It becomes a condition for knowing whether an alliance actually exists.

A model proposed by Robinson (2020), based on emotion-focused therapy with autistic adolescents, suggests yet another reading. Misunderstandings can be used as therapeutic material, rather than being mere obstacles. The model is preliminary and locates ruptures mainly in the client's difficulties, which puts it in tension with the idea of double empathy.

Why we know so little

The scarcity is real and measurable. A recent scoping review identified only 15 studies on the therapeutic alliance and autism, with 657 participants aged between 7 and 77 (Losh et al., 2026). None used standardised alliance measures with adults over 21. For the general population, the meta-analysis by Flückiger et al. brought together almost 300. There is no single explanation. There are several, which add up and reinforce one another.

1. A history that made the relationship unthinkable

Since Kanner (1943), autism has been described as a "disturbance of affective contact". For decades, the very idea of a therapeutic bond seemed to contradict the diagnosis. If the autistic person is defined by difficulty in relating, studying their relationship with the therapist seemed a question without an object.

The theory of mind deficit hypothesis reinforced this reading. The idea that autistic people lack empathy drew interest away from relational processes, which are precisely made of reciprocal empathy.

2. Two fields that do not talk to each other

Research on the alliance was born in the psychodynamic and humanistic traditions. These traditions have a difficult history with autism, including the blaming of mothers in the 1950s and 1960s.

Autism research, for its part, grew mainly in the fields of development, biology and behavioural interventions. In these models, the focus is on techniques and target behaviours, and the relationship is a means rather than an object of study. The result is a no man's land between two fields.

3. Research focused on childhood and biology

Autistic adults were, for a long time, an almost invisible population. Efforts were concentrated on early diagnosis and intervention in children. The existing studies on the alliance themselves are mostly with children and adolescents.

In the United Kingdom, between 2007 and 2011, 56% of autism research funding went to biology, brain and cognition. Only 5% went to services (Pellicano, Dinsmore & Charman, 2014). Autistic people and families, when consulted, asked for exactly the opposite: research with an impact on daily life and services.

4. Instruments that were not made for this population

The most widely used alliance measures ask the client to rate, on abstract scales, the trust and agreement they feel. For a person with alexithymia, with more literal processing or who camouflages, these questions may measure something else: the effort to answer "correctly".

Observational measures have the opposite problem. They rely on non-verbal cues (eye contact, facial expression, tone of voice) calibrated for non-autistic people. An autistic person may be deeply engaged and appear distant. Measuring poorly leads to concluding poorly, and the difficulty of measuring discourages those who want to study the topic.

5. People who do not appear in samples

Many autistic adults are only diagnosed late, especially women and people who camouflage. They are probably present in the samples of general psychotherapy research, but without being identified. At the same time, clinical trials often exclude people with neurodevelopmental conditions in order to obtain "homogeneous" samples.

6. A question that took a long time to be asked by those who live it

For a long time, research was done on autistic people, not with them. Participatory research and the neurodiversity movement brought other questions: what makes therapy safe, what makes it invalidating, what makes a good therapist. Most of the qualitative studies cited in this text are less than five years old.

A different perspective

Some argue that this scarcity is less serious than it seems. If the alliance is a transtheoretical and transdiagnostic factor, the general findings should also apply to autistic people.

The problem is that this generalisation has never been tested. Qualitative data suggest that the same principles apply, but that the way of expressing them changes. And it is precisely this way that we do not know well.

The challenges felt by psychologists and psychotherapists

Therapists also feel the mismatch, and they rarely talk about it. The little research with therapists points to a combination of lack of training, insecurity and relational fatigue.

We lack maps

Many psychotherapists report having had no training on adult autism or on how to adapt their interventions (Lipinski et al., 2022). In a study with community service clinicians, the phrase that gave the article its title was "I wouldn't know where to start" (Maddox et al., 2020).

The lack of training has a relational effect. An insecure therapist tends to take refuge in technique, to avoid the person or to refer them on. The autistic person, who has often already seen several professionals, senses this insecurity.

Losing our usual tools

Much relational work rests on cues we have learned to read without thinking: gaze, expression, tone of voice, the emotional response to our interventions. With many autistic people these cues work differently. The therapist may be left not knowing whether the person is connected, bored, hurt or simply thinking.

This creates a particular insecurity: the feeling of not being "in contact". It is tempting to read this feeling as a characteristic of the client. Double empathy suggests that it is, at least in part, a limit of our own reception.

The therapist's own fatigue and overload

In Brook's (2022) study, relational therapists described sessions dominated by "tangents", by a high density of information or by a pace different from their own. They also described the feeling of sessions fragmented between "talking about autism" and "talking about life". The response they found was to persist and return to basic skills.

The metaphor they proposed is useful: alternating between a zoom in, in which autism comes to the foreground, and a zoom out, in which it stays in the background while the human experience is worked on. The challenge is knowing when to do each.

Tensions with one's own theoretical model

Some adaptations requested by autistic people clash with training habits. Giving a direct opinion, making explicit that one likes the client, providing structure and written instructions, or sharing something about oneself may feel "too much" for those who learned neutrality or non-directiveness.

The reverse also happens. A strongly technique-oriented therapist may feel that the person needs, above all, time to talk and to be heard without hurry to solve things. In both cases, the therapist has to make the model more flexible without abandoning it.

Countertransference and overidentification

Therapists with a personal connection to autism, because they are autistic or have autistic family members, report finding attunement easier (Brook, 2022). They also report the risk of overidentification: assuming that the client's experience is the same as their own.

At the other extreme, the non-autistic therapist may feel frustration, helplessness or even boredom, and feel guilty about it. These feelings are clinical information and deserve supervision, not silence.

Decisions with no obvious answer

There are decisions on which the literature does not help much. One of them is what to do when the therapist suspects unidentified ASD. Naming it may bring meaning and access to support. Not naming it may respect the person's pace. The therapists in Brook's (2022) study made different decisions, and all described the importance of supervision.

Contextual constraints

A study in a university setting, with more than 40,000 clients, suggests that autistic clients benefit from therapy as much as others, but need about twice as many sessions (Anderberg et al., 2017). In a service with a session limit, the alliance may just be starting to form when therapy ends.

Spaces, in turn, have rarely been designed from a sensory point of view. The therapist may want to adapt the room and have no way to do so: fluorescent lighting, noise, shared rooms.

The privilege of not having to learn

Nick Walker (2021), cited in the IACP/AsIAm guidelines, describes neurotypical privilege as that of "not having to learn". Autistic people spend their lives learning to function in a world designed for others. The therapist is asked to travel the opposite path for one hour a week.

Perhaps this is the deepest challenge, because it is not technical. It is accepting that our way of relating is one among many, and not the measure of all the others.

Perspectives in tension

This field has no settled consensus. The positions below are held by serious clinicians and researchers, and they often complement rather than exclude one another.

Is a different relationship needed?	Specific adaptations are needed throughout the process: space, communication, structure (IACP/AsIAm, 2025)	The usual relational conditions are enough, expressed more explicitly and genuinely (Hume, 2022)
Where does the relational difficulty lie?	In the mismatch between neurotypes, shared by both (Milton, 2012)	Also in the client's differences in mentalisation, which can be worked on in therapy (Robinson, 2020)
Talking or solving?	Being heard without judgement is, for many people, the most helpful thing	Talking about what does not change is frustrating; people ask for an honest opinion and solutions (Pappagianopoulos et al., 2025)
Name an unidentified ASD?	It can bring meaning, self-compassion and access to support	Only if it is the person's concern; the label may overshadow the human experience (Brook, 2022)
Which model of understanding?	Clinical model: an ASD diagnosis provides a name, assessment, support and rights	Neurodiversity paradigm: difference, not deficit; disability also created by the environment
Who should provide support?	Autistic or specialised therapists offer an attunement that others can hardly achieve	Any well-trained and humble therapist can build a good alliance; specialisation does not replace the relationship

The tension between the clinical model and the neurodiversity paradigm runs through all the others. In practice, many clinicians use both: the diagnosis as a gateway, neurodiversity as a way of seeing the person.

Final thoughts

The therapeutic alliance with autistic adults is neither impossible nor different in nature. It is an encounter that requires more explicitness, more humility and more attention to what goes unsaid. What we know points in that direction. What we do not know is still a great deal.

For research

• Develop or adapt alliance measures with autistic people, not just for them

• Study adults, including those diagnosed late, women and non-binary people

• Study ruptures and repair: how they manifest and what resolves them

• Listen to therapists, including autistic therapists, about what they feel and what they do

• Carry out participatory research, with autistic people involved in defining the questions

• Study the Portuguese context, where there is practically no data

  
  

For practice

• Ask more and assume less: about language, previous experiences, sensory and communication preferences

• Ask for concrete and regular feedback, in writing when it helps

• Treat ruptures as opportunities and take ownership of our part

• Bring to supervision what we feel, and not only what the client does

• Seek training developed with and by autistic people

Autistic people spent a long time being studied as a relationship problem. Perhaps it is time for us to study the relationship that we, as therapists, are able to build with them.