TL;DR: The double empathy problem (Damian Milton, 2012) reframes social difficulties in neurodivergence as mutual misunderstanding between neurotypes, not as a deficit in the neurodivergent person. When neurodivergent people are forced to mask to maintain relationships, the resulting connection feels fraudulent to both parties, creating an intimacy gap that neither can name.
Mark Said It Without Flinching
“When I’m being true to myself, people ask ‘what’s wrong with you now?’”
He was describing something more precise than social rejection, though rejection was part of it. What he was describing was the double bind that structures the social life of a neurodivergent adult who has spent years learning to perform neurotypicality: the authentic self produces distance, and the performed self produces connection that feels like fraud. There is no third option in the binary he had constructed, which is why the insight carried the weight it did when he said the next part.
“It makes me feel like none of my relationships are real.”
The room was quiet after that. Not the therapeutic quiet of a group waiting for the facilitator to interpret, but the quiet of recognition. Several people in the group had, in their own language and from their own histories, arrived at the same conclusion: that the relationships built on their masked selves were extensive, functional, and hollow in a way they could identify but not resolve.
The Theory That Explains the Feeling
Damian Milton, an autistic sociologist, proposed the double empathy problem in 2012 as a challenge to the prevailing clinical model of autism, which located social difficulty entirely within the autistic person. The standard model, still reflected in DSM-5 diagnostic criteria, describes “deficits in social-emotional reciprocity” and “deficits in developing, maintaining, and understanding relationships.” The word deficit appears as though the problem has a single address.
Milton’s argument was structural rather than individual. When a neurotypical person and a neurodivergent person fail to connect, the failure is bilateral. The neurotypical person cannot accurately predict, interpret, or respond to the neurodivergent person’s behavior any more than the reverse. Both parties experience the other as confusing, unpredictable, socially opaque. The difference is that only one party is diagnosed for the confusion.
The research that followed Milton’s theoretical work confirmed what the theory predicted. Autistic people communicating with other autistic people show no deficit in social reciprocity. The information transfer between autistic interlocutors is as efficient as between neurotypical pairs. The breakdown occurs specifically at the cross-neurotype boundary, which means the breakdown is a property of the interaction, not a property of either individual.
This distinction sounds academic until you sit in a room with someone like Mark, for whom the academic distinction is the difference between “I am broken” and “I am operating in a system that was not designed for the way I process.”
Where the Intimacy Gap Opens
The clinical implications of the double empathy problem extend well beyond social skills training, which is where most intervention models stop. If the communication difficulty is mutual, then teaching the neurodivergent person to perform neurotypical social cues does not resolve the mismatch. It conceals it. The mask produces the appearance of connection while the underlying processing difference remains unaddressed, and the gap between appearance and reality is where the intimacy problem lives.
Consider the structure of a masked relationship. The neurodivergent partner learns to make eye contact at the culturally expected frequency, to modulate vocal tone within the neurotypical range, to respond to bids for emotional connection with the timing and content that the neurotypical partner expects. The performance is convincing. The neurotypical partner feels connected. The neurodivergent partner feels exhausted, because every element of the interaction required explicit computation that the neurotypical partner processed automatically, and because the version of themselves that produced the connection is not the version they recognize as themselves.
The result is an intimacy gap that neither partner can name with precision. The neurotypical partner senses something is off but cannot identify what, because all the surface-level social signals are correct. The neurodivergent partner knows exactly what is off but cannot articulate it without exposing the mask, which risks the very connection the mask was designed to protect. Both partners feel alone in the relationship, and neither has a framework for understanding that the aloneness is structural rather than personal.
The Mask as Relational Corrosive
Masking does not produce bad relationships. It produces relationships built on incomplete information, which is a different and in some ways more painful problem. The neurotypical partner fell in love with a performance. The neurodivergent partner knows this, carries the knowledge silently, and faces a choice that has no clean resolution: maintain the performance and preserve a connection that feels fraudulent, or drop the performance and risk discovering whether the partner can love the unedited version.
Mark’s statement, “it makes me feel like none of my relationships are real,” is the logical conclusion of this bind. If the person your partner connects with is a character you play, then the connection is real for them and fictional for you. You are present in your own relationships as an actor rather than a participant, which means that the more successful the masking, the more complete the alienation.
The clinical mistake is to treat this alienation as a symptom of depression or attachment insecurity, which it resembles on every screening instrument. The interventions for depression and attachment insecurity (increasing social engagement, challenging negative cognitions about relationships, practicing vulnerability) all assume that the self being engaged, challenged, and made vulnerable is a stable entity. For the neurodivergent person whose relational self has been a performance for years or decades, there may not be a stable relational self to make vulnerable. There is a mask and there is the person behind it, and the person behind it has had very little practice existing in the presence of another human being.
Connection as Shared Problem
The reframe that the double empathy problem makes possible is this: connection is a shared problem, not a personal failure. The neurodivergent person who struggles to maintain closeness is not deficient in the capacity for connection. They are attempting connection across a processing gap that neither party created and neither party can bridge alone.
This reframe does not solve the problem, but it changes the location of the work. Instead of social skills training that teaches the neurodivergent person to perform more convincingly, the work becomes mutual: both partners learning how the other processes social information, both partners adjusting expectations, both partners taking responsibility for the gap rather than assigning it as the neurodivergent person’s deficit to remediate.
In practice, this means the neurotypical partner learns that their neurodivergent partner’s lack of eye contact is not disinterest. It means the neurodivergent partner learns that their neurotypical partner’s need for verbal affirmation is not neediness but a processing requirement as real as their own need for reduced sensory input. It means both people stop asking “why can’t you just” and start asking “how does this work for you.”
What the Mask Costs
The double empathy problem clarifies something that decades of social skills research obscured: the neurodivergent person’s social difficulty is not a skills deficit. It is a translation problem, compounded by the fact that only one party is expected to translate. The neurotypical world does not learn neurodivergent communication norms. It does not adjust its pace, its volume, its expectations around eye contact and small talk and the performance of emotional states. The entire burden of translation falls on the neurodivergent person, and the cost of that translation is the slow erosion of the translator’s sense of self.
Mark sat in a group of people who understood this cost without needing it explained, because they had all paid it. The relationships in the room, built between people who processed the world similarly, had a quality that their outside relationships lacked: the quality of being known without performing. For some of them, that group was the only context in which they experienced connection without the tax of translation.
That fact, the fact that connection without masking was available but only in a specific, time-limited, clinical context, was not a resolution. It was the residue of a problem that no individual therapy outcome can fully address, because the problem is not individual. It is the architecture of a social world built for one neurotype and billing the other for admission.
Frequently Asked Questions
What is the double empathy problem?
The double empathy problem, proposed by Damian Milton in 2012, argues that communication breakdowns between neurodivergent and neurotypical people are not caused by a deficit in the neurodivergent person but by a mutual difficulty in understanding each other’s perspectives. Both parties struggle to predict and interpret the other’s behavior, but only the neurodivergent person is typically told they have a social skills deficit.
Do neurodivergent people lack empathy?
No. Research on the double empathy problem shows that neurodivergent people communicate effectively with each other, suggesting the issue is cross-neurotype communication, not a deficit in empathy or social understanding. The perception of empathy deficits reflects a measurement bias where neurotypical social norms are treated as the standard.