Communication and relationships

Damian Milton, an autistic sociologist, set out the double empathy problem in a 2012 paper.^1,c1 He defined it as "a disjuncture in reciprocity between two differently disposed social actors", and wrote that both people experience it, so it is not located in either one.¹ In a 2020 study in which 72 adults passed a story along chains of eight, autistic chains passed on as much of the story as non-autistic chains, while mixed chains passed on less.² A preregistered replication with 311 adults found no difference in how much information passed along single- and mixed-neurotype chains.³

A 2026 meta-analysis of 2,622 studies found a substantial average difference in social functioning between autistic and non-autistic people.²⁶ That compares group averages with non-autistic controls. The double empathy problem is about what happens between two differently disposed partners, and it predicts a difference between mixed and same-neurotype pairs, so a group difference cannot decide between the two views.^26,27 A 2026 systematic review of language and communication research found that about half the studies compared autistic people with non-autistic groups, a share that rose to about three-quarters once intervention studies and studies of people with an intellectual disability or few or no words were excluded; of those that stated a prediction, 60% expected autistic people to do worse.²⁷ Catherine Crompton, who led the two diffusion-chain studies, argues in a 2026 letter that many social difficulties are "a mismatch" between autistic and non-autistic ways of communicating rather than an individual deficit, and that in mixed interactions both parties experience similar difficulty.²⁸

A meta-analysis of 41 studies found that, on average, autistic groups score moderately lower than non-autistic groups on tests of figurative language.⁴ Metaphors were relatively harder for autistic participants than irony and sarcasm.⁴ When the groups were matched on language ability, the difference was small and not significant.⁴ An analysis of metaphor studies matched on age and IQ still found a gap, but the better participants' verbal skills, the smaller it was.⁵

Alexithymia is difficulty identifying and describing one's own feelings.⁶ In a meta-analysis of studies using one self-report questionnaire, about half of autistic participants scored in its alexithymic range, against about 5% of non-autistic participants.⁷ Researchers separate cognitive empathy, understanding what someone else feels, from affective empathy, being moved by it.^8,9 A 2025 meta-analysis of 226 studies found autistic groups lower on cognitive empathy by a large margin on average, while the difference in affective empathy was small and was no longer significant when only high-quality studies were counted.⁸

Megan Griffith describes meltdowns as "intense and uncontrollable experiences that may affect emotions, cognition, and/or sensory processing".^c2 Devon Price calls a shutdown "the quieter, more interior counterpart to an Autistic meltdown".^c3 When 32 autistic adults were asked what a meltdown is like, most described being overwhelmed by information, by the senses, and by social and emotional stress.¹⁰ Researchers rarely study meltdowns by that name, and in the studies the full entry reviews, no validated questionnaire measures meltdowns or shutdowns as autistic people define them.

Autistic people report more loneliness than non-autistic people on average, and a meta-analysis pooling 23 studies of children and adults found a large difference.¹¹ Surveys by four independent groups agree that autistic adults report being bullied, exploited and assaulted more often than non-autistic adults.^12,13,14,15 In one UK online survey, 70% of autistic adults, against 31% of comparison adults, had been bullied by someone they considered a friend.¹² Studies that account for other conditions give a less clear picture of how much of this belongs to autism itself.

A 2011 inquiry into disability-related harassment in Great Britain credited the Association for Real Change, a UK charity, with coining the term mate crime, for "the exploitation, abuse or theft from people with a learning disability, by those they consider as their friends".²³ In South Carolina, children with autism and no intellectual disability had 1.86 times the odds of population controls of being reported for maltreatment, after adjustment.²⁴ A Dutch randomized trial of a sex-education program for 189 autistic adolescents improved knowledge and understanding of boundaries compared with a waiting list; it did not measure victimization.²⁵

A 2021 systematic review starts from the observation that autistic people report as much interest in romantic relationships as non-autistic people, but more difficulty starting and keeping them.¹⁶ In a German sample of 229 autistic adults recruited through a clinic and online, 73% had had a romantic relationship and 7% did not want one.¹⁷ How many autistic adults have a partner depends on who is counted. In a Dutch volunteer register of 675 autistic people over 15, half were in a relationship, against about 70% of their general-population peers.¹⁸