Direct communication is a pared-down, efficient style of communicating, where the surface-level content is the entire content. There is no need to decode subtext, translate implied meaning or peel off layers of formalities to get to the message. The words mean what they mean, nothing more, nothing less.
For many autistic people, this is the default register. It comes from the same place as literal thinking: the tendency to process language at face value, where the explicit meaning is the first and strongest reading. It often gets misread as bluntness or aggression, but the directness is usually doing precision work.
Table of Contents[Hide][Show]
“Autistic bluntness”
Literal thinking describes how incoming language gets processed. Direct communication is the other half — what happens when an autistic person is the one talking.
Let’s look at an example. “Do you think we’ll leave in the next 60 minutes?”
That question has a job. It’s a binary question (yes or no) designed to return one piece of information: can I start a task that needs concentration, or will it be interrupted in ten minutes? The “60 minutes” boundary is there because a bounded question gets a usable answer.
The precision is doing the work. The specificity, the absence of hedging, and the tight framing are all engineering features of a question designed to get the information it needs and nothing else.
This precision is often what gets misread as aggression, confrontation or control, and this is the reason autistic direct speech is often called “autistic bluntness.”
The word blunt is contested within the autistic community. Some autistic people reclaim it proudly. Ginny Grant, reading her diagnostic report’s description of her communication as “blunt and frank,” responded: “I’m proud to be ‘blunt and frank’ – essentially straightforward, yes? – sounds like a great attribute to me.” 1
Others draw a firm line: direct means saying what you mean without relying on the listener to infer it; blunt carries an edge of disregard that most autistic direct communicators don’t feel. Both responses are real. The word appears in diagnostic reports as evidence of disorder and in community spaces as a badge of identity, sometimes in the same breath.
What It Feels Like From Inside
For many autistic people, direct speech doesn’t feel direct; it just feels like talking.
The clear, specific, no-frills register is the baseline, not a setting that gets switched on for effect. When an autistic person says, “I don’t want to go,” they are stating a preference. When they say “that’s wrong,” they are flagging a factual error. When they say “I already know that,” they mean you can skip ahead.
But that’s not how it lands. “I don’t want to go” gets heard as rejection. “That’s wrong” gets heard as combative. “I already know that” gets heard as dismissive. In each case, the listener adds something to the sentence that you didn’t put there: an emotional charge, an implied judgment, an attitude. Your words were neutral. The reception wasn’t.
The gap becomes visible when you see what the “expected” versions of these sentences look like:
- “I don’t want to go” → “I’m not sure I’m really feeling it, but you go ahead!”
- “That’s wrong” → “I wonder if maybe there’s another way to look at that?”
- “I already know that” → “Oh yes, I think I’ve come across that before!”
- “What time does this finish?” → “I’m having a great time, I just need to plan around my evening”
The versions on the right all add a performance layer: softening, hedging, managing the listener’s emotional response. For many non-autistic speakers, this layer is automatic and nearly invisible. If you’re autistic, it probably isn’t automatic at all, and producing it costs significant effort. You don’t arrive at the direct version by removing the softening. The direct version is where you start. The softened versions take conscious construction.
The mismatch starts when you notice the “vibe” change. Something shifts on the other person’s face (a flicker of surprise, a tightening, a pulled-back expression), they become silent, and now you know the sentence landed wrong. You just don’t know how. The words were accurate. The question was reasonable. Nothing in your intention matches the reaction you’re seeing. So you replay it. You run the sentence back through your head, testing it for edges, looking for the part that cut. And there isn’t one, because the problem was never in your words. The listener expected the softening layer, and when it wasn’t there, the bare sentence read as hostile.
This replay loop can run for hours. It runs after work conversations, after phone calls with family, after throwaway exchanges with strangers. Over time, it builds a specific kind of self-doubt: not “am I wrong?” but “is there something about the way I exist in conversation that I can’t perceive and can’t fix?”
Specific words start to follow you around. Intense. Intimidating. Harsh. Cold. Aggressive. They describe your character, not any particular thing you said. They don’t come with a correction you can act on. Nobody says “that question would have landed better if you’d softened it with ‘I was just wondering.'” They say “you’re a lot,” and leave you to work out what that means.
This is different from the harm described in the intolerance of uncertainty entry. Intolerance of uncertainty gets you labelled as someone difficult to accommodate: high maintenance, controlling, rigid. Direct communication gets you labelled as a difficult person. The first is about your needs, the second is about your character, so any rejection hurts even more.
Pattern recognition, close observation, and years of studying how language lands all give you a fine-grained understanding of what would hurt someone most in a given moment. You know which sentence would do the most damage, and you choose not to use it. When someone flinches at one of your neutral sentences, they are reacting as if it were designed to hurt. It wasn’t. You were just talking.
Time and patience can change things. People who know you stop flinching; they slowly learn your register, recalibrate, and hear your questions as questions. You haven’t changed how you speak. What changed is the relationship: the listener learned to hear direct communication as direct communication, not as hostility. That adjustment is the accommodation, and this is evidence that the friction was never in the words themselves.
Why Direct Communication Works This Way
Research on autistic language processing tells a consistent story: the difficulty isn’t with language itself. Autistic adults perform as well as or better than non-autistic adults on structured language tasks — vocabulary, word memorisation, context-independent language knowledge. 2 The gap shows up specifically when a task requires extending meaning beyond the literal, interpreting implied intent, or reading between the lines of what someone said to work out what they actually meant. 2
This is an important distinction. It means direct communication doesn’t come from a limited understanding of language. It comes from a processing style that treats the surface content of a sentence as the primary signal. When an autistic person hears “Can you shut the window?” they are more likely to process this as a question about ability than as a polite request — not because they don’t understand language, but because the literal reading is the one their processing system reaches first. 2
A predictive processing account offers one explanation for why. In social situations with high uncertainty (where tone, context, facial expression and implied meaning are all competing for attention) the literal interpretation is the safest reading. It’s the one that doesn’t require guessing at someone else’s hidden intent. When the social environment is unpredictable, defaulting to what was actually said is a reasonable strategy. 3
The same thing happens in the other direction. Autistic people tend to build sentences that carry their full meaning on the surface, because that is how they process incoming language too. The communication ethic is symmetrical: say what you mean, and assume the other person is doing the same. In autistic social spaces, this works. Among autistic people, asking a clarifying question — “What do you mean by that?” or “Can you say that more specifically?” — is considered polite. It signals engagement, attentiveness, and a desire to understand accurately. 2
Non-autistic people frequently misread these same clarifying questions as rude, challenging, or an attempt to dominate the conversation. 2 This is the double empathy problem operating at the sentence level: two communication systems with different built-in assumptions about what counts as polite, each interpreting the other’s norms as a violation.
The mismatch also generates a specific kind of invisible labour. Many autistic adults develop extensive libraries of idioms, metaphors, and social scripts through deliberate study and memorisation. As one autistic adult described it: “I wish people were more mindful of things that are ambiguous in communication. I’m fortunate to have memorized so many idioms and metaphors that I can instantly translate them in my head.” 4 This translation work happens in real time, every conversation, and it runs in one direction. Autistic people learn to decode non-autistic indirectness. The reverse accommodation — non-autistic people learning to hear direct communication at face value — rarely happens without prompting.
Researchers studying nonverbal communication in autistic adults describe a similar asymmetry. The cognitive load involved in processing and producing the expected social signals — facial expressions, prosody, backchannelling (the nods, “mm-hms” and “yeahs” that signal you’re still listening) — is not a sign that autistic people can’t do it. As one research participant put it: “it’s not so much a case of ‘can/cannot’ read body language, so much as a different way of doing it which has a much higher cognitive load, so is much more tiring.” 5 The effort is hidden precisely because autistic people get good at it. But every unit of energy spent on decoding tone, maintaining eye contact, and manufacturing the expected softening is a unit not available for the actual content of the conversation.
The Double Empathy Problem at a Sentence Level
The double empathy problem, first described by Damian Milton, reframes autistic social difficulties as a two-way mismatch rather than a one-way deficit. 6 Social friction between autistic and non-autistic people comes from two different communication systems (each internally coherent) operating with different assumptions about how conversation is supposed to work.
Studies on autistic social interaction support this. When autistic people interact with each other, the conversation goes just as well as when non-autistic people interact with each other. 7 The signals look different (autistic pairs use less mutual gaze and less backchannelling, but rate their interactions just as positively), and the difficulty shows up specifically in mixed pairs, where each person is reading the other through the wrong framework. 8
In autistic social spaces, a clear question gets a clear answer. A correction is received as helpful information. A “no” is a complete sentence. Nobody is scanning for a softening layer that isn’t there, because nobody expects one. The communication system works because both speakers share the same assumptions about what words are doing.
The friction starts when those assumptions don’t match. A non-autistic listener hears “that’s wrong” and processes it through a system where bare corrections signal hostility. An autistic speaker says “that’s wrong” inside a system where bare corrections signal helpfulness. Both people are using language according to their own rules, and each one experiences the other as rude.
If you are autistic and communicating directly in non-autistic spaces, the words people use to describe you tend to follow a pattern. Rude. Blunt. Harsh. Cold. Abrasive. Aggressive. Sometimes the feedback is subtler: “You don’t have to be like that,” “There’s no need for that tone,” “I just think you could be a bit more…” The sentence trails off because the speaker can’t name what’s missing. What’s missing is the performance layer they expected and didn’t get.
Speaking as a woman…
For autistic women, the penalty has extra layers. The expectations go beyond politeness: autistic women are required to fit societal interpretations of femininity by performing warm, soft-spoken, emotionally attuned, bubbly personalities. An autistic man who skips the softening might get called blunt or awkward. An autistic woman who skips it gets called cold, unfeminine, or aggressive. Research confirms the double penalty: autistic women receive less favourable first impressions than non-autistic women, and autistic girls perceived as less feminine receive worse social ratings. 9 The communication penalty and the gender penalty multiply each other.
And the trap closes from both sides. The cushioning strategies autistic women learn to use in response to the directness penalty produce a different kind of dismissal when performed by a female-presenting speaker. We’ll come back to this in the next section. For now, what matters is that the direct version and the cushioned version are both punished — just in different ways.
The word blunt has accumulated all of this history. It appears in diagnostic reports as clinical evidence. 1 It appears in workplace feedback, school reports and relationship arguments. It also appears in autistic community spaces as something people have reclaimed: a word that once described what was wrong with them, now worn as a description of what they refuse to hide. Some autistic people hear blunt and feel pride. Others hear it and feel the sting of every conversation where their natural register was treated as a character flaw. The word is doing double duty, and the context determines which version you hear.
The Cushioning Labour
Most autistic adults learn, eventually, that direct communication has a social cost. So you start adding things. A “sorry to bother you,” a “just wondering,” an “if you have a sec,” a rising tone at the end of a statement to make it sound like a question. You learn to manufacture the softening layer in real time, and you learn to do it well enough that most people don’t notice it’s a performance.
Cushioning as prevention
But cushioning doesn’t have only one origin. For some autistic people, the softening is something they consciously added in adulthood after years of being told their natural register was rude. For others, the cushioning was built much earlier, in environments where directness was dangerous, where any boundary or clear request could trigger retaliation from a parent, sibling, teacher, or partner. In those environments, the speech patterns a child develops are not really about communication style — they are about survival.
These two pathways into cushioning look the same from the outside. The sentences sound similar. The same softeners get used. But they function very differently from inside.
The clearest way to tell them apart is to ask what the cushioning is for. In one version, you have something to say, and the cushioning is a filter you apply to make the saying go more smoothly — the message comes first, and the cushioning is a delivery mechanism. In the other version, the cushioning comes first, designed to keep the listener calm and you safe, and the message arrives later if it arrives at all. Sometimes the cushioning expands until there is no time left to reach the point. Sometimes you stop yourself before you get there because you sense you are approaching a boundary the listener will not tolerate. The shape of the sentence is similar. The function is completely different.
How cushioning can weaken your position
When the message is still intact, the cushioning mechanism does two things at once: one intended, and one unintended.
The intended thing is what it’s supposed to do: it works. The sentence sounds gentler, less confrontational, more polite. The listener doesn’t flinch, and the conversation continues without friction. From the outside, the accommodation looks successful. Hooray!
The cushioning doesn’t simply soften the sentence, unfortunately, because the softening can also discredit the authority of what you’re saying. All the fillers that take away the “edge” of what you are saying can also take away the confidence you would say them with. They make you more palatable, but also make you sound unsure. They make you seem non-threatening, but also make you seem hesitant.
So now there is a gap between what you actually think and how you sound when you say it. You know the answer. You have the knowledge. But the packaging you have been forced to add tells the listener you might not. The softening layer implies uncertainty, and that uncertainty attaches to you — to your confidence, your knowledge, your competence — even though you only added it to manage the listener’s reaction.
This has consequences. Instead of a clear exchange, you get reassurance. You get explanations of things you already understand. You get your own idea handed back to you slowly, with encouraging nods, as though you needed help arriving at it. For many autistic people, this can add salt into old wounds: sometimes a PDA response, sometimes the long history of being mansplained to, sometimes being patronised or talked down to as though you couldn’t possibly understand. Sometimes all of these at once. The reassurance is a response to a low confidence level you never actually had. You are now dealing with the consequences of an uncertainty you manufactured for someone else’s benefit.
For autistic women, this dismissal has a sharper edge. The cushioning that female-presenting speakers are taught to use — “maybe,” “possibly,” “if it’s not too much trouble,” “I might be wrong, but” — doesn’t get read as accessible communication. It gets read as low confidence, and low confidence in a female-presenting speaker gets dismissed quickly. The strategic stakes are professional and social: not being invited back to the meeting, not being asked again for input, learning over time that your ideas don’t get traction even when you know they’re the right answer. This is the second half of the trap previewed in the last section. Direct speech gets autistic women called cold and aggressive. Cushioned speech gets them dismissed as uncertain. There is no register that is both heard and accepted, and the labour of finding the impossible middle ground is its own exhausting project.
How cushioning dilutes the message
Cushioning also causes precision loss. A direct question is often engineered to return a specific piece of information. “Do you think we’ll leave in the next 60 minutes?” is a binary question with a job: you want to know whether you can start a task that needs concentration, or whether you’ll be interrupted in ten minutes. The boundary of the question is the whole point.
When you cushion it, the boundary disappears. “Do you think we’ll leave in the next 60 minutes?” becomes “What’s your idea of when we’ll leave?” And the answer you get is no longer a number. It’s a monologue about everything still on the to-do list before leaving. Information you don’t need, can’t use, and now have to filter through to get to the thing you originally asked for. The cushioned question got an open-ended answer, because that is what open-ended questions get.
You are now further from your goal than before you asked. You still don’t know if you can start the task. You have spent processing energy decoding irrelevant information. The cushioning cost effort to produce, and it also broke the tool.
This reveals what the directness was actually doing. The direct version wasn’t blunt; it was precise. The precision was the point. Stripping it out changed the tone and disabled the function the question was built for. What gets called “bluntness” is often the precision itself, doing exactly the job it was designed to do.
The mental load of softening direct speech
Producing the cushioning is also expensive. Voice modulation, facial expression, and tonal softening all take energy, and that energy comes from the same budget as everything else you are doing. When you are thinking through a problem, formulating a precise question, or processing information, the budget for the performance layer often runs out. Energy is limited, and because producing the content of the communication uses up most of it, not much remains to allocate for the filter that would make it more consumable. (This connects to flat affect, which describes the same resource allocation showing up in face and voice rather than in word choice.)
This means a direct sentence can misfire for two different reasons. One is the neurotype mismatch: the listener is used to language that includes a softening layer, and when it doesn’t, the absence reads as hostility rather than as a stylistic difference. The other is the flat delivery itself, which can read as cold or hostile regardless of what the words actually say. Both kinds of misreading can happen at once: flat affect plus no cushioning. Sometimes the energy budget has run out. Sometimes there is no particular reason — direct communication is simply how you talk, and the filter was never going to be there in the first place.
When energy runs out, direct speech floods
For people whose cushioning comes from the prevention pathway, there is a pattern that shows up over time. When you spend a long time cushioning, by the time the directness emerges, the weight of everything you held back also comes out with it. You are usually already in distress: tired, dysregulated, overwhelmed, frustrated by being misheard or dismissed. The direct speech that comes out in those moments comes out with an emotional charge. With volume. With tears. With the things you have been swallowing for hours or days or years.
For some autistic people, this creates a feedback loop. If directness keeps emerging in moments of overwhelm, the brain can start to associate the two — until directness itself starts to feel like a warning sign, evidence that you’ve already passed into collapse, even when you haven’t. The directness gets contaminated by the distress that often arrives alongside it. The result is that the very capacity that should be your default communication style starts to feel dangerous to use, and the cushioning gets reinforced as the only safe option.
For female-presenting speakers, this overshoot pattern has an extra penalty. Distressed direct speech from male-presenting speakers can be read as forceful, passionate, even leaderly in some contexts. From female-presenting speakers, it is read as hysterical, unstable, further proof that the speaker should not have been at the table in the first place. So the same overshoot pattern that costs everyone their composure costs female-presenting speakers their credibility on top.
How you can use cushioning to your advantage
All of this might sound like the conclusion is that cushioning is a trap with no escape. It isn’t, but the way out is more interesting than just refusing to cushion.
Cushioning that comes from agency is a different thing from cushioning that comes from fear. When the message is first, when you are the one deciding how much sugar to add to the spoonful of medicine, when the cushioning is a tool you control and deploy on purpose, it stops being a betrayal of your authentic voice. It becomes an accessibility service you offer to a listener whose processing system needs the medicine delivered a particular way. “I am putting this in terms you can hear” is not the same as “I am hiding what I mean because I am afraid of what will happen if I say it clearly.” The first is communication. The second is survival.
The diagnostic test is whether you still have the message. If you can name what you wanted to say, and the cushioning is a filter you applied to that message, you are doing the first thing. If the cushioning has expanded to the point where the message is buried, lost, or never spoken, you are doing the second thing. Both are valid responses to your circumstances. But only the first one is a tool. The second one is a cage that has learned to look like a tool.
The reclaiming doesn’t mean the cushioning becomes free. It still costs energy. It can still break the tool of a precise question if applied carelessly. It can still get female-presenting speakers dismissed. None of that goes away. What changes is the relationship between you and the cushioning. The question stops being “should I cushion or not?” and becomes “what is my cushioning doing right now, and who is it serving?”
What Helps
Direct communication is not the problem. The problem is misunderstanding, and misunderstanding happens between two people. So the things that help are not really about how to make direct communication go away. They are about how to prevent the misreading, or repair it when it happens.
Here’s what you can do if you are autistic (you are probably already doing this)
Most autistic people are already doing a significant amount of this work. The cushioning, the tone-policing of their own sentences, and the upfront explanations are all accommodations the autistic speaker is making for the listener, often without the listener realising it is happening. Anything that helps needs to start from this fact: the labour is already unevenly distributed, and the question is how the listener can meet the speaker partway.
For autistic people, knowing you’re autistic helps. Understanding that your neutral sentences are being processed differently than you intended — and that the difference comes from how two different communication systems meet, not from anything wrong with you — changes the experience of being misread. The hours of replaying conversations afterwards, looking for the part that went wrong, get shorter. The accumulated self-doubt starts to loosen.
For autistic people who got their diagnosis later in life, this can also rewrite years of accumulated evidence. The character labels — intense, intimidating, harsh, cold — start to look less like accurate descriptions of who you are and more like artefacts of a calibration mismatch you had no language for at the time. The people who used those words about you weren’t always wrong about what they were experiencing, but they were wrong about what was causing it.
A short explanation can help in moments of misfire. “Sorry if that sounded blunt, I didn’t mean it like that. Sometimes when I’m focused on something, I don’t have any energy left for voice modulation.” This locates the cause in energy allocation rather than in feeling, and it does it without apologising for the directness itself.
Choosing your channels also matters. Many autistic adults find written communication easier (text, email, messaging) because the asynchronous format removes prosody and facial expression from the equation. Research supports this preference: autistic adults often choose written communication precisely because it gives them more control and reduces real-time social pressure. 10 If you have the choice between a phone call and an email, the email may not be avoidance. It may be the channel that lets you communicate most accurately.
Here’s what you can do if you are not autistic
For non-autistic people in autistic people’s lives (partners, family members, colleagues, clinicians), the most useful thing is also the simplest. When a direct sentence lands in a way that feels off, ask. A genuine check works: “Sorry, I know you probably didn’t mean it like that, but are you mad? Did I do something wrong?” This kind of question does several things at once. It opens the door for the autistic speaker to clarify their intention. It signals that the listener is willing to consider that their first reading might be wrong. And it interrupts the loop where the listener decides the autistic person was being hostile and acts on that assumption without ever checking it.
This is the accommodation that costs the least and helps the most. It doesn’t require the listener to like direct communication, learn a new vocabulary, or change how they themselves speak. It just requires them to treat their first interpretation of a direct sentence as a hypothesis instead of a verdict.
There is relief that comes from being in conversation with people who ask instead of assuming: where direct sentences get received as direct sentences, where a clarifying question gets clarification, where nobody is building a story about your character based on a tone they may have misread. Most autistic people who experience it for the first time describe it as something close to coming up for air.
This is what being understood actually means for many autistic people. Heard accurately, in a language we already speak.
