The Explicitness Gap: When Social Inference Becomes a Processing Cost

Apparently I went down another rabbit hole instead of finishing my Relational Bandwidth series. 😂

I got stuck looping on my own struggles with the difference between implicit and explicit communication—and, more specifically, all the problems that difficulty has probably caused me over the course of my life.

Which, ironically, makes this rabbit hole relationship-series adjacent.

I’ve been looking backward at friendships, relationships, professional interactions, boundary mistakes, misunderstandings, and situations where I could tell that something had changed—but couldn’t necessarily figure out what the change meant, what rule I was apparently supposed to know, or what response the other person expected from me.

And I started wondering whether there might be a much bigger information-processing question underneath all of this:

What if one source of autistic processing cost is the amount of socially relevant information we’re expected to infer that was never explicitly communicated?

That question sent me into the research.

And now I think I may be constructing another offshoot of the Bandwidth Model.

I’m tentatively calling two pieces of it the Explicitness Gap and Inference Load.

These are not established constructs in autism research. They are terms I’m beginning to use to describe a possible relationship between existing research on autistic information processing and the processing cost of navigating implicit communication.

And I think there may be something here.

Autistic Thinking Is Not Simply “Concrete Thinking”

One place to start is research on detail-focused processing.

Historically, the weak central coherence account proposed that autism involved differences in integrating individual pieces of information into broader contextual meaning. Happé and Frith (2006) subsequently reframed the theory toward a detail-focused cognitive style, emphasizing enhanced or preferential local processing rather than simply assuming an inability to process information globally.

That distinction matters.

Autistic people can obviously think abstractly. We can theorize. We can use metaphor. We can recognize extraordinarily complex patterns, build systems, conceptualize relationships between ideas, and become deeply immersed in abstract subjects.

So describing autistic cognition simply as “concrete thinking” may obscure something more interesting.

Perhaps one of the relevant questions is not whether an autistic person can think abstractly.

Perhaps it is how information gets assembled into meaning—particularly when important pieces of that meaning have never actually been stated.

That becomes especially interesting when the information is social.

Knowing the Rulebook Is Not the Same as Intuitively Navigating It

Another relevant research area distinguishes explicit and implicit social cognition.

Keifer and colleagues (2020) examined both processes in autistic youth. Explicit social cognition involved relatively controlled, conscious social processing, while implicit social cognition involved faster and more spontaneous processing. Both contributed to social outcomes, but measures related to implicit social cognition were relatively more predictive of social behavior.

Gates and colleagues (2023) make another important distinction in their critical review of autism research: social knowledge and social performance are not synonymous. Knowing what is socially expected does not necessarily mean that knowledge will translate fluidly or consistently into successful behavior as an interaction unfolds.

That distinction stopped me in my tracks.

I’m 54 years old. I’ve worked in mental health for more than 25 years. I have decades of explicit training and experience in communication, boundaries, validation, active listening, relationships, human behavior, and social interaction.

I know an enormous amount about people.

But I’m increasingly realizing:

Knowing the rulebook isn’t necessarily the same thing as intuitively navigating the invisible rulebook.

I can recognize that something has changed socially without necessarily knowing what the change means.

Someone becomes quieter.

Communication frequency changes.

A relationship feels different.

Someone’s tone changes.

An expectation shifts.

Another person may integrate all of those contextual cues and arrive at something like:

“She needs some space.”

My brain may arrive at:

“Something changed. I now have seven pieces of information that don’t seem to agree with one another. What exactly am I supposed to do with these?” 😂

That is an important distinction.

The difficulty isn’t necessarily failure to notice social information.

Sometimes I notice it intensely.

The difficulty may be figuring out what the collection of information means, which information should carry the most weight, and what reciprocal response is expected.

What Happens When the Information Contradicts Itself?

This led me into predictive-processing research.

One influential theoretical account proposes that some features of autism may involve differences in the way prediction errors—the discrepancies between what the brain expects and what it actually encounters—are weighted.

Van de Cruys and colleagues (2014) proposed that unusually high or inflexible precision assigned to prediction errors could contribute to difficulty developing sufficiently flexible predictions in uncertain environments.

Predictive-processing accounts of autism remain theoretical. They should not be treated as a settled explanation of autistic cognition.

But they raise a fascinating question for relationships:

What happens when the uncertain environment is another human being?

Imagine that someone’s words communicate X.

Their behavior appears to communicate Y.

Your previous experiences suggest Z.

And nobody explicitly tells you which piece of information you are supposed to prioritize.

For me, unresolved discrepancies like that can become extraordinarily sticky.

It isn’t necessarily:

“I don’t understand relationships.”

It can be:

“I cannot make these pieces of information form one internally coherent model.”

And until the discrepancy is resolved, my brain may keep working on it.

When the Unsolved Problem Captures Attention

That brings me to another theoretical framework: monotropism.

Murray, Lesser, and Lawson (2005) proposed monotropism as an attentional account of autism, emphasizing differences in how attention is allocated and distributed. Rather than attention being distributed relatively broadly across many competing channels, monotropism proposes a more concentrated allocation of attentional resources.

Monotropism is also a theoretical framework, not a comprehensive explanation of autism.

But putting these literatures next to my own lived experience raises an interesting possibility.

What happens when socially important information is incomplete, contradictory, emotionally salient, and unresolved?

Maybe the unresolved question itself becomes an attentional object.

And perhaps that helps explain why some social ambiguity can become so difficult to disengage from.

Again, the existing research does not establish that pathway.

That is where my own hypothesis begins.

The Explicitness Gap

I’m tentatively calling the first construct the Explicitness Gap:

The Explicitness Gap is the distance between the information that is explicitly available and the information a person is expected to infer in order to respond successfully.

Consider a simple relational example.

One person says:

“Everything is fine.”

But their behavior changes. They initiate less. Their responses become shorter. They stop making plans.

Perhaps another person easily integrates those contextual signals and concludes that the relationship has changed.

But the explicit information still says:

Everything is fine.

The behavioral information appears to say:

Something is not fine.

Now there is a gap between what has actually been communicated explicitly and what the other person is apparently expected to understand.

That is what I mean by the Explicitness Gap.

And the size of that gap could presumably vary enormously.

Sometimes very little inference is required:

“I’m overwhelmed right now. I care about you, but I need a couple of weeks with less communication. I’ll reach out when I have more bandwidth.”

There is still emotional information to process, but relatively little relational detective work is required.

Other times almost everything important remains implicit.

The person is expected to notice the change, determine which cues matter, infer what they signify, decide whether the change is temporary or relational, determine what response is appropriate, and act on that inference without ever being told whether the interpretation is correct.

That is a lot of invisible cognitive work.

Inference Load

Which brings me to the second construct.

I’m tentatively calling the cognitive work required to bridge the Explicitness Gap Inference Load.

Inference Load is the cognitive work involved in deriving the information necessary to respond when that information has not been explicitly provided.

My working hypothesis is:

As the Explicitness Gap increases, the potential Inference Load increases.

That does not mean the relationship will always be linear, that every autistic person will experience it identically, or that explicit communication eliminates all processing demand.

It is a proposed relationship that would eventually need to be operationalized and tested.

Conceptually, however:

Greater explicitness → less information left to infer → potentially lower processing cost

while:

Greater implicitness, ambiguity, contextual dependence, or contradiction → more information to infer → potentially greater processing cost

And when the available information conflicts:

Words say X + behavior appears to suggest Y → unresolved discrepancy → additional inference demand

What happens after that could vary enormously.

One person might update their interpretation almost effortlessly.

Another might ask for clarification.

Another might withdraw.

Another might repeatedly analyze the interaction.

Another might become distressed.

Another might construct an explicit rule from the experience and carry that rule into the next relationship.

That last one has me looking backward at much of my own life very differently.

Maybe Some of Us Have Been Building the Rulebook Manually

I’m beginning to wonder how much of my social competence was constructed through a process that looked something like this:

Experience → observe outcome → infer rule → store rule → apply rule in future situation → encounter exception → revise rule

Do that for five decades and you can build an extraordinarily sophisticated social database.

But the finished database doesn’t necessarily tell us how the database was built.

Someone watching a socially competent autistic adult may see excellent communication, empathy, humor, reciprocity, insight, and relationship knowledge.

They don’t necessarily see the calculations underneath it.

That leads me to another distinction I think is important:

Social competence is not necessarily the same thing as social automaticity.

And this may be particularly relevant for autistic adults who are highly verbal, highly educated, professionally trained in human behavior, or who have spent decades consciously studying social interaction.

The better someone becomes at running the model, the less visible the model-building becomes.

Where This Reconnects With the Bandwidth Model

This is where I realized I hadn’t actually wandered as far away from Relational Bandwidth as I thought.

Because inference itself requires processing capacity.

The Bandwidth Model is concerned with variable functional capacity and the processing costs created by the demands a nervous system is being asked to carry.

So perhaps we shouldn’t only ask:

“How much demand is this nervous system carrying?”

Maybe we also need to ask:

“How much invisible inference are we requiring this nervous system to perform just to understand what the demand actually means?”

Because the capacity available for that work is unlikely to be constant.

Attention, executive functioning, sensory demands, fatigue, emotional salience, stress, trauma activation, previous experience, and burnout could potentially alter the processing resources available for bridging an Explicitness Gap.

The same ambiguous social situation might therefore be manageable under one set of conditions and extraordinarily costly under another.

This is also where the model could potentially become relevant beyond autism.

Different people may encounter different levels of Inference Load for different reasons.

The construct may eventually prove useful for thinking about ADHD, trauma, cognitive load, language differences, developmental differences, workplace communication, education, relationships, and other situations in which people are expected to act successfully on information that was never fully communicated.

But autism is where this particular rabbit hole started.

The Roadmap Problem

And suddenly my ridiculous roadmap analogy starts looking slightly less ridiculous. 😂

An admittedly oversimplified way I’ve been thinking about it:

ADHD: I have the roadmap, but sometimes I miss the turn. 😂

Autism: Wait. There was a roadmap? Why did nobody give me one? 😅😂

AuDHD: I finally reverse-engineered the roadmap, and then I missed the turn anyway. 🫣😅😂😂

Obviously, that is humor—not a scientific taxonomy of ADHD and autism.

But there is something underneath the joke that I think is worth exploring.

Sometimes the difficulty isn’t simply executing a known rule.

Sometimes the person first has to figure out what the rule is.

And with AuDHD, both problems can potentially exist at once.

This Is Not a Theory of “Impaired Implicit Learning”

There is an important distinction here.

I am not proposing that autistic people are generally incapable of implicit learning.

Foti and colleagues’ (2015) meta-analysis did not find evidence supporting a generalized implicit-learning impairment across the autism literature they examined.

My question is considerably narrower.

I am interested in the processing cost of deriving socially consequential meaning when relevant information remains implicit, ambiguous, context-dependent, rapidly changing, or contradictory.

Those are not the same claim.

Communication Is Also a Two-Way Process

There is another problem with framing all of this as something happening solely inside an autistic person.

The Double Empathy Problem challenges the assumption that autistic/non-autistic communication difficulties necessarily represent a one-directional autistic deficit.

Crompton and colleagues (2020), for example, found that neurotype matching influenced ratings of interpersonal rapport. Mixed-neurotype interactions were rated less favorably on several dimensions than matched-neurotype interactions, while simply being autistic did not predict poorer rapport.

That matters.

Because sometimes the problem may not simply be:

“The autistic person failed to understand the cue.”

Sometimes two people may be operating with different assumptions about what constitutes adequate communication in the first place.

One person believes:

“My behavior changed. I communicated that I needed space.”

The other believes:

“Your behavior changed, but you never told me what the change meant.”

Both may sincerely believe communication occurred—or didn’t.

That is not simply an individual deficit.

It is a communication mismatch.

What If Explicit Communication Is a Scaffold?

Which brings me to the part of this model I’m probably most interested in exploring.

Perhaps the intervention isn’t always:

Teach the autistic person to infer better.

Sometimes it might be:

Reduce the Explicitness Gap.

Say what you mean.

Make expectations visible.

Clarify changes.

Define the boundary.

Give the roadmap.

Because my emerging hypothesis is:

For some people, explicit communication may function as external cognitive scaffolding by reducing the amount of socially relevant information that must be inferred—and therefore reducing potential processing cost.

I want to be very precise about that statement.

The existing research does not establish this hypothesis.

The literature on detail-focused processing, explicit and implicit social cognition, predictive processing, monotropism, implicit learning, and double empathy provides pieces that make the question interesting and plausible enough to investigate.

The Explicitness Gap, Inference Load, and their proposed relationship to available Bandwidth are the model I’m beginning to construct from those pieces.

Eventually, the hypothesis could become testable.

For example: if the objective social situation remains essentially the same, does increasing the explicitness of communication reduce perceived uncertainty, cognitive effort, rumination, processing time, or recovery cost? Is that effect different across neurotypes? Does it become larger when someone’s available functional bandwidth is reduced?

Those are empirical questions.

I don’t have those answers yet.

But I think they’re worth asking.

Because maybe explicit communication isn’t merely a preference for clarity.

Maybe, under some circumstances, clarity changes the amount of cognitive work the nervous system has to perform.

And if that’s true, then communication itself can either consume bandwidth or scaffold it.

Apparently this is what happens when I’m supposed to be finishing the Relational Bandwidth series and my brain decides we need another theoretical framework instead. 😂

To be continued. Because I’m very clearly still building this one.

Darcy Stephens, LPCC

References

Crompton, C. J., Sharp, M., Axbey, H., Fletcher-Watson, S., Flynn, E. G., & Ropar, D. (2020). Neurotype-matching, but not being autistic, influences self and observer ratings of interpersonal rapport. Frontiers in Psychology, 11, 586171. https://doi.org/10.3389/fpsyg.2020.586171

Foti, F., De Crescenzo, F., Vivanti, G., Menghini, D., & Vicari, S. (2015). Implicit learning in individuals with autism spectrum disorders: A meta-analysis. Psychological Medicine, 45(5), 897–910. https://doi.org/10.1017/S0033291714001950

Gates, J. A., McNair, M. L., Richards, J. K., & Lerner, M. D. (2023). Social knowledge & performance in autism: A critical review & recommendations. Clinical Child and Family Psychology Review, 26(3), 665–689. https://doi.org/10.1007/s10567-023-00449-0

Happé, F., & Frith, U. (2006). The weak coherence account: Detail-focused cognitive style in autism spectrum disorders. Journal of Autism and Developmental Disorders, 36(1), 5–25. https://doi.org/10.1007/s10803-005-0039-0

Keifer, C. M., Mikami, A. Y., Morris, J. P., Libsack, E. J., & Lerner, M. D. (2020). Prediction of social behavior in autism spectrum disorders: Explicit versus implicit social cognition. Autism, 24(7), 1758–1772. https://doi.org/10.1177/1362361320922058

Murray, D., Lesser, M., & Lawson, W. (2005). Attention, monotropism and the diagnostic criteria for autism. Autism, 9(2), 139–156. https://doi.org/10.1177/1362361305051398

Van de Cruys, S., Evers, K., Van der Hallen, R., Van Eylen, L., Boets, B., de-Wit, L., & Wagemans, J. (2014). Precise minds in uncertain worlds: Predictive coding in autism. Psychological Review, 121(4), 649–675. https://doi.org/10.1037/a0037665

#BandwidthModel #ExplicitnessGap #InferenceLoad #Autism #AutisticProcessing #AuDHD #Neurodivergence #Neurodiversity #AutisticAdults #LateIdentifiedAutism #ImplicitCommunication #ExplicitCommunication #SocialCommunication #SocialCognition #Monotropism #PredictiveProcessing #DoubleEmpathy #ExecutiveFunction #ProcessingCost #CognitiveLoad #RelationalBandwidth #NeurodivergentAffirming #ActuallyAutistic #MentalHealth