I had already looked at affective touch as a possible tactile response for in-the-moment regulation, but this paper is important because it asks a more specific question: how do autistic adults actually experience affective touch, both subjectively and physiologically?

That makes it a useful counterpoint to the earlier wearable affective-touch paper. It stops me from assuming that because slow stroking can feel calming in one population, the same mechanism can simply be transferred into an autism-focused wearable.

This paper is not testing a wearable intervention. It is helping me understand the sensory mechanism and the variability around it before I decide whether affective touch is an appropriate design direction.

What affective touch means here

The paper focuses on a slow, caress-like form of touch associated with the C-tactile (CT) afferent system.

These nerve fibres are particularly responsive to gentle stroking at slow velocities. In this study, the affective-touch condition used stroking at approximately 3 cm/s, which sits within the range usually associated with CT-optimal touch.

The researchers compared this with a control condition involving tapping.

What matters for my project is that affective touch is not simply “touch that is meant to be calming”. It is a specific type of tactile stimulation with a particular speed and social meaning.

What the researchers tested

The study involved 48 young adults.

Twenty-four participants were autistic adults recruited through an adult autism centre. All met criteria for level one support needs and did not have cognitive impairment. A further 24 participants formed the non-autistic control group.

Participants sat with their left arm hidden from view. Across a series of trials, an experimenter either slowly stroked the back of the participant’s forearm or delivered the tapping control touch.

Each touch lasted around 14 seconds.

Throughout the task, the researchers recorded skin conductance as a measure of autonomic activity. After each trial, participants also rated how affective and how unpleasant the touch felt.

This combination is what makes the paper particularly useful to me: it compares what the body signal appears to show with what the person reports experiencing.

What the physiological data showed

The autistic group had lower overall skin-conductance activity than the control group.

The control group also showed a clearer autonomic distinction between the two types of touch: affective stroking produced greater skin-conductance responses than tapping.

In the autistic group, that distinction was not statistically significant. Their autonomic response was broadly similar across affective stroking and tapping.

This does not mean autistic participants “did not feel” the touch. It means that the specific skin-conductance measures used in this study did not differentiate the two touch conditions in the same way as they did in the control group.

That distinction is very important for my project.

What the participants themselves reported

The subjective results were quite different from the physiological pattern.

Across participants, affective stroking was rated as more affective than tapping.

However, the autistic participants gave higher affectiveness ratings overall than the control group. They also gave higher unpleasantness ratings overall.

So the same group reported touch as both more affective and more unpleasant.

I find this particularly important because it shows that tactile experience is not necessarily a simple positive-versus-negative scale. A sensation can carry emotional or comforting qualities while also being intense, uncomfortable or overstimulating.

For an autism-focused design, that feels much more realistic than assuming that a touch is either “calming” or “not calming”.

The body signal and the subjective experience did not match neatly

The paper describes a discrepancy between the autistic participants’ explicit ratings and their autonomic response.

At the subjective level, participants could distinguish affective touch from tapping and gave strong ratings to the tactile experience.

At the physiological level, skin conductance did not show the same condition-specific differentiation.

For me, this is one of the strongest pieces of evidence against treating biodata as a direct read-out of sensory experience.

A lower or flatter autonomic response does not automatically mean that the person is less affected by the stimulus.

In this study, the autistic participants could report a strong affective or unpleasant experience even when the skin-conductance pattern did not reflect that difference in the same way.

What this means for biosensing

This paper does not use biosensing to control an intervention. Skin conductance is measured as an outcome.

But its relevance to my project is probably greater than that of some explicitly bioresponsive systems because it shows the limitation of the sensing layer itself.

If I used electrodermal activity within Grounded, I could not reasonably interpret a particular skin-conductance pattern as:

“this person likes this touch”

or

“this intervention is calming them.”

The physiological signal may add useful information, but the user’s own perception still has to remain central.

This strengthens my current position that biosensing should contribute to a decision rather than replace the person’s interpretation of their own sensory experience.

It complicates my earlier affective-touch research

The earlier wearable affective-touch study suggested that slow stroking could reduce reported state anxiety during some stressful tasks.

This paper adds an important autism-specific complication.

Here, autistic adults experienced affective touch differently from the non-autistic comparison group, and their subjective experience was not mirrored straightforwardly by the autonomic data.

That means I cannot take evidence for affective touch in a general adult sample and assume that a wearable stroking mechanism will transfer directly into autistic airport accessibility.

If I continue exploring it, the correct position is:

affective touch is a plausible tactile mechanism worth testing, not a pre-validated calming response for autistic travellers.

Human touch is not the same as wearable touch

There is another transfer problem.

The tactile stimulation in this study was delivered by a human experimenter, not by a motorised wearable.

The paper itself discusses literature suggesting that human touch and artificial touch can be experienced differently.

So even though this study helps me understand how autistic adults respond to affective stroking, it does not tell me that a synthetic fur mechanism, textile actuator or robotic sleeve would create the same experience.

For my project, translating this into a wearable would therefore involve two steps of inference:

first, from laboratory social touch to real-world sensory support;

and second, from human touch to mechanically generated touch.

Both would need testing.

Predictability and control become even more important

The finding that autistic participants rated the touch as more unpleasant overall makes predictability and consent especially important.

A tactile response that begins unexpectedly could be intrusive even if its speed or pressure has been designed around an “affective” mechanism.

If I explored slow stroking or moving pressure within a wearable, I would want the traveller to be able to decide beforehand whether that modality is acceptable at all.

They would also need control over body location, texture, pressure, speed, timing, duration and immediate stopping.

This paper strengthens the argument that tactile personalisation cannot be added at the end. It is part of whether the intervention is accessible in the first place.

A caution around the authors’ interpretation

The authors discuss several possible explanations for the mismatch between subjective ratings and autonomic response, including social learning, camouflaging and differences in the motivational or rewarding aspects of touch.

I need to treat these as interpretations rather than findings directly demonstrated by the experiment.

The study shows a measurable difference between groups and a mismatch between some physiological and self-reported measures.

It does not establish why that mismatch exists.

Alternative explanations could include individual sensory sensitivity, differences in baseline autonomic activity, measurement limitations, context, familiarity with touch or other participant characteristics.

For my own work, the safest and most useful conclusion is therefore about variability and measurement, rather than trying to infer an internal psychological mechanism from the physiological result.

Limitations

The sample was relatively small and focused on young autistic adults with level one support needs and no cognitive impairment.

The sex distribution was also uneven, particularly within the autistic group.

The experiment took place in a controlled seated setting and involved repeated touch from an experimenter. That is very different from moving through an airport while processing sound, information, crowds, uncertainty and physical movement.

The study also measures only one aspect of autonomic activity through skin conductance. It does not establish that affective touch would reduce anxiety, sensory overload or distress in a real-world travel context.

Finally, several participants did not complete every trial, and some non-responsive skin-conductance data were excluded from the physiological analyses.

What this means for my project

This paper does not make me abandon affective touch, but it changes the way I would design around it.

The strongest design lesson is that a bodily response and a person’s subjective experience can diverge.

That matters far beyond touch. It supports the wider Grounded principle that the system should not treat physiological data as more authoritative than the traveller.

If affective touch remains in the design space, I would treat it as an optional, pre-selected modality for people who already know that slow touch or moving pressure is comfortable for them.

Biosensing might help indicate that support could be useful, but it should not be used to infer whether the tactile sensation itself is pleasant or effective without asking the user.

This paper therefore strengthens two parts of my project at once:

personalisation of the response, because touch can be simultaneously affective and unpleasant;

and

caution around biosensing, because physiological activity does not necessarily map neatly onto lived sensory experience.

For an airport system, that pushes me further toward:

body signal → uncertain interpretation → optional support → user choice → personalised tactile response

rather than:

body signal → automatic assumption → automatic touch.

Source

Capiotto, F., Romano Cappi, G., Mirlisenna, I., Mazza, A., Cicinelli, G., Lauritano, C., Keller, R. and Dal Monte, O. (2024). Autonomic and hedonic response to affective touch in autism spectrum disorder. Autism Research, 17(5), pp. 923–933.

Key reading from my analysis: participant sample and affective-touch procedure, pp. 924–927; skin-conductance and subjective-rating results, pp. 927–929; discussion of explicit versus autonomic response, pp. 929–931; study limitations and future research, pp. 931–932.