Teaching STEM to Students with ASD: What Gets in the Way, and What Helps

A STEM lesson asks a student to solve several problems before reaching the one the teacher planned. Alongside the science or math, there are spoken instructions to remember, materials to organize, unfamiliar sounds to manage, and group decisions to negotiate. For many students with ASD, one of those other problems is the one that stops the lesson.

That is why teaching STEM to students with ASD can be hard even when the student is interested in the subject. The teacher's job is to work out which demand is in the way, because each one calls for a different support. This post walks through the six demands research points to, and what each one looks like in a classroom.

The short version: keep the thinking, change the conditions. Students with ASD will not all hit the same barriers, so the goal is to make the sensory, organizational, social, and communication load manageable without removing the intellectual work.

1. Interest in STEM does not remove the barriers

A student who loves numbers, structures, or computers has a starting point, not a guarantee. In a U.S. longitudinal study, about 34% of college students with ASD who declared a major chose a STEM field, yet only about 32% of the full ASD sample had enrolled in college at all [1]. High STEM interest and low access can exist at the same time.

Enthusiasm for a subject tells you nothing about whether the student can follow a fast discussion, organize a multi-step build, or explain an idea in the format you expect. Exceptional talent should not be the price of admission to real STEM instruction.

In the classroom: when a strong-interest student disengages mid-lesson, treat it as a signal about the lesson's demands, not the student's motivation. Ask which of the next five demands changed at that moment.

2. Sensory demands compete with the lesson

Hands-on STEM changes the sound and movement of a room. Groups talk, materials shift, equipment makes unfamiliar noises, and classmates work close together. In a study of adolescents with ASD in mainstream schools, students named hearing as the sense that affected them most, and described how it broke their concentration [2]. The sample was small, but the lesson for teachers is direct: students can tell you about conditions adults stop noticing.

In the classroom:

• A student who moves away from the group may be reacting to the environment, not the task. Check the noise and crowding before assuming avoidance.
• A missed instruction may have been lost to competing sound. Pair spoken directions with a visible version.
• Materials matter. Felt, fabric, and soft-surface building pieces produce less noise on contact than plastic or metal and are easier on students who avoid hard, cold, or loud textures. Large magnetic felt panels, such as Superspace, let students build at full scale with quiet connections and no small parts. Sensory preferences differ, so this is a materials choice to test with each student, not a fix for every student.

What you might see: a student who stays with a build for the full work period when the activity moves from a shared table to a quieter corner, or from hard pieces to soft ones.

3. Open-ended work asks for more than a good idea

Engineering and inquiry ask students to plan, compare options, hold constraints in mind, and revise. Those are executive function skills. A meta-analysis of 235 studies found average differences in executive function between people with ASD and comparison groups [3]. A group average says nothing about one student, but it explains why task organization needs as much attention as content.

Uncertainty adds a second load. "Build anything" leaves the starting point, the purpose, and the definition of done unstated. A 2020 review found a substantial link between difficulty tolerating uncertainty and anxiety in people with ASD [4]. The evidence is correlational, so it does not show that open-ended STEM causes anxiety. It does give teachers a reason to check how much uncertainty a prompt introduces.

Creativity and clarity can coexist. Students can reach different solutions while knowing the problem, the materials, and how their work will be judged.

In the classroom:

• Replace "build anything" with a bounded challenge: "Build a shelter that one person can sit inside and that stays up when you let go." The student still decides the shape, the size, and the method.
• Separate the two questions a build raises. "Does this need support?" is a physical observation. "Which piece do I move first?" is a planning step. When a build stalls, find out which of the two the student is stuck on.
• Make revision an expected step, not a failure. Say in advance that every team will change its design at least once.
• Large-scale materials help here because a wall that leans or a roof that sags is visible from across the room. The feedback comes from the structure, not from an adult.

What you might see: a student who states a plan before touching materials, or who names what changed after a collapse and chooses a different next move.

4. Language can hide scientific understanding

STEM runs on relational language: greater than, because, before, unless, similar to. A student can know every vocabulary word and still need help with how a paragraph, question, or spoken explanation connects the ideas. Long instructions add another layer between the student and the problem.

A small study taught three middle-school students with ASD an explicit compare-and-contrast approach to science text. Comprehension improved during instruction and stayed high afterward [5]. Three students limits what this proves, but it is a concrete example of teaching the structure of science language, not only the terms.

Expressing an idea is a separate challenge. A lesson that depends on fast spoken answers gives few openings to a student who communicates differently. ASHA describes augmentative and alternative communication (AAC) as including gestures, communication boards, and speech-generating devices, and its school guidance stresses access to curriculum vocabulary [6].

In the classroom:

• Teach the comparison frame before the content: "Both shapes have four sides. The square's sides are equal; the rectangle's are not." Then hand students two physical shapes and ask for the same frame in their own words.
• Give one instruction at a time, and pair it with the object it refers to. "Put the triangle on top" lands better when the triangle is in the student's hands.
• Communication access in STEM has to cover predicting, questioning, disagreeing, and explaining, not only requesting. A student who can ask for a panel also needs a way to say "that side will fall." Add STEM words (taller, balanced, enclosed, corner, edge) to the student's AAC vocabulary before the unit starts.
• Let the model do some of the explaining. A student who cannot yet say why a structure is stable can point to the piece that is holding it up, then build the sentence from there.

What you might see: a student who makes a prediction with a gesture, a board, or a device before the group tests it, and whose prediction gets treated as a contribution.

5. Group work adds a layer nobody explained

A shared build means deciding whose idea to try, when to speak, how to disagree, and what happens when someone changes the plan. Those rules are rarely stated. In focus groups with 15 school staff and a review of 23 IEPs, staff described difficulties with group projects, asking for help, starting work, and handling changes [7]. The sample was small and selected, but the demands it names are exactly the ones collaborative STEM makes.

Participation has to be intellectually real as well as socially possible. A student whose only job is handing out pieces has no chance to predict or explain.

In the classroom:

• Assign rotating roles that each carry a decision: planner, builder, tester, reporter. The tester decides whether the structure passed. The reporter decides what to tell the class.
• Say the group rules out loud and post them: one idea at a time, test before you argue, anyone can call a pause.
• Build big enough that everyone has a side. Life-size construction spreads a team around a structure instead of crowding four students over one small kit, which lowers the social pressure of working shoulder to shoulder.
• Script the change moment. "We are going to change the plan. Here is what stays the same." This is the step staff flagged most often.

What you might see: a student who disagrees with a teammate's idea by proposing a test instead of withdrawing, or who takes the reporter role and explains a result to the class.

6. Good support keeps the thinking with the student

Several commonly recommended supports have an evidence base. A 2021 systematic review screened 972 articles published 1990 to 2017 and identified 28 practices meeting its criteria, including visual supports, modeling, task analysis, and AAC [8]. The review covers focused practices across many goals; it does not endorse any complete STEM lesson or any particular product.

The principle is to match the support to the demand that is in the way:

The demand	The support	What it looks like with building materials
Holding the problem in mind	Visual support	A sketch or photo of the goal structure stays posted beside the build
Unfamiliar process	Modeling	The teacher connects two panels once, then hands the next two to the student
Too many steps at once	Task analysis	Base first, then walls, then roof, each on its own card
No way to share the idea	AAC	STEM vocabulary loaded on the device before the unit

Assessment needs the same care. CAST's Universal Design for Learning guidelines call for varied ways to take in information, work with materials, and show understanding [9]. UDL is a design framework, not an ASD-specific study, but applied to assessment it asks one useful question: does the required response measure the learning you intended?

In the classroom:

• If the goal is understanding a spatial relationship, accept a model, a diagram, speech, or the student's communication system as evidence. If the goal is a written explanation, writing is part of what you are teaching, so say so.
• Be explicit about which decisions were the student's. A picture or a checklist on the wall says nothing by itself about whether the lesson was accessible; the student's own reasoning does.

What you might see: a student who shows that a wider base is more stable by building two versions and pointing to the one that stands, with no written sentence required.

Where three-dimensional, tactile materials fit

Physical models give spatial ideas a form that can be seen, touched, and changed. A student can walk around a relationship, compare two arrangements side by side, and connect a drawing to an object. That is why construction materials belong in a serious conversation about STEM access, and not only as an engagement tool.

The evidence for spatial instruction is reasonably strong. A 2022 meta-analysis of 29 controlled studies with 3,765 participants found that spatial training improved mathematics performance on average. Programs that used physical materials showed stronger math effects than those that did not, and transfer was greater when the math was closely related to the training [10]. This was general spatial-learning research, not research on students with ASD or on any specific product.

Superspace panels are large magnetic shapes (squares, triangles, rectangles, trapezoids) made of soft felt that connect into structures big enough for children to sit inside [11]. For the demands in this post, four properties of materials like these matter:

Material property	Which demand it addresses	What the teacher can do with it
Full scale	Spatial language, group work	Teach inside, outside, corner, and edge with the student's own body as the reference; spread a team around one build
Soft, quiet surfaces	Sensory load	Build in a shared room without the clatter of hard pieces; offer a tolerable texture to touch-averse students
Few, large pieces	Executive function	Keep the step count low; a structure of 6 to 10 pieces has a plan a student can hold in mind
Visible failure	Open-ended work, assessment	A wall that leans gives feedback without an adult; two builds side by side show which design was more stable

Large materials bring their own considerations: reach, movement, available floor space, and coordination with other builders. Their usefulness depends on the student and on the instruction wrapped around them. The research cited here supports attention to spatial, hands-on instruction; it does not establish Superspace-specific learning or therapeutic outcomes.

What you might see across a unit: a student who uses the words taller, balanced, and enclosed correctly after using them inside a build; who compares two of their own structures and says which held more weight and why; or who draws a plan on paper and then recognizes it in the finished build. Those are the outcomes to record, and they are observable without a written test.

Teachers need time and a shared plan

The same staff study that named group work as a challenge also named time, resources, and communication across adults as constraints [7]. ASHA's guidance points the same way: classroom teachers and communication specialists plan together, the teacher bringing the STEM goal and the specialist bringing access [6]. The adult's job is to enable the student's idea while leaving the decisions with the student.

Progress in STEM for a student with ASD often shows up in small, specific moments: a better prediction, a clearer comparison, a question asked without prompting, a design changed because of something the student observed. Those moments are the point of access, and they are worth writing down.

Next steps

• Browse the STEM lesson plans built around full-scale construction.
• See how Superspace is used in special education and therapy settings.
• Request a quote or check grant and funding options for your program.

References

1. Wei, X., Yu, J. W., Shattuck, P., McCracken, M., & Blackorby, J. (2013). Science, technology, engineering, and mathematics (STEM) participation among college students with an autism spectrum disorder. Journal of Autism and Developmental Disorders, 43(7), 1539–1546. Link
2. Howe, F. E. J., & Stagg, S. D. (2016). How sensory experiences affect adolescents with an autistic spectrum condition within the classroom. Journal of Autism and Developmental Disorders, 46, 1656–1668. Link
3. Demetriou, E. A., et al. (2018). Autism spectrum disorders: A meta-analysis of executive function. Molecular Psychiatry, 23, 1198–1204. Link
4. Jenkinson, R., Milne, E., & Thompson, A. (2020). The relationship between intolerance of uncertainty and anxiety in autism: A systematic literature review and meta-analysis. Autism, 24(8), 1933–1944. Link
5. Carnahan, C. R., & Williamson, P. S. (2013). Does compare-contrast text structure help students with autism spectrum disorder comprehend science text? Exceptional Children, 79(3), 347–363. Link
6. American Speech-Language-Hearing Association. (n.d.). Augmentative and alternative communication (AAC). ASHA Practice Portal. Accessed October 22, 2026. Link
7. Duncan, A., et al. (2023). School challenges and services related to executive functioning for fully included middle schoolers with autism. Focus on Autism and Other Developmental Disabilities, 38(2), 90–100. Link
8. Hume, K., et al. (2021). Evidence-based practices for children, youth, and young adults with autism: Third generation review. Journal of Autism and Developmental Disorders, 51, 4013–4032. Link
9. CAST. (2024). Universal Design for Learning Guidelines version 3.0. Accessed October 22, 2026. Link
10. Hawes, Z. C. K., Gilligan-Lee, K. A., & Mix, K. S. (2022). Effects of spatial training on mathematics performance: A meta-analysis. Developmental Psychology, 58(1), 112–137. Link
11. Superspace. (n.d.). The Big Set (24 Panels). Accessed October 22, 2026. Link