NET in the Wild Natural Environment Teaching & In-Vivo Data Collection
Presented by Sean Yocum
Natural Environment Teaching (NET) provides opportunities for skill acquisition within meaningful activities, naturally occurring antecedent conditions, and functionally relevant reinforcement contingencies. However, effective implementation requires substantially more than conducting therapy outside of a traditional tabletop environment. Behavior technicians must recognize relevant environmental variables, identify changes in motivation, capture or arrange appropriate learning opportunities, implement programmed prompting and reinforcement procedures, and accurately measure behavior while actively interacting with the learner. This interactive virtual training examines the practical implementation of NET and in-vivo data collection from the perspective of the Registered Behavior Technician (RBT). Participants will analyze simulated clinical scenarios to differentiate NET from unstructured interaction and “DTT on the floor,” identify naturally occurring and contrived learning opportunities, recognize motivating operations, evaluate natural reinforcement contingencies, and select measurement procedures appropriate for behavior occurring within dynamic environments. Particular attention is given to maintaining data integrity when traditional trial-by-trial recording may interfere with ongoing interaction. Through polls, scenario-based decision making, knowledge checks, and simulated data-collection exercises, participants will practice identifying the behavioral processes underlying effective NET while maintaining learner engagement, dignity, and meaningful skill development.
Sign in to buy — $10One-time purchase, permanent access. Your BACB PD form is delivered to your inbox the moment you pass the quiz.
Learner ratings
from 15 learners
100% would recommend this PDU to other professionals in their field (15 responses)
“nope!”
— Elexus G.“Amazing”
— Emily G.
About this course
Natural Environment Teaching (NET) is more than playing with a learner and following their lead. Effective NET requires behavior technicians to recognize meaningful learning opportunities, respond to changing motivation, arrange the environment intentionally, use appropriate prompting and reinforcement procedures, and collect accurate data—all while maintaining natural and meaningful interactions. In this interactive virtual RBT Professional Development Unit (PDU), Sean Yocum of the Rad N Bad Podcast and Hickory Learning Group, takes participants beyond the buzzwords surrounding “naturalistic” and “learner-led” intervention and examine what high-quality NET actually looks like in practice. Through interactive scenarios, knowledge checks, simulated clinical situations, and in-vivo data exercises, participants will practice distinguishing purposeful NET from unstructured interaction, identifying naturally occurring learning opportunities, recognizing motivating operations, selecting appropriate measurement procedures, and collecting defensible data without disrupting the natural interaction. The goal is simple: stop simply playing therapy and start engineering meaningful learning opportunities.
From the talk
What was covered
Sean Yocum breaks down what real NET (teaching built into natural play) looks like on the floor. Name what you are teaching, read motivation, match your measure, and never rebuild data from memory.
- Before you call a session NET, name the target. Then name the antecedent (what happens right before), the response, and the consequence (what happens right after).
- When arranged trials get no response, check motivation before you decide the learner lacks the skill.
- If percent correct does not fit the behavior, talk to your BCBA (board certified behavior analyst). A frequency or rate measure may work better.
- Never rebuild data from memory at the end of a session: score what you saw and write up the barrier.
- Keep three or four programs open instead of twenty so the data you turn in stays accurate.
- Honor assent (the learner agreeing to take part). Pivot to a new teaching context instead of waiting out the rest of the session.
NET Is a Teaching Procedure, Not a Spot on the Floor
Sean Yocum opened with the problem he sees in the field. Sessions get called NET when no teaching happens. The tech sits on the mat. The kid plays. It looks like a session. He called that passive hanging out. Then he named the opposite failure. Techs take DTT (drilled trials, one at a time) and move it to the floor. Neither one is NET.
His test has four parts. What is the program target? What is the antecedent? What is the response? What is the consequence? If you cannot name all four, you were not teaching.
He made the room try it. You play trains with a learner for 20 minutes. You narrate and laugh the whole time. Is that NET? The best answer is that it depends. Trains can be a rich block of teaching. It can also be dead time. One choice on the slide was "only if data are collected." He threw that out. Data do not turn play into teaching.
The next scenario flipped it. A learner plays with a farm set. The tech sits close and runs unrelated car flashcards. That is DTT on the floor. Sitting near the toy changes nothing. Reinforcing right answers changes nothing either. Then he named a trap many of us have hit. You see farm play, so you grab the farm animal cards. That is still DTT. If the learner is playing farm, the targets have to use the farm.
The function of NET is we are actually contriving learning opportunities.
From the talk — Sean Yocum
NDBI Roots and the RBT Clinical Lane
NET did not come out of nowhere. He walked through NDBIs, or naturalistic developmental behavioral interventions (play based teaching with real targets). He described a group that came together in 2015 to push for better care. NDBIs get a bad name as going with the flow. He pushed back hard. They are rigorous and evidence based.
Most behavior analysts study behavior analysis, not child development. NDBI blends the two. You still use clean prompting, planned sequences, and contingencies (rules linking behavior to outcome). You also watch social communication and milestones. He named the Denver model as one example that stays behavior analytic.
Then he drew the lanes. The BCBA picks the target behaviors and writes the protocol. The RBT (the front-line technician running sessions) recognizes the moment, runs the program, and records what happened. An RBT should not be flying blind. The written program should say how to set up the chance to teach. Behavior skills training (explain it, model it, then practice) means you get those steps in writing, plus a model.
He asked the room a fair question. Which is easier to run, DTT or NET? Most hands went up for DTT. In NET the structure moves on you. Chances to teach show up and then vanish. He called that part of the job an art.
DTT is easier to implement. Because you're in control of the structure.
From the talk — Sean Yocum
The SCAN Routine: Scan, Check, Arrange, Naturally Reinforce
He gave the room a four step routine built on his own name. SCAN. Scan the room. Check motivation. Arrange the chance to teach. Naturally reinforce. He tied it back to Stokes and Baer on generalization (using a skill in new places). Teach in the real setting and you build that in from day one.
Scan means read the whole room, not just the face. What toys and items are there? What is the learner moving toward? What are they moving away from? Avoiding tells you as much as approaching.
Check means you ask about the MO, or motivating operation (what a learner wants right now). Is it sensory input? Social time? Access to a toy? Or is it escape from something they want away from? He told the room to watch for value dropping off too. Something can lose its pull in seconds.
Arrange means set up the antecedent once you know the MO. One tech in the room gave a clean example. A learner loves the trampoline, so the tech jumps with them. Then the tech stops and waits. The learner asks for more jumping using sign language. Nothing is forced. The contingency is simply there.
Naturally reinforce means the payoff fits the act. They ask to open the container, so you open it. They ask for a push, so the swing moves. Praise can ride along, but praise is not the reinforcer (the thing that actually drives the behavior).
we can have this tunnel vision where we're staring like entirely at this client's face
From the talk — Sean Yocum
Motivation Shifts, and Your Job Is to Follow It
Another scenario. A learner walks up with a clear bin of toy cars and reaches for it. Do you hand it over? Do you ask what they want? Do you require a full sentence? Or do you watch and follow the written program? The last one wins. Handing it over burns the moment. The learner learns that reaching makes adults fetch. Requiring a sentence ignores the plan. That learner may not be vocal at all.
Reaching tells you there is an MO. It does not tell you the target. The target could be a sign, a word approximation, or a picture exchange. It could be "open," or even "all done." Motivation does not pick the program. The written procedure does. If that procedure looks wrong, say so and ask the BCBA to change it.
Five minutes later, the learner walks away from the cars. The MO moved. What is worth working for at 2:03 can be worthless by 2:08. He used his own coffee as the example. At 4:45 in the morning he wants a cup badly. At 10 at night he wants nothing to do with it.
This has a direct fix for a daily complaint. You set up chance after chance and get nothing back. Or the learner just waits for your prompt every time. Check motivation before you call it a skill problem. He tied the same idea to pairing and instructional control. Pairing means building rapport by linking yourself to good things. Instructional control means the learner is ready to follow your lead. You cannot pair with nothing. You pair with what the learner wants right now.
Motivation is a moving target.
From the talk — Sean Yocum
High Engagement Is Not the Same as Quality Teaching
He pointed to a 2012 reference on trial density and fidelity. Trial density (how many teaching trials fit in the session) is one half. Fidelity (running the program as written) is the other half. Then came the hard question. Is high engagement the same as quality teaching? The room said no, and he agreed.
As a supervisor he has watched warm, skilled techs in family homes. Kids smiling. Lots of movement and high fives. One session he reviewed had three trials in four hours. The bond was great. The learning was not there. You cannot score clinical success by how much a kid smiles.
The next week that same tech ran far more trials. Then challenging behavior showed up. Density had swung too far the other way. He called that turning into an interrogator. A learner who keeps asking why gives you real feedback (how the session feels to them). That is your cue to ease off.
The target sits in the middle. Good fidelity, enough trials, and a session that still feels natural. Without that balance, he said, you waste a child's time.
we're missing out on so many natural learning opportunities that can be occurring.
From the talk — Sean Yocum
Match the Measurement Procedure to the Behavior
Then he moved to the part he called the hard part. In vivo data (scored live, during the session) is messy by nature. The learner moves. You move. Motivation shifts. A parent asks you a question. Safety comes first. And you still have to score it. His first question when scoring breaks down: is the problem the tech, or the measure?
Percent correct is the common mismatch. Percent correct assumes you present a trial, get a response, and deliver a payoff. For requests in NET, count the requests instead. Fifteen requests in an hour gives you a rate. Rate shows how fluent (quick and accurate) the skill is. Percent correct hides that.
He and the room listed other tools. Frequency counts. Interval recording (checking for behavior in set time chunks). Permanent product (a lasting result you can count). Rating or magnitude scales for how bad a behavior was, with each level defined in writing. Self scoring for learners who can track their own behavior, checked by you. One tech dropped the tablet app for a clicker. He described a presenter who wears counting beads on a cord and moves them like an abacus.
He asked supervisors to check the measure before blaming the tech for fidelity. He asked BCBAs to open fewer programs. Twenty open goals makes clean data unlikely. Three or four is workable. He gave techs the same advice: pick one or two targets and stay with them.
Record What You Saw and Report What Blocked You
He named the trap the 4:45 problem. The session ends at five. You taught well and scored nothing. So you sit there and put in eight out of ten. That number is not real, and you know it.
He ran it as an ethics check. Several real requests happened during play and never got marked. He gave four options. Guess at 80 percent. Enter about 10 trials from memory. Ask another tech what they recalled. Or score only what you verified and report the barrier. Only the last one holds up. Asking a coworker works only if they truly tracked it. A vague memory is not data.
Reporting a barrier is not a failure. It is information. The barrier field exists so the BCBA can see that live scoring did not work. Then the team can fix the measure instead of blaming the tech. He gave a five step loop for the floor. Spot the chance. Respond the way the program says. Reinforce it. Mark it. Then get back in.
He closed the data block with a short drill. The target was independent functional requests (clear asks the learner makes on their own). Five responses came up. "Bubble" on its own. "Bubble" after a partial vocal prompt. "Help" on its own. A reach with no program behind it. Then "more" on its own. The answer is three. The word does not matter. Function and independence do.
don't guess, don't estimate. And don't fabricate your numbers.
From the talk — Sean Yocum
Assent and Autonomy Without Clinical Negligence
The last block was assent. In NET, assent is ongoing. If you truly follow the learner, you honor it as you go. Behavior is your information. Turning away, ignoring you, or leaving the room all say the same thing. The MO is gone. He also said a full session of adult led compliance is not good therapy.
Then he drew a hard line. Honoring assent is not sitting back while a child runs the house for two hours. Waiting 20 minutes for a learner to come back is not therapy either. Kids are in therapy because they need to learn skills. If they already followed every direction, they would not be in therapy.
So what do you do when a learner says no? Read the room again. Check motivation again. Pivot to a new natural context. Escape itself is a teaching moment. Teach the break request. Say it looks like they want a break, then model asking for one. You cannot expect a learner to sit with a hard task if they have no way to ask out.
That does not mean the hard task disappears forever. You come back and build tolerance later. The real world asks people to wait and to hear no. Both are teachable, and both still run off motivation. He said the job lives in that tension. Respect the voice, stay active, keep teaching.
Common questions
What is the difference between NET and just playing with a learner?▾
NET has a named target, a clear antecedent, a response, and a planned consequence. Yocum's point was that you cannot tell the two apart by watching the form of the session. You tell them apart by whether teaching is actually running.
Is DTT on the floor the same thing as NET?▾
No. Running flashcards next to a toy is still discrete trial work. Sitting on the carpet does not change what the procedure is. If the learner is playing farm, the targets have to use the farm activity itself.
What should an RBT do after missing data during a NET session?▾
Score only what you directly saw, then report the barrier. Do not estimate a percentage or fill in trials from memory. Do not lean on another tech's vague recollection either. A reported barrier gives your BCBA something real to fix.
Why is percent correct a poor fit for request data in NET?▾
Percent correct assumes you set up each trial and score the response that follows. In NET many requests come up on their own, so there is no fixed trial count. Counting requests gives you a rate, and rate shows fluency. Yocum used 15 requests per hour as the example.
Does honoring assent mean ending the session when a learner walks away?▾
No. Walking away tells you the current motivation is gone, not that your job is done. You check the environment again, find new motivation, and pivot to a new teaching context. Yocum called passive waiting clinical negligence, not respect.
How many programs should be open during a NET session?▾
Fewer than most teams run. Yocum suggested three or four open goals instead of twenty. He also told techs to pick one or two targets to focus on in the moment. Fewer targets means the data you turn in is far more likely to be accurate.
About the speaker
Sean Yocum is a behavior analyst who also brings a special education background to his work on child development. He began as a paraprofessional and RBT. He now supervises techs in family homes, where he watches NET and data collection play out in real time. In this 90 minute professional development unit, he taught NET and the SCAN routine. He covered motivating operations, measurement selection, and in vivo data ethics. He also drew the line between honoring assent and clinical negligence.
This summary was generated from the recording’s transcript. Quotes are taken word for word from the talk.
What you'll learn
- 1Upon completion of this training, participants will be able to: 1. Differentiate Natural Environment Teaching, structured teaching, and unstructured interaction by identifying the observable behavioral components of each approach. 2. Identify naturally occurring and appropriately arranged learning opportunities within common play, leisure, social, adaptive, and daily living activities. 3. Identify changes in motivating operations and relevant environmental variables that may influence learner responding and the effectiveness of reinforcement during NET. 4. Describe how antecedent arrangement, prompting, prompt fading, and naturally related reinforcement can be incorporated into NET while maintaining meaningful learner engagement. 5. Differentiate naturally related reinforcement contingencies from arbitrary reinforcement and identify when the outcome of a learner's behavior can function as the reinforcer. 6. Select appropriate in-vivo measurement procedures—including frequency, opportunity-based recording, duration, latency, first-trial probes, task analysis, and permanent-product measurement—based on the behavior being measured. 7. Demonstrate through scenario-based exercises accurate in-vivo data collection while distinguishing directly observed behavior from retrospective estimation or reconstructed data. 8. Evaluate simulated NET sessions to identify missed learning opportunities, measurement barriers, changes in learner motivation, and potential improvements to treatment implementation.
Concepts in this course
Course materials
Related talks
Talks that cover the same concepts
Beyond Table Work
ABA Collective / Camber
Understanding ADHD: An RBT PDU Series | ADHD 101: Inattention, Hyperactivity & Impulsivity
ABA Made Ez
Rad N Bad: Beyond the Blue Square — Demystifying Discrete Trial Training, Stimulus Control, and Generalization for RBTs
Hickory Learning Group
Bite-Size Behavior Analysis: Differential Reinforcement
Candid Contingencies, LLC
Navigating Communication with Clinicians
Candid Contingencies, LLC
Bite-Size Behavior Analysis: Behavior Analytic Definitions of Reinforcement & Punishment
Candid Contingencies, LLC
OpenCEU is supported by advertising from third-party sponsors. Sponsors are clearly labeled, have no influence over course content, instructors, or CEU decisions, and never appear on certificates. The BACB does not sponsor, approve, or endorse OpenCEU sponsors or their products.