Concurrent Chains as a Preference Assessment for Assent

A concurrent chains arrangement works as a preference assessment because the client picks with their behavior, not with a stack of toys. That makes it stronger than a multiple stimulus without replacement (MSWO) test for measuring real assent. A board certified behavior analyst (BCBA) gets to see what the client will actually choose when the choice has weight.

MSWO tells you what they like. Concurrent chains tells you what they choose.#

A multiple stimulus without replacement, or MSWO, lays out items on a table. The client picks one. You move it. They pick again. You rank what they reach for. That gives you a list of likely reinforcers. It does not tell you if they want to work for those reinforcers, or if they want to be in the room at all.

A concurrent chains arrangement is different. You put two paths in front of the client. Each path leads to a clear outcome. The client picks the path. Then they live with what comes next for a set time. You watch which path they pick again, and again, across sessions. That is preference you can trust, because the client is voting with real time and real effort.

This is, without a doubt, my personal favorite preference assessment. It beats MSWO out of the water. From the talk — Matt Harrington

The shift matters for assent work. Assent is the client saying "yes, keep going" with their behavior. A ranked list of snacks does not tell you that. A path the client picks every time does.

Two-button setup for assent: contingent vs non-contingent zones#

Here is the simplest version of the setup. You have two buttons, or two cards, or two small mats. One color goes with one plan. The other color goes with the other plan. The client picks. You run that plan for a fixed block of time, like five or ten minutes.

In a classic concurrent chains arrangement, one path leads to free reinforcement with no demands. The other path leads to demands with reinforcement built in. The client's escape from work tasks is one common reason for challenging behavior, so the two paths give a clean comparison.

Now translate that to assent. Make one button mean "I want to keep going." Make the other button mean "I want a break from this." Now you are measuring assent provision and assent withdrawal as clean, countable behaviors. You are not guessing from facial expressions. You are not relying on a vocal "no" the client may not have.

This works for non-vocal clients too. The button, card, or mat is the response. The picture on it can be a thumbs up and a thumbs down, a green circle and a red circle, or a photo of the work area and a photo of the break area. Pick what the client can already discriminate.

The ongoing room split: a chains arrangement that runs all session#

The two-button version is fast to set up. The room-split version is what makes this a real assent tool, because it runs the whole session, not just at the start.

Split the therapy room into two zones. One zone is for pairing, engagement, calm activities, and alone time if the client wants it. The other zone is for work tasks. The client can walk from one zone to the other any time. Their body is the response. You count time in each zone and count the transitions.

The client could then vote with their feet. They could use dynamically their own behavior to communicate to us when they want to ascent to treatment in one side of the room versus ascent withdrawal from treatment into another side of the room. From the talk — Matt Harrington

This setup turns assent into a continuous measure. You stop asking "did the client assent at the start?" and start asking "how much of this session did the client choose to be in the work zone?" That number changes with your pairing, your task selection, and your reinforcement schedule. It is the cleanest dependent variable for assent work that you can run in a real clinic.

You can layer the in-the-moment assent data on top. If the client walks out of the work zone after a specific demand, that is a moment to mark. If the client stays through a hard step, that is also a moment to mark. The room split gives you the macro number. The in-session notes give you the why.

What the data sheet looks like (and why it's just frequency in two columns)#

The data sheet for a two-button setup is two columns: presses on Button A and presses on Button B. You can add a third column for the time block each press started. That is it. Frequency in two columns and a time stamp.

For the room split, the sheet is time in Zone A, time in Zone B, and a tally of transitions. A simple paper timer works. So does a tablet with two stopwatches. You do not need a separate research-grade tool to run this in session.

A few practical notes from the field. Run a baseline week before you change anything, so you have a stable picture of what the client picks now. Train two technicians on the same client before you trust the data, so you know the count is not drifting with one person. Graph the daily totals each week, not each session, because session-to-session noise is normal and weekly trends are the signal.

If you are billing this under insurance, the data sheet is your defense. You can show a payer that the client's chosen time in the work zone went up week over week. That is a clean line on a graph that maps to what they pay for.

Translating the enhanced choice model to a clinic that bills insurance#

The enhanced choice model is a research arrangement where the client has three options at any time: skills-based treatment, non-contingent reinforcement, or leaving the clinic. The third option, leaving, is the one most BCBAs flinch at.

Here is what the research actually showed.

When I saw this article and I saw what they were doing, my first thought was, well, they probably left the clinic a ton of times. They probably just hung out in NCR. Really amazingly, there wasn't much hanging out in NCR. There was almost no behavior of leaving the clinic. From the talk — Matt Harrington

Clients chose the work, when the work was set up well. That is the headline finding. Giving the option to leave did not produce a stampede for the door. It produced engagement because the client knew the door was open.

In a clinic that bills insurance, the "leave the clinic" option usually is not on the table inside a session. The caregiver brought the client for a reason. So translate the model down a layer. Use the two zones inside the room. The non-contingent reinforcement zone is the "I am taking a break" option. The work zone is the skills-based treatment option. You preserve the spirit of the model: the client always has a clear, observable way to opt out without escalating.

That preserves the assent finding without putting your authorization at risk. You can document the chosen time in the work zone as a measure of treatment engagement. You can also document the drop in challenging behavior, which usually follows when the client has a real path out that does not require challenging behavior to use.

When concurrent chains is the wrong tool#

A concurrent chains arrangement is the wrong tool when the client cannot yet discriminate the two paths. If pressing Button A and Button B both look like the same response to the client, you are not measuring preference. You are measuring random pressing. Teach the discrimination first, then run the assessment.

It is also the wrong tool when the two paths are not actually different. If both zones have the same demands, the same reinforcement, and the same staff energy, the client has nothing to vote on. Make the contrast real before you collect data.

And it is the wrong tool when you need a quick ranking for a small set of items. That is what MSWO is for. Use MSWO to pick the reinforcers. Use concurrent chains to test whether the client wants the whole package of work plus those reinforcers.

Frequently asked questions#

Is a concurrent chains arrangement billable under insurance?

Yes, when you frame it as preference assessment and treatment planning. The codes that cover assessment and protocol design cover the setup, the data, and the analysis. The room split runs inside a regular session, so the session time is billed as it usually is. Keep the data graphs with the rest of the client record so a payer audit can see what you ran and why.

Can I use concurrent chains with a non-vocal client?

Yes, and this is one of its biggest wins over methods that need a vocal "yes" or "no." The response can be a button press, a card touch, a step into a zone, or a reach toward a picture. The only thing that matters is that the response is observable, the client can do it without help, and the two options are clearly different to the client. That is why training the discrimination first is part of the setup.

How is this different from a free-operant preference assessment?

A free-operant preference assessment puts items in front of the client and measures how long they engage with each. It is a great tool for picking reinforcers. A concurrent chains arrangement puts two ongoing plans in front of the client and measures which plan they keep choosing. Free-operant tells you which item is the strongest reinforcer right now. Concurrent chains tells you which intervention the client wants to live with over time.

Watch the full talk#

The recording walks through the full setup with the data sheet on screen, the room split mapped out, and the assent moments called out as they happen. You will also see how the in-the-moment assent data layers on top of the macro time-in-zone measure, which is the part that is hardest to picture from words alone.