Analyzing Assent and Taking Data

Assent Versus Informed Consent: Why the Split Matters

Matt opens with a clean split. Informed consent (legal permission signed by an adult) is the legal side. Most analysts feel at home there. We sign forms. We know the rules. Assent (behavior that shows a client is willing) is the part many of us skipped. It is not a form at all. It is behavior you can see and count.

He calls the ethics code definition a solid one. Then he makes it plain. Assent is the client moving to the next link in the chain. They walk into the clinic. They reach for the pencil. They get out of the car. He uses a therapy office to show it. You drive there. You park. You sign in. You walk into the room. Every one of those steps is an assent behavior.

He also sets the scope up front. One hour does not make you skilled at this. He built the course as a first step, not a last one. Take the tools. Then keep reading. Then shape them to each client you serve.

a vocal or non-vocal verbal behavior that can be taken to indicate willingness to participate in research or behavioral services by individuals who cannot provide informed consent possible because of age or intellectual impairments

From the talk — Matt Harrington

Three Arguments for Putting Assent in Daily Practice

The first argument is legal. He points to the research on how courts treat assent. The short answer is that it is murky. There is a lot of gray. His rule for gray areas is simple. Do more, not less. He would rather over-collect assent than get sued.

The second is ethics and plain morality. Several parts of the code speak to this. They cover involving clients and stakeholders. They cover getting informed consent. They cover our duty to the client. Then he asks a blunt question. In what world is it fine to skip assent? You would not skip it with a coworker. Why is therapy different?

The third argument is behavioral. Assent means choice. Choice has been studied hard for years. Adding a choice of reinforcer can work as a motivating operation (something that raises a reward's value). He uses candy to show it. One bag of Skittles, or a pick from five candies? Most people take the pick. So assent can make the whole session more worth showing up for.

He raises one more thread and steps back from it on purpose. The autistic community has made hard critiques of our field. He says he cannot do that argument justice himself. Instead he points to sources he follows. Those include The Bearded Behaviorist, NeuroClastic, the Do Better Collective, and the PFA and SBT community group. His ask is direct. Go listen to them.

the addition of reinforcer choice can actually act as a motivational operation that increases the value of that reinforcer and instructional efficiency

From the talk — Matt Harrington

Spotting Assent and Assent Withdrawal in Vocal and Verbal Behavior

This is the nuts and bolts part. Any behavior that clearly moves toward an option may show assent. Walking to a work task. Reaching for a pencil. Turning the body toward the tech after a direction. He hedges on purpose with words like may. These behaviors are not the same across clients. Broad definitions will fail you. Write an operational definition (a clear, countable description) for this one person.

For tracking, he gives one rule. Pick a single assent behavior and track that. Pick one that is highly relevant. It should happen right before the hard task. Say a client is working on tolerating ten minutes of homework. The homework happens at one table. Moving toward that table may be your behavior. Pulling the chair out may work too. Now change the target. If the hard part is a group of peers, that chair means nothing. Opening the door might be your behavior. Waving at the other kids might be better.

Assent withdrawal is the flip side. Any clear move away from an option may count. Sitting still and not engaging. Staying outside the session room. Pushing back during a transition. He notes these are often easy to spot and harder to count. They tend to be passive. They are also often a precursor (a smaller behavior that comes first) to challenging behavior. If the hard task keeps coming, the behavior climbs.

His fix is low tech. Watch, and write plain notes. Tech presented task. Client clenched fists. Client frowned. Client stopped moving. Client sat on the floor. Then pick the clean one and count it. Frequency works fine. Turn it into a rate if you want.

Typically, assent provision is something active, right? You're doing something to progress the behavior chain, whereas assent withdrawal, the client's pausing that behavior chain or actively going in the other direction.

From the talk — Matt Harrington

Teaching an Initiation Behavior When Assent Is Hard to See

Sometimes the data get messy. Assent shows up on some trials and not on others. Sometimes the signs are too small for a tech to catch. His answer is an initiation behavior. It is a behavior that starts the chain that holds the demand.

Here is the flow. The client is with preferred items. The tech gives a visual that work is coming. The tech sets out the materials. The client taps the picture card. Only then does the first task come out. The tap says one thing. I am ready.

The catch is obvious. What if the client never taps? Then you reinforce that too. Both sides get reinforced, and they get reinforced the same. Saying no is a fine behavior everywhere else in life. It should be a fine behavior here.

He learned this with a client who had a long history of aggression during demands. A shaping and fading plan got them up to ten minutes of work. Bursts of aggression still showed up. The techs were following the plan. Nothing looked wrong on paper. The real problem was timing. The client was not ready when the schedule said go. So they added a simple tap to start. On good days the client opted out of a fifth of the trials. On rough days it ran much higher. Some days less got done. Goals still fell one after another. At first the client did not tap at all, and they reinforced that anyway. That is how the trust got built.

And what we saw is opt-outs increased and aggression decreased.

From the talk — Matt Harrington

Concurrent Chains: Keeping Two Clear Paths Open All Session

A concurrent chains arrangement (a setup with two clear choice paths) is his favorite preference assessment. He says it beats the multiple stimulus without replacement (a common array-based preference test) by a mile. The basic form is simple. The client presses one button and gets one condition. They press the other and get a different one.

His sample uses two treatments. Press green and you get non-contingent reinforcement (free rewards, no work needed). No demands come out for ten minutes. Press red and the rewards come with the work. Researchers use this to test what clients prefer. Now swap what the buttons mean. One is assent. One is withdrawal. Leave both live for the whole session.

The hard part is teaching the two paths. His trick is to make the switch loud and clear. A client pulls out of work. You sweep the table clear in one big motion. You drop down an obvious signal for play and pairing. The room just changed, and the client can tell. Do the same in reverse when they come back. You can also split the space. One side of the room is work. The other side is hangout. Then clients vote with their feet.

He points to the enhanced choice model as proof this holds up in real rooms. Clients there could do the treatment, hang out with free rewards, or leave the clinic. He expected them to bail often. They mostly did not. Almost nobody left, and there was little hanging out. He thinks the clean, no-strings exit is a big reason those sessions went so well.

Participation, no participation, no strings attached. Choose one and you can always switch between either.

From the talk — Matt Harrington

Degrees of Freedom: Is the Choice You Offer a Real Choice?

This is the part he says made it click. Degrees of freedom (how many real choices a person has) is the core idea. It comes from nonlinear contingency analysis (weighing all possible behaviors at once). He credits a book by Joe Lang and a Facebook group called The Constructional Approach. Look at the costs and payoffs of every possible behavior, not just the one you targeted.

He walks it through with groceries. Drive to the store: thirty minutes, you get food. Walk to the store: ninety minutes, you get food. Stay home: no effort, no food. That looks like three choices. Now say you truly need the food tonight. Staying home drops out. You have two. Now say you cannot leave the house at all. You have one.

The math is easy. Degrees of freedom equals N minus one. N is the count of options with similar payoffs and similar costs. Stay above zero. He likes to see two. Now run a common plan through it. Assent earns candy and praise. Withdrawal earns planned ignoring. The client wants the candy. That is zero degrees of freedom. The withdrawal behavior is on extinction (it no longer earns its payoff).

He got this wrong himself for years. He taught a client both paths in a natural teaching setup. Assent led to lots of one-on-one time with the tech. Withdrawal meant the tech backed off and left them alone. The client loved that tech. So the two paths were never equal. He thought he had one degree of freedom. He had zero. They changed it, and aggression dropped.

All things have to be equal within the degrees of freedom calculation.

From the talk — Matt Harrington

Building a Treatment Plan With Assent at the Center

Part five is where the tools join up. He starts at the widest level with degrees of freedom. Look at the client's whole time in your setting. Say the full enhanced choice model is off the table. Maybe nobody can pick the client up mid-session. You can still build three options.

Here they are. One: take part in programming and get the rewards. Two: skip programming and hang out with the tech. Three: skip both and play alone or go home. Option three still holds the rewards. His logic is neat. If the client turns down tech attention, attention is not what they want right now. So playing alone is not a loss for them. That is two degrees of freedom for every client on your caseload.

Then zoom in. Set a concurrent chains inside the session. Split the room, leave a door open, or just say the reminder out loud. Keep the payoffs even here too. Then zoom in once more and track the moment-by-moment behavior. This feels redundant. It is not. Say a client has three subjects of homework. Assent holds for two of them and drops on the third. Flip the order the next day. If it drops on the same subject again, you learned something real about the task.

For measurement he lists your options. Frequency of assent and of withdrawal. Momentary time sampling (a quick check at set moments). Whole or partial interval. Track how much time the client spends in each zone. Then audit yourself on a schedule. Every week or two, reread your protocols. Before a change goes live, count the degrees of freedom. He caught one of his own this way. A new plan quietly cut a client's choices from one to zero. They paused, rebuilt it, and kept the effect they were after.

Answering the Objection: But Nothing Will Get Done

This is the question he gets most. If they can always opt out, will they ever work? He starts by naming a bad assumption. We treat assent withdrawal as the same thing as escape maintained behavior (behavior that gets someone out of something). It is not. People opt in to hard things all the time. He writes reports he hates. He has laundry he hates. He did grad school homework he hated. Long-term values pull us through short-term pain. Clients deserve that same shot.

For the colleague who will not budge, he splits the group in two. Some were trained a certain way and simply never met this topic. Talk with those folks. Some want to be left alone because their way has worked for years. Show those folks instead. Run it on your own caseload. Bring the data to the staff meeting.

His technical argument uses the response class hierarchy (behaviors ranked by effort, same payoff). A client who destroys property is escaping. It works, but the effort is huge. Look for the cheaper members of that same class. Freezing. Clenched fists. Pushing back from the table. Reinforce those instead. The dangerous end drops out. Clenched fists are not a behavior you need to reduce. Then a calmer room opens doors again. Items you had to pull can come back in.

If a client still does nothing, he flips the question back on you. What does that tell you about the setting you built? Then he asks the question he keeps returning to. What has to be true for participation to occur? He gives one example from his own work. Sorting by feature, function, and class was going nowhere with zoo animals. He swapped in the client's special interest. Assent showed up. The payoffs stayed equal. Only the materials changed.

So you may not be getting the goals on your treatment plan done, but you are still teaching an important skill.

From the talk — Matt Harrington