Article Club: Using Deposit Contracts and Reinforcement Thinning to Increase Physical Activity

Why a Step Count Study Belongs on Your Reading List

Most of our work sits inside autism care. This session pushed the frame wider. The group read a study on daily steps. The people in it were sedentary adults (adults who move very little). Pauli Catanese picked the article on purpose. Behavior is all of life. Our scope is huge. We just tend to stay in one lane.

The health case is blunt. About one in four U.S. adults get almost no activity. Most U.S. adults log 3,000 to 4,000 steps a day. The CDC asks for 8,000 to 10,000. Heart disease is a leading cause of death here. Sitting still is also tied to weight gain, high blood pressure, anxiety, and low mood. So steps are a target with real social value.

The room also had a tie to the work. One of the authors, Dr. Zerger, used to be an analyst at Achieve. She also teaches at USF. The group planned to send her their questions. That closeness made the read feel less like homework. Nothing in the study is far from clinic work either. Reinforcement, motivation, and behavior change all carry over.

aba is the science of behavior

From the talk — Pauli Catanese

What a Self-Tailored Deposit Contract Actually Is

Start with the core idea. You hand over something you value. You get it back only when you hit your goal. That is a deposit contract (you risk your own reward). The term self-tailored (you set your own terms) matters here. You pick the goal. You pick what goes in, and how much.

The group traded plain examples. Think of the coin you drop into a store cart. You get it back when you return the cart. One professor held a student's lipstick against a weekly page count. Another student left his record collection with a thesis chair. He lost one record for a missed deadline. He never missed one again.

Then came a sharp line that is easy to miss. A gym that keeps your money when you skip class is not doing this. That is response cost (you lose what you already had). In a deposit contract the payback stays on the table. You can win your own stuff back. The group also flagged the piece we often skip with clients. In the real version, the person chooses to deposit. We usually tell the learner to hand the toy over.

it's just a response cost procedure

From the talk — Pauli Catanese

How the Study Ran: Goals, Phases, and Payback

Three adults took part. All were women, ages 20 to 45. Each logged under 5,000 steps a day at the start. Some sign-ups were cut because their counts were too high. The design was a non-concurrent multiple baseline across people (people start at staggered times). Steps came from an Apple Watch. You had to own one and wear it before the study began.

Baseline (counts before the plan began) used weekly means. The watch does not hold daily counts that far back. All later phases used daily counts. The first phase paid out each day. Hit the step goal, and get part of the deposit back that same day. That phase ran 14 days, so there were 14 chances to earn.

One woman put in a modest deposit. Her goal was 6,000 steps a day. She earned back a small flat amount on each day she met it, no matter how much she had deposited. Another put in a much larger deposit and still earned back only a bit more each day. The room balked at how small that payout felt. A flat daily payout that small is easy to say yes to up front. It is also easy to shrug off later.

Then came reinforcement thinning (payouts spread further apart). Phase two paid every three days. The bar was two of those three days. That phase held five chances to earn. Phase three stretched out to a week, with a bar of five of seven days. After that the plan came off and the team watched for 12 weeks.

the design was a non-concurrent multiple baseline across participants design and they used an apple watch to count steps

From the talk — Pauli Catanese

What the Data Showed, and What Follow-Up Did Not Hold

All three went up once the plan began. Their start counts sat near 2,000 to 4,000. During the plan they rose into the 6,000 range. They held through both thinning phases. One woman gained about 130 percent over her own start.

The charts had spikes too. Two of them had days over 20,000 steps. One slump traced back to a week of illness. The group liked how clean the write-up was. Every step of the method was spelled out well enough to copy.

Follow-up is the honest part. Steps did not stay at the level seen during the plan. But every follow-up mean sat well above the start. The room called that a win, and that read is fair. Full maintenance (keeping gains after help stops) is rare. Plan for the settle, and say so to families before you begin.

it does mention how it didn't maintain but all their follow-up steps were significantly higher than their baseline

From the talk — Pauli Catanese

The Accountability Confound Nobody Should Skip

Here is where the hour got good. The people in the study already wanted more steps. A real person checked their counts. A real person sent the payback by hand. So what moved the steps? The dollar, or the eyes on them?

The group had a name for part of this: reactivity (we act better when watched). Several said a watching person beats a small payout. That claim did not stay a hunch. Social validity data (what users say about the plan) lined up with it. One woman said she had nobody in her life to keep her on track. Her follow-up steps were the lowest of the three.

This is an internal validity issue (a hidden cause of the change). The paper's own discussion raises it. Feedback from the team may have added push. The first meeting also covered why activity matters, which could lift behavior on its own. There is a clean way to test it. Run the same contract with no human in the loop. Automate the check and the payback, then compare.

The internal validity is when confounding variables may have an effect on the dependent variable rather than the independent variable.

From the talk — Pauli Catanese

Wrist-Worn Step Counts Are Not Clean Data

The team did check the watch. Someone walked a set number of steps to test the count. The write-up does not say whether the arm was swinging. That detail matters. Arm swing is what the watch reads.

The room poked more holes. A watch worn on the ankle logs more steps than one on the wrist. Push a cart or carry a box and the count drops. Sit on the couch and swing your arm and the count climbs. People do exactly that when a step alert goes off.

The push-back was also fair. Someone who put her own money in has little reason to cheat. Buy-in guards the data a bit. Still, treat any watch count as an estimate. If steps are your clinical measure, add a second source. At the least, name the limit out loud in your notes.

they said they tested the step count on the apple watch by like having someone engage in a specific amount of steps to see if it was accurate but it didn't say if their arm was swinging or not

From the talk — Pauli Catanese

Taking Deposit Contracts Into Parent Training

Parent training was the first clinical fit the group named. The plan works at a distance. It runs over weeks. Many parents do want to change what they do at home. What they lack is a push and a check-in.

The money part is where the ethics bite. Do not hold a parent's cash. The room was clear on that. You can still keep the rest of the plan. Set the goal with them. Graph their data. Tell them you will look at it. Then follow up on the day you said you would. If a deposit helps them, let a friend or family member hold it.

For kids the group got creative. The deposit could be a toy, a snack, or a show they love. It has to be something they already own and want. But the teeth of the plan come from real risk. They have to be able to lose it. That is a hard ask for a learner who cannot agree to the terms. So this tool leans toward people who choose the change themselves.

One upgrade from the room is worth stealing. A step app pays a bonus for going past the goal. Beat it all week and you enter a draw for a bigger prize. You can get back more than you put in. The same trick scales down to candy. Trade a few pieces back for a shot at more.

It just has to be something motivating that they want back.

From the talk — Pauli Catanese

Reinforcement Thinning You Already Run Every Day

Thinning is the part we all do, often without a written plan. A token economy (earn tokens, trade them later) starts on every single response. Then it moves to every third, then every fifth. Then it goes variable. The goal is behavior that holds on a thin schedule.

The group listed more cases fast. Manding (asking for what you want) starts with a payoff every time. Then fewer. Then the natural payoff takes over. Shaping (reward closer and closer tries) runs the same way with sounds and signs. Desk time grows that way too: praise at 30 seconds, then a minute, then two. New staff get praise after every session at first, then just a drop-in.

The study used the same ladder: daily, then every three days, then weekly. Copy the structure, not just the idea. Pick your step size before day one. Pick the rule that moves you up a rung. Pick what you will do if the behavior dips, as it did when one woman got sick.

One last point closed the hour. Sitting still has no quick cost. Getting up has no quick payoff. The harm shows up years later. So the plan has to build a reason to move today. In our terms, you are setting up a motivating operation (a reason to want it now). Your clients face the same math with sleep, meds, and food logs.

it's the long-term effects of not engaging in this behavior

From the talk — Pauli Catanese