Self-restraint as positive reinforcement for self-injurious behavior.

The team worked with adults who had intellectual disability and severe self-injury.

They used an alternating-treatments design to test if self-restraint itself could reward hitting or head-banging.

In one condition the person got a jacket or towel to hold right after self-injury.

In the other condition the same items were taken away and never given.

Sessions flipped back and forth to see if self-injury tracked access to restraint.

Self-injury shot up when restraint followed the behavior.

It dropped to near zero when restraint was unavailable.

The pattern showed that self-restraint acted like candy or praise — a positive reinforcer — not just a safety habit.

Kocher et al. (2015) later used this idea to split automatically reinforced self-injury into three subtypes.

They showed that only the group who use self-restraint need treatments that specifically replace or limit that reinforcer.

Hatton et al. (2005) seems to disagree — they found that social attention, not automatic reinforcement, drives early self-injury.

The clash disappears when you note age: toddlers may start for attention, but adults can shift to automatic sources like restraint.

Dawson et al. (2025) put the whole package to work: they identified a teen as subtype-3, swapped in safe “alternative restraint,” and cut both injury and mechanical holds.

If your client wraps a blanket, hugs themselves, or seeks jackets right after self-hits, you may be looking at a built-in reward.

Do not hand over soft restraints as comfort during crises — you could accidentally feed the problem.

Instead, run a brief test: remove the item, offer a non-restraint sensory toy, and track the data.

Use the results to pick the right subtype and build a plan that gives safe sensory input without the dangerous wrap.