On the relationship between self-injurious behavior and self-restraint.

The team ran a functional analysis on three adults with intellectual disability. Each person showed both self-injury and self-restraint, like wrapping their arms inside shirts.

The analysts tested whether the same events kept both behaviors going. They ran the usual FA conditions: alone, attention, demand, and toy-play.

They watched to see if self-restraint rose and fell with self-injury under the same condition.

For some participants, self-restraint and self-injury tracked together. If escape from demands made hitting go up, restraint also went up.

This match meant the two behaviors belonged to the same functional class. The restraint was not random; it was part of the problem skill set.

Holehan et al. (2020) later showed that you can use either isolated or mixed contingencies and still get clear FA results. Their work keeps the 1992 message: test the actual contingencies, don’t guess.

Irwin Helvey et al. (2022) warned that synthesized contingencies can hide true functions. Their caution lines up with G et al.—if you lump behaviors together without checking, you might miss that restraint serves its own purpose.

Matson et al. (2004) added that who runs the FA changes the data. Caregiver-run sessions looked different from stranger-run ones. So the 1992 findings hold, but you must think about the therapist’s identity when you repeat the test.

Before you try to block or fade self-restraint, run a quick FA. If restraint covaries with SIB under escape, teach a safe escape response instead of simply banning the wrap. You might turn one behavior into two replacements and still meet the same need.