The use of conditional probability analysis to identify a response chain leading to the occurrence of eye poking.

The team watched one adult who poked his own eyes. They used conditional probability analysis. This means they checked if eye poking was more likely right after certain stereotypy.

They guessed that the stereotypy came first in a chain and led to the eye poking. To test the idea, they later blocked only the early stereotypy.

When staff blocked the early stereotypy, eye poking also dropped. The person needed less help and the whole chain slowed down.

The simple block worked better than trying to stop the final self-injury itself.

Hagopian et al. (2000) tried the same kind of blocking five years earlier. They saw mixed results: blocking helped two people but not the third. The new study shows that blocking works best when you first prove the responses sit in the same chain.

Lang et al. (2009) flipped the idea. They gave free stereotypy before play time and saw less of it later. One paper blocks the chain; the other drains its value. Both cut stereotypy, but in opposite ways.

Fruchtman et al. (2025) now offer a fuller plan. They start with a performance-based IISCA, build a skill package, and remove dangerous behavior for good. Their method replaces simple blocking with a full function-based treatment.

You can test any odd movement the same way. Record what happens right before self-injury. If the data show a chain, block the first link and watch the last one fade. If the case is tough, move to a full IISCA and skill plan like Fruchtman et al. (2025) describe.