High-probability request sequence

Escape
Delivering several requests the learner reliably completes before presenting a difficult or non-preferred request.

A high-probability request sequence - often called behavioural momentum - involves presenting three to five requests you know the learner will comply with before delivering the one you expect them to refuse. You build a run of successful responding and reinforce each one, then use that momentum to carry the learner through the harder demand.

The behavioural explanation is that a history of reinforcement for responding to your instructions increases the probability of responding to the next instruction, even if it's harder. Think of it like a running start.

You use this when a learner has a specific demand or transition they consistently escape from, and the refusal happens immediately on contact with the instruction - before they've even tried it. The sequence interrupts that pattern by establishing responding as the current activity before the difficult request arrives.

The critical implementation detail is pace. The requests must come in rapid succession with brief, enthusiastic praise in between. If you pause to write data or check your phone between requests, the momentum dissipates. Another mistake is choosing 'easy' requests that the learner actually hesitates on. If compliance isn't immediate and reliable, it's not a high-probability request.

Implementation

  1. Identify 3-5 instructions the learner complies with immediately and reliably.
  2. Deliver them in quick succession, praising enthusiastically after each.
  3. Immediately follow the final high-probability request with the target low-probability request.
  4. Provide highly preferred reinforcement when the low-probability request is completed.
  5. Vary the high-probability requests across sessions to prevent predictability.

Common Mistakes

  • Pausing between requests and allowing momentum to dissipate.
  • Using requests the learner doesn't actually comply with quickly and consistently.
  • Making the low-probability request too far removed in difficulty from the high-probability ones.