MDPE.23:2.5 - Probe context-dependent access before declaring loss

The COIN model treats sensorimotor adaptation as contextual inference over several memories. Broader work on memory also reports context dependence, but one common computational account for all kinds of memory remains an open hypothesis. Use the hypothesis when it changes a cheap discriminating probe.

When a learned response disappears:

1. return briefly to the earlier successful context without reteaching;
2. check whether the response reappears;
3. vary one cue, partner, material, venue, task, or transition at a time;
4. compare blocked and interleaved tasks and an unfamiliar but relevant setting; and
5. record recovery, interference, unintended intrusion, deliberate transfer, and delayed retention separately.

Immediate return in the old context supports a retrieval or context-classification problem over simple capability loss. Finding no context effect supports a different explanation. Use either result in the next development decision; retain the current Method when it remains adequate.