fpf.tools
Library / Music and Dance Practice Engineering Principles Framework
Jump to passage
In this reading
Source · lines 3380–3418

Link to current text

Published source confirmed at last check

Source changed 2026-10-03 07:42:37 UTC · snapshot created 2026-10-03 07:43:27 UTC · last check 2026-10-03 07:45:10 UTC

MDPE.23:1 - The Recurring Difficulty

A physical difference can exist without being available to a particular observer. An observer can discriminate two examples without naming them, can name them without knowing when the distinction matters, or can recognize them in one classroom context and fail in performance. Recognition is therefore not one undifferentiated mental event.

For this pattern, recognition capability belongs to a named holder and a named family of recognition tasks. The holder can be a musician, dancer, partner, teacher, listener, curator, judge, performing Agent, or machine recognizer. A population result concerns a distribution of performances by sampled members unless the population is also an organised Agent performing collective Work.

Useful recognition tasks include:

  • discriminate whether two occurrences differ in a stated respect;
  • identify which distinction or category obtains;
  • locate a difference in time, space, body, sound, or interaction;
  • predict the next relevant change;
  • notice that a learned coordination, concept, or Method applies in the current situation; and
  • act on the recognized difference without the training cue giving away the answer.

These tasks can rely on different capabilities. A dancer can hear that two rhythms differ yet be unable to locate the event that changes partner timing. A teacher can name a style but fail to distinguish its enacted movement from a costume cue. A classifier can label audio without understanding, authority, or cultural participation.

Three failures are especially costly:

  1. Cue dependence. Training and test share a teacher, label, costume, recording, partner, or fixed order, so the cue rather than the intended distinction controls the answer.
  2. Context-bound retrieval. The capability appears lost when music, partner, venue, body state, task, or recent transition changes, although the earlier response can return in its old context.
  3. A bad distinction. Intended recognizers cannot reliably access the signal, the categories overlap for the receiving use, or the representation omits the relation that matters. More repetition then trains noise or a proxy.

Recognition must also stay separate from selection. A listener may recognize a variant and reject it. A platform may amplify material that its users do not recognize as intended. A judge can recognize a difference without the authority to define a culture. Demonstrated recognition supplies evidence to MDPE.13; it does not make the later selection decision.