| Holder and task are named | “The audience should recognize the style.” | Name the sampled people or Agent, recognition task, context, response, and later use. |
| Distinction is grounded | A label or representation becomes the thing recognized. | Name the world-side sound, movement, rhythm, interaction, or relation and how it is accessed. |
| Recognition differs from performance | A wrong movement is treated as proof of failed perception. | Test the smallest recognition response separately, then test performance capability if needed. |
| Cue leakage is controlled | The teacher, costume, file name, or fixed order gives the answer. | Match irrelevant cues, fade prompts, randomise order, and use held-out examples. |
| Variation serves transfer | More contexts are added without a receiving use. | Vary conditions expected in later Work or needed to distinguish rival explanations. |
| Context-bound access is tested | Reappearance in the old context is called relearning. | Return to the old context before reteaching and vary one relevant cue at a time. |
| Human and machine claims retain their kinds | Classifier accuracy becomes cultural understanding. | State the task, holder, data, conditions, error distribution, and non-use boundary. |
| Recognition differs from selection | A capable recognizer is treated as authorised judge or curator. | Pass capability evidence to MDPE.13; establish authority and selection separately. |
| The distinction can be changed | Repetition continues despite ambiguous or inaccessible categories. | Revise the signal, representation, boundary, population, or intended use; abandon the distinction when needed. |
| Direct Methods remain primary | A cross-domain pattern replaces ear training, Dance pedagogy, partner sensing, or ML evaluation. | Use each direct Method for its result and retain only the cross-result Music-and-Dance decision here. |