MDPE.Application:5.3.8 - Where an epiplexity analogy is useful

Computational epiplexity, as treated in C.2.8, uses a bounded model class and a two-part coding criterion to distinguish model structure from residual unpredictability. A mathematical application must state what represents the account, observer, prior knowledge and extraction operation.

For method acquisition, a comparable research question would ask which reusable organization of action a specified learner can acquire from specified material and assistance within a budget, and how that acquisition changes later performance. That question is useful without an established universal numerical characteristic.

Begin with the explicit structure and acquisition contrasts above. Reusing C.2.8 for description recovery and capability-development methods for learning preserves their different bearers and evidence. A new general characteristic, or a conditional epiplexity model of learning, needs additional construction: units of Method structure, correspondence to the learner’s representation, an acquisition criterion, time and assistance conditions, and an account of transfer. A count of instructions, perceived intricacy or hours of lessons supplies none of these by itself.

An agent can extract more already familiar structure without learning anything new. Learning can also change future extraction. That reciprocal relation is one reason to keep prior preparation and the timing of the comparison explicit. It is also a reason not to assign a single intrinsic “learnability” number to a Method independently of its learner and learning arrangement.