TRADITION

Adaptation → constructed structure → later competence

Reference point

Jean Piaget's developmental account treats intelligence as something constructed through successive adaptation, rather than as a finished stock of answers.

Resonance

Arina begins with small domain-neutral mechanisms and accumulates distinctions, relations, concepts, and operations through staged encounters. Later work must preserve or revise what earlier stages earned.

Departure

Arina is an engineered research system, not a model of childhood or biological development. Its stages are experimental and governed rather than a claim about universal cognitive development.

Relational alignment → transfer → concept revision

Reference point

Dedre Gentner's structure-mapping theory and the Copycat tradition place relational structure and flexible analogy near the centre of cognition.

Resonance

Relatia and Arina both privilege relations that survive changes of label or surface form. Repeatable exaptation asks whether a structure earned in one setting can do real work in another.

Departure

Arina requires explicit evidence boundaries, preserved baselines, and independent evaluation. It is not presented as an implementation of structure-mapping or Copycat, nor as a validated model of human analogy.

Anonymous signal → consequential distinction → public handle

Reference point

Stevan Harnad's symbol-grounding problem asks how symbols acquire meaning beyond relations to other symbols. Rodney Brooks pushed AI toward incremental intelligence coupled to perception and action.

Resonance

Arina treats names as public handles rather than meanings. A distinction becomes useful through predictive, operational, communicative, or normative consequences, and the body project tests whether those relations survive distributed action.

Departure

Current grounding is narrow and controlled. It does not establish open-world sensorimotor grounding, human-like semantics, or the representation-free intelligence Brooks proposed.

Prior knowledge + experience → skill acquisition → unfamiliar problems

Reference point

François Chollet's account of intelligence centres the efficiency with which a system acquires skills across tasks, rather than prepared-task performance alone.

Resonance

Arina likewise emphasises unfamiliar problems, constrained resources, declared priors, and whether learned structure transfers beyond the setting that produced it. This is the closest quick way to place the evaluation instinct behind the project.

Departure

Arina is a separate architecture and evidence programme. It is not an ARC system or result, and resemblance to Chollet's framing does not validate Arina's broader claims.

Observed failure → proposed repair → independent court → governed continuation

Reference point

Jürgen Schmidhuber's Gödel machine is a rigorous reference for systems that reason about improving their own problem-solving machinery.

Resonance

Arina can treat a failed improvement process as another object of diagnosis and propose a repair to the process that generated the earlier repair. A longer governed mathematics sequence extends that recursive pattern across many successive research stages.

Departure

Arina cannot authorise its own continuation. Sandboxes, independent evaluation, budgets, and promotion remain external; no optimality, autonomous rewriting, acceleration, or intelligence explosion is claimed.

Developmental cognitive architecture · relational representation · transfer-first evaluation · governed self-repair

Closest short description

A small developmental, domain-neutral learner investigating whether one governed line of relational learning can accumulate and reuse increasingly general cognitive operations.

Not a claim

This map does not make Arina an ARC entry, brain simulation, unrestricted theorem prover, universal reasoner, autonomous self-authorising system, or finished AGI. It locates the questions, not the conclusion.