Theoretical and Pedagogical Reference Framework (Marco de Referencia)

What this is. Where Smoother comes from. A walk through the pedagogical tradition — from antiquity to the contemporary models of instructional design — and, at every stop, an explicit statement of what Smoother takes from that framework, how it uses it, and where it differs. Smoother rests as much on decades of practice organizing learning experiences as on that body of theory and on what new technologies make possible.

From antiquity to the twentieth century

The document starts with the Trivium (grammar, logic, rhetoric) and the Quadrivium (arithmetic, geometry, music, astronomy), and reads them as modern reference models: not as a syllabus to reproduce, but as an articulation of how thinking is built and how quantitative knowledge is structured. It continues with the pedagogical proposals of the nineteenth and twentieth centuries, and with the contributions that fixed the vocabulary of instructional design:

• Skinner — the science of learning and the art of teaching.
• Mager — observable learning objectives and instructional sequencing.
• Bloom — the taxonomy and its domains. Smoother uses it as a classification, not as a hierarchy of quality: the cognitive level of a task does not measure the value of the craft with which it is done.
• Gagné — his taxonomy of outcome types, and his nine events of instruction.
• Fleming (VARK), Felder-Silverman, Gardner (Multiple Intelligences) and Kolb — the traditions from which Smoother’s Personal Preferences (Preferencias Personales) come, used as conceptual reference and never as proprietary instruments.

Dewey and Parker — tradition, never outcome evidence

John Dewey and the Laboratory School. In January 1896, in a small house on Chicago’s South Side, the elementary school that would become the University of Chicago’s Laboratory School opened its doors. The name is the argument: a school conceived as a laboratory, where a pedagogical hypothesis is put to work with real people, observed and corrected. In The School and Society (1899) Dewey left the comparison that has outlived everything else: the difficulty of finding school furniture meant for someone to work at rather than merely to sit and listen at, and the row of identical desks as the visible diagnosis of what was expected to happen there.

What Smoother takes, from the learner’s side. Three things, and none of them is new: they enter as convergence with instruments that already existed.

• Learning by doing → the whole, authentic task. The unit of design is the real task, not the fragment of content — which Smoother already holds from 4C/ID’s component 1 and from the six Learning Tasks (Tareas de Aprendizaje). Dewey is its ancestry, not its empirical support.
• The school as a community → the Environment as a pole of the triad Subject–Object–Environment (Sujeto-Objeto-Entorno), and the requirement to type the Contexts before deriving anything.
• Inquiry as method → the Scenarios and Critical Analysis, and Active Learning, which is how that position reached the twenty-first century.

Francis W. Parker. Dewey called him the father of progressive education, which is why the genealogy Smoother recognizes starts before Dewey. The Francis W. Parker School was founded in Chicago on 7 October 1901, on land donated by Anita McCormick Blaine, and is still running. Its position — centering design on the person who learns rather than on the subject matter, and treating that person as a whole unit rather than as a collection of subjects — converges with two things Smoother already had and reached by its own route: the learning-first guiding principle — “the learner’s experience is the subject; teaching is in service of it” — which is a project-identity principle and a personal conviction of Carlos Miranda Levy, with Parker as a historical convergence and not as its foundation; and the multi-component profile in evaluation, which structurally prevents reducing a person to a single grade.

The honesty caveat. Both enter as design tradition and intellectual ancestry, never as outcome evidence. The Laboratory School was an authored experiment, not a controlled trial, and nothing here licenses saying that a practice works because Dewey proposed it; it is the same register in which the canon treats Papert and Resnick. And a point of vocabulary: when Dewey or Parker are quoted, their words are kept — they say child, pupil, student — because Smoother’s vocabulary rule governs Smoother’s own voice, not the voice of the sources it cites.

Learning styles, and what the evidence says

Pashler, McDaniel, Rohrer and Bjork (2008) reviewed the empirical base of the meshing hypothesis — that teaching each person according to their preferred style produces better learning — and found no solid evidence supporting it. Preferences exist; what does not hold is that matching instruction to them improves outcomes.

That is why Smoother’s catalog is called Personal Preferences and not “learning styles”. Attending to them makes the experience more smoother — more appealing, more comfortable, more memorable — but not necessarily better or more effective learning: those are different things and we do not confuse them. Preferences work as a menu of variety for everyone who is learning, never as an individual prescription that narrows what is on offer. This is a Smoother design position, authored by Carlos Miranda Levy, and the Personal Preferences document holds it in full.

One caveat about Kolb: what this framework takes with confidence is the experiential cycle; the style dimension derived from it is contested in the literature, and Smoother uses it as a derived view and a sequencing recommendation, never as a permanent label on the person who is learning.

Eight modern models, and what Smoother takes from each

The heart of the document is a declared mapping: which concrete part of Smoother reflects each model.

Stanford’s d.school — abilities, not a recipe

Stanford’s Hasso Plattner Institute of Design — the d.school — belongs to the same institutional lineage as P⁵BL, and what Smoother takes from it is not what circulates under its name. What circulates is a recipe: five stages — empathize, define, ideate, prototype, test — run in order, once, with sticky notes. What the institution holds today is something else: design thinking understood as a set of abilities and of non-linear iteration, not as a rigid sequence. That retreat from the linear recipe is recorded in the canon with its declared strength — moderate, secondary verification — and adopted as Adapted, never as a finding.

What Smoother takes. Abilities, not a recipe: design thinking is one of the frameworks a team may declare at the Process Gate (Compuerta de Proceso) of the Challenges, alongside its own named way, which is always valid — and there is the hard consequence, the deliberation about adoption is scored, fidelity to the framework never is. Prototyping and iteration, which converge with Product-Based Learning, the Project Guides (Guías de Proyectos) and SAM: it is convergence, Smoother already prototyped. And the framework museum, prohibited — the anti-ceremony prohibitions exist precisely against the ritual version of design thinking, and that a framework is welcome as a declared option and forbidden as liturgy is how this canon adopts it.

The honesty caveat. The retreat from the linear recipe is an institutional position of moderate strength, not an experimental result, and it is not cited as though it were one. It is not claimed that design thinking produces better learning outcomes than its absence: there is no verified evidence here that would support that. What is claimed is more modest, and does hold: treating it as a recipe is a reading its own institution has been abandoning.

Active learning, and the case of 4C/ID

Active learning involves the person who is learning directly, through discussion, problem solving, case analysis and collaborative projects, and promotes critical thinking and practical application over memorization. It rests on two maxims: commitment to the learning and reflection on the learning.

4C/ID deserves separate treatment because it is the model with which Smoother shares the most machinery. Its four components map one to one; in particular, its part-task practice corresponds to the Riddles and Puzzles (Acertijos), microlearning and nano-learning, which exercise concrete sub-skills to fluency — useful when the automation of a routine component is blocking progress on whole tasks, and never as a replacement for whole tasks. 4C/ID’s separation between supportive information (conceptual models, for reasoning) and procedural information (step-by-step instructions, for execution) answers directly to the difference between germane and intrinsic load in cognitive load theory.

Where Smoother differs or extends. While 4C/ID concentrates on how an instructional sequence is designed, Smoother combines it with (a) the catalog of Personal Preferences as an additional criterion of coverage and variety; (b) the evidence-grounded principles of Memory and Learning, which discipline how information is presented inside each component; and (c) the Task Routers, which select appropriate tasks according to the multidimensional context of the person who is learning, the structure of the Learning Unit and the modality.

What Smoother has adopted since

The historical walk ends at 4C/ID. What came after lives in other documents, and this framework maps it without explaining it a second time: each piece keeps its canonical document and its label.

• Memory and learning (research-backed) — retrieval, spaced practice and interleaving, the family of desirable difficulties; cognitive load and faded worked examples, which work for novices and attenuate as expertise grows; Mayer’s coherence principle, which argues against seductive details; productive failure (Kapur), adopted with the standing constraint that the evidence refutes minimally guided discovery (Kirschner, Sweller and Clark, 2006) — scaffolding must exist and then fade; feedback quality (Hattie and Timperley); deliberate and purposeful practice (Ericsson), with the replication debate about the magnitude of its contribution said out loud; and Vygotsky’s zone of proximal development.
• Creativity and artifact-making — Amabile’s Consensual Assessment Technique as the judging method, with its limits: it requires appropriate judges and its validity rests on the expertise of whoever judges; and the four-C model with its domain-specificity caution. Papert’s constructionism enters as a named design tradition, never as outcome evidence: “wide walls” is Resnick and Silverman’s own addition, and the pairing “low floor, high ceiling” is a Logo-lineage design tradition that is not locatable on a page of Mindstorms.
• AI in learning (mixed label) — the distinction between performance and learning: the “crutch effect” the OECD names in Digital Education Outlook 2026; Bastani et al.’s (2025) randomized field experiment in secondary school, where unrestricted access left the group worse off than the control while a scaffolded tutor largely removed the harm; the “metacognitive laziness” of Fan et al. (2024); and the homogenization of Doshi and Hauser (2024), individual gain paid for in collective diversity. AI competency frameworks — UNESCO, AI4K12, OECD/EC — are referenced, never copied, and none of them is binding.
• Teaching-side frameworks, read from the learner’s side — Two contemporary frameworks whose object is the practice of the person who teaches entered in 2026-09, and they entered turned around, because the learning-first guiding principle requires exactly that: even when the topic is teaching, it is approached from the learner’s side. Both are registered and referenced, never copied — native identifiers and paraphrases authored by us, zero owner text — and none of their nodes is binding on delivery; their evidence labels are those of their own literatures and are not mixed with Smoother’s. The OECD Teaching Compass (2025), which declares itself an aspirational framework and presents itself as an enabler of the Learning Compass 2030 — already present in Smoother as an outcome axis of classification, and therefore referenced rather than registered a second time — converges with the General Learning Definition, with the boundary of the Learning Facilitator, and with the Enable and Connect stages of the CEMI Impact Arc, which remains a design position of Carlos Miranda Levy and which this convergence does not upgrade; one node is declared without convergence, because Smoother today has no instrument at all on the professional well-being of the person who accompanies, and the registry says so rather than fabricating a correspondence. The 5 Dimensions of Teaching and Learning™ (5D™) of the UW Center for Educational Leadership — a registered trademark and an all-rights-reserved work — contribute five convergences, every one of them with instruments that already existed: Purpose ↔ the Learning Object’s Objectives to Achieve; Student Engagement ↔ the six Learning Tasks; Curriculum & Pedagogy ↔ the eight Unit Structures and the fading of scaffolding; Assessment for Student Learning ↔ the analytic rubric anchored to the Learning Object; Classroom Environment & Culture ↔ the Environment pole and the Guidance Guide, with its verified scope said out loud: the physical K-12 classroom, with no transfer claimed to asynchronous or remote settings or to adult learning. And a limit that is one of identity, not of licence: the instrument that made that framework famous is a rubric for evaluating the performance of the person who teaches, and Smoother does not have that instrument and does not want it — Smoother evaluates learning, with a rubric anchored to the Learning Object and a multi-component profile, and does not rate the Learning Facilitator; the Gates verify deliverables and conditions, never a person’s performance.
• Deliberate practice at scale — §12 of Memory and Learning is already canon: expertise is built through structured practice at the edge of capability with informative feedback, with the replication caveat the canon records — magnitude under debate and varying by domain, and the “10,000 hours” as a simplification. On that base, and without touching it, an authored essay is registered: What AI Finally Makes Possible: Deliberate Practice at Scale, by Carlos Miranda Levy. Its thesis is that what AI makes possible for the first time is not content but volume of varied practice with feedback, personalized within a single group, and that the consequence is one of equity: deliberate practice used to be unavailable wherever no expert person was available to give the feedback. It is a Smoother design position, authored by Carlos Miranda Levy, a sibling of the Guiding Position “The Almighty Gym”. It is not new evidence and is not presented here as such: the text declares itself a working draft and cites no studies, and nothing it asserts modifies §12’s label or its caveat about magnitude. Permitted wording: “design position”, “aligns with”, “converges with”; prohibited: “evidence-based”, “derived from”, “proven by”, and any efficacy figure attributed to the essay.
• Onboarding — a systematic review of formal onboarding interventions (Frögéli et al.) found promising effects but rated the overall certainty of the evidence low; hence the model rests on mechanism evidence plus local measurement, and promises no outcomes.
• The design positions, said as such — the Guiding Position “The Almighty Gym”, the Impact Arc (Engage → Enable → Inspire → Empower → Connect), the essay “Deliberate Practice at Scale” and the Experience Layer are Smoother design positions, authored by Carlos Miranda Levy — the Arc created by him and inherited by CEMI as a group. None is an evidence-based framework and none is presented as validated; where they converge with researched constructs we write “aligns with”.

Evidence

This document has no entry of its own in Smoother’s evidence index, and it is right that it should not: it walks through frameworks that belong to other people and other traditions, and each one keeps the label of its own literature. What the index labels is what Smoother builds on top of them — the Learning Object, the six Task Types, the eight Unit Structures, the Personal Preferences, and the rest of the typologies, each with its declared level.

What is our own work — the curation, the mapping of each framework to a concrete instrument, and the critical treatment — the document declares in its own metadata as synthesized.

Three disciplines in this document are a position of our own, and are stated as such: using Bloom’s taxonomy as a classification and not as a hierarchy of quality; the Pashler posture on preferences; and explicitly excluding debunked practices — matching by learning style, the learning pyramid with its fabricated percentages, and the others that Memory and Learning enumerates.

Acknowledgement. The curation and critical treatment of the frameworks covered here draw on Gareth Manning’s open-source education-agent-skills library (CC BY-SA 4.0), whose reading is recommended as a rigorous complement to this historical framework.

Related

Memory and Learning · Principles of Instructional Design · Personal Preferences · The Learning Object · Learning Units · Task Routing · Evidence Framework

Source

protocol/smoother-marco-de-referencia.md — version Unreleased (2026-09-19).