Product Decisions

The Learning Curve developmental ontology covers birth to 60 months as its primary scope. 60–72 months is a transition zone where developmental and early curriculum overlap. Above 72 months is a future adolescent framework, not yet built.

60 months (5 years) is the consensus cap across the major developmental assessment frameworks in use today:

At 60 months, formal schooling begins and the primary acquisition mechanism shifts from informal home scaffolding to structured instruction. The Stanford SHQ dataset extends to 72 months, which is why a transition zone exists — but the developmental claim weakens above 60 months and the canonical schema reflects that.

A skill belongs in the developmental ontology if a parent can scaffold it through everyday home activity without following an instructional sequence. It belongs in the curriculum content library if it requires structured, sequenced teaching to acquire.

The criterion is mechanism of acquisition, not age.

• Developmental: participatory-cultural acquisition — skills that emerge through environmental exposure, shared routines, and guided participation. No instructor required.
• Curriculum: instructional-didactic acquisition — skills that require teacher-designed, sequenced instruction to develop.

Applied to literacy

Applied to numeracy

Research basis: Teale & Sulzby (1986) emergent literacy; Snow, Burns & Griffin (1998); NELP (2008); Gelman & Gallistel counting principles (1978); Clements & Sarama learning trajectories (2004–2023); UNICEF ECDI2030 Technical Manual (2023); Blair & Raver (2015).

Learning Curve will not store, surface, or use sex or gender as a variable in any recommendation, scoring, or content classification system.

Research confirms that sex differences in early developmental timing exist — roughly 1–2 months on some milestones in the 0–36 month window — but the reasoning against encoding this is clear:

1. The effect size is small relative to individual variation. A sex-adjusted window would move milestone display by ±1–2 months while child-to-child variation spans ±6–12 months.
2. Surfacing sex-differentiated milestones to parents risks reinforcing stereotypes and creating self-fulfilling developmental gaps.
3. The mechanism driving observed sex differences is partially socialization, not biology — which means coding it into the product reinforces the cause.
4. Parents already know their child's sex. Adding it to the system adds no new information but adds significant ideological risk.

Instead, Learning Curve uses a movement_required format dimension (high / medium / low) on activities, with a parent-toggled “active learner” preference. This captures the practical relevance of the finding — some children prefer high-movement learning — without demographic encoding. The preference is temperament-based, not gender-based.

Activities target milestones. Dispositions — curiosity, persistence, empathy, creativity — are observed across activities by the Guide Agent. Learning Curve will never surface an activity as “targeting” a disposition.

Dispositions are not discrete skills that can be trained in isolation. They are character tendencies that emerge from the full context of a child's experience. Treating curiosity as a learning objective to “achieve” through a specific activity misrepresents how dispositions develop and risks reducing them to a checklist.

Research basis: Katz (1993) on dispositional learning; Carr (2001) on learning stories.

The three parent-facing categories are Body, Mind, and Heart. The internal field values are physical, cognitive, and social_emotional. Internal field names are never exposed on parent surfaces.

Six subcategories were added in May 2026, as an additive field that does not change existing behaviour or routing:

Items without a subcategory fall back to the parent category.

The Developmental Gantt and all milestone window displays source exclusively from the canonical skills file (approximately 2,135 items, of which around 1,011 are milestones). The legacy framework extraction file — which contains 1,823 unaggregated, framework-specific items with overlapping and conflicting windows — is never used for display.

The canonical file is the output of a normative calibration pipeline that produced empirically validated windows, mastery spectra, and zero true domain gaps. Using the legacy file would regress all window quality work done in that pipeline.

The static activity library is the primary content source. On-demand generation is a fallback only — triggered when a child's frontier milestone has no matching static-library activity.

The fallback chain is: check for a cached generated activity less than 7 days old, then run a generation job, then fall back to a semantic match from the static library. Generated activities are cached for 7 days and labelled “Personalised for [name]” in the Learning Curve interface.

Authored activities have been reviewed for developmental accuracy, age appropriateness, and language quality. Generated activities are best-effort and may have errors. The static library is always preferred when a match exists.