Attention is where identity is built

iOSE is a framework for working with attention and orientation, developed by Eskil Frydenberg — clinical psychologist and Arfur's Chief Education Officer — out of clinical practice. The reasoning below, and the research it draws on, is Arfur's account of why the framework fits what we build.

It describes a short, repeatable cycle: pause before the next demand; presence, noticing where attention actually is, without judgment; orient, identifying what matters now; and one grounded next step. The commitment underneath it is as much a design commitment as a clinical one — technology should return attention to a person rather than take it from them.

iOSE is also being developed as a separate product at iose.io, under the same parent company. Arfur uses the framework, not the product. Nothing in Arfur's storytelling universes is a clinical tool, and iOSE is not sold, embedded, or delivered through Arfur.

Attention research does not treat attention as a single resource that is either present or absent. In the framework most widely used in cognitive neuroscience it is at least three interacting networks: alerting, which sets readiness; orienting, which selects what gets in; and executive control, which holds a goal in place against everything competing with it. These systems change substantially across childhood, and children differ from one another in how efficiently they run.

The practical consequence for anyone building for children is that capturing attention and supporting attention are different engineering problems with different solutions. A product tuned to the alerting and orienting reflexes — motion, novelty, sound, a next item already playing — can hold a child's eyes for a long time while giving the executive network almost nothing to practice on. That is a design choice, not a property of screens.

Long before a child can reason about what another person thinks, they practice holding attention together with an adult on a third thing: a bird, an object, a page. This joint attention is not a by-product of social development — coordinating one's own and another's attention appears to be part of what social cognition is built out of.

It is also the shape of shared reading. Across studies from infancy to early adulthood, how much children are read to and read themselves shows moderate to strong associations with comprehension and reading skill, in a reinforcing spiral rather than a one-way effect. An earlier meta-analysis found the same for the youngest: how often parents read with preschoolers is related to language growth, emergent literacy and later reading achievement.

So the unit Arfur designs for is not a child and a screen. It is a child, an adult, and a story held between them.

The popular claim that heavy media multitasking damages attention is weaker than the headlines suggest, and we will not lean on it. The original finding — that heavier media multitaskers were more susceptible to interference from irrelevant information — has not replicated cleanly. Direct replications found a small association at best, and a meta-analysis of 118 assessments a decade later put the pooled effect at small. A review of the field reads the balance of evidence as pointing to lapses of attention rather than a broad deficit.

What is far better established is the cost of the switch itself. When a person moves from one task to another, part of their attention stays behind on the first — attention residue — and performance on the new task suffers. The cost is measurable, it is incurred per interruption, and it accumulates.

This is why Arfur designs for uninterrupted, bounded attention: not because screens have been proven harmful to children, but because every interruption has a price and continuity has a value we can actually name.

In contemporary developmental and personality psychology, identity is understood substantially as narrative: a person builds an internalized, evolving story of the self that gives their life unity and purpose. That story is constructed with other people, revised over time, and assembled from the cultural material available to tell it with.

Read alongside the attention literature, this has a direct implication. Making a coherent self is not a lookup — it is composition, and composition needs exactly the conditions fragmentation removes: stretches long enough to follow a thread, repetition, retelling, and an adult present enough to listen and ask. Attention continually reclaimed by something else is attention not spent on that work.

This is where the framework's two halves meet. Attention is not only how a child learns; it is the medium in which a child assembles who they are.

Adolescents who know more of their family's history — where their grandparents grew up, what their parents were like as children, what the family has come through — score better on measures of self-esteem, internal locus of control and family functioning, and lower on anxiety and behavior problems. The authors of that work are explicit about what it does and does not mean: reciting facts does not produce the effect. The knowledge is an index of families in which stories are actually told.

The mechanism they propose is the intergenerational self: the child places their own life inside a line that began before them, and that placement is a source of stability.

For children growing up between cultures the pattern extends. In meta-analysis, a positive and settled sense of one's ethnic and cultural identity shows small-to-medium associations with adjustment, wellbeing and academic outcomes. And in acculturation research, the poorest outcomes are consistently associated with marginalization — connected to neither the heritage culture nor the surrounding one.

These are correlational findings, largely from North American samples, and they license no causal promise. They do justify a design priority: make the inherited story available, in the child's own languages, in a form an adult can share with them.

Most children's media is paid for with attention, and that is not a figure of speech. In a study of the apps used by 160 children aged three to five in the United States, only one app in five had no manipulative design features. The rest used pressure from characters, invented time pressure, navigation that makes leaving hard, or lures to play longer or buy more — and children from lower-income families met more of it. An earlier content analysis of 135 apps for children under five found advertising in 95 percent of them, from pop-up videos that interrupt play to characters urging purchases.

The heritage language is lost quietly. When children from minority-language families learn the majority language early, the home language often gives way, and the cost lands on the family's ability to talk across generations. In census data on the grandchildren of immigrants in the United States, the home language had largely given way by the third generation for many groups — more slowly where communities stayed close together. The stories told in that language go with it.

Cultural material is not raw material. International norms already say so: the UN Declaration on the Rights of Indigenous Peoples gives indigenous peoples the right to maintain, control, protect and develop their cultural heritage and oral traditions, and the CARE principles for indigenous data governance — collective benefit, authority to control, responsibility, ethics — set out how data about a people should be handled. As AI learns from the world's stories and languages, Arfur takes these as the floor for its work with indigenous communities.

These studies are largely American, and they describe the market Arfur enters, not Arfur's effect on it. What they justify is the three refusals the product is built on: no manipulative design, a heritage language kept in use rather than replaced, and stories used only on the terms of the people who carry them.

Arfur is an edtech startup building a cultural platform. It is not a clinical intervention, a therapy, an assessment, or a treatment. iOSE informs how the product is designed; it does not diagnose, screen, or measure a child.

The reasoning on this page is a design rationale: a suggestion for how stories can create calm, presence and conversation. It rests on the third-party literature listed below and makes no clinical claims. Where a step in the argument is our inference rather than an established finding, it is marked as such.

This framework exists to be examined, not admired. Pilot measurement is being designed together with the first schools and partners who join — short, child-appropriate methods, teachers and parents as informants, aggregated use data alongside qualitative feedback.

If you evaluate education, integration or cultural-heritage programs, we would rather have the framework tested by someone with no stake in it. Tell us where it is weak.