Kimi

PARTI

2023

Priming Attention for Relational Teaching, Implicitly — an academic study asking whether children’s visual attention style can be shifted to improve their relational reasoning.

RoleLead Researcher
TeamPrincipal Investigator, postdoctoral researcher; managed 4 research assistants
Timeline15 months
Participants200 children, ages 3.5–6
MethodsLiterature review, physical prototyping, in-lab testing, quantitative analysis (R), cross-cultural replication

The question

Can attention be taught to see relations?

People from different cultures show differences in reasoning, and a lot of adult work documents it. There is research on social and language development — but nobody had investigated it from the visual side.

Our hypothesis: children’s attentional style can be manipulated by visual stimuli.

The primes

Two ways of looking at one photo

The manipulation is a script. Children see the same picture, but one condition walks them through the background while the other walks them through the focal object.

Background priming — “There is a yard… paved with stone… Can you see the building with brown doors?”
The priming photo with the whole yard highlighted: a cat sitting on cobblestones before a building with brown doors
Focal object priming — “There is a cat… fur is brown… Can you see its little nose?”
The same photo with the cat highlighted: attention drawn to its fur, its pricked-up ears, its nose

The task

Match the relation, not the object

Between primes, children play a relational match-to-sample game: a card shows a relation — two of the same shape, or two different ones — and the child picks which card matches it. The shapes never match; only the relation does. Try it:

Try it

This card shows a relation. Which card below matches it?

same + same
different
same + same

Different shapes, same answer — that generalization is what the study measures.

  1. A lattice-paved courtyard, the neutral photo shown before any primingBaseline block4 trials
  2. The cat in the yard, one of the priming photosPrime block 18 trials
  3. A matching trial: a source card above two choice cardsRMTS block 18 trials
  4. A dog on a lattice-paved yard, another priming photoPrime block 25 trials
  5. A matching trial with circles, an arch and heptagon, and arrowsRMTS block 28 trials
  6. A goat by a log cabin, another priming photoPrime block 35 trials
  7. A matching trial with stars, parallelograms, and mixed shapesRMTS block 38 trials
  8. A catch trial with everyday objects: a chocolate bar, a fan, a candyCatch trials2 trials

The prototype

A study you can hold

There was no established methodology, so I built one: a physical, high-fidelity prototype — laminated card stacks and a flip binder — that a researcher can run across a table from a three-year-old.

Laminated card stacks, labeled by block
Stacks of laminated stimulus cards, hand-labeled Q1, R1, Q2, on a black table
The flip binder on the testing table
The flip-binder prototype open on a testing table beside hand sanitizer and a pen

Pilot, then iteration

Three problems, three fixes

The pilot was honest with us. The task was boring — children got fussy and couldn’t finish. The prototype was hard to manage — rearranging cards was slow, and children got distracted by them. And research assistants struggled with the most unpredictable participants there are: young kids.

So the discussion guide gained encouraging phrases and small toys — “Give me a high five!” — the prototype got a faster way to reset between trials, and I designed a training session for the RAs, with the difficult child played by a moderator.

The study became 35% faster, and retention improved by 45%.

RA training — role-playing the difficult sessions
Two researchers at a table role-playing a session with the flip binder during RA training

Testing

200 children, ages 3.5 to 6

We recruited from a preschool, the lab database, and a science museum. The age range matters: 3.5 to 6 is a critical period for developing relational reasoning, and 200 participants gave the study more power than previous work on the question.

In the lab
An in-lab session: a masked researcher runs the card task across a small table from a young child
At the science museum
A researcher runs the card task at an outdoor table with a parent and child at the science museum

Across the Pacific

Democratizing research internationally

Collaborators at Chung-Ang University in Korea adopted the protocols and materials I developed. Working closely with international partners — including translating between the teams — made the research more scalable and more cross-culturally relevant.

Coordinating the Korean replication in Slack
A Slack message coordinating the Korean replication, asking Cheng to help translate for the visiting researcher

Findings & impact

Attention shapes reasoning — and it’s malleable

Children primed to attend to background and contextual features demonstrated notably better relational reasoning.

Before, studies showed that U.S. and East Asian children differ in relational reasoning without the causes being well understood. Now visual attention emerges as a key factor — one that can be shifted by priming. For educators, the implication is that contextual, relational talk — not just object-focused talk — can meaningfully enhance children’s thinking and help them become adept analogical thinkers.

Reflection

What I’d do differently

Shorten the task, to keep children engaged and reduce dropout. And given more time and resources, test object-focused priming with Chinese children — to see whether attentional styles can shift in both directions across cultures.