The Invisible Curriculum: How Culture Shapes What We Can Even Imagine

In Tokyo’s Ikebukuro Station, something remarkable happens every weekday morning. Trains arrive empty at the terminal platform, and passengers spontaneously form two queues on each side of the doors. The first queue boards immediately; the second queue shifts forward to become the next “first” queue when the train departs. This isn’t taught anywhere. There’s no official directing traffic. Yet everyone participates flawlessly.

This is culture in action—and it’s far more profound than customs and traditions.

The Core Insight

What gets synchronized across a population isn’t just behavior—it’s the underlying mental frameworks that generate that behavior. Anthropologist Joseph Henrich’s research reveals something unsettling: the differences between cultures run deeper than what people do. They extend to what people can even perceive. Westerners, raised on abstract categories like “fish” and “bird,” literally see the world differently than members of small-scale societies who think in specific categories like “robin” and “jaguar.”

These “framings”—the ontological foundations of how we model reality—are the real currency of culture. They determine not just what decisions we make, but what options we can conceive of making in the first place.

Why This Matters for Technology

Here’s where it gets relevant for anyone building AI systems: if culture shapes perception, and AI systems are trained on human-generated data, those systems inherit our cultural framings. But they do so without the decades of implicit learning that humans undergo.

Consider what this means for AI assistants:
– An AI trained primarily on Western data might struggle to understand contexts where “standing out” carries profound social shame
– The concept of “simpatia” that drives Italian social interaction simply doesn’t exist in Japanese cultural frameworks
– Sarcasm—the bread and butter of Western communication—is largely absent from Japanese cultural framings

The implications are significant: when we build AI systems meant to serve global audiences, we’re asking them to navigate cultural terrains they’ve never actually experienced.

Key Takeaways

• Culture is the mass-synchronization of framings: Not just behaviors, but the mental models that generate those behaviors
• Framings are largely arbitrary and self-reinforcing: They emerge from chance and contingency, then become self-sustaining through feedback loops
• Different cultures literally perceive differently: Research shows Westerners focus on isolated objects while other populations attend more to context and relationships
• AI inherits cultural blind spots: Systems trained on human data absorb our framings without understanding them
• The QWERTY problem applies everywhere: Once a framing stabilizes, it persists regardless of whether it remains optimal

Looking Ahead

As AI systems become more integrated into daily life globally, the question isn’t whether they’ll need to navigate cultural differences—they’ll need to do so with awareness that they’ve been trained on data that reflects specific, arbitrary cultural framings.

The Japanese queuing phenomenon isn’t about rules or DNA. It’s about shared framings so deeply internalized that they feel like second nature. Our challenge with AI is building systems that can recognize and respect framings they cannot see.

Based on analysis of “Culture Is the Mass-Synchronization of Framings”