Egil Diau

I’m rebuilding the cognitive foundation of how we understand humans. Intelligence, society, and power aren’t mysteries—they are structured expressions of cognitive architecture. This is a new theory of human reality—grounded in cognition, tested through simulation. My background spans computer science, economics, and behavioral science. I’m currently based at National Taiwan University.

Email

Research

I'm reconstructing human cognition as a biological extension—demonstrating that core capacities like tool use, cooperation, fairness, and political strategy are not uniquely human, but shared with other primates. This perspective leads us to reexamine intelligence and social cognition, redefining their boundaries through functional and comparative analysis. On this foundation, we propose a new account of society—where economy, finance and social institutions are not cultural constructs, but structured expressions of cognitive architecture.

We propose a simple idea: most intelligent behavior is pattern matching—not symbolic logic, not Bayesian inference. The evidence comes from two sides. First, intelligence is difficult to observe directly—but its failures are not. These failures aren’t random—they follow clear, structured patterns of misalignment. Second, intuition and creativity come from contextual cues triggering stored structure, not explicit reasoning. We show that many core functions—language, reasoning, planning, and social inference—can be explained as pattern matching.

Theory of Mind is conceptually vague and empirically unreliable. Both humans and other social animals show high error rates in mental state inference, suggesting it reflects memory-based social inference rather than true mindreading. We argue that, in contrast, tracking status and navigating power are robust, consistent, and cross-species—offering a stronger foundation for understanding social cognition.

Most theories treat morality, politics, and cooperation as uniquely human inventions. We argue, instead, that these capacities are structured extensions of cognitive functions shared with other primates. To support this view, we propose a comparative framework—organized into six core domains—that bridges biological roots and social complexity, offering a new foundation for cognitive science, social theory, and computational modeling.

Prevailing accounts frame large-scale institutions as products of cultural or social construction. We argue instead that their true origin lies in reciprocity—a behavioral substrate shared with other social animals. Modern systems like credit and money are structured extensions of this logic. We formalize this in a three-stage, simulateable model showing how decentralized reciprocal behavior scales into societal institutions.

Finance is often traced to institutional design or cultural invention. We argue instead that it originates in the same behavioral substrate as economic exchange: the fundamental logic of reciprocity. Trade—commonly seen as finance’s starting point—is the canonical form of reciprocity, from which credit, insurance, token exchange, and investment emerge as structured extensions.

Prevailing accounts trace the origins of economic exchange to barter or symbolic trust—a framing that places its roots in uniquely human invention. We overturn this assumption, proposing instead that economic exchange originates in reciprocity: a social behavior shared with other social animals and sustained by three minimal, simulateable mechanisms—individual recognition, reciprocal credence, and cost–return sensitivity.

We survey evaluation methods for LLMs and propose a two-stage perspective: from core language abilities to agent-level capabilities.

Projects

We reconstruct 3D affordance structures from egocentric demonstrations to support functional understanding and guide general object use.

Adapt from Jon Barron's template