ideasthesia

Sign in Subscribe

Ryan Collison

I write about cool science and share it here.

Network of conscious agents with perception-action loops creating emergent structure

Conscious Agents All the Way Down: Hoffman's Mathematical Framework

Series: Interface Theory | Part: 4 of 10 If reality is a perceptual interface hiding deeper truths, what lies beneath? For Donald Hoffman, the answer isn’t particles or fields or spacetime geometry. It’s conscious agents all the way down. This isn’t mysticism dressed in equations. Hoffman has built

Computer desktop interface in deep space showing perceptual icons above hidden circuitry

The Desktop Metaphor: Why Your Perception Is Like a Computer Interface

Series: Interface Theory | Part: 3 of 10 When you click on a file icon, you don’t see spinning magnetic platters or voltage fluctuations in silicon. You see a blue folder. That folder isn’t lying to you—it successfully guides action. But it’s not revealing truth either. You

Evolutionary game theory visualization showing fitness-tuned strategies winning over truth-tracking

Fitness Beats Truth: The Mathematical Theorem That Undermines Naive Realism

Series: Interface Theory | Part: 2 of 10 We believe what we see is real. The coffee cup on your desk exists the way it appears. The tree outside your window has leaves, bark, and roots exactly as you perceive them. This conviction—that perception reveals reality as it actually is—

Desktop interface metaphor for perception—translucent layers showing interface hiding computational complexity

The Cognitive Scientist Who Says Reality Is a User Interface: Donald Hoffman's Radical Theory of Perception

Series: Interface Theory | Part: 1 of 10 You trust your eyes. Why wouldn’t you? They show you the world as it is—solid objects in three-dimensional space, colors that exist on surfaces, distances that can be measured. When you see a red apple on a table, you’re

LLM neural network architecture wrapped in generative model structure with belief propagation pathways

Active Inference for Language Models: The Next Frontier

Series: Active Inference Applied | Part: 9 of 10 In March 2023, a team at MIT published something quietly revolutionary. They didn’t build another scaling law. They didn’t stack more transformers. They wrapped a large language model in a generative model—gave it beliefs, preferences, and the capacity to

Multi-level hierarchical architecture with stacked manifolds connected by precision-weighted pathways

Hierarchical Active Inference: Scaling to Complex Tasks

Series: Active Inference Applied | Part: 8 of 10 You’re making coffee. Your hand reaches for the pot. Your fingers wrap around the handle. Muscles contract in sequence, joints articulate, neurons fire in cascades. But you don’t think about any of that. You think “make coffee” and somehow your

Two diverging computational pathways - active inference with message passing and RL with value gradients

Active Inference Agents vs Reinforcement Learning: A Comparison

Series: Active Inference Applied | Part: 6 of 10 If you’ve been following the active inference literature, you’ve likely noticed researchers repeatedly positioning it as an alternative to reinforcement learning. Not just a different algorithm, but a fundamentally different paradigm for building intelligent agents. The comparison appears everywhere: papers,