LLMs predict the next token in a sequence. The brain's cortex may function as a general prediction engine capable of "omnidirectional inference"—predicting any missing information from any available subset of inputs, not just what comes next. This offers a more flexible and powerful form of reasoning.

The cortex has a uniform six-layer structure and algorithm throughout. Whether it becomes visual or auditory cortex depends entirely on the sensory information plugged into it, demonstrating its remarkable flexibility and general-purpose nature, much like a universal computer chip.

Counterintuitively, the development of specialized speech pathways involves turning off certain genes. These genes code for "repulsive molecules" that prevent neural connections from forming. By deactivating them in speech areas, the brain allows for the unique and critical connections for vocal learning to be established.

The need for our ancestors to communicate about memories and future plans—the essence of stories—drove the evolution of simple grunts into complex language. Our brains are fundamentally story-shaped because language was built to narrate events.

The brain regions for speech production and hand gesturing are adjacent. Dr. Jarvis suggests speech pathways evolved from older body-movement pathways. This explains why humans instinctively gesture while speaking, even when the other person cannot see them, such as on a telephone call.

Thought is fundamentally non-linguistic. Evidence from babies, animals, and how we handle homophones shows that we conceptualize the world first, then translate those concepts into language for communication. Language evolved to express thought, not to be the medium of thought itself.

The brain regions processing language also control core bodily functions like heart rate, hormones, and the immune system. Consequently, the words you use have a direct, physiological effect on others. A kind word can calm, while a hateful one can trigger a resource-depleting threat response.

The idea that language creates thought is backwards. Pre-linguistic infants already have a sophisticated understanding of the world (e.g., cause and effect). They learn language by shrewdly guessing a speaker's intent and mapping the sounds they hear onto thoughts they already possess.

Paradromics uses LLMs to decode brain signals for speech, much like how speech-to-text cleans up audio. This allows for faster, more accurate "thought-to-text" by predicting what a user intends to say, even with imperfect neural data, and correcting errors in real-time.