Why Every Brain Metaphor in History Has Been Wrong

The Necessity and Danger of Scientific Simplification

Science requires the simplification of complex reality to make it intelligible to human cognition. Because human working memory and attention are limited, researchers build models that intentionally omit details to identify patterns. However, a critical tension exists between two philosophical perspectives on this process:

• Simplicius: The belief that the universe is fundamentally simple and that finding an elegant equation indicates the discovery of an underlying truth.
• Ignorantio: The belief that simplification is a necessity born of human limitation; models are useful fictions or approximations (the map, not the territory) rather than literal truths.

Professor Mazviita Chirimuuta, author of The Brain Abstracted, argues for the "Ignorantio" position, suggesting that successful science proves we are adept at building useful simplifications, not that nature itself is simple. This is described as "learned ignorance"—studying a subject to understand the boundaries of what cannot be known.

The Evolution of Brain Metaphors

Throughout history, the prevailing model of the brain has consistently mirrored the most sophisticated technology of the era. This pattern suggests that these models are analogies rather than literal descriptions:

• Hydraulic Automata: Descartes viewed the nervous system as fluids pumping through tubes to push levers.
• Telegraph Networks: With the discovery of electrical signals, the brain was modeled as a network of wires.
• Telephone Switchboards: The brain was seen as a system of signals routed by operators.
• Computers: The current dominant metaphor treats the mind as software running on biological hardware.

This transition illustrates the "fallacy of misplaced concreteness," where a useful metaphor hardens into a perceived reality. While early cyberneticists like McCulloch and Pitts used logic gates as functional descriptions of neurons, modern discourse often claims the brain is a computer, mistaking the elegance of the description for the structure of reality.

Software, Spirit, and Causal Invariance

Joscha Bach proposes a provocative view that software is literally "spirit," not metaphorically. He argues that certain patterns possess "causal invariance"—meaning they have causal power regardless of the physical substrate (silicon, neurons, or paper money) they inhabit. In this view, software is an abstract mechanism that controls its substrate.

Counterarguments to this position suggest that "sameness" across substrates is a human imposition. For example, the causal power of money exists not in the paper or digital ledger, but in the social substrate of human agreement. Therefore, the invariance is a result of human interpretive practices rather than an intrinsic property of nature.

Prediction vs. Understanding

There is a fundamental distinction between the ability to predict or control a system and the ability to understand it. Nobel laureate John Jumper distinguishes these three categories:

1. Predict: Forecasting a future value or state.
2. Control: Manipulating a system to achieve a specific future value.
3. Understand: Possessing a compact collection of communicable facts that explain why a result occurs.

While AI models (like LLMs) excel at prediction and control, they do not perform the act of understanding. Noam Chomsky argues that a theory that predicts everything (e.g., "anything goes") explains nothing because it fails to answer why things are a certain way and why they are not another way. Relying solely on predictive black boxes creates a risk: when the tool breaks, the lack of underlying understanding means the failure will be unforeseen.

The Illusion of AGI Inevitability

The widespread belief in the inevitability of Artificial General Intelligence (AGI) may be a "cultural historical illusion." This perspective suggests that the belief is not based on a scientific certainty but is a byproduct of a long history of mechanistic views of life and mind. If the hypothesis that the mind is merely a mechanism is wrong, then the inevitability of biological-like AI is unfounded.

Knowledge as Embodied and Perspectival

Knowledge is not a universal, perspective-free repository (like a "god's eye view" provided by the internet or LLMs), but is inherently social, embodied, and perspectival.

Haptic Realism

Professor Chirimuuta proposes "haptic realism," suggesting that scientific knowledge is more like touch than vision. Rather than observing reality from a distance, researchers "poke and prod" the system, meaning the patterns they find are partially created by the process of investigation itself.

The Cognitive Horizon

Noam Chomsky notes that organic creatures have a "cognitive horizon"—inherent bounds to their capacities. Just as a rat cannot be trained to understand prime numbers regardless of the amount of data provided, humans likely have structural limits to what they can comprehend. Recognizing these walls is a vital part of scientific progress.