What do an embryo, an ant colony, and the latest version of ChatGPT have in common? At first glance, there seems to be nothing linking them, but a team of scientists from various fields has found a hidden algorithm that unites them. Research shows that any mind, from cells to complex systems, does not merely process information but navigates through imaginary maps of meanings and forms. We tend to think of intelligence as the exclusive prerogative of the brain, but both biology and artificial intelligence demonstrate that thinking is navigation in a realm of abstract coordinates. Understanding how this internal compass functions reveals the world not as a random set of events but as an impressive geometric problem. This opens the door to a new era of cognitive science, where mistakes are seen as part of the journey, and geometry becomes the essence, notes HighTech+.
Who are the authors of the research?
Michael Levin, the lead author from Tufts University, has become known as one of the most progressive scientists in biology. He is often referred to as the "hacker of the code of life," thanks to his work on creating xenobots — the world's first living beings assembled from frog cells. Levin claims that even the tissues of our body possess intelligence and can "reprogram" themselves. His colleague Benedict Hartl is a physicist and expert in complex systems who translates biological processes into a strict mathematical language. Together, they are developing a theory on how any matter can think.
In their recent scientific paper "Remapping and navigation of an embedding space via error minimization: a fundamental organizational principle of cognition in natural and artificial systems," the preprint of which is available on arxiv.org, they explain their ideas.
Imagine you are lost in an unfamiliar city. To find your way out, you need two things: a map to understand your location and the surrounding space, and a navigator to guide you to your destination.
In their new work, Levin and his team argue that this duo — mapping and navigating — represents a fundamental law of the mind. It does not matter whether it concerns a human, an embryo cell, or an artificial intelligence like ChatGPT. They all engage in the same activity: transforming the chaos of external signals into an organized internal "map" and striving to minimize errors on the way to their goal.
1. Cognitive invariant: universal software of the mind
We often view intelligence from a "brain-centric" perspective: neurons lead to thought, and their absence makes us mechanical beings. Levin and Hartl dismantle this approach. They introduce the concept of a cognitive invariant — a fundamental principle of information processing that remains unchanged regardless of whether it is based on carbon cells or silicon chips.
Every living system faces a similar problem: the world is too complex to respond to each stimulus individually. For survival, the system must become not just a "collection of parts" but a cognitive agent.
- For an embryo cell, "intelligence" means knowing how to create an eye, relying solely on interactions with neighbors.
- For AI, it is the ability to predict the next word, taking into account the context of all literature.
The authors argue that the mind is not an "addition" to evolution for brainy beings but a basic property of organized matter. Every system striving for integrity must engage in cognitive activity, constantly predicting the future and adjusting its state. This shifts biology from "chemistry" to "informatics."
2. Embedding space: the secret map of meanings
If the cognitive invariant is the "engine" of thought, then the embedding space is the "roadmap" along which it moves. This can be envisioned as a system of metaphors.
In a computer or a cluster of cells, there are no specific concepts such as "apples," "fear," or "liver." Instead, there are only signals. The embedding space is a mathematical method that transforms qualitative concepts into numerical coordinates.
How it works:
- Multidimensionality. Each concept is represented as a point in a space that can have thousands of dimensions. For example, the word "dog" has coordinates along the axes of "living/non-living," "size," "loyalty," etc.
- Meaning through proximity. Here, the principle is important: distance corresponds to meaning. In this space, "dog" and "puppy" will be almost close, while "dog" and "cryptocurrency" will be far apart.
- Biological embeddings. Levin applies this logic in biology, asserting that cells also have their own "embedding space." For them, the coordinates are the levels of electrical tension across membranes and the concentration of proteins. A cell "perceives" its position in this space and understands when it is too far from the coordinate "healthy tissue" and needs restoration.
Thus, thinking becomes not just manipulation of real objects but navigation through this vast mathematical map. We (along with our cells and AI) are constantly searching for the "right address," recalculating the route to minimize the distance to the goal.
3. Remapping: when the world changes
If navigation is moving along familiar routes, then remapping is a major change in the landscape. In Hartl and Levin's work, this is a key point of adaptation.
The system does not just move towards a goal; it constantly assesses the adequacy of its map.
- In learning: When you suddenly grasp a complex metaphor, a significant shift occurs in your "embedding space." Points that were previously far apart now appear close. Your internal map is restructured to better minimize errors in the future.
- In evolution: If an organ is damaged to the extent that previous recovery pathways no longer work, cells can "remap" their signaling pathways, finding new ways to restore function.
This explains the remarkable flexibility of life. We are not rigidly programmed mechanisms; we are systems capable of adapting our "internal navigation" in the process.
4. Navigation: how abstract thoughts become actions
If the embedding space is the map, then navigation is life itself. In traditional biology, it is commonly believed that everything operates on the principle of "stimulus — response." However, Hartl and Levin propose a different model: "state — goal — maneuver."
Navigation in biology. Imagine a group of cells needs to form an eye. They do not follow a strict instruction from DNA, like a chef preparing a dish. Instead, they act like an experienced driver who, when faced with a traffic jam (for example, an injury), does not stop but looks for an alternative route to the goal. This is navigation: the system knows the "coordinates" of a healthy organ in space and constantly adjusts its course. This is why embryos have a high resilience to damage — they "navigate around" obstacles to achieve their final form.
Navigation in AI. When a neural network generates text, it does not simply produce random words. It takes small steps in its space of meanings, choosing the next word each time to get closer to the most logical conclusion of the thought. Artificial intelligence "feels" the space in front of it, selecting the most probable path.
5. Minimizing errors: energy and the compass of knowledge
This is the most technical yet also the most fascinating part of the theory. Why does a system move in a certain direction? The answer: free energy or "prediction error."
Mathematically, this is expressed in the desire to reduce the gap that arises from the error that any cognitive system seeks to minimize to zero.
The system constantly compares its current coordinates with the target ones. This creates "tension." Like a ball trying to roll into a dip, a cognitive system seeks a state of zero error.
There are two paths:
- Change the surrounding world (Navigation): act so that reality matches our map (for example, grow a tail or write a text).
- Change oneself (Remapping): acknowledge that the map is incorrect and redraw it (learning).
6. Biological mind in action: how cells "agree" on shape
Michael Levin is not only a theorist but also a practitioner who has observed for decades how living matter solves geometric problems. The examples in the article serve as evidence that navigation in state space is not just a metaphor but a physical reality.
Case #1: Regeneration as a search for the "right address."
A well-known example from Levin's work is the flat planarian worms. If you cut such a worm into 200 pieces, each piece will "know" what part of the body it is in and will be able to reconstruct the missing part. From the perspective of the theory of embeddings, each cell has its own coordinates on the overall map of the body. When the worm is cut, the system registers a huge "error": the current coordinates (a piece of the tail) do not match the target ones (a whole worm). The cells begin to change their bioelectrical parameters, charting a route to the state of "whole organism." Once the "address" is reached, growth stops.
Case #2: Xenobots — intelligence without ancestors.
Levin creates xenobots — tiny beings assembled from frog skin and heart cells. They have no brain, no neurons, and no millions of years of evolution behind them (they have never existed in nature in this form). Nevertheless, these cells demonstrate an astonishing ability to navigate:
- They form groups.
- They "find" ways to move in space.
- They can collaboratively perform tasks, such as cleaning surfaces. This proves that cells possess an "innate navigator": when faced with a new situation (a new embedding space), they can remap their actions on the fly and find ways to survive and work together.
Case #3: Bioelectrical memory.
The article mentions an experiment that seems fantastic: scientists altered the bioelectrical code on the body of a worm without touching its DNA. As a result, the worm grew two heads. The astonishing thing happened later: when this two-headed worm had its heads cut off again, it grew two heads once more! The worm's DNA was set for one head, but the cognitive map of the system had changed. The cells "remembered" the new coordinates in the embedding space and now strive for this state as a target.
Why this matters
These examples demonstrate that life is not a rigidly programmed conveyor belt but a dynamic agent. Cells are not bricks but builders who have a tablet with a map. If the tablet is changed or the goal rewritten, the builders will alter the entire project. This opens new horizons in the medicine of the future: instead of attempting to rewrite DNA (changing bricks), we could simply redraw the map in the "mental" component of tissues, prompting them to heal cancer or restore organs on their own.
The article by Hartl and Levin represents an attempt to create a "unified field theory" for intelligence, combining developmental biology, cognitive psychology, and computer science. It asserts: we are capable of thinking not because we have a brain, but because we are systems capable of traveling through the maps of our states.
