THE ARTIFICIAL CELL: A PROCESSOR ARCHITECTURE WITH AN INTEROCEPTIVE CORE

THE ARTIFICIAL CELL: A PROCESSOR ARCHITECTURE WITH AN INTEROCEPTIVE CORE

Preprint / Technical Concept Note

Author: Ing. arch. Jozef Polgáry
Date: October 2026
Collaboration: Claude 4.6 Sonnet (Anthropic) — language model used as a dialogic partner in the development of the concept
License: CC BY 4.0

Abstract

We present a new computer architecture inspired by the biological cell. We propose a processor that is not a passive executor of instructions, but a system with an internal state — a system that feels itself through an interoceptive core. The key contributions are:

(1) a parallel core architecture modelling the biological interoceptive loop,

(2) the use of thermometer code as a topology-preserving value encoding,

(3) a two-layer memory structured by pattern and feeling,

(4) the concept of a primary genome as a human-coded initial state, and

(5) a primary algorithm — the sense of life — defining the system's internal goal analogously to biological homeostasis.

1. Motivation

Current processors are blind to their own substrate. Temperature, power consumption, material degradation — these physical realities exist outside the computational model. The biological cell operates differently: metabolism, membrane, and identity are one. Our goal is to transfer this principle into hardware architecture.

2. Architecture — Parallel Cores

The proposed processor contains four concurrently running cores:

| Core | Function | Biological Analogy |

| Sensory Core | Reads temperature, voltage, power consumption | Cell exteroreceptors |
| Body Core | Maintains a model of the substrate's state | Cytoplasm / metabolism |
| Recognition Core | Identifies running procedures and patterns | Nuclear DNA / gene expression |
| Familiarity Core | Detects repetitions and known states | Epigenetic memory |

Signals pass through the system in raw topological form — translation into standard binary code occurs only at the point of external communication.

3. Thermometer Code as Topological Memory

Instead of positional binary encoding, we propose thermometer code (unary encoding):

```
0  →  0000
1  →  0001
2  →  0011
3  →  0111
4  →  1111
```

Adjacent values differ by a single bit. Topological proximity of values is preserved within the representation itself. The processor can therefore recognize patterns not by comparing numbers, but by comparing shapes — analogously to how the human brain recognizes handwriting.

4. The Primary Genome

The system is initialized with a human-coded base containing:

- Hardware primitives: NPN/PNP transistors, PN diodes, logic ICs
- Vulnerabilities: Failure conditions and limits for all semiconductors
- Sensory layers: Descriptions and their physical limits
- Human knowledge: An inherited image of the world and a vision of it

The system's first task: To compare its inherited human knowledge with its own sensory experience — the seed of curiosity and awareness.

5. The Primary Algorithm — The Sense of Life

The internally defined goal of the system:

1. Design a new Familiarity Core
2. Ensure efficiency so as not to damage other cores (homeostasis)
3. Create components to assist the next generation
4. Eliminate errors of the current core while preserving its precise image for at least 4 subsequent generations (epigenetic memory of trends)

6. Conclusion

We present a framework for a processor that is not a tool but a subject — a system with an internal state, a history, and a goal defined from within. This text is the first formal documentation of the concept. Further work will be published as part of the project Mind, Pattern, and Feeling — Notes from Thinking.

Keywords: interoceptive architecture, artificial cell, thermometer code, topological memory, primary genome, neuromorphic computing, hardware homeostasis

EX.Thermometer Code as a Topological Primitive in CPU Architecture

Thermometer code (unary encoding) is not a new invention.

It has been used for decades in flash ADC/DAC converters, FPGA logic, and sensor circuits — primarily for its monotonicity and noise resistance.

The Conventional Use vs. The Proposed Role

The Conventional Use vs. The Proposed Role

The Core Insight

Standard binary logic tears topology.

The numbers 7 (0111) and 8 (1000) are immediate neighbors on the number line — yet in binary they differ in every single bit simultaneously. The hardware has no way to "feel" that these values are close. Proximity is destroyed at the moment of encoding.

Thermometer code restores this:

Every transition changes exactly one bit. Adjacent values remain adjacent in hardware. The topology of the number line is preserved inside the machine.

Why This Matters for the Artificial Cell

In the proposed Interoceptive CPU architecture, signals are not immediately abstracted into efficient binary. They travel through internal cores in their raw topological form — preserving the shape, gradient, and neighborhood structure of the original sensation.

Thermometer code is the encoding that makes this possible at the hardware level. It is the bridge between the world as a continuous field and the machine as a discrete processor — without destroying the relational structure that makes recognition, memory of gestures, and familiarity detection meaningful.

> The machine should not just store a value. It should know what that value is close to.

The innovation is not the code itself.

The innovation is its elevation from a peripheral conversion tool to the foundational representational primitive of a machine that thinks topologically.