🚀 Quick links: License, ReadMe, Installation, Contributing, Innovation-Lab, Philosophy, Genesis, Architecture

Manifesto: For a more organic and sustainable AI

░                     ░░░░░░░░░░▒▒▒▒▒▒░░
           ░░░░░░░░░▒▒▒▒▒▓▒▒▒▒░░░░░░░░░░▒▒▒▒░                                                          ░░░░░░░░░░░
░░░░░░░░░░░░░░░░▒▒▒▒▓▓▓▓▓▓▓▓▓▓▓▒░░░░░▒▒▒░░░░▒▓▒░░                      ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
░░░░░░░░░░░░░░▒▒▒▓▓▓▓▓▓▓▓▓▓▓▒░░▒▒▒░░░░▒▓▓▓▓▓▓▒▒▒▒▒░     ░░░░░░░░░░░░░░░░░░░░░▒▒░░▒▒▒▓▓▓▓▓▓▒▒▒░░░░░░░░░░░░░░░░░▒▒▒▒
▒░░░░░▒▒▒▒▒▒▓▓▓▓▓▓▓▓▓▒░         ░░▒▒▒░▒▒▒▒▓▓▓▓▓▓▓▒▒░░  ░▒▒▒▓▒▒▒▓▒▓▒▓▒░░░░░░░▒▓▓▓▓▓▓▓▓▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒░░░░░░░░░░░░▒▓
░▒▒▒▒▓▓▓▓▓▓▓▓▓▓▓▓░                   ░░ ▒▒▓▒░▒▓▓▓░▒▒░░           ░▒░░░▒▓▒▒▒▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▒▒▒▒▓▓
▒▒▓▓▓▓▓▓▒▒▒░░                           ░▓▓▒░░▒▓▓░ _    _    _ ░▒░░▒▓▒▓▓▓▓▓▓▓▓▓▓▒░░░░░░░░░▒▒▒▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓
▓▓▓▓▓▒░AI inspired by natural plasticity ░░   ░░░  a    N    A  ▒▓▒▓▒▒▒▓░Autonomous Neural Architecture v5.2   ░▒▓
░                                                  ‾    ‾    ‾ ░▓▒▓░░▒▓░

"The Creation" —Michelangelo

Important

This project has evolved. It has been significantly optimized. Access the full update here 👉🏻 v5.3-beta(/)

The quest for AI reflecting life

Current artificial intelligence, while powerful, relies on massive, rigid architectures that remain disconnected from biological reality. Simultaneously, the AI community is increasingly concerned with the rising carbon footprint and unsustainable computing costs of these models. aNA (Autonomous Neural Architecture) AI Project breaks with these paradigms. Inspired by the organization of the six-layered cortical_columns(.py), synaptic plasticity, and the precise management of the thalamus(.py), hippocampus(.py), amygdala(.py), cerebellum(.py), limbic system(.py), and five key neuromodulators(.py) (dopamine, adrenaline, nitric oxide, acetylcholine, and serotonin). This project aims to create not a mere computational simulation, but an organic resonance. Far from being a static data repository, aNA AI is a dynamic system that learns, forgets, adjusts, and focuses—much like our own minds.

Dynamic Cognitive Architecture

• Asynchronous Orchestration: The pulse(.py) module acts as the system's biological pacemaker, regulating non-blocking cycles according to the organism's homeostatic state.
• Thalamic Filtering: Sensory inputs are gated and processed through specific nuclei to isolate relevant signals from stochastic background "noise".
• Adaptive Fidelity: aNA dynamically simulates information fidelity based on dopamine levels, modulating the quality of internal representations in real-time.

The imperative of energy sobriety

We are at a tipping point. The frantic race toward ever-larger and more energy-intensive models is placing an unsustainable burden on our environment. The aNA AI project proposes a radical alternative: efficiency through targeted plasticity. By mimicking the economical functioning of the human brain—which achieves cognitive feats with a mere 20 watts—we are developing algorithms that activate only the neurons(.py) necessary to process specific information. Learn less to understand better; filter more to compute less. This is the path forward toward a sustainable, responsible AI.

Towards the next generation of intelligence

We are embarking on the "next wave" of artificial intelligence—an AI that does not simply predict the next token, but truly integrates context, attention, and emotional modulation to act with discernment, even amidst "noise". This project is more than a technical achievement; it is a social commitment. By designing transparent, explainable, and resource-efficient systems, we are laying the groundwork for a more harmonious coexistence between humans and the machines of tomorrow.

🧪 Running the Biological Simulations (Demos)

To witness aNA v5.1's neural processing in real-time, you can run the integrated tests(/) suites. These simulations demonstrate how sensory data is transformed into emotional importance and cortical action.

1. Limbic Resonance (Test)

This demo simulates how the amygdala(.py) and hippocampus(.py) collaborate to filter critical information from routine noise.

# From the project root
python3 src/tests/test_limbic_system.py

What to look for in the output:

• Routine Scenario: Low arousal levels (Cortisol/Adrenaline) leading to standard memory encoding.
• Shock Scenario: High arousal triggering a "System Breach" alert and prioritized memory storage.

2. Cortical Column Cascade (Test)

Validation of the 6-layer signal flow (L4 → L2/3 → L5) with real-time precision monitoring.

• The L6 Feedback Loop (cortical_column(.py) Layer 6) This is what allows the neocortex(.py) to “say” to the thalamus(.py): “I recognized this signal, you can lower the volume (gain)”. This is the basis of the selective attention.

# From the project root
python3 src/tests/test_cortical_column.py

📚 Cognitive Mapping (The Digital Bridge)

• Amygdala(.py) (Priority Filter & Interrupt Controller): Manages emotional valence and high-priority signals. It acts like an interrupt controller that can override standard processing cycles when "critical events" (stress or high-reward stimuli) are detected, ensuring immediate system response.

• Cerebellum(.py) (Timing Engine & Output Calibration): The specialized unit for fine-tuning motor and cognitive outputs. In aNA, it ensures that the "Thinking Shell" operates with perfect mathematical synchronization, acting as a calibration layer for fluid, real-time interaction.

• Cortical Columns(.py) (Hierarchical Data Modules): The standard vertical organization of the mammalian neocortex(.py). In aNA, these 6 cortical layers define the functional hierarchy: Layer IV (input), Layers II/III (association/prediction), and Layer V/VI (motor output).

• Limbic System(.py) (Emotional Valence & Memory Orchestrator): The vital bridge between sensory impact and long-term storage. In aNA, it evaluates incoming signals via the amygdala(.py) to assign an "emotional weight", ensuring that critical experiences are prioritized by the hippocampus(.py) for deep encoding and empathic recall.

• Lobe, Frontal(.py) (Executive Logic & Command Center): The primary site for high-level decision-making and motor control. In aNA, it acts as the central executive that orchestrates complex task sequences and manages the "top-down" attention directed to other modules.

• Lobe, Occipital(.py) (Visual Stream Processor): Dedicated to the decoding of visual information. It functions as a specialized GPU-like buffer within the architecture, transforming raw sensory "pixels" into structured spatial patterns before they are analyzed by association layers.

• Lobe, Parietal(.py) (Spatial Mapping & Data Integration): Manages the integration of sensory information from various parts of the system. It acts as a multi-modal coordinate system, allowing aNA to understand the "where" and "how" of data points in a unified 3D-like internal workspace.

• Lobe, Temporal(.py) (Semantic Storage & Pattern Recognition): The hub for processing auditory signals and high-level linguistic or object recognition. In the digital model, it serves as the semantic engine that links sensory inputs to long-term "concepts" stored in the memory hierarchy.

• Hippocampus(.py) (Dynamic Buffer & Indexing System): Functions as a high-speed memory buffer for short-term data. It manages the temporary storage of information and coordinates its eventual "migration" (consolidation) into long-term cortical databases, preventing immediate system saturation.

• Neocortex(.py) (The "Thinking" Shell): The structural integration of the 4 Lobes and 6 Cortical Layers. While the neocortex(.py) handles complex reasoning and prediction, the cerebellum(.py) (in aNA) fine-tunes motor outputs and timing, ensuring the system’s actions are fluid and mathematically synchronized.

• Neuromodulators(.py) (Global State Tags): Chemical core "gain controls" (Dopamine, Adrenaline, Nitric Oxide, Acetylcholine, Serotonin) that regulate the global state of the network. They don't carry specific data but adjust how the brain processes information (e.g., focus, reward, stress response).

• Neurons(.py) (Atomic Processing Nodes): The three functional types (Sensory, Interneurons, Motor) operate as the basic logic gates of the architecture. Their Synaptic Plasticity represents a self-modifying code capability, where connection weights evolve dynamically based on the frequency and timing of data flow.

• Thalamus(.py) (Sensory Gateway & Packet Router): Acts as the system's central hub for incoming data. It filters and directs sensory signals to the appropriate cortical layers, preventing "system overflow" by dropping irrelevant background noise.

Note: These definitions are adapted to the specific metabolic and cognitive constraints of the aNA v5.0 architecture and above.

*Every measurement reflected here is a digital bridge to biological reality, designed to synthesize the fundamental principles of living systems.

# From src/anatomy/thalamus.py
# Highlighting the proactive "Sensory Gateway" logic
def create_sensory_thalamus(position: np.ndarray = None) -> Thalamus:
    """Creates a thalamus optimized for clarity and sensory focus."""
    thalamus = Thalamus(position)
    # Enhance specific nuclei to sharpen digital perception
    thalamus.nuclei[ThalamicNucleusType.LGN].config.size = 800  # Vision focus
    thalamus.nuclei[ThalamicNucleusType.MGN].config.size = 600  # Hearing focus
    return thalamus

aNA sharpens digital perception by prioritizing meaningful signals, ensuring a clear and focused cognitive flow.

░▒▓ BT 2026-04-07