Systems Framework: From Natural Epistemology to Artificial Intelligence

A structured architecture mapping human cognition to AI systems through properties, parameters, and attributes.

Core Philosophical Framework: From Natural to Artificial

Natural Epistemology (Human): Senses + Intelligence + Objects → Perception & Knowledge

Artificial Intelligence (System): Sensors + Algorithms + Data → Models & Actions

Key Insight: Properties, Parameters, and Attributes serve as the formal, quantifiable representations of the qualities perceived by senses and reasoned about by intelligence.

Systems Architecture: A Three-Layer Model

This model maps directly to the hardware, firmware, software paradigm, creating a coherent information pipeline.

Layer 1: Hardware Layer (The "Body" & Raw Interface)

Correlate to:

The Five Senses + Physical Objects.

Role:

Transduces physical phenomena (objects, events) into structured digital data.

Incorporating Properties, Parameters, Attributes:

Attributes (Intrinsic to Objects/Signals): These are the raw, measurable qualities of the physical world.

Example (Vision): Pixel luminance (brightness), wavelength (color), spatial coordinates.

Example (Audio): Frequency, amplitude, phase.

Example (Touch Sensor): Resistance, capacitance, pressure (psi).

System Format: These are low-dimensional, physically-grounded data vectors from sensors. They are the atomic primitives of the system's perception. A parameter here might be the sampling rate (firmware-defined) that governs how these attributes are captured.

Layer 2: Firmware / Middleware Layer (The "Perceptual Spine")

Correlate to:

Lower-level, quasi-reflexive perception and signal processing (the "hardwired" parts of intelligence).

Role:

Converts low-level attributes into higher-level properties and features. This layer performs invariant detection and filtering.

Incorporating Properties, Parameters, Attributes:

Properties (Derived & Relational): These are computed interpretations of combined attributes.

Example: From pixel attributes (color, brightness), compute the property texture = {"rough", "smooth"} or edge_strength = 0.87.

Example: From audio attributes (frequencies), compute the property pitch = 440Hz or phoneme = "/ae/".

Parameters (The Tunable Knobs): This layer is parameter-heavy. These are the fixed or tunable settings that control how attributes are synthesized into properties.

Examples: Edge detection kernel coefficients, filter cut-off frequencies, time-window sizes, noise-floor thresholds, activation functions in a neural net layer.

System Format: A pipeline of parameterized transforms (e.g., DSP filters, convolutional kernels, spectral analyzers). Its output is a feature vector—a structured set of properties ready for cognitive software.

Layer 3: Software / Cognitive Layer (The "Mind")

Correlate to:

Higher-order Intelligence + Synthesis.

Role:

Uses properties and features from the firmware layer to form abstract representations, make decisions, learn, and act. This is where epistemology becomes explicit.

Incorporating Properties, Parameters, Attributes:

Attributes (in the software sense): Now become symbolic or semantic labels attached to conceptual objects.

Example: An object in a knowledge graph has attributes: {"type": "cat", "size": "medium", "affectionate": True}. These are high-level assertions.

Properties are used as evidence to assign these attributes via classification (if "furry" and "meows" then type:cat).

Parameters (The Learned & Adaptive Core): These are the learnable weights of models (e.g., weights in a Deep Neural Network, probabilities in a Bayesian network, rule weights in an expert system).

These parameters encode the system's epistemology—its "beliefs" about how sensory properties correlate with conceptual attributes and categories. They are updated via learning algorithms.

System Format: Models (e.g., neural networks, probabilistic graphs, symbolic KBs) defined by:
1. Architecture/Logic (the fixed structure of reasoning).
2. Parameters (the malleable knowledge within that structure).
3. Input/Output Schemas (mapping perceptual properties to cognitive attributes and actions).

Unified Systems View: The Information Flow

Natural Epistemology Component	AI System Component	Form of Representation (Properties, Parameters, Attributes)
Object in the World	Data Source / Target	Has physical attributes (mass, reflectivity, etc.).
Sense Organ (e.g., Eye)	Hardware Sensor (e.g., Camera)	Outputs signal attributes (pixel arrays). Governed by physical parameters (exposure, gain).
Perceptual Processing	Firmware/Middleware Layer	Transforms signal attributes into perceptual properties (edges, textures). Uses algorithmic parameters (filter coefficients).
Intelligence (Understanding)	Software/Cognitive Layer	Maps properties to semantic attributes (identity, intent, risk). Uses model parameters (neural weights) refined by learning.
Knowledge	Internal Model State	A structured network where entities have attributes and relations have properties. The model's parameters are the encoded knowledge.
Action / Expression	Actuators & Outputs	Commands defined by control parameters, which are functions of the system's state (attributes + properties).

Pedagogical Value of This Framework

Demystifies AI

It shows AI not as magic, but as a systematic engineering implementation of the natural process of knowing.

Clarifies Terminology

Attribute: A qualifier. Can be low-level (sensor data) or high-level (semantic label).

Property: A descriptive characteristic derived from relationships or computations. It often sits between raw data and abstract knowledge.

Parameter: A system variable that controls a process. It can be fixed (design choice), tunable (knob), or learned (the essence of AI).

Emphasizes the Pipeline

Students see that intelligence is built on a layered transformation of representations, each with its own type of parameters and attributes.

Unifies Symbolic and Sub-Symbolic AI

High-level symbolic attributes (e.g., dangerous) can be grounded in sub-symbolic properties (e.g., rapid_looming_motion = true) via parameterized models.

In essence: By adopting this systems format, you teach that building an AI is the process of designing a pipeline that transforms physical attributes into cognitive attributes, mediated by parameters that are either engineered or learned. This perfectly captures the transition from natural epistemology to artificial intelligence.

Systems Framework for AI Education | Natural Epistemology to Artificial Intelligence