✦ ANN Module 00 · Amygdala AI

McCulloch–Pitts Neuron
The 1943 Origin of Neural Computation

The first mathematical model of a neuron — binary threshold logic, an absolute inhibitory veto, and the proof that networks of simple units can compute any Boolean function.

Historical & Biological Foundations
In 1943, Warren McCulloch (neurophysiologist) and Walter Pitts (logician) published "A Logical Calculus of the Ideas Immanent in Nervous Activity," proposing the first formal model of a neuron.
Key Contributions 1943

The MCP paper gave the first mathematical model of an artificial neuron expressed in propositional logic, showed that networks of binary threshold units can compute any function expressible in propositional calculus, and laid the foundation for Hebbian learning (1949) and Rosenblatt's Perceptron (1958).

📌 The MCP neuron predates all learning algorithms — weights and thresholds were fixed by design, not learned from data. Learning only arrived later, with Hebb and Rosenblatt.
Biological Analogy MAPPING

The model draws an abstract correspondence between real neurons and its computational parts:

Dendrites receive signals→ Inputs x₁, x₂, …, xₙ
Synaptic strength→ Fixed weight (+1 excitatory / inhibitory)
Cell body (soma) integration→ Weighted summation Σwᵢxᵢ
Firing threshold→ Threshold θ
Axon fires (all-or-none)→ Binary output y ∈ {0,1}
Inhibitory synapse blocks firing→ Inhibitory input forces y = 0
DIAGRAM — BIOLOGICAL NEURON → MCP ABSTRACTION
BIOLOGICAL NEURON dendrites soma axon MCP NEURON x₁, x₂ … xₙ (inputs) Σ wᵢxᵢ (weighted sum) y ∈ {0,1} (binary output) Σ ≥ θ ? y
The "all-or-none" law of neuronal firing — a neuron either fires fully or not at all — directly motivates the binary threshold output.
The Six Classical Assumptions RULES

1. Binary inputs/outputs — xᵢ, y ∈ {0,1}.   2. Equal excitatory weights — all wᵢ = +1.   3. Inhibitory veto principle — any active inhibitory input absolutely blocks firing.   4. Fixed threshold θ, not learned.   5. Synchronous, discrete-time operation.   6. No learning mechanism at all.

Mathematical Formulation
The inhibitory veto principle, expressed both as a piecewise rule and as an algebraic product.
Firing Rule DEFINITION

For m excitatory inputs x₁…xₘ and k inhibitory inputs z₁…z_k, all binary:

y = 1 if (Σxᵢ ≥ θ) AND (zⱼ = 0 for all j)   else   y = 0

Equivalently, using the Heaviside step function Θ(·):

y = Θ( Σxᵢ − θ ) · Π(1 − zⱼ)
Inhibitory Veto Principle: if any zⱼ = 1, the product Π(1−zⱼ) = 0, forcing y = 0 no matter how large the excitatory sum is. This is a strict, absolute veto — unlike modern networks where negative weights merely subtract.
FIGURE 1 — GENERAL MCP NEURON ARCHITECTURE
x₁ (+1) x₂ (+1) ⋮ xₘ (+1) z₁ (inhibit) ⋮ z_k (inhibit) Σxᵢ ≥ θ ? veto if any zⱼ=1 Θ(·) y threshold θ
Firing Algorithm PSEUDOCODE
// Require: x1..xm excitatory, z1..zk inhibitory, threshold θ S ← Σ xᵢ // sum of excitatory inputs if ∃ j : zⱼ = 1 then y ← 0 // veto: any inhibition blocks firing else if S ≥ θ then y ← 1 // excitatory sum meets threshold else y ← 0 return y
Logic Gates Lab — Build a Gate with an MCP Neuron
Toggle inputs, tune the threshold, and watch the excitatory sum, veto, and output update live — exactly as Section 8 of the handbook defines each gate.
Interactive Neuron LIVE
NEURON DIAGRAM
Toggle inputs above.
FULL TRUTH TABLE
Classical MCP (1943) vs. Modern Threshold Neuron
Textbooks often teach NOT/NAND/XOR with signed weights (e.g. w=−1, θ=−0.5) — that's a later, algebraically-equivalent generalization, not the original 1943 formulation.
General Correspondence TABLE
FeatureClassical MCP (1943)Modern Threshold Neuron
Excitatory strengthFixed, wᵢ = +1Arbitrary wᵢ ∈ ℝ
InhibitionAbsolute veto (separate synapse)Negative weight wⱼ < 0
Decision rule(Σxᵢ≥θ) ∧ (no inhibitor active)Σwᵢxᵢ ≥ θ
Bias / constant driveExplicit "always-on" bias neuronImplicit inside threshold value
LearningNoneNone (learnable in Perceptron)
Θ(S−θ)·Π(1−zⱼ)  ⇔  Θ(Σwᵢxᵢ − θ′)
NOT Gate: Bias+Veto vs. Signed Weight EQUIVALENCE

Classical: bias b=1 (excitatory, +1) supplies default firing; input A is a pure inhibitory veto, θ=1 → y = 1−A.

Modern: y = Θ(−A − θ′), with θ′ = −0.5, w = −1 → identical truth table y = A̅.

FIGURE 7 — NOT GATE: CLASSICAL vs. MODERN
CLASSICAL (1943) b=1 A (veto) θ=1 veto y MODERN (SIGNED WEIGHT) A w=−1 θ′= −0.5 y
NAND: Two Neurons vs. One COMPLEXITY CHANGE

Classical needs a two-neuron cascade (AND θ=2, then NOT θ=1) because a single equal-weight, veto-only unit cannot invert its own decision. Modern collapses this into one neuron: y = Θ(−x₁−x₂−θ′), θ′=−1.5, w₁=w₂=−1.

FIGURE 5 — NAND: TWO-NEURON CASCADE (CLASSICAL)
x₁ x₂ AND θ=2 inhibit b=1 NOT θ=0 y Neuron 1 (AND) feeds Neuron 2 (NOT) as inhibitory veto
XOR: No Complexity Change INVARIANT

XOR is not linearly separable in either formulation — a topological property of the function itself. Both the classical (OR, AND, veto-output — 3 neurons) and modern (2 signed-weight hidden units + OR output — 3 neurons) formulations require a hidden layer.

FIGURE 6 — XOR: TWO-LAYER MCP NETWORK
x₁ x₂ OR θ=1 AND θ=2 excite inhibit (veto) OUT θ=1 y hidden layer output layer
Summary of the Evolution
GateClassical countModern countChange?
NOT1 (bias+veto)1 (signed weight)No — same count
NAND2 (AND→NOT)1 (signed weights)Yes — classical needs +1
XOR3 (OR, AND, out)3 (hidden+out)No — both need hidden layer
Worked Examples
Threshold logic beyond the basic gates — majority voting, compound Boolean expressions, and the inhibitory veto in action.
Majority Function 3 INPUTS

Three excitatory inputs x₁,x₂,x₃ (weight +1 each), θ=2 — fires if at least 2 of 3 are active.

x₁=1, x₂=1, x₃=0 → S=2 ≥ θ=2 → y=1 (majority active)
Compound Expression: F = x₁·x₂ + x₃ 2-NEURON NETWORK

Neuron 1 (AND): x₁,x₂, θ₁=2, computes x₁·x₂. Neuron 2 (OR): output of Neuron 1 and x₃, θ₂=1, computes F.

x₁=1,x₂=1,x₃=0 → AND=Θ(2−2)=1 → S₂=1+0=1≥1 → F=1
x₁=0,x₂=1,x₃=0 → AND=Θ(1−2)=0 → S₂=0+0=0<1 → F=0
Inhibitory Veto Example SAFETY OVERRIDE

Four excitatory inputs x₁..x₄ (weight +1, θ=3) and one inhibitory "safety override" z.

x₁=x₂=x₃=x₄=1, z=0 → S=4≥3, no veto → y=1
x₁=x₂=x₃=x₄=1, z=1 → veto active → y=0  (even though S=4 ≫ θ=3)
Even a hugely excess excitatory sum is unconditionally overridden by a single active inhibitory input — the veto is absolute, never graded.
Limitations & Applications
Limitations ⚠️

• No learning mechanism — weights/thresholds are hand-designed
• Binary inputs/outputs only — no continuous signals
• Equal excitatory weights — biologically unrealistic
• Absolute (all-or-nothing) inhibition, not graded
• Linear separability constraint (e.g. XOR needs multiple layers)
• No rich temporal dynamics beyond synchronous discrete steps
• Fully deterministic — no probabilistic/noisy firing

Applications ✓

• Theoretical foundation of neural computation
• Digital logic circuit design (AND/OR/NOT/NAND/XOR ↔ neurons)
• Precursor to the Perceptron & deep learning
• Finite automata theory — MCP nets model finite-state machines
• Educational modeling of thresholding & decision boundaries

Summary

The McCulloch–Pitts neuron proved that networks of binary threshold units are computationally universal for propositional logic — a single neuron realizes AND, OR, NOT directly, while layered networks realize NAND and the non-linearly-separable XOR. Though superseded by learnable, weighted, continuous-valued models, it remains the first formal bridge between neuroscience, logic, and computation.