The journey from a biological brain to an artificial one begins with a simple but powerful idea: a neuron can be modeled as a basic computational device. The McCulloch-Pitts neuron, or M-P neuron, was the first formal attempt to do this. It strips away the complex biology to its logical core, treating a neuron as a simple decision-maker that produces a binary output, either a 1 (fire) or a 0 (don't fire), based on binary inputs.

Inputs to an M-P neuron are not all created equal. They fall into two categories: excitatory and inhibitory. Think of excitatory inputs as votes for firing. The more excitatory inputs that are active (equal to 1), the more likely the neuron is to fire.

Inhibitory inputs work differently. They are absolute vetoes. If even a single inhibitory input is active, the neuron will not fire, regardless of how many excitatory inputs are screaming "yes."

The M-P neuron's decision-making process is a straightforward two-step rule. First, it checks for any vetoes. Are any inhibitory inputs active? If so, the output is immediately 0. End of story.

If there are no active inhibitory inputs, it moves to the second step. It counts the number of active excitatory inputs. This is the neuron's aggregation function, a simple summation. It then compares this sum to a predefined value called the threshold, often represented by the Greek letter theta ($\theta$). If the sum of active excitatory inputs is greater than or equal to the threshold, the neuron fires and outputs a 1. If the sum is less than the threshold, it remains silent and outputs a 0.

An important feature of the original M-P model is that the connections, or weights, are fixed. The model doesn't "learn" by adjusting weights like modern neural networks. The excitatory inputs are typically treated as having a weight of 1, and the inhibitory inputs have their own separate logic. The threshold $\theta$ is also set in advance. This makes the M-P neuron less like a learning cell and more like a pre-programmed logic gate, capable of performing basic operations like AND, OR, and NOT.