Understanding vs. Computation
Physicist Roger Penrose argues that human consciousness—specifically our ability to understand mathematics—cannot be replicated by a computer. To understand why, we must look at how computers actually work.
1. The Formal System
A computer operates within a "formal system." It is given a set of axioms (starting facts) and a set of rules. It blindly applies those rules to generate new facts. It does not know what the symbols mean; it only knows how to move them.
Axiom: 0
Rule: To get the next truth, add 1.
If we ask this machine to prove that 4 is a valid truth, it simply runs the rule until it hits 4. This is computation. It is predictable, mechanical, and entirely contained within its rules.
2. Gödel's Trap
In 1931, Kurt Gödel discovered a fundamental limit to these mechanical systems. He proved that for any set of rules, you can construct a mathematical statement that essentially says:
"This statement cannot be proven by these rules."
Let's call this statement G. What happens if we feed G into our formal computer and ask it to prove if it is true or false?
Statement G: "I am not provable by the Machine."
Notice the trap. If the machine proves G is true, then G is provable, which makes the statement false. A contradiction. If the machine proves it false, it means it can be proven, another contradiction. The machine is paralyzed.
3. The Human Leap
The machine cannot compute an answer. Its rules fail it. But now, read statement G again yourself.
"This statement cannot be proven by the Machine."
Because you just watched the simulation, you know for a fact that the machine cannot prove it. Therefore, looking at the statement from the outside, what is its status?
This is Penrose's argument. A computer operates entirely within a system of rules. Human intelligence has the capacity to step outside the system, look at the rules themselves, understand their meaning, and see truths that the rules cannot generate.
Understanding is not a computational process. It is something else entirely.