9. A Logical Calculus of the Ideas Immanent in Nervous Activity
McCulloch and Pitts were apparently influenced by Russell and Whitehead's Principia.
This is an interesting paper because it represents the other direction that is taken in computer science when trying to model intelligence. There is the logic first approach, which you see from Boole, Turing, etc., where you start with the basics but have utmost certainty. The other is trying to go from the nature first approach. Nature can mean human thinking, thinking as such, etc.
The authors try and build an abstraction of how neural nets work, way ahead of their time in a sense, that can perform computations. From here and here, as some recent examples I saw, the brain is extremely complex.
It is also unclear that the brain actually computes, if what the brain does can be called computing. The logic, math, and philosophy papers we have seen here that outline what it means to compute work. But it is unclear still how the brain works, there may be some aspects that do computation, but the brain is inherently fuzzier in terms of logic than the logic we do. The brain is quite capable of being illogical.
Computers do roughly two things, calculations like we see in math but also logic like we see in Boole or Aristotle. These things collapse at an abstract enough viewpoint into the same thing and are enabled via Shannon's circuits, von Nuemann architecture, etc.
But so much abstraction has already taken place that we forget that this might just be a partial picture. In the realm of philosophy, for something to be computable it must be intelligible. But even to be intelligible potentially discounts much of human experience and thought. Already, this is a contentious philosophical position. Philosophy aside, computation does work, it can be proved, it needs no philosophical defender.
The closer we come to approximating human thinking, the more it seems folks forget this. This forgetting is very similar to Heideggerian forgetting. I don't really know what the ramifications are of this, but it is all very interesting and I don't think there are clear answers yet theoretically or practically.
I do strongly believe that human experience or thinking in the broadest sense is not encompassed by the LLM or the computer.
Author: Warren McCulloch and Walter Pitts
Original date: 1943