Confusion about computation and brains

In The empty brain Robert Epstein claims that the brain isn’t a computer and it doesn’t process information. He is wrong and he is quite badly confused about computation and information and I’m going to explain why.

Epstein writes some stuff about babies and then makes his main argument:

Senses, reflexes and learning mechanisms – this is what we start with, and it is quite a lot, when you think about it. If we lacked any of these capabilities at birth, we would probably have trouble surviving.

But here is what we are not born with: information, data, rules, software, knowledge, lexicons, representations, algorithms, programs, models, memories, images, processors, subroutines, encoders, decoders, symbols, or buffers – design elements that allow digital computers to behave somewhat intelligently. Not only are we not born with such things, we also don’t develop them – ever.

We don’t store words or the rules that tell us how to manipulate them. We don’t create representations of visual stimuli, store them in a short-term memory buffer, and then transfer the representation into a long-term memory device. We don’t retrieve information or images or words from memory registers. Computers do all of these things, but organisms do not.

Epstein states that the way the brain works is very different from the workings of any computer we can currently construct. For example, I can store a picture of a dollar bill on my computer and it will be exactly the same every time I retrieve it. A person’s memory of a dollar bill doesn’t work the same way. A person will in general interpret and recreate his memories of a dollar bill each time he thinks about a dollar bill. Epstein claims that large regions of the brain light up when any particular memory is recalled and so the information about the memory isn’t stored in a particular place, which he claims makes the brain different to a laptop. And since the brain works in a different way from a laptop it’s not a computer. He also states that talk of the brain as a computer is just an analogy.

Epstein doesn’t understand the theory of computation or the theory of information. He states that the brain doesn’t contain information, but he doesn’t bother to discuss what this would entail. 

A physical system instantiates information about X when there is a correspondence between X and that physical system. For example, if you write a book about how to make jam, there is a correspondence between the book and the physical properties of fruit, sugar and so on that allows you to use the book to gather the ingredients used to make jam and then make it successfully. The instruction manual for a motorcycle doesn’t have much correspondence with the properties of fruit, sugar and so on and reading it will not enable you to make jam, so it contains no information about jam. 

Once you have learned how to make jam you won’t need the book as much as you did when you first learned to make jam and you might not need it at all. Something in your brain corresponds to the properties of fruit, sugar and so on in a way that allows you to make jam. So then your brain instantiates information about making jam. All of your senses, reflexes, learning mechanisms and so on are also information about what is happening in the world or ways of solving problems. So your brain instantiates information.

A physical system processes information if it takes that information and produces new information. For example, the following Clojure function takes two numbers a and b and processes them to produce the sum of those numbers:

(defn add-two-numbers [a b] (+ a b))

The main problem with Epstein’s argument is that the theories of information and computation don’t refer to the physical location of the information or the physical means by which it is processed and there is a good reason for that: it doesn’t matter very much. Once a device can store information and perform some very simple computations on that information can do any information processing that doesn’t involve quantum mechanical effects like entanglement and single particle interference – classical information processing. And not only can such a device produce the same output as any other classical physical system it can process the information by performing the same series of steps. Your brain is a warm wet lump of matter that constantly interacts with its environment and such interactions suppress quantum mechanical effects in the brain so it’s only doing classical information processing. As such a computer can simulate the information processing your brain is doing given enough resources. So unless Epstein can refute the laws of physics, the brain’s information processing can be simulated and understood in terms of classical computation.

There are many reasons why brains haven’t been simulated very successfully. We may not yet have enough resources to do a good job. Also, simulating a human brain requires understanding how people create knowledge. We understand that knowledge is created by evolutionary processes that produce variations on existing knowledge and select among those variations, but the details of how that works aren’t understood. Our understanding of programming – actually writing descriptions of information processing that can be executed by computers – is also bad enough that there are large disagreements among programmers about the best way to do it. We may simulate brains in the future. We may also be able to make knowledge creating programs without understanding how the brain works well enough to simulate it because many of the details of the brain’s activities may be irrelevant to the knowledge creation algorithm.