A response to “A response to `The Empty Brain` — The Information Processing Brain”

It’s a funny feature of the scientific community that they often behave so dogmatically; which almost always helps them get work done, but inevitably holds back a quantum leap in progress until the dogmas are rejected.

Normal science, the activity in which most scientists inevitably spend most all their time, is predicated on the assumption that the scientific community knows what the world is like. Normal science often suppresses fundamental novelties because they are necessarily subversive of its basic commitments. As a puzzle-solving activity, normal science does not aim at novelties of fact or theory and, when successful, finds none. — Thomas Kuhn

As a computer scientist, I have long been skeptical of the notion that our brain “was basically a computer”. This came about mostly because things that are super-duper easy for us subjectively aligned, ego-bearing “minds” — are extremely difficult for computers, and vice versa.

An infant can recognize its mother’s face, but it took clever scientists and a TON of computing power decades to recognize faces; while an ancient computer can calculate large (but trivial) equations with ease, while its easy to construct one that is virtually impossible for any given genius to do in their mind.

In short, they (our minds, and computers) seem at essentially opposite ends of a spectrum in terms of the types of things they are good at.

So I was happy to see the aeon article by Robert Epstein, talking about how our brain is actually not a computer. https://aeon.co/essays/your-brain-does-not-process-information-and-it-is-not-a-computer

I was annoyed slightly (in that way that happens when people fail to see your point of view) when a philosophy-friend of mine responded with Pendergrass’ response, defending the “Information Processing” model with a point by point breakdown that surely pleases those within the same paradigm, but does little to really defend the case. http://lukependergrass.work/blog/the-information-processing-brain

REASON NUMBER ONE: WE HAVE INNATE REFLEXES.

Pendergrass concludes this point with “ Why couldn’t it be the case that this reflex phenomenon is explained by an innate heuristic for processing stimuli? Why can’t that be a computational process?”

I concede that there is no good reason why this could not be based on some innate heuristic; but would counter with the equally-valid, “why couldn’t it not be the case?”

There is no valid criticism here either way, just both sides claiming that their paradigm does not fall to this criticism.

Pendergrass 0 — Epstein 0

REASON NUMBER TWO: COMPUTERS CODE INFORMATION INTO BYTES. WE DON’T.

A computer codes information into bytes. These byte patterns represent information. Computers then move patterns place to place. Therefore: “computers really do operate on symbolic representations of the world. They really store and retrieve. They really process. They really have physical memories.” But, concludes Epstein, we don’t do this. Therefore, the IP thesis is wrong.

Nowhere does the IP thesis assert that we represent the world in specific, byte-like patterns.

This looks like an extremely weak counter to me by Pendergrass. It doesn’t appear to me that Epstein is focused on specifically “byte” representation; but rather “the data is represented in patterns” and, “those patterns actually move around the machine to different parts”, and “we can point at these symbolic representations of data”, and therefore, “we can actually point at the processing of information”.

Pendergrass goes on to say, “ An analog amplifier processes and manipulates information, but by definition doesn’t represent that information in bytes.”

But this in no way supports the idea that our brains process and manipulate information that is symbolically stored.

Of course, we have “neural networks” in the computing sense, but lets not take that metaphor too seriously either; there is no (known) isomorphic relationship between our own neuronal structure and the well-known method of problem solving in computer science known as neural networks.

Given that symbolic representation of information moving around different components inside a computer is a primary factor in what makes a computer work, and there is zero ability on the part of neuroscientists to point to symbolically represented data in our brains in any way (at least, currently), this point has to be rewarded to Epstein.

Pendergrass 0 — Epstein 1

REASON NUMBER THREE: THE IP METAPHOR WILL EVENTUALLY BE REPLACED BY A BETTER METAPHOR OR BY ‘ACTUAL KNOWLEDGE’.

This, in my opinion, is the best part of Epstein’s article, as it really shows how paradigms and dogmas insidiously affect even the most honest, earnest scientists and knowledge seekers.

Pendergrass concedes, “ we should not allow the IP metaphor to blind us to other possible empirical explanations.”

Smartly though, he goes on

But what exactly is “actual information”? Are we to understand the brain only in terms of rushes of ions moving in and out of neurons that result in action potentials and a cascade of synapses? Is this what actual knowledge looks like? Abstraction and metaphor will always be a critical part of science, and it shouldn’t be a point against the IP thesis that it is only “the best we have so far.”

I agree, it shouldn’t be a point against, but certainly not a point for.

Pendergrass 0 — Epstein 1

REASON NUMBER FOUR: THE IP THESIS IS BASED ON BAD LOGIC.

I will take Pendergrass at his word here that his colleagues would not make the simple logical mistake that Epstein claims, but do have to take issue with claims about neuronal modeling.

It’s much more natural to say that the motivation for the IP thesis comes from our ability to model the behavior we’ve seen from neurons — all the way down to the atomic level.

There are several reasons why this is a little sketchy:

• The IP thesis was dominant before our ability to model the behavior of neurons.
• Though we have models of neurons and synapses, we have no holistic model for how these things play together to create a mind.
• It is very likely that our minds are working “below” the atomic level (http://phys.org/news/2014-01-discovery-quantum-vibrations-microtubules-corroborates.html)

In any case, this looks to me like another moot point.

Pendergrass 0 — Epstein 1

REASON NUMBER FIVE: WE CAN DRAW A DOLLAR BILL BETTER WHEN WE HAVE ONE AS A TEMPLATE TO WORK OFF OF.

This is one of the stronger arguments against the IP paradigm in my opinion.

Pendergrass whinily responds: “First of all, why would the IP thesis have to hold that any representation is perfect?”

Well, because, you are using a computer metaphor. Imagine if your photos on your hard drive slowly faded and contorted, until they became completely different things that never even happened, or so down-resed they were not recognizable.

Further, he says, “ Even more concerning is the claim that the IP thesis implies the precise localization of individual representations.”

If the IP thesis does not claim this, then what does it claim? It appears to me if you are going to claim a metaphor is so poignant you will use it as a thesis, some parts of the metaphor ought to hold up to scrutiny and this one does not.

Computers store data symbolically, it is localized, and unless there is corruption or bitrot, it is as a rule, a perfect storage.

Our minds are exactly the opposite, data is not stored symbolically (at least, we cannot point to it), it is not localized (at least, we cannot point to it), and its retainment as a general rule is imperfect.

Pendergrass 0 — Epstein 2

REASON NUMBER SIX: COGNITIVE FUNCTIONS ARE REALIZED BY SPATIALLY DISTRIBUTED SYSTEMS IN THE BRAIN.

Again, Pendergrass seems to think “Why can’t X” is a positive argument toward X, when in reality it is at best a neutral statement.

Why can’t information processing happen in virtue of a lot of different distributed components? Put another way, what about the IP hypothesis commits it to a simplistic and specific view of functional localization?

It appears to me, using a metaphor of a computer is what commits you to it. Other than, perhaps, brains — all known information processing devices absolutely have localized functions. Each part of the device very clearly has a job, and data flows from one thing to another, each doing necessary, separate functions.

Even more interestingly: “no computer scientist advances that an individual jpeg is stored in a particular circuit.”

This is just silly, because

1. It would be trivial to construct a circuit to hold an individual jpeg.
2. We don’t need to do that, because we have non-circuit ways of storing individual jpegs; in a localized place, on a device with a specific function.

Pendergrass 0 — Epstein 3

REASON NUMBER SEVEN: IT IS MORE ACCURATE TO SAY THAT THE BRAIN ‘CHANGES’ RATHER THAN THE BRAIN ‘STORES’.

If the IP thesis is wrong, what alternative does Epstein offer? His alternative model:
“(1) we observe what is happening around us (other people behaving, sounds of music, instructions directed at us, words on pages, images on screens); (2) we are exposed to the pairing of unimportant stimuli (such as sirens) with important stimuli (such as the appearance of police cars); (3) we are punished or rewarded for behaving in certain ways… the brain has simply changed in an orderly way.”

A proponent of the IP thesis could just as well agree with this abstract picture of learning and brain development.

I agree with Pendergrass here; Epsteins alternative model is (at least stated here) almost perfectly vacuous.

Indeed, it would be easy for an IP proponent to agree with this, because it doesn’t really say anything at all.

Pendergrass 1 — Epstein 3

REASON NUMBER EIGHT: THERE HAVE BEEN NO INTERESTING FINDINGS FROM THE IP THESIS.

Pendergrass responds

This is, frankly, nonsense. If cognitive neuroscience has been a slave to the ‘useless’ IP metaphor for so long, how can the massive leaps and advances in science and psychology throughout this period be explained?

This is, frankly, nonsense. As Kuhn showed us, we can do science while within a paradigm. It is unlikely that Epstein is arguing that neuroscientists are doing nothing at all, but rather, that IP Metaphor is contributing nothing to that work.

A neuroscientist that adheres to the IP metaphor-paradigm, and is doing work looking at individual neurons with some specific protein binding is advancing knowledge and contributing greatly; but no where in the work does it necessarily dictate that the brain is acting like a computer.

Pendergrass 1 — Epstein 3

In conclusion

I’ll leave it to the reader to decide whether or not the IP metaphor holds weight.

For myself, I find it to be an extremely weak metaphor that, at least to laymen, does more harm than good. We know computers are automatons, incapable of a myriad of internal states that us subjective beings experience.

When we have people believing that we are just computers as well, it unnecessarily invalidates or paints over a huge spectrum of the human experience. It leads people to assume some of their most basic, intrinsic experiences are meaningless/random/preordained because “if we are a computer than they must be — and scientists say we are a computer.”