Your Brain Is Not An Information Processor

Except in the trivial sense that its a blood flowing machine, an electrical signal generator, and a glucose consuming system

The Brain As An Information Processor

There is this idea in cognitive science and the functionalist literature, that the mind/brain is fundamentally an ‘information processor’. This is meant as an ontological claim, not a descriptive one (see here about this distinction).

Whether this claim is meant ontologically or descriptively has serious ramifications. If meant descriptively, it is of little consequence. Sure the brain processes information, but it does a lot of other things too.

The weight of the ontological claim, however, is that it claims to tell us what a brain fundamentally is. All sorts of further implications follow from this too, such as that we will have explained all there is to know about the mind/brain by showing how the everything about the mind ‘reduces’ to processing information.

Another way of putting this point is that the ontological ‘information processing’ view claims that its way of viewing the brain ‘carves the brain at its joints’, as its were, namely, that it picks out something real and fundamental about the nature of the brain, and of neural activity. The brain is an information processor

So, here is the thing: this claim needs defending. Because it is not a trivial claim.

Why Believe In This Ontology?

The core argument for a belief in this ontological view is—dare I say it— a belief in functionalism: The view that the brain is a complex input output system, in which consists, if the system has the right functional organization, minds.

If you believe that much about the brain, then you can say that it generates the following basic picture of the brain: the brain is a complex input-output system that receives information, stores and uses that information in various ways, and retrieves that information to produce behavioral outputs.

Of course, one does not need to commit to the information processing view if they are a functionalist (there is an open question how to cash out the claim that a system must have the right ‘functional organization). In practice, however, a lot of functionalists do uphold this view of the brain, and an interesting question is why.

I suspect it has to do with the history of how a variety of different fields, including computer science, engineering, and philosophy intersected in a variety of different ways and borrowed conceptual vocabulary from one another, and in ways that to this day, are not made explicit or often justified.

All this to say, the ontological claim that the brain is fundamentally an information processor is rarely, if ever, defended on its own terms.

Instead, it sort of piggybacks on functionalism (a view that I believe is flawed in many ways and have written about here). If you already believe that minds consist in the right kind of causal-functional organization, then describing the brain as a system that receives, stores, and retrieves information can seem like a natural and innocent further step. Except that it isn’t.

It is a substantial additional commitment, one that turns a specific computational vocabulary into an ontological claim about neural reality and brains.

Functionalism by itself does not tell you that ‘information processing’ is the right way to cash out functional organization. But even if it did, functionalism is a view that I think we should reject for other reasons.

The problem with the ontological view of the brain as an information processor, is that it is deeply confused. This should only be understood as descriptive claim.

The Descriptive View vs The Ontological View

Let’s consider an easier case: the heart. We can all agree that the heart pumps blood. We can helpfully describe the heart as a pump— it receives blood, propels throughout the body, and maintains the blood in circulation. None of this problematic as a descriptive claim about hearts. But this claim about hearts does not exhaust what the heart is. Nobody thinks that this description of the heart implies that the heart is a blood pumping system, and that to understand everything about the heart we need to show how that thing reduces to blood flow.

Rather, it depends on what you are trying to understand about the heart whether you choose to focus on that description of the heart, and not another one. The heart, for example, is also an evolved biological organ with a developmental history, it is embedded in a hormonal and nervous system, it is capable of being damaged by emotional stress in medically documentable ways, and so on.

None of this is captured in the description of the heart as a ‘pump’ which is only problematic if we understand that descriptive claim as an ontological one.

The same is true of the brain. As a descriptive claim, it is of course true that the brain processes information, but it also does a lot of other things.

It also (1) flows blood through it (2) consumes glucose (3) produces heat (4) uses oxygen (5) generate electric signals

I could go on and on…

None of these claims are very interesting as descriptive claims, because there is nothing to understand or ask about them. They are all trivially true.

Now, let’s translate them to ontological claims and see whether anyone would be willing to defend them…

ontological claim1: the brain is a blood flowing system

ontological claim2: the brain is a glucose consuming system

ontological claim3: the brain is an oxygen using system

ontological claim4: the brain is an electrical signal generator

If all of these ontological claims sound ridiculous to you it’s because they are.

Nobody finds any of these claims worth basing a theory of the mind/brain on, because we would not find any interesting results about the mind/brain if we did this. But here’s the punchline: this is also true of the ontological claim that mind/brains are information processors.

The only difference between the information processing view and the blood flow view, glucose consuming view, or oxygen using view, is that our obsession with AI’s and machines obscures from view that there is, in fact, no difference.

A Speculative History of the Ontological View

Why are we convinced of the information processing view as an ontological thesis? It’s an interesting question that I won’t get into here (I need to do more research on the topic).

But as already mentioned, I suspect it has a lot to do with the history of computers and the idea that, by comparing computers and brains, we might be able to better understand the brain. As a descriptive claim, I think that is true. I have some examples below about how thinking of the brain as capable of computing has vastly advanced our understanding of the brain. As an ontological claim, however, it is plain misleading, confused, or false.

Three important points of history seem to be (1) in mid 20th century, people made such claims as the brain is a digital computer (this was called the strong identity thesis). Digital computers were new and interesting. Mind/brains were poorly understood. So some thought, why not claim there is a relation of identity between the two, that a brain is a computer. One hypothesis is that identity relations get us into ontological rather than descriptive territory, get us into confusion.

(2) at around the same time, the work of Marr, Newell, and others appears to have been interpreted as saying that we can think of the brain as an information processor in the ontological sense, though I think these authors should be read as claiming we can helpfully understand the brain as a information processor in the descriptive sense (perhaps a matter of interpretation I won’t get into here).

(3) Turing, Church, Godel and others did very important work understanding computation, work that, as we will see shortly, was essential to making advances in how we understand the brain (though not, I believe, in the ontological sense)

Case Studies: Descriptive vs Ontological Claims on Brains

As mentioned, in saying all of this, I am not saying that Turing and others work on computers was not relevant to making great strides in understanding the brain. It certainly was. One can do excellent scientific work by assuming that the brain is a computer in Turing’s sense (Gallistel and King’s work, which I discuss below, believes this), but this claim should not be interpreted ontologically.

It should be interpreted descriptively and as a useful methodological tool.

Because here is the thing: any serious theory in the sciences of the mind/brain takes for granted that the brain processes information. That’s a trivial uncontested and uninteresting fact. The interesting facts emerge when you develop a theory of some particular capacity of the mind/brain and use that theory to learn something about how the works. Here are two examples.

Example 1: Understanding Human Language

Merge (which I’ve written about here) is a biological computation of the human mind/brain, a generative procedure that produces in human beings an infinite set of hierarchically structured linguistic expressions.

This computation (Merge) is well described, and there is now neurobiological work showing that there is a neural basis for this computation.¹

It is also trivially true that Merge involves ‘processing information’ . The relevance of this fact, however, is precisely zero for offering insights or explanation into how human language works.

Presumably, blood also flows through the brain as Merge produces linguistic thoughts, electric signals also get generated, and so on. The facts are true but irrelevant to actually understanding how the brain works.

The point of ontological interest is not ‘information processing’ but Merge, the distinct computation that renders language acquisition and use possible, and that is not known to exist in any other organism or machine.

Example 2: Understanding Human Memory

Randy Gallistel and Adam King (a neuroscientist and mathematician respectively) are interested in understanding how human memory functions.² Their work is an excellent example of what working science does and does not commit us to about brains. They argue that there must be a symbolic read/write memory mechanism in the brain — analogous in important ways to the memory systems of a computer — that encodes information from past experience into symbols, stores it, and retrieves it in a computationally accessible way when needed.

The motivation for this view is straightforward: the authors argued that the information required to act effectively in the world is not explicitly present in the raw sensory signals the brain receives. Something in the brain must therefore make that information explicit, carry it forward in time, and make it available to whatever computational machinery needs it for memory to work.

They invoke Shannon’s theory of communication to define information precisely, and assume the brain implements something like Turing computation. But notice what they are actually doing: they are using these computational concepts as descriptive and methodological tools to understand a specific capacity of the brain (memory). They are not claiming that ‘information processing’ tells us what the brain fundamentally is. They are working scientists trying to understand a particular system, and computation is their most useful tool for doing so.

This is the ways in which work in computer science and engineering has been an imperative advance for understanding the brain. The advance was methodological.

The relevance of believing in Turing computation is descriptive, not ontological.

The authors are not concerned with claiming anything about minds. They, like most working scientists, are interested in understanding the brain.

Similarly, one generally assumes that regardless of how exactly this memory mechanism is instantiated in brains, the brain will be ‘processing information’ along the way, as will it be ‘flowing blood through it’, ‘generating electricity’, and so on, for the many different things brains do on a daily basis.

Why Confusing Description for Ontology Is A Disaster

Confusing descriptive claims for ontological ones is not just a philosophical error, it has very real and practical consequences. In AI and philosophy of mind discourse, this pattern of mistake is very common. Importantly, it forms the bedrock on which the overwhelming majority of claims about AI minds are built.

Once you treat ‘the brain is an information processor’ as an ontological truth rather than a useful description, the conceptual work appears to be done. There are no harder questions left to ask. It is all just science now. But nothing could be further from the truth.

The irony is that this move, which presents itself as scientifically rigorous, actually prevents us from doing the kind of conceptual thinking that serious science absolutely requires. AI debates conducted in this register are not just making philosophical errors. They are misinterpreting the scientific work they claim to be grounded in. They also do so with irreverent confidence, dismissing in the process entire scientific disciplines, philosophical traditions, and hard-won ways of thinking about what minds and brains are, and how they work.

That is not a small cost. It is a big one. Soon, it will likely be an expensive one too.

Photo credit

Image by Gerd Altmann from Pixabay

Thanks for reading! If you found this useful, please consider subscribing! I post every Wednesday on AI consciousness and philosophy of mind.

Want to support this work? Consider a paid subscription. Currently, all my posts are free because accessibility matters to me, but paid subscriptions allow me to keep writing rigorous, informed philosophy that cuts through AI hype. Every paid sub helps

1. See, for example Zacarella, Emiliano, Meyer Lars, Makuuchi Michiru, and Friederici, Angela D. ‘‘Building by Syntax: the Neural Basis of Minimal Linguistic Structures’’. Cerebral Cortex 27, no.1 (January 2017): 411-421

2. Charles Randy Gallistel and Adam Philip King, Memory and the Computational Brain. Why Cognitive Science will Transform Neuroscience (Malden, MA: Wiley-Blackwell, 2009)