Evolutionary Philosophy
  • Home
  • Worldview
    • Epistemology
    • Metaphysics
    • Logic
    • Ethics
    • Politics
    • Aesthetics
  • Applied
    • Know Thyself
    • 10 Tenets
    • Survival of the Fittest Philosophers >
      • Ancient Philosophy (Pre 450 CE)
      • Medieval Philosophy (450-1600 CE)
      • Modern Philosophy (1600-1920 CE)
      • Contemporary Philosophy (Post 1920 CE)
    • 100 Thought Experiments
    • Elsewhere
  • Fiction
    • Draining the Swamp >
      • Further Q&A
    • Short Stories
    • The Vitanauts
  • Blog
  • Store
  • About
    • Purpose
    • My Evolution
    • Evolution 101
    • Philosophy 101

Consciousness 13 — (Rethinking) The Attention Schema

4/8/2020

8 Comments

 
Picture
Graziano with his ventriloquist puppet orangutan named Kevin. Consciousness studies sure do draw renegades.
In the last post, I noted that Dr. Ginger Campbell conducted one-on-one interviews with three prominent neuroscientists during the final episodes of her Brain Science podcast series on consciousness. We've already covered the first interview with Joseph LeDoux. Today, I'm going to go over the second interview with Michael Graziano about his book Rethinking Consciousness: A Scientific Theory of Subjective Experience. Graziano is currently a professor of Psychology and Neuroscience at Princeton University where he has had a lab studying consciousness since 2010. Here are the highlights from his interview:
  • In 10 years of lab work, I have worked to put my ideas into an evolutionary context (i.e. how they developed), in order to give us an idea of the components that go into this thing we call consciousness.
  • More and more, people in the science of consciousness are beginning to coalesce around a coherent set of ideas. My work fits into this growing standard model of consciousness. This core set of scientists realise that we are machines and the brain is an information processing machine that thinks it has magic inside it because it builds somewhat imperfect models of the world inside it. This includes Higher Order Thought Theory, Global Workspace Theory, and even some Illusionists who talk of consciousness as an illusion. My theory is not a rival to these. We are moving past rivalry and towards an integrating picture of it all.
  • The realisation is coming that everything you think derives from information. No claims can be put out by the brain without information upon which to base it. This is just basic logic. The question then is how and why did the brain construct a particular piece of information? The brain can construct all sorts of seemingly crazy ideas (e.g. “I have a squirrel in my head instead of a brain.”)
  • I study movement control, which requires a whole model. If the brain wants to control the arm, it needs a model of the arm. It needs an internal model, a simulation of what an arm is and where it is at any one time. This is an engineering perspective, which is useful for the study of consciousness. Similar to the moving arm, the brain is continually shifting its focus of attention. So, how do you control that? The same way as the arm. The brain needs a model or simulation of attention, of what it means to focus resources on something.
  • This is called “attention schema theory”, which follows the “body schema” developed 100 years ago. Phantom limbs are good examples of “body schema”. By analogy, there must be a schema for attention—the brain's model for seeing information and processing it deeply.
  • Like all complex traits, you can go back very, very far and see this gradual transition where it becomes impossible to draw a line and say “the trait exists after this but not before this.” For example, you couldn't draw clear lines in evolution for hands, feet, and flippers. Consciousness is the same.
  • I start with attention—a basic ability of a nervous system to focus on a few things at a time and process them deeply. Some forms of this attention go back possibly all the way to the beginnings of nervous systems. Attention is at the root of intelligence. At the heart of intelligence is a very pragmatic problem: you only have so much energy and space for a brain, but you need to use it as efficiently as possible to process deeply and intelligently. How do you do that? Don’t occupy the brain with processing all of the million and a half things going on around you. Focus on one or two things at a time. Without that level of attention, any kind of intelligence is impossible.
  • Attention comes in very early in evolution, and over time it becomes more and more complex. There’s central attention, sensory attention, more cognitive kinds of attention, and they emerge gradually over this sweep of history from about half a billion years ago up to the present. Piggybacking off of this, what people call consciousness also emerged, and also as a gradual process.
  • Attention can be separable from consciousness. At what point might it be consciousness?
  • Bodies have been involved from the beginning. Schemas only came once nervous systems were capable of building models of these bodies. A body schema stands hierarchically above the body. It isn’t the same thing, and they can be dissociated (e.g. phantom limbs). Similarly, this is the relationship between attention and consciousness. Attention is literally what the brain is focusing its resources on. The Attention Schema is what the brain thinks it is focusing its resources on, what the brain thinks focusing is, and what the brain thinks the consequences of focusing are. And those are dissociable too. Typically, they don’t. Typically, they track quite well (like the body schema), but you can trick them and get them to peel off from one another.
  • Global Workspace Theory is basically a theory about attention. How do you become conscious of an apple you are looking at? GWT says you attend to the signals. They become stronger from your visual system at the expense of other signals. At some point, the signals become so strong that they reach a state called “ignition” when they can then influence wide networks around the brain. Now attention has been reached, you can talk about it, you can move toward it, you can remember it later. The apple information reaches the global workspace and becomes available all around the brain systems. GWT says that is consciousness. The weakness of GWT is that it doesn’t explain why we claim to have a subjective experience. It doesn’t say why I have an inner experience of the apple.
  • The attention schema says great for GWT, but you need one more component—a system in the brain that says “Ah, I am attending to the apple. I have a global workspace that has taken in that apple information.” You need something in the brain that can model itself and build some kind of self-description. GWT is the attention. Attention Schema is the consciousness riding on top of that.
  • To control something, you need a model of it. But an overly complicated one is wasteful. A “cartoonish” one is good enough.
  • Why does it feel non-physical? This is one of the most successful points about the Attention Schema. The brain models itself, but it doesn’t need to include little physical details. It doesn’t need to know anything about the little implementation details. Therefore, the brain’s self-models depict something that has no physical components. It depicts a vague non-physical thing that has a kind of location within us, but that’s the only physical property it has. Efficiency dictates the models be as stripped down as possible. This is why introspection, informed by internal models, tells us there is something inside us but it feels like a non-physical essence.
  • With this Attention Schema, we don’t need another explanation for the philosopher’s qualia because there it is. Chalmers, after the Hard Problem, now talks about the Meta Problem. The Hard Problem is how do we get qualia, or that inner subjective feeling. The Meta Problem is why do we think there is a Hard Problem? The Attention Schema solves the Meta Problem. It explains why people think there is this magical non-physical thing inside us. It does an end run around the Hard Problem.
  • The ability to attribute consciousness to others is important. In this evolutionary process, we start out evolving an ability to model and keep track of ourselves, which helps make predictions about ourselves and control our behaviour. At some point, as social interactions become more sophisticated, we develop the ability to use the same machinery to model others. This social use probably came in very early in evolution. There is a lot of sophistication in reptiles, birds, and mammals. We not only keep track of and model our own attention, but we keep track of and model others’ attention. That allows me to predict your behaviour.
  • Ventriloquist dummies are great examples of our souped-up drive to model conscious minds in the world around us.
  • We seem to model attention as if it were a fluid flowing out of their eyes, which explains all kinds of folk beliefs about feeling eyes on the back of the neck, telekinesis, the Force in Star Wars, the evil eye, etc., etc.
  • Integrated Information Theory is kind of the opposite of this. IIT belongs to theories where you start with an axiomatic assumption. IIT starts with “consciousness exists” stating there is this non-physical feely thing inside us. The magical thing is there, so how does it emerge and under what conditions? So right from the outset there is a divergence. On my end, the starting point is that the brain cannot put out a claim unless there is information for that claim on which it is based. There is no reason to assume this information is accurate. When people feel they have magic, the job of scientists isn’t to find out how the brain produces magic; it’s to find out why the brain builds that model to describe itself. IIT is a fundamentally magical theory.
  • According to IIT, consciousness arises from information and everything in the universe has some information in it. So, you end up with panpsychism that consciousness exists in everything and everywhere. That seems like you’ve used faulty logic to paint yourself into a corner. If everything is conscious, what does consciousness even mean anymore?
​
(Not So Brief) Brief Comments

When Graziano opened his interview talking about putting consciousness into an evolutionary context, he had me hooked. When he stated the field was coalescing around a growing standard model of consciousness that brought together Higher Order Thought Theory, Global Workspace Theory, and even some Illusionists, I got excited because those were the theories I most agreed with in the prior posts in this series. When Graziano said this core set of scientists think that we are machines and the brain is an information processing machine that thinks it has magic inside it because it builds somewhat imperfect models of the world inside it, this made a lot of sense. But when Graziano tried to offer his picture to integrate all of this, he finally lost me. To see why, let me go through some of his points one by one.

>>> "No claims can be put out by the brain without information upon which to base it."

This is an excellent place to start. I'll use this later in the series when making connections between the evolution of consciousness and evolutionary epistemology, which charts the way knowledge-gathering has grown incrementally over evolutionary history.

>>> "If the brain wants to control the arm, it needs a model of the arm. It needs an internal model, a simulation of what an arm is and where it is at any one time. This is an engineering perspective, which is useful for the study of consciousness. Similar to the moving arm, the brain is continually shifting its focus of attention. So, how do you control that? The same way as the arm. The brain needs a model or simulation of attention, of what it means to focus resources on something. ... By analogy, there must be a schema for attention—the brain's model for seeing information and processing it deeply."

I believe Graziano is making a poor analogy here. When an arm moves, it moves through space and time by contracting muscles that cannot see anything. When a focus of attention shifts, no such physical movement or navigation issues occur. I think it's a mistake to think of models being required to control both of these different things in the same kind of way.

>>> "Attention is at the root of intelligence. At the heart of intelligence is a very pragmatic problem: you only have so much energy and space for a brain, but you need to use it as efficiently as possible to process deeply and intelligently. How do you do that? Don’t occupy the brain with processing all of the million and a half things going on around you. Focus on one or two things at a time. Without that level of attention, any kind of intelligence is impossible."

This isn't the way evolution works. It doesn't start with information about "a million and a half things" and then pare back from that. Early nervous systems would have begun by sensing just one or a few things, with lots of trial and error going on about which few things. The most successful senses would have been naturally selected for, and then gone on to (blindly) experiment with adding a few new bits of information to sense and process. This evolution never stops, but it only gets as far as it needs to in order to remain alive and reproduce. As Michael Ruse wrote in The Oxford Handbook of Philosophy of Biology, "Consider the much-discussed example of the frog, which snaps at anything suitably small, dark, and moving, regardless of whether it is frog food. A frog cannot discriminate between moving flies and small plastic pellets tossed in front of it no matter how many pass its way."

So, contrary to Graziano's claims, attention is NOT at the root of intelligence. And intelligence IS possible without attention. Intelligence can be very slowly built up by very narrow increments of additional information. Attention — the way that Graziano is using it — is really another word for choice, i.e. choosing which stimuli to "pay attention" to. But such choices do not need control; they can be made non-consciously by simply responding to the loudest signals, where evolutionary trials and errors shape what "loud signals" actually are. Think of the bees flying back from explorations for nectar and doing their wiggle dance to "convince" others to "listen" to them. It's just the most excited dances that "get paid attention to" by the rest of the hive. That doesn't require conscious choice. So, it's not obvious to me that attention is what consciousness is or is required for.

>>> "A body schema stands hierarchically above the body. It isn’t the same thing, and they can be dissociated (e.g. phantom limbs). Similarly, this is the relationship between attention and consciousness. Attention is literally what the brain is focusing its resources on. The Attention Schema is what the brain thinks it is focusing its resources on, what the brain thinks focusing is, and what the brain thinks the consequences of focusing are."

I think there is an excellent point here about body schemas and brain schemas both being separate from the actual bodies and brains. I just don't think attention is at the heart of it.

>>> "Global Workspace Theory is basically a theory about attention. How do you become conscious of an apple you are looking at? GWT says you attend to the signals. They become stronger from your visual system at the expense of other signals. At some point, the signals become so strong that they reach a state called “ignition” when they can then influence wide networks around the brain. Now attention has been reached, you can talk about it, you can move toward it, you can remember it later. The apple information reaches the global workspace and becomes available all around the brain systems. GWT says that is consciousness. The weakness of GWT is that it doesn’t explain why we claim to have a subjective experience. It doesn’t say why I have an inner experience of the apple."

>>> "The attention schema says great for GWT, but you need one more component—a system in the brain that says “Ah, I am attending to the apple. I have a global workspace that has taken in that apple information.” You need something in the brain that can model itself and build some kind of self-description. GWT is the attention. Attention Schema is the consciousness riding on top of that."

See. Graziano unwittingly contradicts himself here by describing GWT as the attention without the consciousness. All of the choices of attention can be made (through evolutionarily-learned ignition) without a schema sitting on top of it and controlling it. Again, I think he's right that a schema is needed, but it isn't about attention alone.

>>> To control something, you need a model of it. But an overly complicated one is wasteful. A “cartoonish” one is good enough.

I think this may be a big source of Graziano's errors on this. He is thinking like an engineer who is concerned with top-down "control" rather than thinking like an evolutionary biologist who sees bottom-up emergence. There is no top-down control or design in nature.

>>> "Why does it feel non-physical? This is one of the most successful points about the Attention Schema. The brain models itself, but it doesn’t need to include little physical details. It doesn’t need to know anything about the little implementation details. Efficiency dictates the models be as stripped down as possible."

This is more thinking like an engineer. Nature doesn't strip down; it builds up. And if more building provides an advantage, then that building up gets selected for. Why wouldn't an Attention Schema ever build up these little physical details? Graziano raises an excellent point, but I think there's a better answer just ahead.

>>> "The ability to attribute consciousness to others is important. In this evolutionary process, we start out evolving an ability to model and keep track of ourselves, which helps make predictions about ourselves and control our behaviour. At some point, as social interactions become more sophisticated, we develop the ability to use the same machinery to model others. This social use probably came in very early in evolution. There is a lot of sophistication in reptiles, birds, and mammals. We not only keep track of and model our own attention, but we keep track of and model others’ attention. That allows me to predict your behaviour."

Making models is vital, but I think Graziano has it backwards here. Life wouldn't have started with models of itself; it would have started with models of the outside world, with models of others. As we saw in my post about Antonio Damasio, "Valence / value evolved much earlier. Even bacteria can go toward food and away from danger." What is a model other than a set of if / then rules? What rules would a bacteria have in place about itself before it developed rules for going towards food and away from danger? I can't think of any.

Graziano says that "at some point, as social interactions become more sophisticated, we develop the ability to model others." But long before social interactions mattered, the predator / prey relationship would have dominated the natural selection of minds that could make models of others. And here is a big realisation. Those models....would not have had any physical inputs for them! To say it like a philosopher, I cannot know what it feels like to be a bat, but I may need to know how a bat might attack or elude me, so I will build a model in my head of that bat, even though I have no physical inputs into that model. In more philosophical jargon, the epistemic barrier created by living in a physical world where mental phenomena do not just leap across organisms is exactly the reason why our theories of minds have to feel non-physical.

[I feel like I hit on something big there.]

By the time our model-building of others could turn inwards, these models would have experienced a runaway arms race between predators and prey that shaped them into sophisticated, but non-physical, models. Such sophisticated external models would do just fine for understanding our internal selves, so there would be no need to develop a new model using all of the internal physical processes going on. In fact, there would likely be evolutionary harm to even try because the resources expended on such a project would be wasted with no chance to catch up to the existing model-making skill. (Note: even if the internal models were being built at the same time, the external ones would have faced much stiffer competition and developed more rapidly.)

>>> "With this Attention Schema, we don’t need another explanation for the philosopher’s qualia because there it is. Chalmers, after the Hard Problem, now talks about the Meta Problem. The Hard Problem is how do we get qualia, or that inner subjective feeling. The Meta Problem is why do we think there is a Hard Problem? The Attention Schema solves the Meta Problem. It explains why people think there is this magical non-physical thing inside us. It does an end run around the Hard Problem."

As we saw in my post about Chalmers, that's not an accurate description of the Hard and Meta problems. You can't make an "end run" around the Hard Problem. Chalmers doesn't consider the Meta Problem to be beyond it. (He called it another "easy" problem about behaviour.) I think my explanation works better as to why this magical thing inside of us feels non-physical. And it's an impossible question to ever answer all the whys behind the Hard Question.

>>> "We seem to model attention as if it were a fluid flowing out of their eyes, which explains all kinds of folk beliefs about feeling eyes on the back of the neck, telekinesis, the Force in Star Wars, the evil eye, etc., etc."

I think Graziano is mixing up the possible uses of attention here. His Attention Schema is about choosing to pay attention to *some* senses rather than others. Modelling the attention of another being is about modelling *everything* that that being can see. We model the fluid as if it were on all the time, not as if it were being paid attention to only occasionally. My idea — lets call it an ExteroSchema for now — may still build its model of vision as a fluid flowing out of others' eyes. That might be the easiest way to do it and it's a cool explanation of that range of folk beliefs.

>>> "IIT is a fundamentally magical theory." 


Finally, Graziano finishes with a critique of Integrated Information Theory that sounds pretty dismissive. Our next post will be all about IIT though, so I look forward to diving into it and seeing how it is presented by a strong proponent.

What do you think? Do you agree with me that Graziano has some evolutionary ideas backwards? Does my explanation of modelling others first make more sense? I'd love to hear what you think of this in the comments below.

--------------------------------------------
Previous Posts in This Series:
Consciousness 1 — Introduction to the Series
Consciousness 2 — The Illusory Self and a Fundamental Mystery
Consciousness 3 — The Hard Problem
Consciousness 4 — Panpsychist Problems With Consciousness
Consciousness 5 — Is It Just An Illusion?
Consciousness 6 — Introducing an Evolutionary Perspective
Consciousness 7 — More On Evolution
Consciousness 8 — Neurophilosophy
Consciousness 9 — Global Neuronal Workspace Theory
Consciousness 10 — Mind + Self
Consciousness 11 — Neurobiological Naturalism
Consciousness 12 — The Deep History of Ourselves
8 Comments
SelfAwarePatterns link
4/9/2020 07:29:44 pm

I think Graziano comes across a bit more rigorous if you read him at length. For example, he sees the AST evolving concurrently for both external and internal purposes.

It seems to me that something like the AST must be true. Where I think Graziano goes wrong is characterizing it as something separate and apart from attention. To me, it's part of top down attention. In other words, it's a model used by executive systems to "load the dice" in who wins the global competition.

On the relationship between attention and consciousness, I think we have to be careful in regarding these things as just one thing apiece. Both are enormously complex. Attention happens at several levels and in many disparate brain regions. The contents of consciousness and the mechanisms of attention are separate, but they are so intertwined, I'm not sure how much sense it makes to characterize them as completely separate.

Graziano's attempts to portray his theory as being at the center of everything is annoying. But if we ignore that, all in all, I think he's making an important contribution.

I have serious issues with IIT, but I'll save those for that post.

Reply
James of Seattle
4/9/2020 08:00:09 pm

Oy. Here’s what went through my mind while reading the interview: yes, yes, yes, Yes, Yes, YES, NOOOOOOO!!!, whatever,whatever

Here’s what went through my mind while reading your commentary (approx.): yes,yes,no,yes,yes,no, yes, YES, NO, etc.

So for the interview, I commend him for recognizing the compatibility of all those theories, albeit failing to recognize compatibility with IIT.

The biggest reveal I got from this interview is an understanding of what “attention schema” could be, and that it’s an analog of the body schema. The body schema is a set of models about sensations from the body. The sensory schema is the set of models about sensations from outside the body. [I dont know who it was, but some evolutionary scientist suggested that the very first models were interoceptive, and exteroceptive models came after.] I think Mike (SelfAwareP) would lump these two together as the sensorium. Another schema would be the action schema, or motorium. These models are about physical/voluntary muscle actions.

So what is the attention schema? I think it is the executorium, i.e., a set of models about goals. Sensory models can influence motor models, but they can also influence the global workspace. Attention models can influence which sensory models actually do influence the global workspace.

So what caused the big “NOOOO!!!!!”? It was this: “GWT is the attention. Attention schema is the consciousness riding on top.” Note that he completely left out HOT. As Dennett would say, getting into the global workspace is all well and good, but then what happens? If nothing else happens, is there any consciousness to speak of? I, like Dennett, would say not.

*
[I intend to reply to your comments in another reply]
[Did I mention that I find these posts invaluable?]

Reply
James of Seattle
4/10/2020 10:19:23 pm

So, lots of good comments. Here are the ones I disagree with (or want more argument for anyways).

“When an arm moves, it moves through space and time by contracting muscles that cannot see anything. When a focus of attention shifts, no such physical movement or navigation issues occur. I think it's a mistake to think of models being required to control both of these different things in the same kind of way.”

Obviously different kinds of things can be controlled in the same way. I can have two buttons on the screen of my car. One button turns on the windshield wipers, another button brings up a radio station menu. One -> external movement, another -> internal visualization . It’s not that models are required for both of these, but models are capable of both of these.

“This isn't the way evolution works. It doesn't start with information about "a million and a half things" and then pare back from that.”

Er..., um ..., isn’t that exactly how evolution works. It starts with a million things, and selects the ones that work better than the others. I agree that attention is not necessary for intelligence. Attention, however, can provide better intelligence. As you say, nature tends to make incremental changes, but the incremental ones are usually just adding more. At some point more is no longer better. But adding the ability to ignore one in favor of another is a step change.

[at the risk of losing you] You can think of it in computational terms. The ability to respond to a stimulus looks like “If A then do B”. You can always add “if C then do D”. Adding a new operator like AND gives a step change. “If A AND C then do E”. Adding attention (if you squint and look sideways) is like adding a whole new NOT operator. “If A AND NOT C then do G”. Adding operators changes the computations you can do by magnitudes.

“There is no top-down control or design in nature.”

Obviously there is top down control in nature, because people are in nature. But the question is is there top down control within the brain. Control means having a goal and controlling things toward that goal. So are there parts of the brain that can have goals, and can they act on other parts to achieve those goals? Graziano (I think), and I (for sure), would say yes. Graziano might say the action of those parts on the other parts *is* consciousness. I would say it just sets up consciousness.

“Life wouldn't have started with models of itself; it would have started with models of the outside world, with models of others.”

As I said in my other reply, I don’t think this is true. I think the first *models* were internal. I can’t support this right now, but I’ll try to find a link. In any case, having a model should not be confused with having valence. So, what’s a model? I think that could be a separate discussion. I’m game if you are.

*

Reply
Ed Gibney link
4/11/2020 04:42:19 pm

Thanks guys! I was waiting for James' second comment to respond.

So I do think it'd be good to discuss what a model is, but I want to push on something else first. That's the idea of "top-down control." The way I'm considering that when I say there is no top-down control in nature is the way that Dan Dennett describes "skyhooks" acting from above with nothing anchoring them from below. I don't take you guys as believers in skyhooks so I'll move on.

Mike,
You said something like AST is "part of top down attention. In other words, it's a model used by executive systems to "load the dice" in who wins the global competition."

Given my aversion to top-down skyhooks, can I restate that as a pattern recogniser that learns to respond to the various ways the dice come up again and again? I think of this sometimes as parts of Daniel Kahneman's fast and slow (system 1 and system 2) modes of thinking. The system 2 can slowly put past experiences together into a fast system 1 responses. I don't consider that top-down control; just more bottom-up (and up and up) organising and re-organising.

James,
It's probably impossible to draw an early distinction between internal and external senses and models. Is the first sensation: 1) "I'm missing a carbon atom" or 2) "That is / is not a carbon atom out there." I still stand by my parenthetical note that "external ones would have faced much stiffer competition and developed more rapidly."

As for the buttons, yes, models *can* work for both types of car buttons, but I wouldn't necessarily expect nature to build the same button for two very different kinds of things. I'm not saying that it *couldn't* have happened, just that an argument from analogy doesn't *necessarily* hold for that since the analogy is a poor one. (And yes I know nature makes due with what it has, I'd just like to see it argued for by more than a poor analogy.)

For evolution's million and a half things, I think we're talking about two things here. A million and a half individual trials get tried out during evolution, but each trial doesn't contain a million and a half variables. (Not at first anyway.) I'll put my own ideas down soon, but I'm seeing a very wide development of consciousness as something like a growing ability to sense the world and respond to it. Non-living things sense and respond to precisely zero things. The very first living organisms would have sensed and responded to maybe one thing (or as few as you could draw a circle around during some kind of continuum between non-life and life). Graziano's characterisation of attention as both going back to the beginning of life yet also choosing between a million and half things doesn't make any sense in this evolutionary view of life. Do you see what I'm getting at?

I can see (sideways and squinting!) how the NOTs can add up to "attention" and enable big step changes in processing capability, although it's hard to see that coming in anytime too soon after the beginning of developing AND's. That's why I'm saying that attention wouldn't be first and therefore deserving of being called the most important thing. (Although it certainly is one important part of model building, which is what I think is actually the most important thing.)

Finally (for now), you say "Control means having a goal and controlling things toward that goal. So are there parts of the brain that can have goals, and can they act on other parts to achieve those goals?" That seems like a loose use of the word control to me. To me, control is only something I would use from an engineering perspective where people or guidance systems control some process from *outside of* that process. I don't see where such external control can come into play within people without resorting to skyhooks. That's more or less a determinist point of view (or at least that of a free will skeptic). If you want to say a river bank "controls" the flow of a river, I'll understand what you mean, but I think it gets dangerous to use that word in humans. It's too easy to misunderstand it as a coming from a homunculus.

[Thinking out loud...have I ruled out "self-control" there? I think I have.]


Thanks again for all the inputs and pushback! I'm really finding the interaction invaluable to hone my ideas.

Reply
SelfAwarePatterns link
4/11/2020 06:09:18 pm

Ed,
I'm fine with your restatement. Another way to look at it is systems that increase the causal scope in time in the global workspace competition, an enhancement of the increase in spatial scope from distance senses.

That said, I should note that the term "top down attention" isn't mine. It's a term from neuroscience, although sometimes the more technical "endogenous attention" is used instead, in contrast to bottom up, or "exogenous attention." They are actually distinct (albeit overlapping) networks.

My reference to "loading the dice" comes from William James' attempt to get at what is happening with top down mechanisms.

Reply
James of Seattle
4/11/2020 08:31:44 pm

Ed, I’m glad you like the pushback, because I’ve got more. :)

But first, I am definitely with you that attention had no part in the early development of Consciousness. [quick aside: Boolean logic and information theory concludes all computation is a combination of 4 operations: COPY, AND, OR, and NOT. Trust me, I saw it on YouTube.] So the first operations was COPY’s. This is just communication of information from one place to another. AND’s and/or OR’s would come later, as well as NOT. I am not in position to say when. Now attention mechanisms certainly developed, but these mechanisms are not a schema.

So now for pushback. I think analogising the attention schema with the body schema is an excellent analogy. What is the body schema? Correct me if I’m wrong, but this refers to an area of cortex which gets interoceptive inputs from the body. You could say this area consists of models of the body, or as you suggested to Mike, pattern recognition units. The sensory areas are likewise schemas. They consist of pattern recognition units with inputs from their respective sources. But each “schema” is located in a specific patch of cortex with it’s respective inputs and outputs. I think it is reasonable to guess that, because the morphology of every part of the cortex is the same (with some tweaking, i.e., subtle differences), every part of the cortex is a schema, a set of models, a set of pattern recognition units.

[quick aside: my knowledge of evolution is mostly associated with my background in immunology. I have a master’s degree (didn’t want to finish after 4 yrs into PhD.) It’s fairly well accepted that much of the evolution of the immune system occurred by duplicating certain genes followed by tweaking of the redundant duplicates. It’s believed viruses played a role in this because viruses have mechanisms for duplicating there own genes within a genome, and sometimes this duplication brings endogenous genes along for the ride. Several immune system genes have parts that look “viral”.]

So how did nature create the frontal cortex? I have to assume it did so by just making more of what it was already making: schemas, models, pattern recognition units. But then these would be redundant (and metabolically expensive) unless/until tweaks produced a good function for them. Executive control of the other schemas turns out to be a useful function. What would be the output functions of these recognition units? Lots of NOT’s. And the recognition units could be agents without being full-on homunculi.

By the way, Ruth Millikan has a name for all these recognition units, although she never made this actual connection. That would be my addition. But she coined the word: “unitracker”, and wrote a book about it: Beyond Concepts.

*

Reply
Ed Gibney link
4/12/2020 09:58:50 am

Thanks Mike. I'm glad to hear we seem to be saying the same thing just with different words. That's always a difficult thing to suss out so I really appreciate your ability to see things through different lenses. I also get that you've taken your terminology from neuroscience, but it seems to me that the language there (like in much of society) is infected with folk notions of free will which I'd prefer to avoid in my own writing. (At the risk of alienating the folks!)

James — excellent that we agree about the attention schema (no) vs.attention mechanisms (yes). And I trust that Boolean recap, although I can't say I've heard COPY in that list before. I'm not sure that's exactly a logic gate, but it certainly is an important mechanism.

I don't want to get too bogged down in the body schema vs. attention schema debate (I'm working hard on getting the IIT post out soon!) but while there are certainly similarities between them (making them analogous) I see the dissimilarities as significant enough to make them disanalogous. Your mileage may vary in the weight you put on similarities vs. differences. However, you said "the body schema gets interoceptive inputs from the body", but, crucially, it also has to integrate them with exteroceptive inputs about the space around the body in order to physically navigate that space. The "movement" of attention from one choice to another is not an analogous type of "movement". If interoceptive inputs were enough to give us schemas, couldn't we also hypothesise a heart schema to control its beating? Why didn't that develop? Well, because it runs fine on autopilot. I don't see why attention couldn't be run similarly. In fact, I think Sam Harris would argue that it is, and that you can notice this through mediation and the close observation of the way that things just pop into our minds. Any perceived control of attention can be found to be caused by an uncontrolled precedent.

[Very weirdly, I had a nightmare last night that my wife and I were in a room and an attacker came in to get us. I stabbed him in the chest with a giant stake and then saw that it was....Sam Harris! Soon, he pulled the stake out of his chest and started chasing us, zombie-like, down a stairwell. Then I woke up. How on the nose is my subconscious?!]

Regardless, I'm in agreement with you that executive schemas are built, and that it makes evolutionary sense that they were built the way you describe them using the same schema-building methods that already work.

Thanks for sharing that background about immunology. I'm probably going to write something in the penultimate post in this series about "the parable of the immune system" that David Sloan Wilson used in his book The Neighborhood Project. I look forward to a fight about how analogous that parable is! : )

And thanks for clearing up the source of unitrackers. I briefly googled that term when you first used it here but did not find it. I'll check out Millikan's work when I have a moment.

Reply
James of Seattle
4/12/2020 07:31:20 pm

One last comment. I think there are currently attention mechanisms running on autopilot. When you’re hungry, the appearance or smell of food has a definite effect. When you’re walking in a forest with probable predators, the snap of a twig will change whatever you might have been thinking about. I’m confident (not knowledgeable) that the drivers of these mechanisms are sub-cortical. The new thing that came with the PFC, and the attention schema, is the ability to override or otherwise control those pre-existing mechanisms.

*

Reply



Leave a Reply.

    Subscribe to Help Shape This Evolution

    SUBSCRIBE

    RSS Feed


    Blog Philosophy

    This is where ideas mate to form new and better ones. Please share yours respectfully...or they will suffer the fate of extinction!


    Archives

    January 2023
    August 2022
    July 2022
    June 2022
    March 2022
    February 2022
    January 2022
    December 2021
    November 2021
    October 2021
    August 2021
    June 2021
    April 2021
    March 2021
    February 2021
    January 2021
    December 2020
    November 2020
    September 2020
    August 2020
    July 2020
    June 2020
    May 2020
    April 2020
    March 2020
    February 2020
    January 2020
    May 2019
    March 2019
    December 2018
    July 2018
    March 2018
    February 2018
    January 2018
    December 2017
    November 2017
    October 2017
    August 2017
    July 2017
    June 2017
    May 2017
    April 2017
    March 2017
    February 2017
    January 2017
    December 2016
    November 2016
    October 2016
    September 2016
    August 2016
    July 2016
    June 2016
    May 2016
    April 2016
    March 2016
    February 2016
    January 2016
    December 2015
    November 2015
    October 2015
    September 2015
    August 2015
    July 2015
    June 2015
    May 2015
    April 2015
    March 2015
    February 2015
    January 2015
    December 2014
    November 2014
    October 2014
    September 2014
    August 2014
    July 2014
    June 2014
    May 2014
    April 2014
    March 2014
    February 2014
    January 2014
    December 2013
    November 2013
    October 2013
    September 2013
    August 2013
    July 2013
    May 2013
    April 2013
    March 2013
    February 2013
    January 2013
    December 2012
    November 2012
    October 2012
    September 2012
    August 2012
    July 2012
    April 2012


    Click to set custom HTML
Powered by Create your own unique website with customizable templates.