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 9 — Global Neuronal Workspace Theory

3/31/2020

8 Comments

 
Picture
Stanislas Dehaene — if I'm implicitly biased towards him, I now know why.
For the rest of the research in this series, I'm going to be going over the work of neuroscientists. This is because, as Patricia Churchland stated in the last post, "Philosophy is a proto-science that must remain in touch with empirical discoveries." As a philosopher, however, my goal here is not to gain or present a detailed lesson of all the most complicated inner workings of the brain (read neuroscientists for that). Nor is it to get into a deep debate about the methodologies, assumptions, and conclusions of the people working in this field (read philosophers of science for that). What I'm looking for in this series are findings or hypotheses which have implications for the rest of my philosophical worldview. Is that going to require *some* knowledge of brain anatomy and mechanisms? Yes. But it's not that scary or difficult.

One of the best guides for this world is Dr. Ginger Campbell, whose podcast Brain Science is up to 170 episodes now as of this post. Recently, Campbell posted an incredible four-part series on consciousness that was really a key inspiration for me to finally tackle this subject as well. In the first of these podcasts (called What is Consciousness?), Campbell gave her own summaries of some of the latest and best books on consciousness. Before she dives into them, Campbell notes that while they do have their differences, there are still three concepts they all share:


  1. Consciousness requires a brain
  2. Consciousness is a product of evolution
  3. Consciousness is embodied

While I'm always happy to hear from people with an evolutionary perspective, previous posts in this series make it clear that there are enough quibbles about the term "consciousness" to remain wary of saying it is a coherent enough concept to deserve a label. That throws into question whether a brain is required for it or not. But, if you grant each neuroscientist their hypothetical definition of consciousness, then we can understand what they are talking about and the rest of their claims remain valid within that perspective.

Okay. Time for the first summary. Campbell kicked off her series by discussing Stanislas Dehaene's book Consciousness and the Brain: Deciphering How the Brain Codes Our Thoughts. Here are the most important points from that: 
  • Three key ingredients were required to move the study of consciousness into the lab: 1) a better definition of consciousness; 2) methods to manipulate consciousness experimentally; and 3) a new respect for the study of subjective phenomena (compared to behaviorism).
  • The definition Dehaene uses is called Global Neuronal Workspace Theory (an offshoot of Bernard Baars' Global Workspace Theory)
  • GNWT states: consciousness is global information broadcasting within the cortex.
  • Consciousness adds functionality, ability to hold information in mind, and flexible behavior.
  • Wakefulness, vigilance, and attention enable conscious access, but they are separate things.
  • Some of the main methods used to study this are: binocular rivalry, attentional blink, and masking.
  • No amount of introspection can tell us how our brain works.
  • Most of what our brain does is outside of our conscious access. Many phenomena do not require consciousness to occur. We drastically underestimate this.
  • If our brains can do so much without consciousness, then what is it for?
  • Brains make unconscious predictions as if they were using Bayesian logic, but seem to need consciousness to interpret ambiguous images. Also, consciousness plays a very important role in learning (e.g. subliminal learning doesn’t work).
  • Only consciousness allows us to entertain lasting thoughts. It also allows us to create algorithms, a step-by-step way of solving a problem. It allows for flexible routing of information, and appears to be necessary for making a final decision.
  • Consciousness is an important element of social information sharing. It condenses information, [making it easier to transfer].
  • Our self is just a database that is filled through social experience. Consciousness is the mind’s reality simulator.
  • When conscious access occurs: brain activity is strongly activated when a threshold of awareness is crossed. At that point the signal spreads to many brain areas. There are four highly reproducible signals associated with this. Signature 1: activation in parietal and prefrontal circuits. Signature 2: a slow wave called P3 that pairs late, approximately 1/3 sec after stimulus (i.e. consciousness lags behind the world). Signature 3: deep brain electrodes detect late and sudden bursts of high frequency oscillations. Signature 4: information exchange across distant brain areas.
  • Virtually every circuit in the brain, cortical and subcortical, can participate in conscious and unconscious processes.
  • In Global Neuronal Workspace Theory, conscious access  occurs when perception, or any other signal, crosses a critical threshold and is broadcast across the brain.
  • 50 milliseconds seems to be a limit for the shortest exposure to a signal that we can detect.
  • We can only perceive one signal at a time. And there is a 1/3 second time lag. Error prediction makes up for this.
  • Consequences of consciousness include: the ability to respond, the ability to hold ideas in our mind, and the ability to act flexibly.
  • Dehaene does not show mere "correlates of consciousness" because correlation does not show causation. Correlation just finds things that are present when consciousness is perceived, and absent when it is not perceived. Dehaene's four signatures fit this. Causation would require recreation of conscious states using artificial means and this is now being done using deep brain stimulation.
  • Higher brain regions do appear to be essential.
  • Putting together all the evidence inescapably leads to a reductionist conclusion. The electrical activities of neurons can create a state of mind, or equally destroy an existing one.
  • Dehaene thinks Chalmers swapped the labels. It is the easy problem that is hard, while the hard problem seems hard because it engages ill-defined intuitions. Once our intuition is educated by cognitive neuroscience and computer simulations, he thinks Chalmers' hard problem will evaporate.

​Brief Comments
Dehaene offers lots of persuasive evidence for the brain activities that occur during events that we humans can report (i.e. conscious vs. unconscious activities). It is fascinating to see the list of functions this enables as that presumably provides some guides about what is likely to have evolved later as the long evolutionary history of consciousness has unfolded. For example, it seems plain to me that there would be a massive evolutionary advantage for a brain to be able to predict reality rather than wait 1/3 of a second for the processing of inputs. So far, that seems like a good candidate to help answer the question of what consciousness is for. I'll wait to look at more evidence from other scientists, though, before proclaiming too much. Stick around for that in the next few posts.

What do you think? Is Chalmers' hard problem fading away as our understanding of the correlates of consciousness grows? Or as we even begin to dabble in the causation of our conscious experience itself? If this is all too new or confusing to give an answer to that, I recommend trying a short video on Global vs. Local Theories that is part of a recently released introductory course on the brain and consciousness. Let me know in the comments below if that helps or if anything else would.


--------------------------------------------
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
8 Comments
James of Seattle
3/31/2020 05:53:58 pm

I think the hard problem will fade as the meta-problem (why do we think there is a hard problem) gets answered. Interestingly, Keith Frankish blogged just today that figuring out the illusion responsible for the hard problem is just as deserving of the money/effort going into the contest currently running between IIT and GWT.

Looking at that video you posted at the end reminded me of another problem that will cause difficulty in the contest just mentioned. What if both the local and the global theories are mostly correct? By that I mean what if the thing that is Consciousness is happening both locally and globally, and the main difference is just the audience.

As an analogy, consider Twitter. What if Consciousness is just like tweeting? There are three kinds of tweets: original tweets (or just tweets), reply tweets, and retweets. So locally, you have people tweeting to a small audience. But what if you have an account which follows millions (but not everyone) and has an audience of millions (but not everyone)? Suppose that account does nothing but retweet using some method to determine which tweets get retweeted. This account works a lot like a global workspace.

The point is, what if it’s all just tweeting?

*

Reply
SelfAwarePatterns link
3/31/2020 10:48:43 pm

Anyone interested in the brain should definitely be following Ginger Campbell's Brain Science podcast. I've been clued in to more books from her than any other source, and her interviews are awesome!

I currently think something like global workspace theory is most likely true, and Dehaene's (and Changeaux's) global neuronal workspace seems to be the most empirically supported. Although some aspects of it, such as the P3 wave correlating with consciousness, have been called into question. And it does gloss over a lot of implementation details. It's not the full picture, just a well supported step on the way.

On global vs local theories, I think it comes down to how we define "consciousness". We know image maps can form in local sensory regions, and can be recalled for a brief time. But they can't be used for report (verbal or behavioral) until and unless they trigger recurrent activity throughout the cortex.

Complicating things, a past image in a local region can be recalled after a brief period and *then* reported on. So, was that initial image map a conscious one? Or did it only enter consciousness with the global activation?

Dennett makes the point that there is no one consciousness finish line. We can only designate something to have been conscious retrospectively. Of course, if you think consciousness "overflows" report, then you might see the unreportable stuff as a transient form of consciousness. I'm not sure there's a fact of the matter on this.

Reply
Ed Gibney link
4/1/2020 11:25:06 am

Another reader recently sent me this intriguing article:

https://singularityhub.com/2020/03/12/these-two-brain-networks-arent-active-at-the-same-time-but-theyre-both-key-to-consciousness/

I don't fully understand the relationship between these two networks (DMN and DAT) and the P3 wave or the other signatures that Dehaene mentioned, but they sure seem like global phenomena. Any thoughts or insights on that?

Reply
SelfAwarePatterns link
4/1/2020 02:22:02 pm

The way I understand it, the DMN (default mode network) essentially processes imagination and the DAT (dorsal attention network) processes sensory perception.

I was a bit surprised that it involved separate networks, but it makes sense if you think about it. Although images form in the higher order sensory regions for each, the global processing is pretty different, and you can't imagine a flying pink elephant while paying attention to something directly in front of you, so the fact that they're anti-correlated makes sense.

And it supports the canonical interpretation of global workspace, that that what matters is the systemic effects of the broadcast, not the location of the broadcaster, which is understood to be distributed throughout the frontoparietal network, and this study shows it isn't even one consistent network.

Reply
James of Seattle
4/1/2020 05:02:23 pm

Because Ed may not be aware of this difference of opinion between Mike and myself, and perhaps out of sheer orneriness, I feel obligated to point out that when Mike says “the broadcaster, understood to be distributed throughout the frontoparietal network”, some of us (that is, I) have a different understanding. My current thought, until I see evidence to the contrary, is that the broadcaster is located more centrally, namely in the thalamus. This idea is arguably consistent with all of the above, and to my mind, more workable.

*

Reply
Ed Gibney link
4/2/2020 10:06:24 am

The importance of this might have got lost amid all the bullet points, but in my notes on the Dehaene summary, I wrote an exclamation point next to this one because Dr. Campbell reiterated it a few times:

"Virtually every circuit in the brain, cortical and subcortical, can participate in conscious and unconscious processes."

This sounds like evidence for a distributed broadcast rather than a centralised one. Distribution makes sense to me from the evolutionary history of previously separate entities coming together during the major transitions of biology (see book by John Maynard Smith and Eros Szathmary on this). Separate things come together and keep their same "voice" with some new mediator deciding which one gets listened to, for the benefit of both.

The DMN vs. DAT article made me sketch out a (very oversimplified) model that tries to makes sense of this for me. Since I can't attach pictures here, I'll describe it briefly. The environment surrounding an organism is all "external stimuli." The body of the organism is a growing and changing "internal storage of responses" to this stimuli. The DAT would be the brain interface for the range of senses that focus on the "external stimuli" and get that information to the "internal storage of responses" for appropriate reaction. The DMN would then be the brain's internal regulation system, monitoring the "internal storage of responses." This network looks for patterns and learns what responses lead to the best outcomes (best being defined by the abilities of this system to incorporate and grasp more and more information over longer and longer stretches of time). When the DMN and DAT interact with each other, they produce a holistic view of the external world and the internal responses to that world. That's the kind of conscious awareness researchers seem to be going after. Of course, lots of it is going on non-consciously, but we're discovering which bits consciousness helps.

That's just a rough sketch I've got going, but it seems helpful to me at the moment.

Reply
James of Seattle
4/2/2020 07:42:50 pm

“Virtually every circuit in the brain ... can participate in conscious and unconscious processes."

Ed, I don’t see how this comment favors distributed v. centralized broadcast. Virtually every circuit in the brain could participate in a centralized broadcast by contributing content to and/or receiving content from the central broadcaster.

Also, I’m not saying decentralized distribution doesn’t happen, because I’m sure it does. I’m just saying it makes more sense to me that the “global workspace” be centralized, for reasons of efficiency if nothing else. In a distributed system to get information from A -> Z, that might look like A->C->F->R->W-Z. With a centralized distribution you get A->[GlobalWorkspace]->Z.

Maybe I’m misunderstanding your view of centralized v. distributed.

I like the model you scetched out, largely because it seems compatible with mine, so let me show you how I translate yours to mine.

“Internal storage of responses” is unitrackers, i.e., distinct pattern recognition mechanisms. The responses are responses to recognizing the pattern. The Dorsal Attention Network is essentially tracking the (distributed) flow of information from simple unitrackers to compound unitrackers. For example, there are unitrackers in V1 (I think, my neuroanatomy is not great) for specific pixels from the retina, and these feed into unitrackers for features like textures, edges, borders, etc., which feed into unitrackers for objects, etc. These activations pretty much follow a path from the posterior forward, like the DAT. So each unitracker has (at least one) response. Some of these responses are to activate other unitrackers down the line. Some will be feedback to prior unitrackers. Some will be to motor control, some to hormonal control (amygdala?), and some will be to the global workspace, where ever that is. Those unitrackers sending to the global workspace could be said to be competing to be represented in the global workspace, so, competing for “fame in the brain”.

So I agree that the DMN would be the brain’s internal regulation system. Tellingly, you say this network looks for patterns, almost as if this network were made of ... wait for it ... unitrackers. What if the major response of these unitrackers was suppression, namely suppression of those other unitrackers competing to be represented in the global network? Those unitrackers not so suppressed would be the ones that successfully get represented in the global workspace.

Certainly my model is not well-established, but it’s my work in progress. I think one thing in it’s favor is that it makes the unitracker the unit of the cortex. The expansion of the cortex thru evolution was simply the addition of more unitrackers, a common theme in evolution, and then the tweaking of the purpose of those new unitrackers.

*

Reply
Ed Gibney link
4/3/2020 10:46:23 am

Fair enough James. I admit I don't have a firm grasp of your centralised model to say for sure what rules it in or out. I would definitely not expect evolutionary products to be streamlined though. The giraffe's laryngeal nerve is the classic example of such "unintelligent design".

How about you write that blog post of yours about unitrackers et al and we discuss it there. Such conjectures are interesting enough to debate, but seem a bit too in the weeds for the series I've got my hands full with at the moment. I don't mean to put you off, and others are certainly welcome to debate you here. I just know I can't devote enough time to "grok" it right now. Maybe in a few weeks when I'm done though. Looks like we'll still be locked down with time on our hands then!

Reply

Your comment will be posted after it is approved.


Leave a Reply.

    Subscribe to Help Shape This Evolution

    SUBSCRIBE

    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

    February 2025
    August 2024
    July 2024
    June 2024
    April 2024
    December 2023
    November 2023
    October 2023
    September 2023
    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

Powered by Create your own unique website with customizable templates.