The global playground

(Warning: neuroscience weeds)

Stanislas Dehaene recently called attention to a paper in Nature studying the brain dynamics of something becoming conscious. The study supports the global neuronal workspace theory that consciousness involves “bifurcation” dynamics, an “ignition”, a phase transition between preconscious and conscious processing. Prior to the transition, the processing is feedforward and fleeting. After the transition, it becomes more recurrent and sustained. (Across the span of hundreds of milliseconds.)

One issue that has become an ongoing controversy is to what extent the sustained activity represents report or behavioral decision making instead of just conscious access of the information. In order to control for that, various no-report paradigms have been developed. This study uses one. The percept being studied was a French vowel sound communicated at various signal to noise ratios. In half the trials, the subjects were asked to report when they heard the vowel. In the other half, they weren’t asked to report and were given unrelated tasks to perform. The order was varied up in order to control for report behavior from the active report trials spilling over into the passive ones.

In both the active and passive trials, the bifurcation dynamics were observed. (Although the “observation” involved a lot of analysis, which is described in the paper.) However, what I find interesting is the variances in activation between the active and passive trials, a difference that seems compatible with the results of other recent studies showing that the P300 event isn’t a clear indicator of conscious access.

Active vs passive EEG readings. From the paper. Source:

The activations from the passive trials seem much more localized and briefer. To some extent, this should be expected since the participants’ attention was focused on an unrelated task. If the sound made it into their consciousness, we wouldn’t expect it to be there for long or make much of an impact. On the other hand, it seems to indicate that the idea of an all or nothing global dynamic as to what makes it into consciousness is too simple.

Active vs passive activation patterns, from the paper. Source:

To be clear, while it’s more localized than the active version, it’s still much broader than just the local auditory cortex, so this doesn’t seem like a validation of local recurrent processing theory. There seems to be widespread activation of the temporal lobe (likely including the hippocampus) and even some of the frontal lobe, although the authors admit that the frontal lobe activity could be an artefact of their analysis.

The authors, noting this difference in scope, suggest that prior to the global workspace, there may be a smaller scoped range of activity, which they suggest calling the global playground.

The present results indicate that covert conscious access might be subtended not by a global workspace, but rather by a subset of it, which we may term a “global playground”. This “global playground” would be a broad network of areas among which sensory representations are shared and maintained for several hundreds of milliseconds, thus offering wider cognitive possibilities than automatic unconscious processing, but with no specific agenda. When a task is required, this global playground is augmented by decision-making processes and turns into a global workspace.

This is an interesting idea, but I’m left wondering just how global the playground is, and whether there couldn’t be multiple playgrounds happening at the same time. The bifurcation dynamics seem to pull us away from Dennett’s multiple drafts model, where there is no clear consciousness finish line, and whether something makes it into consciousness is an assessment we can only make after it’s become reportable. On the other hand, the limited scope seems to indicate his version might yet have more life in it.

In any case, it seems to indicate that things are more complex than straightforward global workspace theory implies. Unless of course I’m missing something?

13 thoughts on “The global playground

  1. To some extent, this should be expected since the participants’ attention was focused on an unrelated task. If the sound made it into their consciousness, we wouldn’t expect it to be there for long or make much of an impact.

    This consideration seems to block any further inference, unless the global workspace hypothesis makes specific predictions for affected-localities and intensities of the global broadcast. The fact that there’s a strong difference near the crown of the head, might be explained away as task-relevant activity.

    Liked by 1 person

    1. One thing I didn’t mention is that the participants were randomly probed during the passive trials to see what might be on their minds. 19% indicated the sounds, which makes you wonder whether inattentional blindness could be an issue here. However, it turns out whether they indicated the sound could be predicted based on neural activity.

      Overall, the sound constituted 19% of the reported mind-wandering contents, and this proportion increased with the intensity of the sound that preceded a mind-wandering probe (Supplementary Fig. 5). We thus assessed if we could predict whether or not a participant’s mind-wandering was directed toward the sounds or something else on a particular trial, based on the preceding neural activity. We used the same method as for the active sessions: we fitted the bifurcation model to the neural data, independently of any behavioral response, and used this fit to compute the probability that neural activity evoked by passive listening to a sound on each trial belonged to the “high” versus “low” state of the model. As shown in Fig. 5C, activity recorded between 250 ms and 700 ms after the sound predicted accurately whether the random mind-wandering probes included consciousness of sound or not, given the same external audio stimulation. In other words, with the bifurcation model of conscious access, we could use neural activity to predict whether participants were spontaneously aware of sounds in a passive listening condition.

      Liked by 1 person

  2. Check me on whether I’ve got this paper right Mike.

    The authors seek to advance the idea of neural signatures found in various visual studies. But instead of standard overt stimulus here, such as face recognition, they wanted to see if they could find these correlates among people tasked with distinguishing and identifying a subtle sound. Then at least as importantly they wanted to see if they could find such correlates among people with various tasks who haven’t been instructed to notice the sound. They’re claiming that they have found such correlates in both regards, and refer to the non acted upon perception case as merely “a global playground”, since the subjects do nothing with the information and presumably forget it immediately. Furthermore if validated they hope for the clinical application of identifying patients who are “locked in” rather than merely non-conscious.

    Does that sound right?

    Liked by 1 person

    1. That generally matches my understanding Eric. I would just note that participants in the passive trials were randomly probed about what was on their mind, and some of them reported that the sound was. Whether it was largely seemed correlated with the bifurcated activity. So if they were conscious of it, they didn’t necessarily immediately forget about it.

      Since I didn’t mention the clinical part, sounds like maybe you skimmed the Discussion section of the paper?

      Liked by 1 person

      1. I did skim the paper at least to the point that I might have a reasonable assessment of their thesis to state here. I missed that the non report group was asked what they were thinking and that the bifurcation correlated with reports of that perception. This makes sense since surely each subject should perceive the sound to a chance variable magnitude, though if the pre existing bifurcation correlates with reports then that could be evidence of a neural correlate of consciousness. And yes given that a sound does happen to be reported I’d expect some memory of it to be retained as well, though I doubt it would be retained unless actively considered, such as for report. Far more of these studies should be needed for verification of this particular NCC. Hopefully something along these lines will stick pretty soon though.

        As you know I suspect that as better and better NCCs are found we’ll progressively find that a very small percentage of them function synchronously in order to set up the right EM fields which exist as subjective experience itself. That’s all ahead to be empirically verified or denied however.

        Liked by 1 person

        1. It’s certainly an ongoing process. Dehaene and colleagues thought they had found the NCC (or at least an NCC) with the P300 wave, but that turned out to be a bit too optimistic. This appears promising, but further research is definitely needed. Their results depend on a lot of analysis, which seems like it leaves opportunities for wrong turns. And the authors called for follow up work with fMRI to confirm and enhance the data on the locations of the activity.

          Liked by 1 person

          1. Yes baby steps Mike. To me these EEGs of potential NCCs are somewhat like trying to use a huge crayon to write detailed notes in a tiny notebook. Surely the overwhelming majority of NCCs will not in themselves constitute the subjective. It’s like how I say whatever processing that the brain does, the conscious form of function should do less than 1000th of one percent as much, or that the brain exists as a vast non-conscious supercomputer which also creates the tiny teleological computer by which existence is experienced. It seems to take virtually no conscious processing for me to move my thumb for example (as I just did), though I presume that in order for this to happen as I instruct that a vast array of non-conscious processing is required to help incite my thumb to function as such. Though perhaps neural correlates, I don’t think they should be considered subjective experience itself.

            Liked by 1 person

          2. Eric,
            EEG has excellent precision on when something happens, but its precision on where things happen tends to be limited by how many electrodes can be put on someone’s scalp, although clever analysis of the gradients can often provide insights. fMRI is the opposite. Its spatial resolution is much better, but it’s time precision is across the span of seconds (as opposed to milliseconds with EEG). Unfortunately, they can’t be done together since that would involve bringing metal into the MRI scanner.

            “Surely the overwhelming majority of NCCs will not in themselves constitute the subjective.”

            Interestingly, a large portion of what shows up in EEG surges is positive (hence the “P” in P300) and so is inhibition of much of the circuitry. So you’re right to some degree, although maybe not on the reason, or the magnitude. The actual action going on is the small sliver of circuits that are not being inhibited.


  3. I have been somewhat skeptical on what these sort of studies are actually showing.

    You provide an auditory stimulus to a subject and notice some correlation in EEG, fMRI, or whatever. So was the subject unconscious before and after the stimulus and the correlation? Of course not. The person was perfectly conscious so what in the EEG or fMRI correlated with that? Apparently nothing? What got detected was the incorporation of new information into consciousness, an awareness of a change, the cognition of something new, but not the core consciousness itself.

    Liked by 1 person

    1. An awake brain is always active so there’s a persistent churn of activity. I think neuroscientists have to use statistical methods to subtract that out when capturing this kind of data. In the passive trials, the participants were engaged in an assigned task, which I’m sure was producing a steady stream of activity. I think there’s some baselining involved, probably before, between, and after the trials. It’s definitely not as simple as just giving them a stimulus and a previously quiet brain kicking into gear.

      Liked by 1 person

      1. The “quiet” brain is conscious too. It may be the most common form of consciousness with the addition of new information and high amount of activity being frequent but less common. The “quiet” brain may just be playing out its own internal model of the world using as little energy as possible and waiting on something new. So these studies are capturing the something new part but not the “quiet” part.

        Liked by 1 person

Your thoughts?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.