An introduction to LWS-N

Contents

• Introduction
• LWS
• Purpose
• Some facts and figures
• Major issues
• Minor issues
• Conclusions
• References

Introduction

Quite a lot of posts on this blog mention LWS, LWS-W or LWS-N. It seems time, once again, to post something by way of an introduction, an extended and updated version of reference 15.

Our concern is with consciousness, what it is and how it comes to be. The consciousness of humans in particular, although it seems very likely that higher animals have something of the sort (on which see reference 10). The consciousness which an adult, awake human can usually readily report on. A subject which was not, until recently, considered a proper subject for scientific research – not that this need concern an amateur for whom considerations of proper do not carry the weight that they might for a professional, with grants, hours and posts to secure.

LWS originally stood for local workspace, in contrast to the well known global workspace promoted by Baars and his colleagues (see, for example, references 1 and 2), but later it seemed that layered workspace might be just as useful a name. Also that complementary might be a better word than contrast: the two hypotheses are addressing different aspects of the problem of consciousness.

The present hypothesis is that consciousness arises from the electrical activity in a two dimensional patch of cortex, the LWS, perhaps just a few square centimetres in size, but containing millions of neurons. This patch of cortex is supposed to live somewhere in the middle of the brain, probably above the brain stem but below the cerebral cortex proper.

Part of thinking is, by analogy with getting fusion out of a tokamak, we need to concentrate a lot of activity in a small space if we are to get consciousness out of neurons. Not that all this activity need sum to anything much that can be detected or seen from a distance.

While the global hypothesis is much more global than local, with consciousness arising from the joint activity of a number of wide area networks spanning a large proportion of the brain. Or, according to Dehaene and his colleagues, with the neural correlates of consciousness including the activity of a number of wide area networks. See reference 5.

LWS

LWS-W stood for the version of the LWS which took a Microsoft Excel worksheet view of the world, with the LWS being made up of a small number of large rectangular arrays of small integers. Also known as layers. With the number of layers being in the ten to twenty range, with the dimensions of these arrays being in the small thousands and with the small integers being between zero and something like 20 or 30, the binned version of the non-negative, real valued amplitude of something or other.

LWS, despite its layers, is essentially a two dimensional world, in the same way that the cerebral cortex is essentially a two dimensional world, this being reflected in the experimental habit of mapping fMRI scans of entire brains onto a standard planar map of same. There are plenty of examples of this sort of thing at reference 8.

The idea of layers is taken from their widespread use in technical drawing packages, with the layers there being used to partition, to reduce the complexity of a complex world. So if we were drawing a diagram of an animal, perhaps pinned out on a dissection mat, we might have a layer for muscles, another layer for the pipework of the vascular system and yet another layer carrying a photographic image. For more detail on this see reference 4.

Layers contained layer objects which had parts, perimeters and interiors and layer objects could be linked together, across layers, by means of column objects. Part of the LWS solution to the binding problem, for which see reference 11.

LWS-N stands for the current version of the LWS, one which takes a more neural view of the world, but a version in which we still have the same, relatively small number of layers. We might, for example, have several layers about the visual scene, another layer about the sounds and yet another about the smells.

Our layers are still organised into layer objects, which may have parts and which usually have perimeters and interiors – interior in the two dimensional sense that the interior of a circle is a disc, rather than the three dimensional interior of a cardboard box. Interiors which may well have interesting properties, perhaps colours or smells. We see the perimeters as being rope-like collections or bunches of neurons, loops of rope, while we see the interiors being relatively sparsely connected, planar collections of neurons, with data being carried in the way that those interiors are tiled over. Tiling in squares is different, for example from tiling in hexagons or tiling in triangles. Where by a planar collection of neurons we mean a collection of neurons where the neurons and the connections between them can be mapped onto a plane without crossings.

We suppose that our patch of cortex is organised topically, that is to say, for example, that the data expressing the visual scene is recognisably organised in the same sort of way as it is on the retina, with the parts of objects being very like the triangulations of objects used in various computer graphics applications (see reference 12). We find it plausible that to generate a conscious field about objects in the real world, we start from a foundation organised in the same sort of way.

A layer might also include a background and a foreground, after the manner of an old fashioned theatre.

Layer objects can be linked across layers to form larger composites.

Our hypothesis also includes the idea that consciousness is organised in time into the scenes, takes and frames which were discussed at reference 3. The basic unit of consciousness is the frame (lasting of the order of a second or so) and it is frames of consciousness which are delivered by the LWS.

Purpose

The purpose of the LWS is to provide some explanation of, a springboard for consciousness and the purpose of this note is to describe that springboard. This does not include attempting to say what consciousness might be for. Nor do we address the many and complicated processes which would be needed to put the LWS together, or to put it to work, to get some value from it, beyond saying that our model is that of compilation: back in the 1970’s the form was that you compiled your Fortran program (say) from source code into machine code and then you ran that machine code, unchanged for the duration. The data might change but the program does not. And while these compilation processes may well need to create or adjust synapses, the present hypothesis is that such adjustment is not part of consciousness itself. A frame of consciousness is delivered from a population of neurons and synapses which is fixed for its duration – for a second or so.

Note that we do not exclude all movement from frames. A frame might include, for example, the regular, the predicted movement of a foreground object across a background.

Some facts and figures

We suppose that the patch of cortex making up LWS is of the order of 2cm by 2cm, 4 square centimetres in total, say 5 out of the 2,500 square centimetres of hemispheric surface (or about two square feet, taking the two hemispheres together) of an average person or 0.2% of the total.

We suppose that this patch might contain of the order of 100 million neurons – about 0.5% of a total of 20 billion (with the other 80 billion or so being in the cerebellum). Rather higher than the average, as befits its central role.

We locate places on that patch by means of x and y coordinates. Following the arrangement of an Excel worksheet, we count from the top left hand corner, with the x-coordinate mapping onto the column across the screen and the y-coordinate mapping onto the row down. It is sometimes convenient to think in terms of both the x and y coordinates varying between 0 and 100.

But if we were to think in terms of their varying between 1 and 1,000, each of the cells so defined would contain of the order of 100 neurons. Enough to make analysis of aggregate firing rates a reasonable proceeding.

We will talk of the activity in space and time. This might be simply derived from the actual number of firings in some interval in space and time or it might be defined some point in space and time in some more complicated way. We might talk of the activity of individual neurons. But we do not, for the present anyway, talk in more detail, of spike trains and spike timing, let alone spike timing dependant plasticity (STDP), for which see reference 9.

In LWS-W, the space could be thought of as a three dimensional array, 1,000 by 1,000 in space by around 10 in layers, what we have called elsewhere two and a half dimensions. So each such layer could contain up to about as much information as is contained in a picture taken by my telephone.

The layers were defined by frequency bands, with the oscillations of activity having been analysed into about 10 such bands. Such an analysis limits the resolution in both space and time.

Our patch of cortex is around 20,000 microns by 20,000 microns in size – with a micron being one millionth of a metre or one thousandth of a millimetre. We are unlikely to be interested in distances of less than 20 microns, aka 20μm. By way of comparison, writing paper might be 100μm thick. While the voxel of an fMRI scan might be as small as a cubic millimetre, but still rather too big to be able to show the workings of LWS in any detail.

In LWS-N, we imagine that the layers are defined by the connectivity of their constituent neurons, with connections within layer being much stronger than connections between layers. It might also be that most of the connections between layers are rope-like, concentrated in space, like the column objects of reference 13. These layers are a rather looser construct than those of LWS-W, where we were looking at a stack of identical layers. We can be more flexible here about the number of layers and about the extent of those layers.

Frames of consciousness are supposed to last for of the order of a second. We might be interested in time intervals as short as 10ms, but more usually in time intervals of some hundreds of milliseconds.

Major issues

Defining layers

The idea was that all the layers were defined over the whole of our patch of cortex. That if we have feature A in position P on layer B and feature X in position P on layer Y, then the features A and B are expressed on the same bit of cortex and are bound together to that extent.

The original idea, in which the mass of neural activity had been integrated into signals defined on a rectangular array of cells or points on our patch, was that we differentiated between layers by the frequency of the oscillation of the signal. There was enough signal and enough time at any one cell to analyse that signal into its component frequencies. Bin the frequencies into ten or twenty bins and we have ten or twenty layers.

The current idea is that our layers are defined by their being order of magnitude differences between the amount of synaptic connection between neurons on the same layer and that between neurons on different layers. This concept of layer is more flexible, and, within limits, one can have as few or as many of them as suits.

Part of the price for this flexibility is that we no longer have a natural ordering of the layers. There is no longer a top layer and a bottom layer. One can come up with some algorithm which gives us such an order, but this is not going to be quite as simple or natural as the order by frequency which we had before.

With the word price reminding us that the brain is a piece of engineering, neither magic nor a conjuring trick cooked up by the Divinity. A piece of engineering in which competing interests or pressures have to be reconciled, in which balances have to be struck and in which there will be a mixture of good and bad design decisions.

Meaning out of the void

A large part of LSW is trying to bring meaning out of the void, this in the sense that our LWS is self contained. By definition, it cannot draw its meaning from elsewhere, from anywhere else, say from the paint manufacturer’s colour chart at reference 14, it has to have meaning of itself. So how do we do that, starting from an empty data structure?

Thinking of LWS-N in particular, how do we organise the linkages of neurons to code up for a scene from ‘War and Peace’, the elephant in the room – or even something as apparently straightforward as a patch of red? This last being the subject of reference 16.

This has been the focus of much of the work to date.

Consciousness out of the meaning

The LWS hypothesis includes the notion that consciousness arises from waves of activation repetitively scanning its layers and the objects on those layers, this scanning being both systematic and straightforward. And this activation, being electrical, will generate a time varying field in space – and it is this field which is proposed amounts to consciousness. And with repetitive scanning doing much the same work as the meta-stability of neural activity talked of by others.

This scanning takes account of those objects. In the case of something like a circular disc, represented in LWS by something like just that, one part of the scan will be going around the perimeter, another part will be going across the interior, an interior which might, for example, be coded for colour and texture, possibly varying in some smooth way across that interior.

A process which one if vaguely conscious of when looking at something of this sort; a perimeter and an interior.

Note that, unlike some others, we do not say anything about what we are conscious of. We do not assert, for example, that consciousness depends on a sense of self. Or on any sense of all the work going on in the background to keep the host organism ticking over, on the maintenance of homeostatis. At least to the extent that there are not any problems in that department.

Minor issues

Location

We do not have a location for the LWS, beyond supposing that it is a central rather than peripheral, lying somewhere between the top of the brain stem and the cerebral cortex proper. The claustrum had been a candidate, but this now seems to be ruled out by reports of soldiers wounded in the Vietnam war, whose claustra have been destroyed but whose consciousness is not impaired.

Nor do we know whether the LWS is a single organ or whether, for example, there is one of them for each half of the brain, as is the case for most brain structures above the brain stem. And if there are two, what is the division of labour between them? Does one back the other up? Does one specialise in one aspect and the other in another? All kinds of possibilities here.

However, the subjective experience is singular, there is just one of them per person at any one time, an assertion which given the short duration of a frame is not disturbed by even fanciful ideas about multiple personalities. So our hypothesis is that consciousness is generated by the electrical activity in a small patch of cortex does not sit very well with this small patch becoming two patches. So for the present, we proceed as if the LWS is just one patch of cortex.

We need to put the LWS into context. Work out how it fits into the brain and body. Into the evolution and development of same. Into all the work on consciousness and unconsciousness done elsewhere, by others.

Composition and connectivity

There are many different sorts of neurons and many different sorts of connections to other parts of the brain. Different regions of the brain differ significantly in these matters, differences about which we say nothing here.

Defining the shapes of objects

Thinking in terms of the visual field, we might think of the objects on our layers as two dimensional, with perimeters and interiors.

Generalising, we might retain planar interiors, but allow perimeters to be assembled into, to represent a tiling of three dimensional objects. Computer vision systems might tile the surfaces of the three dimensional objects in their worlds in this way. We touched on some of this at reference 6.

But there are other ways of doing things. We might have:

• Stick men. Various people have worked with the images obtained from attaching lights to the joints of moving people, elbows, knees, feet and so on. It is quite surprising how much can be deduced about the movement and the mover by the sight of the lights alone. See reference 7
• Sausage men, with quite realistic bodies of men being assembled from roughly spherical balls and roughly cylindrical sausages – rather like the things one might make with balloons. See, for example, the ‘Animal Zoo Balloon Kit’ from Amazon
• Vertical symmetry. Lots of things in the world are symmetrical about a vertical axis, including many plants and animals, which means that we only need specify one side.

Just three devices which enable more or less massive compression of the specification of the shape of a possibly moving animal.

Detection

While we do not have a location for the LWS, there is a presumption that it is deep inside, where it is going to be difficult to look at, to get something better than the spatial resolution of the likes of fMRI and the temporal resolution of EEG. One might be able to look at what goes into the LWS and what comes out of it, but looking at the thing itself is going to be hard.

Frames

A presently open question is the conflict between the idea of a frame of consciousness, more or less fixed for its duration of a second or so, and the fact that a lot of the subject matter of consciousness – for examples railway trains and music – is not fixed in that way. We work on various dodges to square this circle.

Language

Humans are the only animals with language. The position seems to be that while there is a strong interaction between consciousness and language, one can be conscious without language – leaving aside the important qualification that someone without language cannot report on their consciousness, at least not in the usual way.

Conclusions

We have drawn together some of the threads which have gone into our speculation, if not hypothesis, about consciousness being the product of a patch of cortex which we have dubbed the LWS, with the present version being dubbed the LWS-N.

References

Reference 1: https://en.wikipedia.org/wiki/Global_Workspace_Theory.

Reference 2: Global workspace dynamics: cortical binding and propagation enables conscious contents - Bernard J. Baars, Stan Franklin and Thomas Zoega Ramsoy – 2013. Open access.

Reference 3: http://psmv3.blogspot.co.uk/2017/06/on-scenes.html.                               

Reference 4: http://psmv3.blogspot.co.uk/2017/04/a-ship-of-line.html.

Reference 5: http://psmv3.blogspot.co.uk/2018/01/what-is-consciousness-and-could.html.

Reference 6: http://psmv3.blogspot.co.uk/2017/09/sensing-spheroids.html.

Reference 7: https://youtu.be/TrIwxKBMDO0. And searches for ‘point light animations human kinematics’ and ‘skeleton animation’ will turn up related material.

Reference 8: Natural speech reveals the semantic maps that tile human cerebral cortex – A. G. Huth, W. A. de Heer, and others – 2016.

Reference 9: https://en.wikipedia.org/wiki/Spike-timing-dependent_plasticity.

Reference 10: The Cambridge Declaration on Consciousness -  Philip Low and others – 2012.

Reference 11: https://en.wikipedia.org/wiki/Binding_problem.

Reference 12: https://en.wikipedia.org/wiki/Surface_triangulation.

Reference 13: http://psmv3.blogspot.co.uk/2017/07/binding.html. From the days of LWS-W.

Reference 14: http://www.farrow-ball.com/colours/paint/fcp-category/list.

Reference 15: http://psmv3.blogspot.co.uk/2017/04/its-chips-life.html.

Reference 16: Seeing red - a study in consciousness - Humphrey – 2006.

Group search key: srd.