In praise of the homunculus

The homunculus in question being the little chap who sits somewhere in the middle of the brain, who has all our experiences and who is generally in charge – the Wooster to the Jeeves which is the rest of the brain, all those parts of the brain which do the leg work. The chap who was created to provide a home for all our experiences and, for those that believe in such things, for our soul – and who is apt to fall prey to infinite regress. He has a generally bad reputation, with reference 1 offering a fair sample of popular report.

Another aspect of this bad reputation is that no-one has been able to find him, to come up with a bit of brain which does consciousness, a bit which is necessary and sufficient for there to be consciousness. The most promising candidate of recent years has been the claustrum – until someone came across some veterans whose claustra had been damaged in combat. See reference 9.

With the result that most people who are looking for consciousness are looking for it in the increasingly complicated networks of brain activity which are being uncovered by big projects with clever experiments with clever scanners. See reference 5 for the big EU project and reference 6 for the big US project.

Notwithstanding, one can see why the homunculus attracts. One works away at taking consciousness to pieces, at taking the brain and its works to pieces. But one wants there to be a single place, a single process which is the continuing me. That is how it feels to be me. One doesn’t want to be some transient bit of electrical activity in some complicated network, creating the illusion of a continuing me. In the language of many religions, one wants to have a spirit or a soul.

Or perhaps one is working away at building replicas in computers of all the things that humans do. One gets them to drive cars, to diagnose diseases, to play chess, poker and go. One can give them bits of program which dart about, mimicking the way that the attention of humans dart about. One can give them bits of program which mimic the workings of emotions. There will soon be plenty of machines which pass the Turing test, machines which trick people into thinking that they are people too. Indeed, it turns out that passing the Turing test is neither particularly difficult nor particularly instructive, in that it turns out that it is quite easy to trick people into thinking that the chat box on their screen is being serviced by a real person, at least for a while. However, no-one is claiming that any of these computers are conscious, or have a soul or anything like that – one just feels that maybe they ought to have.

Then some of us try to write a soul or something like that into the code in a more direct way. At its crudest, one might just have a non-negative, real valued variable called ‘consciousness’ which is zero when the computer is resting and which takes some high value when it is really up and running. Such a variable might be well correlated with the apparent wakefulness, alertness or consciousness of the computer – but it has no life. It is just a number. One can do the same sort of thing with a non-negative, real valued variable called ‘pain’. Or with components of personality like conscientiousness and extraversion – see reference 7 for a sample of such attributes, with google offering plenty more of the same.

One might try to go one better, to make these variables both an indicator of the state of the system and a lever with which to tweak it. So, to give a fanciful example, one might find that the concentration of lactic acid in the thalamus was a good indicator of how extrovert someone was. More interesting still, one finds that one can make someone more extrovert by injecting lactic acid into their thalamus. Making lactic acid both an indicator and a lever, a mechanism which one’s computer could mimic.

But these variables would still have no life. We really do seem to need is a homunculus, a deus ex machina – a phrase which wikipedia says that the Romans pinched from the Greeks – and which demonstrates once again that the ancients knew a thing or two.

So, arguments against notwithstanding and as previously explained at reference 2, we are still backing the homunculus living in a single small place in the brain. A single place which still seems to us to be more likely, less likely to require bizarre physics, than our soul manifesting itself in the workings of a wide area network (aka WAN, as opposed to a local area network, or LAN). Being an emergent property (as in reference 11, from where the illustration of an emergent termite mound is taken) of a wide area network. So we are backing a small sheet of cortical material, in the upper brain stem or lower brain, the activity of which both expresses the data content of consciousness and projects that data into consciousness. The activity is, or rather the activation processes on that data amount to consciousness. Or, still having a penchant for field theories of consciousness, perhaps the activation processes generate an electrical field of some sort which amounts to the conscious experience. A field, the non-negligible parts of which occupy little more space than that of the small sheet from which it is generated, and which would be rather difficult to detect.

We are not saying anything about whether consciousness is for anything, beyond allowing a suspicion that it is for something.

The gross organisation of our patch of our homunculus, in time and space, is described at reference 3.

Conclusions

We claim that this cortical sheet, together with its activation processes, is conscious, is brought to the boil. But we do not claim that it has agency or that it has memory. Indeed, it is a rather limited sort of homunculus and a great deal of machinery is needed to prep him, to convert action, sensation and memory into something that he can be conscious of. To generally prop him up. The good bit is that, in so far as the generation of consciousness is concerned, we do not need to know that much about all this prepping and propping; all that can be put aside. We do not need to know about all the clever systems which turn agitation on the retina or the agitation in the inner ear into images that can be experienced. We do not need to know whether the image is true or not, whether the brain has been tricked or not. But the difficult bit is that the buck stops here. We are no longer in the business of saying that this or that bit of data or process goes towards consciousness, is necessary for consciousness. We are saying that our data structure is conscious.

A corollary of which is that the homunculus has to be self contained; everything that it needs has to be on the spot. It can’t be whizzing off somewhere else to be told what it means – to use the famous example of reference 4 – to have the colour red, or indeed any other colour. We have to demonstrate that we can put all of what is needed into our data structure. Into one place at one time.

Afterthoughts

Testing the homunculus theory might be a bit of a problem, but without which testing it scarcely counts as more than speculation. The part of the brain likely to be involved is not presently very accessible to scanners and probes are, in the normal way of things, only an option with animals. It does not offer any obvious route to testing whether someone is conscious. And animals, with their limited capability for self-report are even harder. But perhaps a way in would be to try to devise cognitive tests which probe the proposed data structure, which attempt to identify the layers and their coding regimes. And perhaps time to take another look at reference 10.

And we wonder whether the sort of mechanism we are proposing could be replicated in a computer. Could the sort of conscious field we are suggesting might be generated by concentrated neural activity, also be generated by the activity on chips? There was a demonstration back in the 1980’s of how easy it was to reproduce the contents of a VDU from its electrical emanations (see reference 8) and security people still worry about the electrical emanations of their computers and go to a lot of trouble to stop people collecting same. A catch for present purposes, might be, as with the brain, that what you can collect from the outside is only a pale reflection of what is going on in the inside. But see all the stuff turned up by google for the search term ‘tempest security’ for an expert view.

We might add that we are a bit soft on agency and free-will altogether. Terms which are convenient and which have their place, but which are apt to get difficult on closer inspection.

References

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

Reference 2: http://psmv3.blogspot.co.uk/2017/04/its-chips-life.html.

Reference 3: http://psmv3.blogspot.co.uk/2017/05/recap-on-our-data-structure.html.

Reference 4: Seeing red - a study in consciousness - Humphrey – 2006.

Reference 5: https://www.humanbrainproject.eu/en/.

Reference 6: https://www.braininitiative.nih.gov/.

Reference 7: http://www.appletreehealthandwellness.com/the-big-5-aspects-of-personality/.

Reference 8: Electromagnetic Radiation from Video Display Units: An Eavesdropping Risk? - Wim van Eck – 1985.

Reference 9: The effect of claustrum lesions on human consciousness and recovery of function - Aileen Chau, Andres M. Salazar, Frank Krueger, Irene Cristofori, Jordan Grafman – 2015. With the abstract including the sentence: ‘claustrum damage was associated with the duration, but not frequency, of loss of consciousness, indicating that the claustrum may have an important role in regaining, but not maintaining, consciousness’.

Reference 10: http://www.glasgowcomascale.org/.

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

Group search key: src.