Tuesday 22 July 2014

The Trouble with Brain Science - History suggests possible solutions


Professor Gary Marcus, Professor of Psychology, At New York University has just written an interesting  article "The Trouble with Brain Science" in the New York Times. This relates to the to the open letter sent by hundreds of scientists who consider the latest EU funding proposals for brain research are unsatisfactory. Gary's article clearly states views about the current state of brain studies.
Neuroscience awaits a similar breakthrough. We know that there must be some lawful relation between assemblies of neurons and the elements of thought, but we are currently at a loss to describe those laws. We don’t know, for example, whether our memories for individual words inhere in individual neurons or in sets of neurons, or in what way sets of neurons might underwrite our memories for words, if in fact they do.
This is just one of the hundreds of articles that have appear in the media in the last few weeks - and relates back to my post several tears ago under the title Brain Storms - 2 - The Black Hole in Brain Research.

Perhaps the difficulty is that when a lot of scientists have been looking very hard for a solution to an important problem it is assumed that it is a very difficult problem and more and more scientists pile in to dig the hole deeper. This approach only works if everyone is digging in the right place ...

There are some interesting precedents if one looks back in history

The Ptolemaic system of epicycles attempted to explain the apparent movement of the planets in the night sky, and tried to  explain their movements in terms of a number of linked circular motions. As measurements became more accurate the number of these epicycles needed to increase and the problem of working out what was happening - although if powerful computers had existed at the time we might still be stuck in the scientific past with epicycles within epicycles within epicycle to the nth degree.  The only reason the Ptolemaic model lasted as long as it is is that we are not dispassionate independent observers - and as the big-headed animals that we are  we considered that we must be at the centre of everything. All that was needed was to step back a bit and put the Earth on which we live in a wider context.

Ambix, cucurbit and retort
The Eighteenth Century chemists had a different problem in that they believed their eyes. At the time they were trying to isolate and identify the elements that were the building blocks of the world we live in. One of the most easily observed and spectacular elements was Phlogiston. There was no doubt that it was important and they needed to get a sample so that they could examine its properties in detail. It was very easy to make, a very large number of everyday materials contained it in profusion, and if you put animals in jars with air containing Phlogiston they died. Try as hard as they could they repeatedly failed to isolate it and put it in a bottle. However the experiments by Joseph Priestley produced a gas which was called dephlogisticated air because it didn't contain any Pholgiston. This gas caused substances which contained Phlogiston to release it very readily.

In fact the search for the element Phlogiston was fruitless – because what appeared to be Phlogiston being created was the light and energy when things burn in oxygen, and they were trying to capture heat and light as if it was a solid substance.

Perhaps the current difficulties in understanding in understanding how the brain holds and processes memories is that we have been asking the wrong questions. As regular followers of this blog will know I am currently drafting a paper describing a model which promises to bridge the gap.  Some aspects of the model have already been aired on this site, and more is in the pipeline, but the two historical examples point to two "outside the box" areas I am exploring.

Self importance

Judy Horacek(C)  www.cathtatecards.com
We are very special. Just look at all the things we can do that other animals living on this planet can't do. We are so successful that our activities are rapidly altering the climate, and we know how to build magnificent weapons of mass destruction. But of course, if we succeed in changing the world so much that we (and many other plants and animals) become extinct, we will just be another dead end branch of the evolutionary tree of life. We have already dismissed the Ptolemaic system and realised that we are not at the centre of the universe. What we need to do is to climb down from the pedestal we have created and  try and take an objective view.

In my model of the brain I start by assuming that our brains are really no better than those of at least the higher animals, except that we have more capacity than most, if not all, other mammals. I then look at how an animal with such a brain might evolve to have the information processing powers that we have - and find that the very simple protocol the neurons in an animal brain need to process enough information to survive is sufficient to support virtually everything we do. The main difference is due to improvements in the speed of learning which allows us to exchange and remember cultural knowledge more easily. Better cultural knowledge means better tools - and the most important tool, which improves generation by generation is language.

Uncertainty and form

The Phogiston story shows we can be looking for the wrong kind of information. Phlogiston turned out to be energy waves and not a solid substance. Perhaps the questions "Where are memories stored?" and "How are they formatted?" are not the best questions to ask if you want an answer. Having got a Ph.D. in theoretical chemistry I accept without question the ideas behind Heisenberg's Uncertainty Principle and have never lost a minute of sleep over wondering whether Schrodinger's Cat is alive or dead.  After all an electron is both a particle or a wave, and only "makes up its mind" when it knows what question I am asking about it.

My brain model takes the physicist's "Ideal Gas model" as a guide - and assumes the brain can be considered as a sea of identical neurons, with different and variable strength links connecting them. You see a "Rabbit" and the eye "in effect" drops a suitable pebble of activity into the sea, generating a suitable wave of activity to spread over the surface of the neuron sea. At the same time your stomach drops a pebble of "Hunger" into the sea. The two wave trains intersect, and collapse onto a single neuron which becomes active and initiates a new wave "Rabbit Pie". The various groups of ripples represent the human short term memory - and it it is inappropriate to ask where a particular memory resides in the brain when the sea is perfectly calm. For instance my memory of my mother only exists when I am thinking about my mother - when there is an appropriate set of ripples in my short term memory.

And what about the EU proposals

Now at the start of this post I mentioned the protest letter about the E.U. funding plans. What we know is that over the years billions of man-years of effort appear to have been invested it trying to do something which every new-born child does automatically. The E.U. (and the U.S. government) have decided that we need to invest even more following our noses in the directions we have been looking because if we spend enough money going in a single direction we will eventually find the answer.

The examples of epicycles and Phlogiston show that if you have the wrong model it is very easy to waste good research resources trying to go down blind alleys. My research is trying to escape from the establishment brain research boxes which assume that we are more intelligent than animals and that you can find memory in specific locations if you look hard enough. O.K. As a scientist I know my model could be wrong (although many establishment scientists behave as if they knew they were right) but my experience with my research suggests that the current scientific world is very bad in recognising that significant out-of-the-box ideas can originate in unexpected out-of-the-box places.

Whatever your views on the E.U. proposals remember that this generation of scientists is not guaranteed to be right all the time, any more than of the previous generations were. The fact the Richard Dawkins called one of his books on evolution "Climbing Mount Improbable" should be warning enough that if we find ourselves facing a major scientific brick wall we should be prepared to step to the side and look for an out-of-the-box way round rather than waste money trying to scale impossible heights.

2 comments:

  1. Chris,
    Thank for this very interesting post. I recently read a paper, “A computational theory of higher brain function” by Leslie Goldschlager [ http://i.stanford.edu/pub/cstr/reports/cs/tr/84/1004/CS-TR-84-1004.pdf ] which goes some way towards proposing “a lawful relation between assemblies of neurons and the elements of thought”. His ideas seem to have something in common with yours; in particular he proposes a uniform mechanism that would apply to the cortex of both humans and other animals. This paper was written in 1984 by one person on sabbatical, rather than emerging from a large research project, and doesn’t seem to have been followed up.
    Attempts to simulate the brain from the ground up, such as the current Blue Brain project, probably have value, but only if the emerging behaviours serve to confirm or refute a pre-existing theory. The theory seems to be what’s missing.
    Best wishes,
    Roger

    ReplyDelete
    Replies
    1. Leslie Goldschlanger's paper is interesting and approaches the problem from a very different angle to mine. He is starting by trying to model the mathematics of how the neurons exchange messages and store information at the conceptual level. I started by trying to model how humans and computers could work together to process very complex applications and then reverse engineered to draw up a specification of how neurons needed to work. I don't ask exactly how neurons pass messages - rather I suggest what information the messages should convey.

      I am clearly going to have to study the paper in more detail - but there appears to be one very important difference. Leslie's model appears to be two dimensional while my model is also two dimensional when you window into the brain at the high level concept level. However my model includes a powerful underlying network involving recursive hierarchies which locally work in the same way as the top level but which are below the level of our conscious thoughts.

      This difference is interesting. The very first simulation of CODIL used a very flat model, and each advance was as a result of discarding some "computer think" concept and replacing it with a new recursive route through an extended network.

      Delete