The Name of the Game

mervyn at xs4all.nl mervyn at xs4all.nl
Sun Apr 27 14:06:16 EST 1997


THE NAME OF THE GAME - Mervyn van Kuyen - mervyn at xs4all.nl

INTRODUCTION

The model for the brain's workings that is presented here, provides the
brain with a seemingly useless and unattainable goal state. However, this
goal turns out to be attainable, albeit for a limited period of time. The
required tactics even render the goal very useful indeed. The metaphor of a
game is used to introduce its features to the reader in an accessible way.
Games often have useless and almost unattainable goals.

[ html version, including images:  http://www.xs4all.nl/~mervyn/servo.html ]

THE GAME

The game that will be introduced here is a little weird, but fun. This is
because the spectators play an active role in it, although they have remain
at their seats. The game is played on some sort of soccer ground, surrounded
by a ditch.

   * The goal of the game is simple: Limiting the inflow of new balls.

It is the spectators that keep shooting new balls onto the ground. The only
way to prevent a ball from reaching the field is to hit it in mid-air with
another ball, causing both balls to fall into the ditch. Does the
interception fail, then the public will shoot this ball onto the field as
well. Failure makes things worse.

Because a lot of spectators are drawn toward this game, spectators can only
shoot in balls from six spots. This way the players on the field have a fair
chance to intercept all balls, assuming they master the skill of aiming
their shots with the required precision.

As they master this skill a new problem arises. Imagine that a game is being
played. There are more than enough balls on the and these balls are used for
succesful interceptions of incoming balls. At some point a shortage of balls
has to arise, causing the inflow of more balls. And it was just that what
should be prevented in this game!

In the past some strong teams did not play well enough to intercept all
balls, but a shortage of balls did arise sometimes. To solve the problem
they installed a machine that produced new balls. This way they could meet
their demand for balls without any effort.

Even smarter, however, were teams that recognized that they were not perfect
and that they did let a ball come onto the field at some times. They
splitted these balls into tiny fractions and all they had to do was shoot
these fractions at the balls from the spectators a bit (or a lot) harder.
These teams didn't have to buy and maintain a magic ball machine, while
these fractions were a lot easier to exchange between players as well.
Exchanging balls was often necessary because sometimes the spectators kept
shooting balls from one point. From one goal, the players would say...

In the course of the centuries this game was being played, the players got
lazier and lazier. But this didn't stop them from playing well; often easier
is better, is was demonstrated by the splitting of the balls. For example,
the players never became any faster runners, instead new generations of
players had longer arms and legs. Even more arms and legs than previous
generations!

Weren't players that could walk faster a lot easier to find than players
that had longer legs and arms or even more legs and arms? No, the players
could not walk in the first place and would never learn to anyway. This
story is about neurons, in fact.

THE PARADOX

Despite this strange development in our story, I advise you to maintain the
mental picture that has been cultivated sofar. It contains an essential
element of this model and the game can be played by neurons indeed. Computer
simulations proved that they can become quite good at it, despite the fact
that they're deaf, blind and stupid. This issue will be addressed by the
next paragraph however.

The essential element is the problem of a shortage of balls, or in fact the
solution of the smartest teams: the splitting of unintercepted balls in
order to keep stopping balls for a prolonged period of time. This innovation
is responsible for a change in the appearance of the game:

A team of beginners is obviously being buried with balls from the tribune
and just kicking them back, trying to hit some incoming balls. A team of
highly skilled players on the other hand seems to be playing with their own
(fragments of) balls against those of the spectators! An illusion that
becomes stringer as fragmentation increases.

Illusion or reality, fact is that the goal of the game is responsible for
this isolation of the team. In terms of a team of neurons: allowing as
little activity (from our senses) to enter the brain. Isolation of the
brain. Not an intelligent goal to persue, one would think.

But is not so bad at all. Remember that every shot from the tribune has to
be hit by the players to stop it. This implies that all information that our
senses receive has to be sent there, 'thought' by the brain at the same
time!

   * This is a paradoxical form of isolation: an isolation from the outside
     world with preservation of knowledge of the same outside world.

This is a nice alternative for the metaphor of the information processing
brain. The main occupation of the brain could well be the creation and
cultivation of a living world inside our scalp. A fantasy that is playing a
game with the observable world with paradoxical isolation, developing with
as its ultimate goal.

THE PLAYERS

Hopefully the concept of paradoxical isolation, with a little help from the
game metaphor, is now within your grasp. Probably only because of human
players that are still running around in your imagination, but that's
alright. Just concentrate on the movement of the balls as the game is being
played, then I'll replace the human team with a bunch of stupid neurons.

         [Image: http://www.xs4all.nl/~mervyn/GRAPHICS/star.gif]

        Fig.1 Three simple neurons with just twee arms (dendrites).
                 Real neurons have thousands of dendrites.

Let's assume, for now, that we can turn these three neurons on and off for
our own good. 'On' means that a pulse can move along the arms of a neuron. A
neuron has to energize such a propagation because of the electric resistance
the pulse meets. So, 'off' means that a pulse is no longer energized by a
neuron and that it wil die out. The speed of propagation is not to be
influenced by any of this: a pulse always covers a fixed distance in a
certain period of time.

In Fig.1 three neurons form a star shaped network with their arms. This form
can be drawn in various ways:

   * 6 interlocking small triangles
   * 3 superimposed parallellograms
   * 2 superimposed large triangels
   * 1 hexagon with continued lines

Imagine that these arms are initially full of pulse traffic. Would we then
turn 'on' the three neurons one at a time (1, 2, 3, 1, 2, etc.) then only
certain pulses would remain running around the small triangles (in the same
direction and in phase). That is, if we maintain the right rhythm: a pulse
should be allowed exactly the right amount of time it needs to travel along
one of the triangle's sides, then jump on the next neuron, and the next,
etc.

Would we have maintained a much slower rhythm, then pulses could have
started to run around the large traingles instead. At an intermediate rhythm
pulses could even have started to run around the hexagon. Note that it would
take two series of pulses (1, 2, 3, 1, 2, 3) to propagate a puls aound the
hexagon once!

By turning just two neurons on and off, the three parallellograms can also
be made passable for pulses. So, by means of smart switching of the neurons
a lot of magic pulse manipulation can be achieved, maybe even the required
game playing.

THE TRAINER

But who is the magician in that case? According to this model it is the
hypothalamus. The hypothalamus could well be receiving information on how
well the game is being played and send waves of certain chemical agents to
the neurons in order to optimize their performance.

These 'magic' waves could be injected into the cortex by a number of special
networks that are interwoven with it. These networks are calles the diffuse
systems. These systems are known to dispurse chemical agents that affect our
mood, our emotions. According to this model these agents act in two ways:

The first manipulation causes permanent changes by means of agents called
neurotrophics:

   * Is the 'team' playing better than before, then growth-inducing agents
     are released, causing new (weak) connections to be formed. All existing
     connections will benefit as well.
   * Is the 'team' playing worse than before, then less growth-inducing are
     supplied, causing the destruction of the weakest connections, mostly
     just the new ones that caused the weaker performance in the first
     place.

The result of this manipulation is the trialing new connection and
maintaining them when performance improves. How well the team is playing
could be signalled to the hypothalamus by the hippocampus. It turns out that
certain hippocampal neurons are only active in the event of novelty. Also,
hippocampal damage results in the loss of the ability to acquire new
knowledge.

Besides this permanent manipulation a temporary variant is suggested to be
active. The diffuse systems also release agents that have temporary effect
on large groups of neurons. These agents are called neuromodulators. The
release of these agents has a very special effect:

   * Weak connections will be disabled first, followed by stronger (older)
     ones.
   * Stronger connections will recover first, followed by weaker (younger)
     ones.

The result of this manipulation is that the network is seemingly partially
broken down and rebuild again. In reality however connections are only being
temporally turned off and on. So, this manipulation has a lot in common with
the manipulation of the neurons that could select only certain pulses,
running around certain pathways!

So, for the hypothalamus it is the trick to keep injecting the right doses
of neuromodulators. This action should cause a rhythmically changing set of
available connections to arise on which pulses run around, 'shooting down'
incoming pulses.

Of course it is necessary that enough pulses are available and this these
pulses are at the right spot at the right time. For this reason stronger
modulation can sometimes be required. For example when one 'enters' a new
context (situation, location, group of people).

During stronger modulation older connections have relatively large influence
on the allocation of pulses inside cortex. Stronger modulation is comparable
to a slower rhythm in our star shaped network: pulses tend to travel longer
distances. Using older connections our 'troops' could be said to be
repositioned along the 'fronts'. Weaker modulation results a priori in more
local 'swarms' of pulses. In these swarmes fast interaction with the outside
world could be supported.

In other words, stronger diffuse neuromodulation enables better adaption to
a different (but known) context, and weak modulation allows for potentially
optimal performance within a stable context.

For this reason, sleeping and dreaming are suggested to be related to
stronger neuromodulation, while they are known to be related to slower
brainwaves. Dreams could be interpreted as attempts to tune in to a new day.
A new fantasy world is being build and matched with the perceived world. At
night these attempts are doomed to fail and each time this world is
destroyed within minutes, and for a reason. Also in the game that our mind
plays it is suggested that any ball that misses an incoming ball is shot
back onto the field (also when there was no incoming ball in the first
place). These balls are indistinguishable from real, novel events and for
this reason cause the perception of fear and halucination.

Naast ons geestelijk functioneren regelt de hypothalamus ook nog vele
aspecten van ons lichamelijk functioneren. De hypothalamus regelt wat ons
niveau van 'opwinding' (arousal) genoemd wordt: Besides our mental
functions, hypothalamus is known to regulate various interconnected physical
functions. The hypothalamus regulates what is called our level of arousal:

       Level of arousal:  Low  High
       pulse rate:        slow fast
       respiration:       slow fast
       brainwaves:        slow fast  (the four modulatory networks)
       heat release:      slow fast  (controls our body temperature)
       digestion:         fast slow
       growth:            fast slow

THE SPECTATORS

Despite the fact that this model suggests that the brain is indeed playing
the game that was introduced in the first paragraph, the opponent is, of
course, not a tribune filled with the spectators. However, the pulses do
enter at six spots: our five senses and our body. On an ordinary stroll this
outside world fires some 50 billion pulses per second at our brain, 90% of
which are coming from our eyes. Not a particularly simple game that brain
would play, that's for sure.

However, the advantage of the outside world over a tribune filled with
people is that there are a lot more invariances in the average observable
world than there are in a the behavior of a large crowd. Invariances in the
shape of things, faces, behavior of falling and rolling objects like a ball
and even in the behavior of soccer players.

It is not my intention to predict to what extent the brain can incorporate
these invariances in a world model. However, computer simulations have shown
such a system's tendency toward paradoxal isolation and in some cases even
reached that state (for a limited period of time, that is). For example, a
network of 36 neurons could learn a sequence of four 6-bit patterns or in
game terms: succesfully keep it of the field.

Besides this, simulations demonstrated such a network's ability to develop
useful behavior. A network turned out to rather perform a certain action
than change its internal fantasy world. Such a tendency is a great
evolutionary advantage: it forces an organism to maintain the (physically
and socially healthy) environment in it grew up.

The more complex its internal model becomes, the more complex behavior it
will be able to develop before it starts changing it. Whether we humans, as
consciously living creatures, are happy with these conservative tendencies
remains the question. Evolution, on the other hand, is blind for such
issues, as long as we survive.

----------------------------------------------------------------------------

BIBLIOGRAPHY

 Dennett, D.C. (1995) Darwin's Dangerous Idea, The Penguin Press

 Gray, J. (1994) 	The Contents of Consciousness, Cambridge University Press

 Penrose, R. (1989)  The Emperor's New Mind, Oxford University Press

 Purves, D. (1990)   Body and Brain - A trophic theory of neural connections
                     Harvard University Press

 Rao, R.P.N. (1996)  The Visual Cortex as a Hierarchical Predictor
                     Technical Report, University of Rochester

 Saper, C. (1995)    Control of Sleep and Body Temperature
                     Harvard Mahoney Letter: On the Brain

 Simpson, J.I. (1995)  On Climbing fiber signals and their Consequence(s)
                     Cambridge University Press

 Tenner, E. (1996) Why Things Bite Back - The revenge of unintended consequences
                     Alfred A. Knopf, New York

 Van Kuyen, M. (1997) Feedback in Knowledge-Oriented Neural Networks
                     Page 33-35, GRONICS'97 Proceedings.

==========================================================

Copyright (c) 1997 Mervyn van Kuyen -- All rights reserved.
        
Updated April 22 (1997) 

==========================================================

More info:

"Feedback in Knowledge-Oriented Neural Networks", page 33-35
Gronics '97 proceedings. Using the metaphor of the servo.
http://www.xs4all.nl/~mervyn/vankuyen.html (postscript available)

"Selfconstriction: A smart trade-off?", introducing the
model by means of the 'evolution' of an imaginary organism.
http://www.xs4all.nl/~mervyn/selfconstriction.html

"The Name of the Game", introducing the model by means
of a crude observation of a full-grown system using the metaphor
of a strange ballgame. http://www.xs4all.nl/~mervyn/servo.html





More information about the Neur-sci mailing list