Doing the neural walk

Mon, 14 Sep 1998 08:43:50 -0400 (EDT)

Subject: Doing the neural walk
Date: Mon, 14 Sep 1998 08:43:50 -0400 (EDT)

Some of my constructive critics (Mario, Mike, Mark and Bob, and perhaps
Richard too if I remember rightly) have pointed out that I need to pay
more attention to what is _actually known_ about what goes on in the
brain during execution of behaviour. With this in mind the following
is my (admittedly unsteady) attempt to 'do the neural walk'.

My source is Eccles (1991, principally pages 46 and 58).

Take a simple motor behaviour, such as snapping one's fingers. This is
a nice concise behaviour; it would be difficult to mistake anything else
for the snapping of fingers. It may occur in isolation or embedded in
a more complex behaviour, such as flamenco dancing or grabbing a waiter.
It is a learned behaviour (I think - is there a society without finger
snapping? Bill is quite good on questions like that) and so would seem
to be ideal memetic material.

So what are the neural events leading to finger snapping? If we are to
have an 'internal' meme-as-neuronal-pattern (as I think Mario, and perhaps
Mike, are suggesting) then we need to have an answer to this question.

There is an area in the motor strip of the cortex at approximately the
2 o'clock position, or the 10 o'clock position on the other side, where
there are pyramidal cells which fire to set off the efferent pathway
leading to finger snapping. But since this a voluntary action and not
a reflex, the pyramidal cells are not the ultimate cause of the behaviour.
They themselves receive input from the supplementary motor area and also
from the thalamus (specifically the ventroanterior and ventrolateral
thalamic nuclei), and also from the association cortex. The thalamic
nuclei don't just send signals to the motor strip pyramidal cells, but
also 'backwards' in the 'pathway' to the association cortex. The thalamic
nuclei are themselves also receiving input from the basal ganglia and the
lateral cerebellar hemispheres. These two latter areas are themsleves on
the receiving end of stimulation from the association cortex and the
supplementary motor area. During execution of the motor stimulus, the
pyramidal cells are receiving input from the pars intermedia of the
cerebellum which travels via the ventrolateral thalamus (not too important
for finger snapping but utterly crucial for motor behaviours that require
balance, like riding a bike).

I'd really need to email the diagram, but suffice to say that the following
areas of the brain are all involved in some way or another in the execution
of a simple finger snapping behaviour.

pyramidal cells of the motor cortex
pars intermedia
ventrolateral thalamic nuclei
ventroanterior thalamic nuclei
lateral cerebellar region
basal ganglia
association cortex
supplementary motor area

[For those of you interested in system theory, there are 8 nodes and 14
connections in Eccles' diagram]

Eccles' estimate of the number of neurons that are involved in the
execution of a trivially simple motor behaviour is _'hundreds of thousands'_
(p58 fig. 3.13 legend). Eccles adds that the pathway he presents is by
no means complete - the supplementary motor area is not the 'ultimate'
root of the behaviour but only the furthest point on the pathway that we
can currently go back to. Beyond that we are into the realms of speculation
about the 'free will' underlying voluntary movements. Since this is also
likely to have a neurological basis, the massive net of hundreds of
thousands of neuronal connections that Eccles presents may only be the
tip of an even more massive iceberg of neural complexity.

So we have a nice concise easily studied behaviour, and we have the
prospect of untangling the hundreds of thousands (or more) of activated
neurons which underlie it. Is my methodological pessimism concerning
'internal' memetics so reprehensible in the light of this? What I propose
may seem a cop-out, but the 'neural walk' is so difficult that I think it
is less of a cop-out and more of a simple realism.

There is one other point which bears making again:

Even if one could identify this array of hundreds of thousands of neurons,
there is no guarantee that it would be the same hundreds of thousands in
the brain of two different individuals. The same general areas would be
involved but the same hundreds of thousands, probably not.

And another 2 points which I haven't made before:

1) The above pathway in its anatomical detail, applies to all motor
activity involving the hands, so snapping one's fingers, eating with a
fork, making a rude sign and playing the piano are all going to run along
this 'route' to a certain extent. It may be very difficult to work out
neurologically whether a given identified neural pattern is specific for
playing the piano, or for making a rude sign. One would probably only be
able to tell by _looking at the resulting behaviour_. I therefore plead
again: let's study the behaviour.

2) On Saturday lunchtime I cut my finger opening a tin of cod's roe
(serves me right for having peculiar carnivorous tastes). Afterwards,
I was trying to snap my injured finger, and I found it impossible to
snap it with the same vigour as I could the other hand. Some afferent
pain signal was inhibiting my motor output. More neurons had joined
the 'internal' meme, ie. motor inhibitory ones. Therefore, not only
are 'internal' memes likely to differ considerably _between_ individuals,
but _even in one individual_ it is highly unlikely that the 'internal'
meme will be consistent from one execution of behaviour to the next.
The hundreds and thousands of neurons that snap my finger today may not
be the ones that do it tomorrow. And yet the behaviour is still 'snapping
the finger', regardless of how it is generated.

Is all this trouble worth going to when we could more easily study the
behaviour or artefact? The notion that internal neuronal states can be
replicating, except in the roughest and most approximate manner, seems
to me to be implausible in the light of the complexity of the material
in question. It's the behaviour that replicates, and that is where our
attention should be concentrated.

Of course, all the above, complicated as it is, is just about the execution
of the behaviour, the finger snapping event. It says nothing about learning
the behaviour or where the memory of that behaviour is stored so that it can
be recalled and re-performed at will. Eccles gives an even bigger diagram
(p.168 Fig. 7.13) which sketches the other areas involved in learning such
a behaviour. The principal extra component is feedback from the motor
action to the primary sensory cortex. This feedback is in the form of
visual, auditory and somatosensory input. That is, when we snap our
fingers for the first time ever, we see ourselves do it, or we see someone
else do it, we hear ourselves do it and perhaps most importantly, we feel
what it is like to do it. From the sensory cortex, there are connections
to the cerebral association areas, namely the prefrontal, prestriate,
periauditory etc. It becomes a little pointless to describe this without
the aid of the diagram, but a list of the other areas involved includes the
pontine nuclei, the inferior olive, the dentatus and the interpositus.

The hippocampus also seems to be important as individuals with
hippocampectomy are severely deficient in learning. However, they do
have adequate recollection of skills and events learned prior to their
operation, so the hippocampus does not seem to be a storage area but
something which is required for the establishment of memory. Whether
hippocampal neurons therefore constitute part of an 'internal' meme is
debatable. Eccles states (p.167) that 3 years is the crucial time,
during which hippocampal activity is required for the maintenance of memory. Memories older than 3 years attain a sort of hippocampal independence.

This raises an important point, that the storage of the 'ability to
perform' a behaviour (and note this is exactly the kind of thing that
Cavalli-Sforza and co. usually seek out when looking for their 'cultural
traits') may take different forms at different times during the lifetime
of the individual. The first time one learns a behaviour there is neural
activity and information storage of a certain kind. But, with subsequent
practice, that storage _changes its pattern_. Which of these subsequent
neural patterns is _the_ 'internal' meme? This, in fact, seems to call
the whole notion of internal neural replicators into doubt, the 'cerebral
code' changes for a behaviour within one individual's lifetime. The
behaviour is the constant thing (and therefore also the replicator).
There is no host-parasite relationship because the alleged 'parasite'
_has no constant form._ (and please nobody mention Trypanosomes,
because they only change their surface antigens, and that's to avoid
the host immune system)

There's also the issue of more abstract thoughts as memes. Visual memory
tasks such as re-running a familiar journey in one's imagination, and doing
mental arithmetic or silently repeating a string of nonsense syllables
(talking to oneself as Mario said), produce a wide activation of cortical
areas, which must involve even many more hundreds of thousands of neurons
than the simple finger-snapping behaviour, probably into the millions.
For those who can't get hold of the Eccles book, the reference given for
this part is Roland and Friberg (1985).

Those who wish to do the neural walk properly will have to devise means
of coping with these complexities. Aside from my methodological pessimism,
which would apply even if internal memes were static invariant structures,
there is also the issue of the extreme variability of the 'internal' meme,
not just between individuals but _within_ individuals.

Eccles JC (1991) Evolution of the Brain: Creation of the Self. Paperback ed. Routledge, London.

Roland PE and Friberg L (1985) Localisation of cortical areas by thinking.
J. Neurophysiol. 53, 1219-1243.

This was distributed via the memetics list associated with the
Journal of Memetics - Evolutionary Models of Information Transmission
For information about the journal and the list (e.g. unsubscribing)