Friday, June 09, 2006

Primitives_0

Primitives _0

http://andreswhy.blogspot.com 
 
Dated 7/06/2006
 
Quantity has a quality of its own (Stalin)

Background:

An example illustrates  :
The visual system works well by having multiple small , very rapid recognition centers that recognise edges , corners , curves , etc . These are then assembled into further complex images further up the line to the brain .

A single neuron one or two layers away from this first primitive can handle surprisingly sophisticated pattern recognition ( eg bees trained to recognise human faces , only one neuron activating on fMRI scan on recognising a face)

The Advantages:
  1. Speed . This is all-important in a survival driven organism .
It is more important to have a 80% picture of the environment in half a second than a 98% picture in 2 seconds . By moving the eye , the recognition system is also tweaked . In other words , you get as good or even better picture of the environment by the assembly of rapid exposures at a lower definition at different viewpoints than a long exposure at a single viewpoint .

Note that the latest findings indicate that the only difference between animals and humans is the speed and capacity at which we perform tasks like planning , memory , tool usage , etc .

Also note that there is a fundamental correlation on neural-network level between speed and capacity . ( Hence the time-constraints in exams , IQ tests , sports ,ets)

  1. Reduction in complexity . You need fewer neurons to do the same task . Most of the number of input data are used in the first primitive layer , which is usually hardwired (ie retina) to be very fast . The subsequent neural-network layers need orders of magnitude fewer nodes . The assembly over time of different viewpoints in these primitive snapshots ( ie the differences) will pick up the elements missing in each snapshot due to poor resolution .

Since we know that the minimum necessary sufficient number of dimensions in an Euclidean space is three , we can say that any visual system would work which has three layers . The more data is processed in each succeeding layer , the better , as long as there is a time-difference between viewpoints .





The Disadvantage
 
Definition :
A system lacking a primitive needed to recognise a new factor in the environment might not pick it up . Most primitives are learned in very early childhood and fixated by paring of neural connections on a hardware level . Eg Chinese “r” and “l” problems .

In other words , it is difficult for an adult to learn new languages , accents or to correct faulty visual systems .

Examples of Human Primitives:

Existing primitives:
1.Visual ( Retina-brain)

  • Pattern recognition (lines , edges curves,etc)

  • Peripheral movement (snapshot differences )

  • Linear Trajectory (integration of eye and body movement in snapshot differences. Very high speed required . ) This is the ur time-binding primitive . How the past and the projected future connect .

This ability would normally evolve into intelligent planning capability if there is sufficient number of neurons . Birds were first , but are constrained by a terrible evolutionary trap : in a low density atmosphere like Earth , their maximum weight ,and hence cranial capacity , is limited . They have a ceiling . If they go flightless , without hands they will lose the trajectory primitive , since the ability to compute trajectories will have no survival value . They optimise (like smart crows ) , but generalization is constrained by cranial size .

Start Speculation
The KT impact would have made no difference in Terran civilization , except that the civilized species would have descended from tree-dwelling dinosaurs . (See argument about insolation fibrillations and Gaia driving intelligence development. ) . Alien intelligent avians would possible if the atmospheric density is high enough to support birds with big brains . An estimate would be 6 to 8 times Terran densities at the moment . A planet like Venus would always have a denser atmosphere ( no large moon) . If there is life in a solar system , it would be seeded on all the planets by meteor impacts . Hence we can speculate that intelligent life on Venus evolved first from an avian base . Since the insolation fibrillations would have hit Venus first (because of its thicker atmosphere) , intelligence would have evolved there first in the solar system . Estimated time about 600 – 1000 million years before present . So where are they now ?
They are either extinct (most likely) , or the planet-bound ones are on Saturn . The clouds of Saturn would be very attractive to an avian species .
End Speculation


Apes coming out of the trees already had hands that could throw projectiles and had to have the trajectory primitive .

Hence , the evolution of upright stance ( to keep the hands free for throwing stones ) and stone using ( leading to neolithic technology ) came first and then led to further time-binding developments like planning , language and intelligence .

Notice that any human society , present and historical , spent an inordinate percentage of resources on trajectory games ( basically ball games) . It is pure Darwinnian statistics : good ballplayers have better time-binding primitives and are better at estimating future effects .

A good ball-player is not more successful because he is a team-player . He is a better team-player because he binds time better  . Competition within co-operative boundaries .

Arrghh!

See Itarin for the nerd’s revenge .

2.Language : the primitives ( Broca’s area) are well established .
3.Fairness : Hardwired even into monkeys .
4.Math : count to 3 , rapidly and iteratively .
5.Threat avoidance : Hippocampus (Stress syndromes)
6.Balance (walking . ref vibrating soles.)
7.Creativity (not just LH/RH brain : it is a true primitive)
8.Music (inherent)
9.Risk/Reward : Profit . Making money (there has been heavy selection for this since the invention of agriculture ( ie surpluses)
10.Shapes (female hip , breast .Male buttocks , legs)
11. Pheromones
12. Immune system.
13. Religion

I’m sure there are more , but we know that these true primitives . Small , fast  , multiple (mostly hardwired) data-handling nodes feeding into more sophisticated feedback systems.

2. Primitives we would like to have .

A wish list.
Note that we might have the sensors for the desired characteristic , but not the primitive . The primitive would only evolve if there was an evolutionary pressure for it .


  • Spin
Computing the flight of a ball due to aerodynamic factors . Trainable (See Itarin)

  • Controlling the potentiating of hippocampus , immune system or any other primitive . A meta-primitive .

  • Recognising and controlling quantum events .

  • Recognising and controlling disequilibrium events

  • Recognise and predict chaotic attractors.

  • Sense and manipulate magnetic fields .

  • Sense and manipulate nuclear fields .

  • Sense and manipulate gravitational fields .

The general principle is that , if a primitive has not evolved , it can be created by using some existing structure and building on that .

Andre

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

No comments: