Is It A Number Line Or A Number Generator?


by

Huen Y.K.

CAHRC, P.O.Box 1003, Singapore 911101
http://web.singnet.com.sg/~huens/
email: huens@mbox3.singnet.com.sg

(A short communication - 1st released: 28/7/97)


Abstract

The July '97 issue of Discover has an interesting article by Robert Kunzig on "A Head for Numbers". Here is what he wrote: "Every body has one - even rats and pigeons, to say nothing of people. The ability to grasp small numbers and map them onto a "number line" in the brain is an evolutionary birthright of ours. Arithmetic, of course, is another matter" [1]. This prompted me to contribute my own thoughts on this topic. No one has yet detected the number line. Does it really exist? John Allman and Jon Kaas discovered that an image falling onto the retina is mapped point-for-point onto neurons in the visual cortex without disrupting the image's geometry [2]. And numbers, Dehaene thinks, is no different from the spatial relationships of objects in the visual field, or their color [1]. Well I think Dehaene did not quite hit the target. I think that the brain handles numbers differently from visual images. I postulate that the brain counts numbers by firing an impulse train with fixed time intervals. For example, one pulse can represents a count of one time interval, two pulses means two time intervals and so on. In other words, the number line is actually the time line. We cannot visualise time, so why should should we expect to find a number line translated into the time domain? Using sequence algebraic generating functions, the author explains the probable mechanism use by the brain to count numbers [6 to 8]. He hopes that neurologists will one day be able to measure the firing by neurons to represent numbers. However, present day measurement techniques might not be sophisticated enough to do it.


1. Introduction

In 1998, the author came upon an idea on how one could plot graphically on computer screens point information above 2-dimensions. In order to do this he postulated that points should be divisible. This violated Eulcid's definition of the indivisibility of a pont. Nevertheless, the idea gelled into the invention of the Hgram graph for which a copyright was lodged with the Library of Congress, U.S.A including several published papers [3 to 5]. Research on Hgram graphs continued for a period of nine years until he retired from academic teaching. He was highly excited when he came across two years later Mongotmery's article entitled "The Mind's Eye" in the Discover (May 1991) issue. In the development of the Hgram graphs, he was not awared that considerable progress has been made in brain probing by Allman and Kaas in which the main conclusion is that visual image from the external world is mapped point-by-point onto the visual cortex like the working of a pin-hole camera. The geometry of the image is practially preserved in this mapping. The part which particularly interested the author was their discovery that the brain have multiple maps for the same point. He immediately made the connections between his Hgram graphs and these multiple maps in the brain. He postulated that in order to see beyond 2-dimensions, points must be split into subpoints with different attributes. The brain solves this by creating multiple regions, each sensitive to a different attribute. For example one region might be sensitive to motions, another to a specific colour, another to a specific pattern and so on. Thus one could build up multiparametric information on a point via these multiple maps just like the way sublayers are used to represent subpoints in the Hgram graph [4,5].

This month, the author encounters an article in the July '97 issue of Discover in which Robert Kunzig related the work of the French neuropsychologist Dahene on how the brain processes numbers. Because of my past interest on multidimensional graphical visualisaton and my present interest in sequence algebra, it gives me the opportunity to contribute some thoughts on how the brain handles numbers. There is a big difference between mapping of visual images and mapping of numbers. An image is mapped into the visual cortex with little geometrical distortions. However, Dahene has not found any such map related to numbers. Neither has he found the number line although he and Cohen have narrowed the region to the inferior parietal cortex. It is not clear what functions are performed within this area in the processing of numbers. The author postulates that there must exist in the brain special generators which can fire the "number line" as a "timing line". He speculates that we count numbers by time intervals. This paper cites examples to support his theory.

2. Number Generators In The Brain

Neurologists already know for a long time that neurons can fire impulses at fixed frequencies but as they are not mathematicians, they do not associate this action with mathematical functions, especially generating functions. A mathematical function is one which gives a single output value for a single input value. Depending on the complexity of the relation, the processing could be as simple as 1+1 or as complicated as finding the 20th Fermat's number. Obviously the difference between these two task is that the former takes very little time whereas the latter takes an inordinately long time. There are things which the brain is not designed to do well whereas a digital computer might succeed. Why? Because the brain is built to solve survival problems. Computing a large Fermat's number is not in the agenda of survial.

The fact that a neuron can fire an impulse train of fixed frequency must implies that there exist an organic analogue of an impulse generator which takes the place of a mathematical function. Imagine a mapping whereby when a subject receives the command "6", a neuron cluster will fire six impulses at equal intervals and likewise for other integer numbers. This can easily be simulated by a sequence algebraic generating function called Nat(z) [6]. Equation (1) shows the firing of eleven impulses by Nat(z) by series expansion.

................................1.........1........1........1........1........1.........1........1...........1
Nat(z) := 1 + 1/z + ---- + ---- + ---- + ---- + ---- + ---- + ---- + ---- + O(---) .............(1).
..................................2........3........4.........5........6........7.........8.........9...........10
................................z.........z.........z.........z.........z........z.........z..........z...........z

The can be liken to the firing of impulses by a neuron as shown below:

----------|----------|----------|----------|----------|------

The author postulates that there might be neurons which could fire a fixed number of impulses by intention. If this is the case then it is easy to associate the firing of one impulse with the integer 1 and two with the integer 2 and so on. This is the case of numbers mapped into the time domain and not a spatial domain. However, it is still a type of mapping. It is known that human baby, and for that matter all higher animals, could count small numbers from birth. As we grow older, we begin to use other parts of the brain to reinforce our arithmetic system, such as language, patterns and so on. For example, at one time we are made to memorizing the multiplication table by rotes. Here we are not using the number line any more. Furthermore, it is found that a person with brain lesion to specific areas, finds it difficult to do precise counting and interpolations. For example one might response that 1+1 = 3. Another might find difficulty to ascertaining whether 5 falls between 3 and 7. If one assumes that damage to the brain could cause damage to some generators then such observations could be explained using sequence algebra [6 to 8].

For example, suppose in a healthy individual, the generator is Nat(z). However for someone with brain lesions, we could simulate, just for illustration, damage to Nat(z) by a new generator z^3/(z^3-1) which can be expanded as follows:

........................1..........1.........1............1
XNat(z) := 1 + ---- + ---- + ---- + O(---) ....................................(2).
..........................3..........6.........9...........12
.........................z..........z..........z...........z

Now the subject can only fire multiples of 3 time intervals and some multiples of 2. He can still recognise numbers in the set such as 1, 3, 6, 9 and so on but he has no clue at all what you are talking about when numbers such as 2, 4, 5, 7, .. are presented to him. He cannot place such numbers in his "number line" .

As to why human babies can only count small numbers has also an explanation from the generator model. Young children sense of time is very short. If you ask a child to wait for you whilst you go to the loo, 5 minutes is an inordinately long time to him. Most probably, young children fire timing impulses at a higher rate than adults. Most probably the timing line is extended as we grow up. Young babies are only precoccupied with things immediate to them, like hunger .. cry for milk. Notice a baby becomes very impatient within seconds if the first cry is not quickly responded by the mother. Their time line is very short. This explains why their number line is also very short.

What about some of these calculating prodigies? Are they born or trained? If the above generator model is correct, they are trained. Through concentrated exercises, one could extend the range of expansion of the generator and do marvellous arithmetic with them. This is evidenced by what sequence algebraists have done with generators. The only difference is that sequence algebraists rely solely on digital computer to do the calculations. Why are calculating prodigies mostly autistic people. A generating function is a very limited domain and since it is endowed from birth, autistic people are on par with other new borns. These are the only people who are willing to concentrate on using this limited resource. Great mathematicians are more likely to be distracted because they have other more interesting explorations to do. By working a tremendous amount of time with the generators, they are able to attain an expert level above average persons. This could only occur if there is an ingredient of work versus reward and this already exists to fire up their passion with the encouragement of parents. Autistic people never graduate into great mathematicians probably this way of doing arithmetic is very resource intensive. Normal people tend to migrate to link numbers with languages and use other parts of the brain for arithmetic.

The question is whether all children should be stimulated to develop the generator components? I would think it is a waste of time since we now have handheld calculators and other training methods. That this is highly inefficient could be explained using the equation of divisibles as shown in equation (3) [6].

...................................................................1
.........................................Comp(z) := ------------- ........................................(3).
...............................................................i....i
.............................................................z..(z - 1)

Finite generation of seven composite numbers starting from 4 is shown in equation (4).

....................1........2........2........1........2........4.......2..........1
Comp(z) := ---- + ---- + ---- + ---- + --- + --- + --- + O(---) .....................(4).
.......................4.......6........8........9.......10.......12....14..........15
.....................z........z........z.........z.........z........z.......z...........z

Presumably few are trained to write down a series of composite numbers by immitating the principle of the equation of divisibles given by equation (3). And yet, if the brain counts numbers by time intervals, it would be possible to train one to write down composites numbers and with difficulties prime numbers this way. And yet, the process will get increasingly difficult as we extend the range of composites or primes as we need to generate impulse trains with increasingly large intervals. If you are a baby, you will not be able to cope. Even with adults, you will soon loose track because you cannot generate intervals precisely with large intervals. Thus doing arithmetic by counting numbers along a line has limitations whether it is a time line or a number line.

Dahene's concept of a number line has another flaw. Evolution will never retain body parts which are not repeatedly used. Long idleness will cause unused parts to atrophy. Imagine a number line with a millioin integers assigned to the neuron clusters. How many of these numbers will ever be used throughout one's life? Those not used will atrophy. On the other hand, a neuron impulse generator model is highly economical. It certainly does not tie up a lot of neuron clusters. As to why the number line is different from the mapping of visual images onto the visual cortex, we have a parallel here in the Computer Monitor. All the pixels are repeatedly refreshed up to 60 times per second. In other words they are all actively firing away whenever a person open his eyes.

3. Going over past experimental evidence.

This section is based on meagre information available from the article in Discover [1].

(i) The Case Of Monsieur N: (reported by Larent Cohen, neurologist)

2+2 = 3 or 5 but never 9
N has no precise knowledge of the meaning of "9".
A dozen is either 6 or 10.
Can't tell whether a number is odd or even.

My diagnosis:

Perhaps in this case only the generator for Odd(z) is working as Nat(z) and Even(z) had been damaged. He has some inkling of even numbers from the language part of the brain but he is unable to generate a pulse train with even intervals. When asked what is a dozen, his answer are in even numbers but rather imprecise. He seemed to fair better with odd numbers such as in approximating 2 as between 3 and 5. This might be because he could generate 3 and 5 as odd numbers but he is hazy about the even numbers except that his language association enables him to approximate the interpolations. He can't distinguish between even and odd because Even(z) generator is not functional. So he cannot do comparisons.

(ii) Robert Moyer and Thomas Landauer (Standford psychologists):

The smaller the number difference between two single digit integers, the longer it takes a person to choose the large of two numbers.

My diagnosis:

Assume that a healthy person fires his Nat(z) to compare maginitude of numbers by time interval counts. He can only fire one number at a time. Maybe there are multiple generators by which he could fire both numbers. Time interval estimates are not very precise. If the time interval difference is large, it is easy to pick the longer time interval but if these are close, he might have to repeat the firing and use polling to decide which is the larger.

(iii) The concept of number line by Dahene

Probably both left and right brain have number lines since Monseur N has a large part of the left brain destroyed. Joself Gerstmann tracked the region where the number line could occur to the inferior parietal cortex of the left hemisphere on the side of the brain, above and behind the ear. Its function is not clearcut. Whether the right or the left parietal cortex is involved is dependent on whether someone is right- or left-handed. Dahene mentioned that whilst we know which area is involved, there is absolutely no evidence of the neural code in that area. He speculated that there is probably a number line hard-wired into our brains, each number corresponding to a dedicated cluster of neurons.

My comments:

Probably Dahene was wrong. It is an uneconomical way of representing numbers. There is probably no physical number line but clusters of cells which functions like pulse generators with fixed intervals. Then counting numbers is reduced to counting time intervals. We don't see time. Therefore why should we see numbers. Is this the reason why no one has detected a line with neurons firing away when we say "6" or "7" to the test subject? I think so. Other evidences mentioned above suggest also time interval counting rather than number mappings. A number line with millions of assigned neuron clusters must be very uneconomical since most clusters will never fire throughout a person's life.

(iv) Counting in lower animals

Even rats, pigeons and chimps can count quite well. Rats can learn to press a lever 4 times or 16 times to get food. This must be related to interval counts, not symbolic number processing. Pigoens can learn to peck at a target 45 time rather than 50 times. Chips will pick a tray with 7 chocolates over one with 6 although they too suffer from the distance effect when trying to choose the larger of two numbers [1]. Thus elementary ability to perceive and manipulate number is part of our evolutionary heritage - we are born with it.

My comments:

The ability to count time intervals is absolute essential to survival. For example, if a monkey jumps 90 feet from one tree top to another, he has got to know before landing the estimated time for doing various things like grasping a branch at the landing site, how long a branch takes to flex and so on. Young babies could count probably only a few numbers say up to 4. That is why they are so impatient when hungry. 4 seconds is eternity to them which all mothers are awared of. However if you are fiftyish and ask a young child of 3 or 4 years old to guess your age he is likely to speculate: "a hundred?". All animals can count using time interval counts. The interception of a gazelle's escape path by a cheetah has to do with time measurement. The number line as such does not exist although one could say that a time line is also a number line. But the time line is not a visual image. Hence the number line does not exist, it is only conceptual.

(v) Child Prodigies Or Idiot Savant

A few of these calculating prodigies have also been great mathematicians - Carl Friedrich Gauss for example but this is rare. As we grow up, our ability to generate impulses with increaing time range will improve. So with concentrated training, it is possible to produce calculating genius. If the premise that we compute by using time intervals is correct then one should expect that autistic calculating prodigies should have a more precise sense of time than average children. But I have no facilities to test this hypothesis. One could devise quite elaborate arithmetics using time interval counting. Idiot savants are those who spend a lot of time on this confined resource and develop it to an expert level. Good mathematicians have too much distractions to do this for long.

(vi) Finding the number line by brain probing

Just because Allman and Kaas succeeded with visual images in the visual cortex, it does not follow that the number line could be probed using the same procedures. You can probably detect firing clusters but it is connected with the firing in short burst with finite number of impulses. The time taken could be very short, say less than a millisecond. The signal might be swamped by a lot of background noise of longer durations. So detecting these would be quite a feat. I don't think present instrumentation is sophisticated enough to detect these machine gun bursts.

(vii) Damaged Brains can self-repair over time

The brain is highly redundant. There must be other sites with these number generating clusters. So I expect the brain could find an alternative site given the time to reconnect. If from early days, a person is trained on mental calculations using both hemispheres, he might recover from partial damage faster than those who are not thus trained. The firing of neurons with fixed number of impulses probably can be intentional and therefore trainable. Some people might be able to fire accurately small burst like expert machine gunners. But then we await much refined measurement techniques to detect these.

4. Sequence Algebra To Simulate Number Defective Brains

Computer simulations might overcome some measurement difficulties. For example, we would like to predict the arithmetic performance of a patient with a deficient generator of the z^3/(z^3-1) type which means that he cannot generate the numbers 1,3,6,9 and so on. Please read previously published papers in section 8 of this URL site concerning the generation of number sequences. All you need is an algebraic package to do the simulation.

5. Summary

In this paper, the author forwards the thesis that the number line is not a space line which could be identified by brain probing such as being done by neurologists in image mappings to the visual cortex. Numbers are translated into time intervals and the ability to count time intervals is inborn. Young children or animals are not born to visualise numbers along a line, rather to count simple time intervals. Nature is very pragmatic. Any endowment handed down through evolution are always connected with survival. For example, colour visions of some mammals enable them to identify potential food sources more clearly at night against fuzzy backgrounds. Forward vision is another. So why should we have a number line when ability to count time intervals is inborn which also serves survival. The whole basis of computer technology is based on precise timing generated by the crystal clock and not on storing every integers into the memory for processing. We think the same reasoning applies to the time line against the number line.

5 References

1. Kunzig R. : A Head for Numbers, Discover July '97, pp108 to 115.

2. Montgomery G. : The Mind's Eye, Discover May '91, pp 51 to 56.

3. Huen Y.K. (1988): An introduction to the HGRAM (an n-Dimensional Graph Paper). Copyright application, July 1988, USA Library of Congress Copyright Registration No. Txu 354026.

4. Huen Y.K.: (1991): The HGRAM graphical format - its geometrical interpretation and its applications, Int. J. Math. Educ. Sci.Technol., 22,No.3, 403-418.

5. Liyaw A.P., Kiu C.M., Pok Y.M. and Huen Y.K.(1997): Hgram patterns of Routh stability zones in linear systems, INT.J.MATH.EDUC.SCI.TECHNOL, 1997,VOL.28,NO.2,225-241.

6. Huen Y.K.: A Simple Introduction To Sequence Algebra, URL site: http://web.singnet.com.sg/~huens/.

7. Huen Y.K. : Composite Number Sequence Challenge 1/97. URL site: http://web.singnet.com.sg/~huens/ .

8. Huen Y.K. : Lemmata, Corollaries, And Theorems In Sequence Order Analysis. URL site: http://web.singnet.com.sg/~huens/ .

======================== END OF PAPER ======================