# How can I familiarize elementary school students with infinities larger than $\aleph_0$?

Cantor's discovery of the existence of more than one infinity was a revolutionary change in human knowledge. He defined the notion of counting by bijections and showed that one can use infinities as numbers for counting mathematical objects.

I am planning a "Set Theory for Children" project to familiarize elementary school students with the basic notions of set theory.

There is no problem with giving an intuitive definition of infinity. Children in elementary school are already familiar with counting by natural numbers. One can convince them that the set of all natural numbers is infinite because it "never ends" and one can find a number larger than any given number by adding 1 to it.

But there are few concrete examples for infinities larger than $\aleph_0$. Also introducing basic transfinite arithmetic in a naive intuitive way can easily lead to paradoxes like Hilbert's Hotel.

Question. How can I explain the existence of infinities of different sizes and transfinite arithmetic so elementary school students will understand it?

Remark. I believe transfinite numbers and their arithmetic are as natural as finite numbers and finite arithmetic. The fact that they seem strange even for professional mathematicians has a root in our elementary school education and our incorrect/incomplete basic intuition about the notions of "number", "enumerating" and "arithmetic". So there is a natural way to teach these natural notions to students even in elementary school. We just need to discover such a teaching method.

• Is there a concrete reason or theory for your assumption that transfinite numbers are as natural as finite numbers? – Anschewski Mar 23 '14 at 14:37
• Citing two contradictory books from Maddy is a bit odd! I may reply on that when you post in more detail. – user173 Mar 23 '14 at 21:54
• If infinities are natural, why does that mean there is a natural way to teach them? Try replacing "infinity" with "Christoffel symbol" in that argument. – user173 Mar 23 '14 at 21:57
• Elementary school students could be anywhere from 5 to 11 years old. What age do you have in mind? Are you talking about a complete cross-section of such students, or the top 5% who show up to some sort of enrichment activity for gifted kids? – Ben Crowell Jun 25 '14 at 2:29
• The fact that you name Cantor (1845-1918) and not "ancient lore, already well-known by the Babilonians...") should give you a clue that this isn't "simple," "easy," or in any form "natural." – vonbrand Jun 25 '14 at 17:41

I think that actually trying to get students at this age to contemplate infinities in a rigorous way is probably ill advised.

I do think that exploring counting from both an "ordinal" and a "cardinal" point of view is probably a good idea.

Example for a 5 year old: Something you could do is have 20 stuffed animals, only 18 of them wearing hats. You can ask: are there more hats or stuffed animals? Get them to explain their reasoning: How could you know there are more stuffed animals than hats without counting them? These kinds of questions introduce them to the ideas of injection, surjection, etc, at a level appropriate to the age.

You can easily make them draw $\aleph$s; however the rest is much more demanding. There is a nice analysis from a researcher group taking a constructivist perspective. They distinguish potential infinity from actual infinity. The first one is covered by the idea of "counting on forever" the second one needs infinitely many things to exist in your thinking at the same time. Taking the step from potential to actual infinity seems to be even more demanding than taking the step from natural numbers as ordinals to natural numbers as cardinals (both times, the view has to change from process to object). The latter step takes children (roughly) two years. Consequently, the literature they cite does not find people up to the age of 15, who manage the transition to understanding actual infinity. This might be due to teaching as you indicated.

I believe that understanding different infinities needs at least understanding $\aleph_0$ as actual infinity. So your goals seem very ambitious.

Literature

Dubinsky, E., Weller, K., Mcdonald, M. A., & Brown, A. (2005). Some Historical Issues and Paradoxes Regarding the Concept of Infinity: An Apos-Based Analysis: Part 1. Educational Studies in Mathematics, 58(3), 335–359. doi:10.1007/s10649-005-2531-z

Dubinsky, E., Weller, K., McDonald, M. A., & Brown, A. (2005). Some Historical Issues and Paradoxes Regarding the Concept of Infinity: An Apos Analysis: Part 2. Educational Studies in Mathematics, 60(2), 253–266. doi:10.1007/s10649-005-0473-0

• Concerning "You can easily make them draw $\aleph$s," here's a quote from Math Reviews on Bourbaki's set theory volume, Chapter 3, 2nd edition (MR0154814) "In the first edition all alephs except those appearing in exponents were printed upside down; in the new edition the exception has been removed." – Andreas Blass Jan 15 '19 at 22:39

Consider a Sumerian person, living around 2500BC, who owns a flock. She hires shepherds to take the flock to pasture. Not being a scribe, she does not know Sumerian numbers- instead, for each sheep that passes her, she places a stone in an urn. When the shepherd returns, for each sheep that comes back, she removes a stone from the urn. If the last sheep passes and the urn still has stones in it, the shepherd has been doing something funny. If the stones run out, but there are still sheep, it's her lucky day. If stones and sheep run out at the same time, all is in order.

The Sumerian person in question can compare urns in the same way- remove one stone from each, at the same time. A finite urn can be emptied by removing stones one at a time. An infinite urn cannot.

Consider now two infinite urns. If we can match each stone in urn A to a stone in urn B, then urn B is at least as large as urn A. But if, after matching up each stone in urn A to a stone in urn B, there are still unmatched stones in urn B left over, then urn B is larger than urn A. You can visualize elements in urn B as stones and elements in urn B as Greek-style points with no volume to add conceptual intuition.

• It is a very nice imaginative example. Thanks and Welcome to MESE Daniel! – user230 Mar 25 '14 at 5:57
• I don't see the point of using examples they are not familiar with to explain concepts they are not familiar with. It would make more sense to me to use sweets in a cup versus icing sugar versus soda. – Jessica B Jan 15 '19 at 7:08

I would expect children can understand the idea that there are as many even numbers as there are natural numbers, as long as it's presented in a lively style, asking them questions and drawing pictures.

How many numbers are there? Infinity. (Write the first few naturals)

Can you tell me what an even number is? What are the first few? (Write them down) How many even numbers are there? Infinity!

Are there more numbers than there are even numbers? Yes, hmm, they're both infinity

But look, if I write the even ones under the ordinary numbers like this...

The trouble with higher infinities is that there are no uncountable sets that children are familiar with. Even middle school students may be uncomfortable with the idea that there are infinite decimals. The simplest example would be the set of all infinite sequences of ones and zeroes, but I suspect that would be too abstract for small children (and possibly some adults) to grasp without some difficulty. You might replace the ones and zeroes with something more colorful, ideally something that could be wrapped up in a story, but the concept of "an infinite set of infinite sets" might never stop being difficult. The diagonalisation argument is accessible for high school students and probably middle school students, but I can't imagine an 8 year old being able to maintain all the structures in their head through to the conclusion.

So if we're resigned to the idea that proving the existence of uncountable sets is too hard, can we at least communicate what their existence means? I think that to do this, you should first establish the idea that there are a lot of countably infinite sets. I'd expect kids to be surprised that there are as many even numbers as natural numbers (surprise them further by repeating the argument with multiples of a billion million trillion, rather than multiples of two). What about an infinite checkerboard (with a top-left-hand corner)? First number the first row with the natural numbers, to make it look like there are way more squares than natural numbers. Then ask if they can think of a way to number the squares so that they all get a number, and if they don't come up with anything, show them the correct answer. The crucial part is to set it up so that they're surprised that an infinite checkerboard is countable.

There might be a few other simple countable sets, and once that's done, you can ask them if they think all infinities are the same size. Then you can drop the existence of uncountables on them. I would finish with something like "these bigger infinities were first discovered by a man named Georg Cantor about a hundred and fifty years ago". I feel it's important that they understand that the fact that certain sets are uncountable is something that a person can prove, and simply giving the name of the first human to do so might be enough to communicate that, on an emotional level at least.

• I think you point at a very important issue: What would be the purpose of this? If they won't use the concepts involved, they will just get filed away (probably in the round filing cabinet) under "weird stuff some teacher got all flustered up about." – vonbrand Jun 30 '14 at 1:45
• @vonbrand Which part of the answer are you referring to? – Jack M Jun 30 '14 at 6:41

Power sets, maybe? I was just pondering this because of course preschoolers don't understand decimals.

Los Alamos National Laboratory wrote a nice lesson plan in the mid-90s around an embellished version of the Hilbert Hotel story, including the alignment to the 1989 NCTM standards:

Hotel Infinity (you'll need to click around, as the interface is very 90s...) http://www.c3.lanl.gov/mega-math/workbk/infinity/infinity.html

If you prefer a youtube video (which didn't exist in the 90s when LANL wrote their lesson plan), here's a 4 part version of the tale sketched out in pen. Here's part one: https://www.youtube.com/watch?v=ZZ7ipq9T_bw and here's the video author's website: http://www.hotel-infinity.com/

• But the Hilbert Hotel doesn't address what the OP wants to do, which is to distinguish different sizes of infinities. – Ben Crowell Jun 25 '14 at 2:31
• At least in LANL version, it grapples with different sizes of infinity by having a bus with an infinite number of people arrive, then two buses with an infinite number of people, and eventually an infinite number of buses with an infinite number of people. I think it gets at a sense of sizes of infinity, as in, there could be a way to construct the set of buses such that you get a different size of infinity. Additionally, the activity introduces the tools such as one-on-one correspondence that are needed for talking about sizes of infinity. – James S. Jun 25 '14 at 14:34

One can give them some intuition.

Draw a line segment of length 'one'. Ask them how long it is.

Then draw a single point. Ask them how much room it takes up (the answer is zero).

Draw two points; how much room do they take up? (Zero) Keep drawing poiints one by one. Say, 'No matter how far I go, even if I go forever, to normal infinity, they take up no room.'

'So there must be an even bigger infinity of points in the line segment'.

• I am not sure that the intuition of elementary school children about a "line segment" is something similar to $\mathbb{R}$ (a continuous line) or $\mathbb{Q}$ (a dense line). I think what they have in mind about a line segment is much more similar to $\mathbb{Q}$. This makes things hard. – user230 Mar 23 '14 at 14:20
• That's a very good point. I think this example relies on measure, more, than topology. I think measure is much more intuitive than things like completeness. – Brian Rushton Mar 23 '14 at 14:43
• @SaintGeorg, for the ancient Greeks it was a huge scandal to find out not all numbers where rational... and those weren't schoolchildren, far from it. – vonbrand Jun 30 '14 at 1:42

To get playing with one-to-one correspondence, The Cat in Numberland, by Ivar Ekeland, is marvelous. Of course that only helps with ℵ0, but it's a step in the right direction. I wonder if there's a way to tell a story about ℵ1...

• Thanks for your useful answer. As you correctly mentioned working in the direction of "one-to-one correspondence" seems the only possible natural way. – user230 Mar 25 '14 at 3:17

If the idea of "larger than infinity" caused a riot among grown up mathematicians, accustomed to rarefied abstract thinking for years if not decades, this will just make their little heads explode.

In my automata theory classes I have to get my students (by then mangled though school and two years of mathematics at college level) to grasp Cantor's diagonal argument, and I feel I'm unsuccessful most of the time.

(Yes, it is an extremely beautiful result. But not everyone is prepared to appreciate it as such. Tough luck, move on.)

It's actually a very hard topic to teach. We don't want to teach them the Naive set theory definition of a cardinal number because Naive set theory is contradictory. In Zermelo-Fraenkel set theory, it's easy to define the property of there existing a bijection from one set to another set but the axiom of choice is not provable so it's very hard to define an object for each set such that two sets are assigned the same object when there's a bijection from one set to the other although that can be done using Scott's trick. Alternatively, you could create a provably consistent extension of ZF where you just declare that the cardinality of each set is a set such that two sets have the same cardinality when there is a bijection from one to the other. With either way of introducing them, you can show that there is in fact no set of all cardinal numbers.

Ordinal numbers are even easier to teach. First we have all the finite ordinals. Next comes $$\omega$$. Now starting from it, you can apply the successor function as many times as you want. The next larger ordinal is $$\omega \times 2$$. The supremum of all the ordinal numbers that can be expressed as $$\omega \times$$ a finite ordinal is $$\omega \times \omega$$. If you then again apply the successor function as many times as you want, the next ordinal after all of them is $$(\omega \times \omega) + \omega$$. We also have $$(\omega \times \omega) + (\omega \times 2)$$ We can keep going and get $$(\omega \times \omega) + (\omega \times \omega)$$ which can also be written as $$(\omega \times \omega) \times 2$$. We can also add to that any ordinal number less than $$(\omega \times \omega) \times 2$$. The supremum of all ordinal numbers that can be gotten that way is $$(\omega \times \omega) \times 3$$. The supremum of all ordinal numbers that can be expressed as $$(\omega \times \omega) \times$$ a natural number is $$(\omega \times \omega) \times \omega$$.

This is really a totally different concept of an ordinal number than the ZF definition and probably easier to teach than ZF. It includes only those ordinal numbers that are small enough to use to create a proof system of pure number theory. By Godel's incompleteness theorem, you can always think of proof systems corresponding to larger ordinal numbers than you could have before. Yet, those ordinal numbers will always be smaller than the Church-Kleene ordinal. How can that be when there's no end to constructing ordinal numbers useable to create a proof system? Because the Church-Kleene ordinal is unfathomably large. Once you create a system of ordinal numbers useable in a proof system, you still don't have all of the ones smaller than the Church-Kleene ordinal so it doesn't necessarily mean that when you make a major extension to that system, you will get one large enough to include the Church-Kleene ordinal. This last paragraph can probably only be understood by adults and probably is not worth teaching to elementry school kids. I wrote it just to explain how the concept of ordinal numbers they could be taught really is a completely different concept from the ZF definition and that concept is the one that should be taught.

• Before ZF I suppose we need to teach first order logic, and before this we need to teach classical propositional logic, and before this we need to teach classical implicational propositional logic, and before this we need to teach various fragments of implicational propositional calculus (e.g. my answer here). – Dave L Renfro Jan 16 '19 at 15:07
• @DaveLRenfro ZF is built on intuition. I never learned how to write a formal proof in ZF but have some intuition for what I think is provable in ZF. People who don't know how to write a formal proof in ZF are probably better at guessing whether a statement is provable in ZF than people who don't know how to write a formal proof in NF are at guessing whether a statement is provable in NF. That's why I think elementry students might be able to be introduced to the concept of those specific cardinal numbers and understand it. – Timothy Jan 17 '19 at 0:44
• @DaveLRenfro I guess you're right. That last part of this answer that I later added in is too hard for them to learn so I deleted it. – Timothy Jan 22 '19 at 2:03

Depending on their age, they may not have reached what Piaget called the “abstract operational” stage yet. If they’re still in the concrete operational stage, they will need physical or otherwise concrete explanations. This is one reason why we don’t attempt algebra at 5th grade—their minds haven’t gotten there yet.