What is Mathematics?
I will here present my own personal account on what mathematics is.
I think that in the broader sense mathematics is the "exercise of formal
thought", hence it is a sort of mental sport, like Chess, but on symbolic
level, i.e. a symbolic game.
Here formal theory is taken to mean any set of strings of symbols
governed by string manipulation rules, in such a manner that
we can construct further strings of symbols from prior ones using
these string manipulation rules, and such that the whole symbolic
streaming is not affected by what those symbols stand for.
For example take ordinary arithmetic that we were taught at primary school,
one can see that we infer string of symbols from prior ones without us
really knowing what for example the symbol 1 or even symbol +
stands for, we have rules about addition and multiplications and equality
and those would produce strings of symbols from prior ones without any
need for us to care about what those symbols exactly stand for. Such
a symbolic inferential system is what I label as FORMAL. The merit of
those systems is that they provide a descriptive account of rule following
that can be applicable to any empirical discipline, i.e. they are
empirically neutral, I can substitute anything instead of 1,
and the system reach into the same results. So to give a specific example lets take 2+2=4, this is a string of symbols dictated by the axioms of arithmetic, we reached that by strict rule following string manipulation rules 4 is the name given to the term symbol 1+1+1+1, and 2 is the name given to the term symbol 1+1, and by equality rule any string can be replaced by its name without causing additional change to the prior to replacement status, so 1+1+1+1=4 leads to 2+2=4. No particular meaning of 1 is required nor of +, they can be taken to be blind symbols (non referring symbols), the flow of inference follows the rules of the game.
A formal system can be taken to mean a system in which all of its symbols
are non referring to any particular object, actually we can even consider
all non logical symbols of it to have no reference at all, and still the whole
system will work, since the inferences in the system has nothing to do with
the reference of its non logical symbols.
However such an extreme non referring formal symbol would be hard to verify if
it is consistent, and actually one of the most important aspects in examining these
symbolic games is their syntactical consistency, for if they are inconsistent then they would prove everything and won't be worth considering other than knowing that it must be avoided.
The simplest criterion for determining significance (non triviality)
of those games is to pose the syntactical criterion of non existence
of a trivial proof of inconsistency of the theory, this doesn't prove
the consistency of the system, but at least tell us that it is possibly
consistent, or even if inconsistent still some non trivial fragments
of it might be consistent. However if we say that mathematics is just those
non trivial symbolic games, then this would constitutes what is meant by
mathematics in the more general sense. It would include all fun symbolic
games, and definitely not all such games would be of scientific (factual)
value. So this is not how mathematics is usually tackled.
The customary mathematical experience is about formal systems
in the above sense that have some apriori justification of being
consistent. This is what I call as justified mathematics.
This will focus our examination of symbolic games to those
that are highly likely to be consistent and likely being of use in the scientific process.
An apriori justified matter is usually something that reflects a basic rule of
our thought whether analytic or synthetic. Analytic rules which are the ones
that can be known from just knowing the meaning of the words, pose no real problem.
However the apriori synthetic ones are somewhat vague,
I would consider those as basic elements of our thought
about basic observations or intuitions about basic observations around us. And I think they are
widely acceptable that requires no empirical validation,
since they are rules of our thought by which we can discover
things, so it in some sense comes prior to empirically
validated (a posteriori) material.
Now if a formal system can be a part of a discourse about
apriori justified concepts that didn't require the consistency
of that formal system in the first place, then this part-hood
will be considered as providing a justification for consistency
of this formal system. By then this formal system would be enrolled
in the mathematical systems, also the apriori justification of it
would be part of mathematics, even the contentful axiomatic system
having that particular justifying semantic is also part of mathematics.
Example of apriori synthetic justified piece of thought is that about
"succession", this came from observations around us, we see that things
can follow things in succession, so the imagery grasp of having a starting
object then having a successor to it, then a successor to that successor
and so on.. is a basic figment of our perception about the world; Now
the ordinary +1 interpreted as denoting the successor, i.e. x+1 means
just Successor or x, this is an analytic fix and is not a cause
of any problem because it didn't add anything other than
extra-wording spelling succession, now same applies to
multiplication which is just abbreviatory over addition.
Now examination of basic thought about succession as presented
to our sensorium reveals that any finite iterative successor of the
first object is reachable by a predicate that holds of
the first object and that transits from every object to the
next, this is trivially true, and is acceptable as a basic rule
of thought about finite successions from a starting object,
there is no qualms about it really, and thus is justified to
be held as non contradictory (doesn't give rise to inconsistent
formalization) when added to the basic axioms characterizing succession,
addition and multiplication. The whole system is pretty much justified
apriori, through a mix of apriori synthetic and analytic,
and so arithmetic which is a part of this apriori investigation
about succession, would be a formal system whose consistency is
justified apriori.
Now there are two significant aspects about mathematics, the first is
it being formal thus enabling it to describe themes of rule following
that can be inhabited by any sort of objects, and the other part
is that of its apriori justification which will make it
about a principal aspect of our thought, thus giving strong grounds
for justifying its consistency and also for probability of having
applications in empirical sciences.
Now being apriori justified wont' restrict the aposteriori applicability
of the justified formal system, so the justified system won't be so limited
as to being just applicable to narrow rule following scenarios about
its justifying grounds, on the contrary being apriori is
"aposteriori neutral" or "empirically neutral". So this
justification won't limit the empirical applicability of the justified
system as compared to aposteriori justified formal theories (see below).
The apriori justification would be important for:
1. Justifying the claim of consistency of an already defined formal system
2. Can lead to the production a formal system from the outset
3. Can be important in extending the formal system by adding new axioms.
4. Provide semantics for the formal system that might be important in applications of it.
Examples of 1 is interpreting ordinary arithmetic in Peano arithmetic or simply
in more trivial kinds of arithmetic + the omega rule.
Example of 2 is Geometry where the formal system was
produced form the outset by basic thought about
spatial entities as trivially presented to our sensorium, which
are justified as apriori synthetic. Then after Geometry
was defined, one can go and use all of the formal system
of it (its syntax) as a non referring symbolic game the
symbols of which can stand used to refer to any empirically
validated object.
Example of 3 is extending the ordinary Peano arithmetic by an omega rule.
Example of 4 is definitely the use of arithmetic in most
of scientific disciplines, this can sometimes benefit
from basic thought about succession as an explanation
for the naturals.
One important issue is about aposteriori formal systems
and non mathematical apriori formal systems. Now for the
former ones, the formal system is already required to be
justified apriori before using it to explain empirical material,
no scientist would produce a theory of physics lets say,
that uses a formal system that he doesn't have any
justification of it being non contradictory in the first place,
that's why aposteriori axiomatic systems are NOT included in
the mathematical justification record because they didn't make
that justification by themselves, those systems only uses
these formal systems by assigning particular meanings to
their symbols as related to the subject matter of their investigation,
and the whole axiomatic system is actually fully contentful
rather than being formal, but say even if we took the formal
system used (the syntactical part) and make its symbols non
referring then still the formal system was not justified in
the first place by the aposteriori account of truth.
Should that be the case then it would become a legitimate
part of mathematics, but this never occurred so far,
and even if it occurs, it possibly won't have much
applications outside that particular aposteriori field
in which they were born.
Similarly apirori formal systems like formal ethics, formal ontology and
formal epistemology etc... all of those are just applied formal,
i.e. they require the formal system they use to be apriori
secured consistent before using it, and so the apriori material related to epistemology, ontology and ethics itself didn't contribute to the
justification of consistency of the formal systems they use.
So they are not part of mathematics. However should they be
the only source of justification for the formal systems they use
(i.e. they didn't require beforehand that the formal system
to be secured consistent by some other apriori investigation)
then they should be a legitimate part of mathematics,
but non of those is present so far, and even if that happened
it would likely be too particular that it won't have much
use outside the field in which it thrived.
In nutshell mathematics is about
"Non trivial symbolic games that have some apriori justification of being consistent"
However what remains unresolved is the exact nature of
mathematical objects, those objects that directly give
rise to the apriori justified formal thought?
what are those objects, are they present in reality or
are transcendental as rules of our thought needed prior
to observation? or in the admixture between those two?
are they totally dispensable by concrete objects and
so they are nothing but names of states about concrete
objects? are they fictions? this is another issue.
But I tend to think some of them are abstract objects
that do have a casual relationship with concrete objects!
they actually frame them into what they are and we
come to know about them from the effect of this framing
of those concrete objects, natural numbers might be
just objects that enforce concrete material to be
distributed as multiplicities, and us being concrete
objects sensed what happens to the distribution of
that concrete material by the abstract objects,
and thus came to know about them in an indirect manner,
so there is a causal relationship from those objects onto
the concrete world which in turn is transmitted to our sensorium.
The argument for mathematical objects being abstract
is that for example number 1 can be present "in whole"
at two places at the same time, this makes it in some
sense spatially ubiquitous, however that doesn't
mean that it is totally alien to space-time, it is still
spatio-temporal! but not delimited in the way concrete objects
are (cannot present "in whole" at two places at the same time,
a property we may label as "spatial-delimitation").
I don't see a clear argument that necessitates that causality
requires spatial-delimitation, in order to be possible?
We can on the other hand have objects that are spatio-temporally
delimited, i.e. BOTH cannot be present in whole at two times at the same place
and cannot be present in whole at two places at the same time, now
still those can effect casual relationships, now causality
is not thought to be decreased when objects become just spatially delimited
(i.e. can present in whole at two times at the same place but not in two distinct places at the same time), which are most of the usual objects, so
dropping the requirement of temporal delimitation didn't affect
the possibility of effecting casual relationships, on the contrary there is a sense that it would increase the ability to effect casual relationships, since it gives the impression of more presence in space-time than the fleeting rapidly transiting objects, and so there is more chance of collision with concrete objects and thus more possibility of effecting casual relationships; so why dropping
spatial delimitation is expected to make effecting casual relationships
impossible? I don't see a clear argument for that? It actually also appears to have more presence in space-time than the concrete objects, although of course not touchable nor can be heard, so in this sense concrete objects no doubt are much easier to know of, but still I don't see a clear argument supporting that mathematical objects are 'outside of space-time', since this is a too vague statement to make. If they are so then how come we came to know about them being spatio-temporally non delimited? why should they have any such relationship to space and time if they are external to it? An object that do not effect casual relationships would be imagined as having Ghosty nature, something that is like zero dense, or in other words some object that no concrete object (i.e. spatially delimited) can collide with, so every concrete object will pass through it, since it cannot collide with it since it is zero dense, such an object cannot be seen because photons for example cannot be reflected from it, nor can be touched nor can be heard nor can it be presented to any of our five senses, such a ghosty object would be of the nature that it cannot effect casual relationship with any concrete object, and it can in principle be thought of being able to permeate through matter, passing through walls and barriers like ghosts in Cartoons. I don't think that any mathematician when he says that mathematical objects are abstract I say I don't think he means that they are so ghosty, because simply such objects won't be discover-able. On the other hands I think it is imaginable to think of objects that have ghosty nature and yet still are spatio-temporally delimited! i.e. being ghosty doesn't entail being spatio-temporally non delimited. The main point that I'm raising here is that when one says that an object cannot effect casual relationships with any concrete object then this entails that it has a ghosty nature as described above, now I don't see how being not spatially delimited (i.e. presence at two places at the same time) implies having such a ghosty nature? I think it is possible to be BOTH "not spatially delimited" and "not zero dense", why not? I don't see any reason to forbid that. It is true that being not spatially delimited can give the impression of being ghosty since the usual spatial barring on concrete objects is breached, this barring strangulate concrete objects to exist in just one place at a given moment of time, so if this is breached, thus it would indeed give the impression of permeating through spatio-temporal concrete material, and thus having a ghosty nature, but this impression is not a proof that this must be the case, and it is in some sense mysterious, there is no clear 'concrete' barrier that was breached (by non-spatial-delimitation) in order for us to claim that a ghosty nature is necessary for such a breach, and after all there is no clear evidence that being ghosty would enable such non spatial delimitation in the first place? what has been breach is a certain spatial barring property that pertains to concrete objects, but breaching a barring property is different from permeating through concrete objects, so that impression is not grounded at all; actually those non-spatially delimited mathematical objects do not have full ghosty material, because we do acquire knowledge about mathematical entities, so if those were ghosty then no such knowledge would have been possible unless there are routs of income of knowledge other than our sensorium, which is unlikely.
One implication of what I said above is that basic mathematical objects might be physically sensed though not as solid as concrete manner, but yet still they might be physically sensed through concrete material reflected from them reaching to our sensorium thus conveying information about them, this might be possible? so if we imagine for example the universals as non-delimited spatio-temporal objects that are not zero dense, attached to the concrete objects that they predicate, then number can be considered a higher level universal attached to the lower level universals they would predicate (since number is a property predicative of predicates in terms of their fulfillment grade, so 2 can be imagined as an object attached (in whole) to all predicates having double fulfillment, and each predicate having double fulfillment can be imagined as an object attached (in whole) to each of the two objects it predicates. Similarly sets can be imagined along the same vain, as predicates of predicates of .... etc. Other definable parts of mathematics are just analytic conservative extensions over those basic sensed objects. This seems rather awkward and unbelievable. However the above line of reasoning presented here shows that this cannot be ruled out, still I think that an empirical way to test it is rather very difficult if possible at all.
In nustshell I don't see that being spatially non-delimited implies not effecting casual relationships, and even the opposite direction is not justifiable. That said, I still see that it is possible to have a casual interaction between concrete objects (at least spatially delimited objects) and abstract objects (i.e. spatio-tempral objects that are not spatio-temporally delimited) and of course this will reach our sensorium since the effect of that interaction on the concrete material will be transmittable to our sensorium, since our sensorium is concrete; so the possibility of indirect casual relationship between us and those spatio-temporal non delimited objects can still hold, and this might explain at least why it is not impossible to know of those spatio-temporal non delimited objects (i.e. the mathematical objects) which are called usually as abstract, should those be truly existing.
I'm not sure if this makes sense. But even if so, surely not
all of mathematics can have this esteemed ontological status,
many parts of it are just games or fictitious conservative extensions or
even false!
A recent account:
What is Mathematics?
For any syntactic formal theory T, let T* be the class of all models of T. Now what is shared between all and only models of T (i.e. elements of T*) can be said to be the 'essence' of T, so here we think of the essence of T as a universal shared between all and only models of T, then we can in some sense define the mathematical realm as the class of all essences of formal theories. Of course mathematics would then be naturally defined as the investigation of possible essences of formal theories. Now an 'essence' of a formal theory is an abstract notion, so it cannot be seen, a better visualization of it would be the syntactical part of that theory itself with all of its primitive symbols not holding any particular reference, so take any formal theory T, take away all particular reference from its symbols, and we'll have a symbolic system that would stand for any relations having the same arity of its primitives as far as those satisfy its axioms, so the system can be said to be general over all such relations that satisfy its axioms. So for example Geometry is satisfied in some imaginary space with its points being atoms and the rest of its primitives being the obvious spatial interpretation given to them, this can be said to be the canonical model, i.e. the meant model, but still we can have exactly the same model being composed from pairs of natural numbers, and since natural numbers and their pairs are both abstract non spactio-temporal objects, then this model of Geometry would be non spatio-temporal in nature, but still it is a model of it, to mathematics it makes no difference which is the model, Geometry can be satisfied by any of those models. Now the particular assignment of primitives of Geometry to a specific model like for example the customarily thought spatio-temporal model, is to be called an instance of APPLIED MATHEMATICS, more specifically APPLIED GEOMETRY. So applied mathematics is the investigation of particular assignment to mathematical objects, like for example formal physics, formal ethics, formal ontology, formal epistemology, etc..
Clarification
Putting the definition of mathematical realm in terms of classes is a convenience rather than a necessity. One can simply coin matters in term of predicates, like in saying, a mathematical object is the essence of some formal theory, and then define mathematics as the investigation of mathematical objects. Let me put it formally:
Q is a mathematical object <-> Exist T for all M (Q(M) <-> M|= T)
or one can put it as:
Q is the essence of T <-> for all M(Q(M) <-> M|= T)
Q is mathematical object <-> Exist T (Q is the essence of T)
I would also note, that one can present a definition of the mathematical realm in a full extensional manner, instead of the purely intentional one presented just above, like in saying that the mathematical realm is the class of all equivalence classes of models of theories, formally this is:
Mathematical realm={T*| Exist T for all M (M in T* <-> M|=T)}
And mathematics would be the investigation of what could possibly be in that realm.
However to me it makes no big difference which way one want to present matters, since they are equivalent.
Now about models, models need not be sets, they can be graphs, mereological totalities, the Cosmos, parts of the Cosmos, actually what is meant here by "model" is any semantic the explains the syntax of the theory.
Any object that can satisfy all sentences of theory T is here referred to as a model.
The idea here is that a single model wouldn't really be the essence of what T is speaking about, since it has many extra-properties not related to the model, to filter out all of these properties and just focus on what the model is saying, we need to take what is shared by all and only those models satisfying the concerned formal theory, and this can be thought of as a kind of a universal ranging over models of formal theories, and that's in my opinion the essence of what a formal theory is speaking about, and mathematics is about those essences.
In some sense I consider "essences" of formal theories as reflective of
their syntactical parts, so take any formal theory and take away any particularity given to the interpretation of its symbols, then you get the syntactical part of the formal theory which I sometimes call it as the "form" of a formal theory. To me "essences" mirror "forms" and vise verse, in other words "forms" and "essences" (of formal theories) are counterparts, each mirror the other on the syntactic\semantic levels respectively.
I'm pretty much sure that all of this can be done in a purely non mathematical framework, and I think Zalta had done such work regarding structuralism, in his "Foundations Mathematical Structuralism".
I wanted to present a main difference between applied formal theories and pure ones, the applied formal theories have a particular interpretation of their symbols, for example formal physics would have the Cosmos as its model, it is not just the syntax but the syntax as interpreted being about the physical objects, which are empirically validated objects, so it is a particular assignment of the symbols of some formal theory to a specific model, it is not about the form of its formal theory, so it is not a mathematical object in the above-mentioned sense, however it does belong to applied mathematics. Same thing to be said of formal ethics, formal epistemology, formal Ontology, etc... all of those theories have particular modular assignments and specific interpretation of their primitives. While formal theories belonging to pure Mathematics would be understood as speaking about what is shared between all models satisfying them. Admittedly we do have modular restrictions in PURE mathematics. That would be an application of the formal theory but it is still a pure application! this looks different from the application of formal theories to physics, ethics, ontology etc... so there is something missing. Pure mathematics must be contrasted with "external applied mathematics", i.e. applications of formal theories to matters external to mathematics. I think the real matter is the way of modular assignments, in pure applications the assignments are still done within some formal theory, so it is formally definable or at least formally characterized, so for example restriction of arithmetic to the standard model, is something that can be done explicitly in a formal theory. But with external applications the modular assignments are made with a "non formal" element, i.e. it is 'external' to the formal theory! for example assigning certain formal theory to the physical world, this require empiricism, but empiricism itself is not a formal matter, there is no formal theory that dictate our five senses, so that restriction to the COSMOS as a model cannot be captured formally. I think this is the essential difference between pure mathematical theories and the external applied mathematical theories, the later ones are not purely formal!
An opposing view
Now an opposing view might be held that mathematics is not about essences of formal theories, but rather it is about canonical models of formal theories, here a specific description of what constitutes a canonical model of a formal theory must be put forth, in order to discriminate it from non canonical models. The idea is that it is the standard model of a formal theory that counts, the other models are just external to the purpose of the theory, so only the standard model reflects the essence of the formal theory and so it is the canonical model of the formal theory. Here to enact that and make it compatible with mathematics still one must take out all characteristics not related to the syntactical part of the formal theory, and then assign a model to it that just realize that syntax. So the mathematical realm would then be defined as the class of all canonical models of syntactical parts of formal theories. However though this is tempting, yet I don't see if there can be a uniform way of identifying canonical models of formal theories, in addition restriction to a standard model must be made explicit in the formalization of the theory itself, and not just being implicit. That said I tend to favor the above more general 'essence' line of definition,
The above 'essence' line of definition of mathematics is basically 'structuralism' in other terms.
I think this line of thought is the nearest to what mathematics is.
Now a basic objection to the above is that it looks too complex a definition, and in some sense it is circular, since it defines mathematics by terms that presuppose mathematics in the first place, like "formal" theories which takes one back to theory of "recursion" which again is mathematical in nature.
The answer to this objection is that the above line of thought can be rigorously formulated in such a manner as to avoid any circular reference to mathematics. "Formal" theories can be defined after 'string manipulation themes" giving rise to "symbolic inferential systems" in a careful manner without requiring any prior reference to mathematical objects. And so we can build symbolic inferential systems like languages then define assigned languages by assigning any two distinct letters (like T and F) to particular string of symbols in them, all these systems would have symbols having no meaning attached to them other than their role in enacting specific rules of assigned string manipulation (i.e. symbolic inferential rules) in the respective system, from that we can define the known logical systems, then we go upwards to the mathematical systems which can be envisioned as all possible combination of sentences having extra-logical symbols, that can extend these logical languages, with all their symbols standing for no particular reference.
In practice we cannot produce all those mathematical systems by mere combinatorics of symbols, we actually start with a certain concept and then lay down rules about it in form of axioms that capture its basic characteristics, then with the logical machinery we'll be having a full formal system. Now take this formal system, and take away all particular external reference to its symbols, then it would be the syntactical part of the formal system, and its essence would be what mathematics aims at, and not just that particular concept it started with. However the initial concept that lead to birth of that formal theory would be mostly the canonical line of models of that theory, although not the only one being so. So mathematics is meant with the investigation of the 'essences' of those formal systems and not just with the particular main applications of them.
Of course "traditional arithmetic" is formal, according to the definition of formal presented here, since it is a symbolic inferential system, but it is an implicit symbolic system since the logical axioms and the inference rules are made implicit, but the whole system is no doubt a formal system, here we'll label such systems as: "implicit formal systems". Even the modular restriction of arithmetic to the standard model of arithmetic is still applied mathematics, but a pure applied mathematics, since it can be the essence of some formal theory explicitly stipulating that restriction.
All of mathematics is presented as formal systems whether implicit or explicit, and the basic difference between those formal systems and those of formal physics, formal ethics, etc.. is that the later ones are particular assignments of formal systems, while mathematics is concerned with the essence of formal systems rather than their particular assignments. However Applied mathematics would be that part of mathematics that is concerned with particular assignments to formal systems, and so formal physics, ethics, metaphysics, epistemology, ontology, etc.. all would be classified among applied mathematics. So mathematics in the broader sense is the investigation of formal systems. Pure mathematics is the investigation of possible essences of formal systems, while applied mathematics is the investigation of particular assignments to formal systems. External applied mathematics, is the applied mathematics whose particular assignments includes extra-formal fixes, i.e. fixes that cannot be
made by a syntactic formal system with all interpretations given to its symbols taken away.
As regards the nature of essences of formal systems, then some of those might be real (Platonism, Ante Rem Structuralism), some might be indirectly real, like in being fictional conservative extensions over concrete reality (Fictionalism;Nominalism), others might be simply false in the sense that they won't have any real world application.
Zuhair