1

A Critique of Absolutist Philosophies

of Mathematics

1. Introduction

The main purpose of this chapter is to expound and criticize the dominant
epistemological perspective of mathematics. This is the absolutist view that
mathematical truth is absolutely certain, that mathematics is the one and perhaps the
only realm of certain, unquestionable and objective knowledge. This is to be
contrasted with the opposing fallibilist view that mathematical truth is corrigible, and
can never be regarded as being above revision and correction.
Much is made of the absolutist-fallibilist distinction because, as is shown
subsequently, the choice of which of these two philosophical perspectives is adopted
is perhaps the most important epistemological factor underlying the teaching of
mathematics.

2. The Philosophy of Mathematics

The philosophy of mathematics is the branch of philosophy whose task is to reflect

on, and account for the nature of mathematics. This is a special case of the task of
epistemology which is to account for human knowledge in general. The philosophy
of mathematics addresses such questions as: What is the basis for mathematical
knowledge? What is the nature of mathematical truth? What characterises the truths
of mathematics? What is the justification for their assertion? Why are the truths of
mathematics necessary truths?
A widely adopted approach to epistemology, is to assume that knowledge in any
field is represented by a set of propositions, together with a set of procedures for
verifying them, or providing a warrant for their assertion. On this basis,
mathematical knowledge consists of a set of propositions together with their
proofs. Since mathematical proofs are based on reason alone, without recourse to
empirical data, mathematical knowledge is understood to be the most certain of all
knowledge. Traditionally the philosophy of mathematics has seen its task as
providing a foundationfor the certainty of mathematical knowledge. That is,

3
The Philosophy of Mathematics Education

providing a system into which mathematical knowledge can be cast to

systematically establish its truth. This depends on an assumption, which is widely
adopted, implicitly if not explicitly.

Assumption

The role of the philosophy of mathematics is to provide a systematic and absolutely

secure foundation for mathematical knowledge, that is for mathematical truth.1

This assumption is the basis of foundationism, the doctrine that the function of the
philosophy of mathematics is to provide certain foundations for mathematical
knowledge. Foundationism is bound up with the absolutist view of mathematical
knowledge, for it regards the task of justifying this view to be central to the
philosophy of mathematics.

3. The Nature of Mathematical Knowledge

Traditionally, mathematics has been viewed as the paradigm of certain knowledge.

Euclid erected a magnificent logical structure nearly 2,500 years ago in his Elements,
which until the end of the nineteenth century was taken as the paradigm for
establishing truth and certainty. Newton used the form of the Elements in his
Principia, and Spinoza in his Ethics, to strengthen their claims to systematically
expound truth. Thus mathematics has long been taken as the source of the most
certain knowledge known to humankind.
Before inquiring into the nature of mathematical knowledge, it is first necessary
to consider the nature of knowledge in general. Thus we begin by asking, what is
knowledge? The question of what constitutes knowledge lies at the heart of
philosophy, and mathematical knowledge plays a special part. The standard
philosophical answer to this question is that knowledge is justified belief. More
precisely, that prepositional knowledge consists of propositions which are accepted
(i.e., believed), provided there are adequate grounds available for asserting them
(Sheffler, 1965; Chisholm, 1966; Woozley, 1949).
Knowledge is classified on the basis of the grounds for its assertion. A priori
knowledge consists of propositions which are asserted on the basis of reason alone,
without recourse to observations of the world. Here reason consists of the use of
deductive logic and the meanings of terms, typically to be found in definitions. In
contrast, empirical or a posteriori knowledge consists of propositions asserted on the
basis of experience, that is, based on the observations of the world (Woozley, 1949).
Mathematical knowledge is classified as a priori knowledge, since it consists of
propositions asserted on the basis of reason alone. Reason includes deductive logic
and definitions which are used, in conjunction with an assumed set of mathematical
axioms or postulates, as a basis from which to infer mathematical knowledge. Thus

4
 A Critique of Absolutist Philosophies

the foundation of mathematical knowledge, that is the grounds for asserting the truth
of mathematical propositions, consists of deductive proof.
The proof of a mathematical proposition is a finite sequence of statements ending
in the proposition, which satisfies the following property. Each statement is an axiom
drawn from a previously stipulated set of axioms, or is derived by a rule of inference
from one or more statements occurring earlier in the sequence. The term ‘set of
axioms’ is conceived broadly, to include whatever statements are admitted into a
proof without demonstration, including axioms, postulates and definitions.
An example is provided by the following proof of the statement ‘1+1=2’ in the
axiomatic system of Peano Arithmetic. For this proof we need the definitions and
axioms s0=1, s1=2, x+0=x, x+sy=s(x+y) from Peano Arithmetic, and the logical
rules of inference P(r), r=t ⇒P(t); P(v)⇒P(c) (where r, t; v; c; and P(t) range over
terms; variables; constants; and propositions in the term t, respectively, and ‘⇒’
signifies logical implication).2 The following is a proof of 1+1=2: x+sy=s(x+y),
1+sy=s(1+y), 1+s0=s(1+0), x+0=x, 1+0=1, 1+s0=s1, s0=1, 1+1=s1, s1=2, 1+1=2.
An explanation of this proof is as follows. s0=1[D1] and s1=2[D2] are
definitions of the constants 1 and 2, respectively, in Peano Arithmetic, x+0=x[A1]
and x+sy=s(x+y)[A2] are axioms of Peano Arithmetic. P(r), r=t⇒P(t)[R1] and
P(v) ⇒P(c)[R2], with the symbols as explained above, are logical rules of inference.
The justification of the proof, statement by statement as shown in Table 1.1.
This proof establishes ‘1+1= 2’ as an item of mathematical knowledge or truth,
according to the previous analysis, since the deductive proof provides a legitimate
warrant for asserting the statement. Furthermore it is a priori knowledge, since it is
asserted on the basis of reason alone.
However, what has not been made clear are the grounds for the assumptions
made in the proof. The assumptions made are of two types: mathematical and logical
assumptions. The mathematical assumptions used are the definitions (D1 and D2) and
the axioms (A1 and A2). The logical assumptions are the rules of inference used (R1
and R2), which are part of the underlying proof theory, and the underlying syntax of
the formal language.

Table 1.1: Proof of 1+1=2 with justification

5
The Philosophy of Mathematics Education

We consider first the mathematical assumptions. The definitions, being explicit

definitions, are unproblematic, since they are eliminable in principle. Every
occurrence of the defined terms 1 and 2 can be replaced by what it abbreviates (s0
and ss0, respectively). The result of eliminating these definitions is the abbreviated
proof: x+sy=s(x+y), s0+sy=s(s0+y), s0+s0=s(s0+0), x+0=x, s0+0=s0, s0+s0=ss0;
proving ‘s0+s0=ss0’, which represents ‘1+1=2’. Although explicit definitions are
eliminable in principle, it remains an undoubted convenience, not to mention an aid
to thought, to retain them. However, in the present context we are concerned to
reduce assumptions to their minimum, to reveal the irreducible assumptions on
which mathematical knowledge and its justification rests.
If the definitions had not been explicit, such as in Peano’s original inductive
definition of addition (Heijenoort, 1967), which is assumed above as an axiom, and
not as a definition, then the definitions would not be eliminable in principle. In this
case the problem of the basis of a definition, that is the assumption on which it rests,
is analogous to that of an axiom.
The axioms in the proof are not eliminable. They must be assumed either as
selfevident axiomatic truths, or simply retain the status of unjustified, tentative
assumptions, adopted to permit the development of the mathematical theory under
consideration. We will return to this point.
The logical assumptions, that is the rules of inference (part of the overall proof
theory) and the logical syntax, are assumed as part of the underlying logic, and are
part of the mechanism needed for the application of reason. Thus logic is assumed as
an unproblematic foundation for the justification of knowledge.
In summary, the elementary mathematical truth ‘1+1=2’, depends for its
justification on a mathematical proof. This in turn depends on assuming a number of
basic mathematical statements (axioms), as well as on the underlying logic. In general,
mathematical knowledge consists of statements justified by proofs, which depend on
mathematical axioms (and an underlying logic).
This account of mathematical knowledge is essentially that which has been
accepted for almost 2,500 years. Early presentations of mathematical knowledge, such
as Euclid’s Elements, differ from the above account only by degree. In Euclid, as
above, mathematical knowledge is established by the logical deduction of theorems
from axioms and postulates (which we include among the axioms). The underlying
logic is left unspecified (other than the statement of some axioms concerning the
equality relation). The axioms are not regarded as temporarily adopted assumptions,
held only for the construction of the theory under consideration. The axioms are
considered to be basic truths which needed no justification, beyond their own self
evidence (Blanche, 1966).3 Because of this, the account claims to provide certain
grounds for mathematical knowledge. For since logical proof preserves truth and the
assumed axioms are self-evident truths, then any theorems derived from them must
also be truths (this reasoning is implicit, not explicit in Euclid). However, this claim
is no longer accepted because Euclid’s axioms and postulates are not considered to
be basic and incontrovertible truths, none of which can be negated or denied
without resulting in contradiction. In fact, the denial of some of them, most notably

6
 A Critique of Absolutist Philosophies

the Parallel Postulate, merely leads to other bodies of geometric knowledge (non-
euclidean geometry).
Beyond Euclid, modern mathematical knowledge includes many branches
which depend on the assumption of sets of axioms which cannot be claimed to be
basic universal truths, for example, the axioms of group theory, or of set theory
(Maddy, 1984).

4. The Absolutist View of Mathematical Knowledge

The absolutist view of mathematical knowledge is that it consists of certain and

unchallengeable truths. According to this view, mathematical knowledge is made up
of absolute truths, and represents the unique realm of certain knowledge, apart from
logic and statements true by virtue of the meanings of terms, such as ‘All bachelors
are unmarried’.
Many philosophers, both modern and traditional, hold absolutist views of
mathematical knowledge. Thus according to Hempel:

the validity of mathematics derives from the stipulations which determine

the meaning of the mathematical concepts, and that the propositions of
mathematics are therefore essentially ‘true by definition’.
(Feigl and Sellars, 1949, page 225)

Another proponent of the certainty of mathematics is A.J.Ayer who claims the

following.

Whereas a scientific generalisation is readily admitted to be fallible, the

truths of mathematics and logic appear to everyone to be necessary and
certain.
The truths of logic and mathematics are analytic propositions or
tautologies.
The certainty of a priori propositions depends on the fact that they are
tautologies. A proposition [is] a tautology if it is analytic. A proposition is
analytic if it is true solely in the virtue of the meaning of its consistituent
symbols, and cannot therefore be either confirmed or refuted by any fact of
experience.
(Ayer, 1946, pages 72, 77 and 16, respectively).

The deductive method provides the warrant for the assertion of mathematical
knowledge. The grounds for claiming that mathematics (and logic) provide absolutely
certain knowledge, that is truth, are therefore as follows. First of all, the basic statements
used in proofs are taken to be true. Mathematical axioms are assumed to be true, for
the purposes of developing that system under consideration, mathematical definitions
are true by fiat, and logical axioms are accepted as true. Secondly, the logical rules of

7
The Philosophy of Mathematics Education

inference preserve truth, that is they allow nothing but truths to be deduced from
truths. On the basis of these two facts, every statement in a deductive proof, including
its conclusion, is true. Thus, since mathematical theorems are all established by means of
deductive proofs, they are all certain truths. This constitutes the basis of the claim of
many philosophers that mathematical truths are certain truths.
This absolutist view of mathematical knowledge is based on two types of
assumptions: those of mathematics, concerning the assumption of axioms and
definitions, and those of logic concerning the assumption of axioms, rules of
inference and the formal language and its syntax. These are local or micro-
assumptions. There is also the possibility of global or macro-assumptions, such as
whether logical deduction suffices to establish all mathematical truths. I shall
subsequently argue that each of these assumptions weakens the claim of certainty for
mathematical knowledge.
The absolutist view of mathematical knowledge encountered problems at the
beginning of the twentieth century when a number of antinomies and
contradictions were derived in mathematics (Kline, 1980; Kneebone, 1963; Wilder,
1965). In a series of publications Gottlob Frege (1879, 1893) established by far the
most rigorous formulation of mathematical logic known to that time, as a foundation
for mathematical knowledge. Russell (1902), however, was able to show that Frege’s
system was inconsistent. The problem lay in Frege’s Fifth Basic Law, which allows a
set to be created from the extension of any concept, and for concepts or properties
to be applied to this set (Furth, 1964). Russell produced his well-known paradox by
defining the property of ‘not being an element of itself. Frege’s law allows the
extension of this property to be regarded as a set. But then this set is an element of
itself if, and only if, it is not; a contradiction. Frege’s Law could not be dropped
without seriously weakening his system, and yet it could not be retained.
Other contradictions also emerged in the theory of sets and the theory of
functions. Such findings have, of course, grave implications for the absolutist view of
mathematical knowledge. For if mathematics is certain, and all its theorems are
certain, how can contradictions (i.e., falsehoods) be among its theorems? Since there
was no mistake about the appearance of these contradictions, something must be
wrong in the foundations of mathematics. The outcome of this crisis was the
development of a number of schools in the philosophy of mathematics whose aims
were to account for the nature of mathematical knowledge and to re-establish its
certainty. The three major schools are known as logicism, formalism and
constructivism (incorporating intuitionism). The tenets of these schools of thought
were not fully developed until the twentieth century, but Korner (1960) shows that
their philosophical roots can be traced back at least as far as Leibniz and Kant.

A. Logicism

Logicism is the school of thought that regards pure mathematics as a part of logic. The
major proponents of this view are G.Leibniz, G.Frege (1893), B.Russell (1919),

8
 A Critique of Absolutist Philosophies

A.N.Whitehead and R.Carnap (1931). At the hands of Bertrand Russell the claims of
logicism received the clearest and most explicit formulation. There are two claims:
1 All the concepts of mathematics can ultimately be reduced to logical
concepts, provided that these are taken to include the concepts of set theory
or some system of similar power, such as Russell’s Theory of Types.
2 All mathematical truths can be proved from the axioms and rules of inference
of logic alone.
The purpose of these claims is clear. If all of mathematics can be expressed in purely
logical terms and proved from logical principles alone, then the certainty of
mathematical knowledge can be reduced to that of logic. Logic was considered to
provide a certain foundation for truth, apart from over-ambitious attempts to extend
logic, such as Frege’s Fifth Law. Thus if carried through, the logicist programme
would provide certain logical foundations for mathematical knowledge,
reestablishing absolute certainty in mathematics.
Whitehead and Russell (1910–13) were able to establish the first of the two claims
by means of chains of definitions. However logicism foundered on the second claim.
Mathematics requires non-logical axioms such as the Axiom of Infinity (the set of all
natural numbers is infinite) and the Axiom of Choice (the Cartesian product of a
family of non-empty sets is itself non-empty). Russell expressed it himself as follows.
But although all logical (or mathematical) propositions can be expressed
wholly in terms of logical constants together with variables, it is not the
case that, conversely, all propositions that can be expressed in this way are
logical. We have found so far a necessary but not a sufficient criterion of
mathematical propositions. We have sufficiently defined the character of the
primitive ideas in terms of which all the ideas of mathematics can be defined,
but not of the primitive propositions from which all the propositions of
mathematics can be deduced. This is a more difficult matter, as to which it is
not yet known what the full answer is.
We may take the axiom of infinity as an example of a proposition which,
though it can be enunciated in logical terms, cannot be asserted by logic to
be true.
(Russell, 1919, pages 202–3, original emphasis)
Thus not all mathematical theorems and hence not all the truths of mathematics can be
derived from the axioms of logic alone. This means that the axioms of mathematics are
not eliminable in favour of those of logic. Mathematical theorems depend on an
irreducible set of mathematical assumptions. Indeed, a number of important mathematical
axioms are independent, and either they or their negation can be adopted, without
inconsistency (Cohen, 1966). Thus the second claim of logicism is refuted
To overcome this problem Russell retreated to a weaker version of logicism called
‘if-thenism’, which claims that pure mathematics consists of implication statements of
the form ‘A T’. According to this view, as before, mathematical truths are
established as theorems by logical proofs. Each of these theorems (T) becomes the

9
The Philosophy of Mathematics Education

consequent in an implication statement. The conjunction of mathematical axioms

(A) used in the proof are incorporated into the implication statement as its
antecedent (see Carnap, 1931). Thus all the mathematical assumptions (A) on which
the theorem (T) depends are now incorporated into the new form of the theorem
(A T), obviating the need for mathematical axioms.
This artifice amounts to an admission that mathematics is an hypotheticodeductive
system, in which the consequences of assumed axiom sets are explored, without
asserting their necessary truth. Unfortunately, this device also leads to failure, because
not all mathematical truths, such as ‘Peano arithmetic is consistent,’ can be expressed in
this way as implication statements, as Machover (1983) argues.
A second objection, which holds irrespective of the validity of the two logicist
claims, constitutes the major grounds for the rejection of formalism. This is Godel’s
Incompleteness Theorem, which establishes that deductive proof is insufficient for
demonstrating all mathematical truths. Hence the successful reduction of
mathematical axioms to those of logic would still not suffice for the derivation of all
mathematical truths.
A third possible objection concerns the certainty and reliability of the underlying
logic. This depends on unexamined and, as will be argued, unjustified assumptions.
Thus the logicist programme of reducing the certainty of mathematical
knowledge to that of logic failed in principle. Logic does not provide a certain
foundation for mathematical knowledge.

B. Formalism

In popular terms, formalism is the view that mathematics is a meaningless formal

game played with marks on paper, following rules. Traces of a formalist philosophy of
mathematics can be found in the writings of Bishop Berkeley, but the major
proponents of formalism are David Hilbert (1925), early J.von Neumann (1931) and
H.Curry (1951). Hilbert’s formalist programme aimed to translate mathematics into
uninterpreted formal systems. By means of a restricted but meaningful meta-
mathematics the formal systems were to be shown to be adequate for mathematics,
by deriving formal counterparts of all mathematical truths, and to be safe for
mathematics, through consistency proofs.
The formalist thesis comprises two claims.

1 Pure mathematics can be expressed as uninterpreted formal systems, in which

the truths of mathematics are represented by formal theorems.
2 The safety of these formal systems can be demonstrated in terms of their
freedom from inconsistency, by means of meta-mathematics.

Kurt Godel’s Incompleteness Theorems (Godel, 1931) showed that the programme
could not be fulfilled. His first theorem showed that not even all the truths of
arithmetic can be derived from Peano’s Axioms (or any larger recursive axiom set).

10
 A Critique of Absolutist Philosophies

This proof-theoretic result has since been exemplified in mathematics by Paris and
Harrington, whose version of Ramsey’s Theorem is true but not provable in Peano
Arithmetic (Barwise, 1977). The second Incompleteness Theorem showed that in the
desired cases consistency proofs require a meta-mathematics more powerful than the
system to be safeguarded, which is thus no safeguard at all. For example, to prove the
consistency of Peano Arithmetic requires all the axioms of that system and further
assumptions, such as the principle of transfinite induction over countable ordinals
(Gentzen, 1936).
The formalist programme, had it been successful, would have provided support
for an absolutist view of mathematical truth. For formal proof, based in consistent
formal mathematical systems, would have provided a touchstone for mathematical
truth. However, it can be seen that both the claims of formalism have been refuted.
Not all the truths of mathematics can be represented as theorems in formal systems,
and furthermore, the systems themselves cannot be guaranteed safe.

C. Constructivism

The constructivist strand in the philosophy of mathematics can be traced back at

least as far as Kant and Kronecker (Korner, 1960). The constructivist programme is
one of reconstructing mathematical knowledge (and reforming mathematical
practice) in order to safeguard it from loss of meaning, and from contradiction. To
this end, constructivists reject non-constructive arguments such as Cantor’s proof
that the Real numbers are uncountable, and the logical Law of the Excluded Middle.
The best known constructivists are the intuitionists L.E.J.Brouwer (1913) and A.
Heyting (1931, 1956). More recently the mathematician E.Bishop (1967) has carried
the constructivist programme a long way, by reconstructing a substantial portion of
Analysis, by constructive means. Various forms of constructivism still flourish today,
such as in the work of the philosophical intuitionist M.Dummett (1973, 1977).
Constructivism includes a whole range of different views, from the ultra-intuitionists
(A.Yessenin-Volpin), via what may be termed strict philosophical intuitionists
(L.E.J.Brouwer), middle-of-the-road intuitionists (A.Heyting and early H.Weyl),
modern logical intuitionists (A.Troelstra) to a range of more or less liberal
constructivists including P.Lorenzen, E.Bishop, G.Kreisel and P.Martin-Lof.
These mathematicians share the view that classical mathematics may be unsafe,
and that it needs to be rebuilt by ‘constructive’ methods and reasoning.
Constructivists claim that both mathematical truths and the existence of
mathematical objects must be established by constructive methods. This means that
mathematical constructions are needed to establish truth or existence, as opposed to
methods relying on proof by contradiction. For constructivists knowledge must be
established through constructive proofs, based on restricted constructivist logic, and
the meaning of mathematical terms/objects consists of the formal procedures by
which they are constructed.
Although some constructivists maintain that mathematics is the study of

11
The Philosophy of Mathematics Education

constructive processes performed with pencil and paper, the stricter view of the
intuitionists, led by Brouwer, is that mathematics takes place primarily in the mind,
and that written mathematics is secondary. One consequence of that is that Brouwer
regards all axiomatizations of intuitionistic logic to be incomplete. Reflection can
always uncover further intuitively true axioms of intuitionistic logic, and so it can
never be regarded as being in final form.
Intuitionism represents the most fully formulated constructivist philosophy of
mathematics. Two separable claims of intuitionism can be distinguished, which
Dummett terms the positive and the negative theses.

The positive one is to the effect that the intuitionistic way of construing
mathematical notions and logical operations is a coherent and legitimate
one, that intuitionistic mathematics forms an intelligible body of theory.
The negative thesis is to the effect that the classical way of construing
mathematical notions and logical operations is incoherent and illegitimate,
that classical mathematics, while containing, in distorted form, much of
value, is, nevertheless, as it stands unintelligible.
(Dummett, 1977, page 360).

In restricted areas where there are both classical and constructivist proofs of a result,
the latter is often preferable as more informative. Whereas a classical existence proof
may merely demonstrate the logical necessity of existence, a constructive existence
proof shows how to construct the mathematical object whose existence is asserted.
This lends strength to the positive thesis, from a mathematical point of view.
However, the negative thesis is much more problematic, since it not only fails to
account for the substantial body of non-constructive classical mathematics, but also
denies its validity. The constructivists have not demonstrated that there are
inescapable problems facing classical mathematics nor that it is incoherent and
invalid. Indeed both pure and applied classical mathematics have gone from strength
to strength since the constructivist programme was proposed. Therefore, the negative
thesis of intuitionism is rejected.
Another problem for the constructivist view, is that some of its results are
inconsistent with classical mathematics. Thus, for example, the real number
continuum, as defined by the intuitionists, is countable. This contradicts the classical
result not because there is an inherent contradiction, but because the definition of
real numbers is different. Constructivist notions often have a different meaning from
the corresponding classical notions.
From an epistemological perspective, both the positive and negative theses of
intuitionism are flawed. The intuitionists claim to provide a certain foundation for their
version of mathematical truth by deriving it (mentally) from intuitively certain axioms,
using intuitively safe methods of proof. This view bases mathematical knowledge
exclusively on subjective belief. But absolute truth (which the intuitionists claim to
provide) cannot be based on subjective belief alone. Nor is there any guarantee that
different intuitionists’ intuitions of basic truth will coincide, as indeed they do not.

12
 A Critique of Absolutist Philosophies

Intuitionism sacrificed large parts of mathematics in exchange for the

soothing reassurance that what remained was justified by our ‘primordial
intuition’ (Urintuition). But intuition is subjective, and not intersubjective
enough to prevent intuitionists from differing about what their ‘primordial
intuitions’ should enshirine as the basis of mathematics.
(Kalmar, 1967, page 190).

Thus the positive thesis of intuitionism does not provide a certain foundation for
even a subset of mathematical knowledge. This criticism extends to other forms of
constructivism which also claim to base constructive mathematical truth on a
foundation of self-evident constructivist assumptions.
The negative thesis of intuitionism (and of constructivism, when it is embraced),
leads to the unwarranted rejection of accepted mathematical knowledge, on the
grounds that it is unintelligible. But classical mathematics is intelligible. It differs from
constructivist mathematics largely in the assumptions on which it is based.4 Thus
constructivism is guilty of what is analogous to a Type I Error in statistics, namely the
rejection of valid knowledge.

5. The Fallacy of Absolutism

We have seen that a number of absolutist philosophies of mathematics have failed to

establish the logical necessity of mathematical knowledge. Each of the three schools
of thought logicism, formalism and intuitionism (the most clearly enunciated form of
constructivism) attempts to provide a firm foundation for mathematical truth, by
deriving it by mathematical proof from a restricted but secure realm of truth. In each
case there is the laying down of a secure base of would-be absolute truth. For the
logicists, formalists and intuitionists this consists of the axioms of logic, the intuitively
certain principles of meta-mathematics, and the self-evident axioms of ‘primordial
intuition’, respectively. Each of these sets of axioms or principles is assumed without
demonstration. Therefore each remains open to challenge, and thus to doubt.
Subsequently each of the schools employs deductive logic to demonstrate the truth
of the theorems of mathematics from their assumed bases. Consequently these three
schools of thought fail to establish the absolute certainty of mathematical truth. For
deductive logic only transmits truth, it does not inject it, and the conclusion of a
logical proof is at best as certain as the weakest premise.
It can be remarked that all three schools’ attempts also fail to provide a foundation
for the full range of would-be mathematical truths by these means. For as Godel’s
first Incompleteness theorem shows, proof is not adequate to demonstrate all truths.
Thus there are truths of mathematics not captured by the systems of these schools.
The fact that three schools of thought in the philosophy of mathematics have
failed to establish the certainty of mathematical knowledge does not settle the
general issue. It is still possible for other grounds to be found for asserting the
certainty of mathematical truth. Absolute truth in mathematics still remains a

13
Introduction

mathematics also have a powerful impact on the way mathematics is taught (Davis,
1967; Cooney, 1988; Ernest, 1988b, 1989c). One influential study concluded:

The observed consistency between the teachers’ professed conceptions of

mathematics and the way they typically presented the content strongly
suggests that the teachers’ views, beliefs and preferences about mathematics
do influence their instructional practice.
Thompson (1984, page 125)
Such issues are central to the philosophy of mathematics education, and have
important practical outcomes for the teaching and learning of mathematics.

3. This book

The first part of the book treats the philosophy of mathematics. It contains both a
critique of existing approaches, and a new philosophy of mathematics. For although
the traditional paradigm is under attack, the novel and promising ideas in the Zeitgeist
have not yet been synthesized. Social constructivism is offered to fill this vacuum.
The second part explores the philosophy of mathematics education. It shows that
many aspects of mathematics education rest on underlying philosophical assumptions.
By uncovering some of them, the aim is to put a critical tool into the hands of
teachers and researchers.

Notes

1 A systematic ambiguity should be signalled. The philosophy of mathematics is the overall field of
philosophical inquiry into the nature of mathematics. In contrast, a philosophy of mathematics is a
particular account or view of the nature of mathematics. In general, these meanings are signalled by
the use of the definite or indefinite article (or the plural form), respectively.
2 It should be mentioned that a more negative attitude to mathematics was associated with view (B)
of the SMP students.

xiv
 A Critique of Absolutist Philosophies

postulates, definitions, and logical reasoning) to replace the absolute truth

point of view.
(Stabler, 1955, page 24).
What we have called pure mathematics is, therefore a hypothetico-deductive
system. Its axioms serve as hypotheses or assumptions, which are entertained
or considered for the propositions they imply.
(Cohen and Nagel, 1963, page 133).
[W]e can only describe arithmetic, namely, find its rules, not give a basis for
them. Such a basis could not satisfy us, for the very reason that it must end
sometime and then refer to something which can no longer be founded.
Only the convention is the ultimate. Anything that looks like a foundation
is, strictly speaking, already adulterated and must not satisfy us.
(Waismann, 1951, page 122).
Statements or propositions or theories may be formulated in assertions
which may be true and their truth may be reduced, by way of derivations
to that of primitive propositions. The attempt to establish (rather than
reduce) by these means their truth leads to an infinite regress.
(Popper, 1979, extract from table on page 124).

The above criticism is decisive for the absolutist view of mathematics. However, it is
possible to accept the criticism without adopting a fallibilist philosophy of
mathematics. For it is possible to accept a form of hypothetico-deductivism which
denies the corrigibility and the possibility of deep-seated error in mathematics. Such
a position views axioms simply as hypotheses from which the theorems of
mathematics are logically deduced, and relative to which the theorems are certain. In
other words, although the axioms of mathematics are tentative, logic and the use of
logic to derive theorems from the axioms guarantee a secure development of
mathematics, albeit from an assumed basis. This weakened form of the absolutist
position resembles Russell’s ‘if thenism’ in its strategy of adopting axioms without
either proof or cost to the system’s security. However this weakened absolutist
position is based on assumptions which leave it open to a fallibilist critique.

6. The Fallibilist Critique of Absolutism

The central argument against the absolutist view of mathematical knowledge can be
circumvented by a hypothetico-deductive approach. However, beyond the problem
of the assumed truth of the axioms, the absolutist view suffers from further major
weaknesses.
The first of these concerns the underlying logic on which mathematical proof
rests. The establishment of mathematical truths, that is the deduction of theorems
from a set of axioms, requires further assumptions, namely the axioms and rules of
inference of logic itself. These are non-trivial and non-eliminable assumptions, and

15
The Philosophy of Mathematics Education

the above argument (the ultimate irreducibility of assumptions on pain of a vicious

circle) applies equally to logic. Thus mathematical truth depends on essential logical
as well as mathematical assumptions.
It is not possible to simply append all the assumptions of logic to the set of
mathematical assumptions, following the ‘if-thenist’ hypothetico-deductive strategy.
For logic provides the canons of correct inference with which the theorems of
mathematics are derived. Loading all logical and mathematical assumptions into the
‘hypothetico’ part leaves no basis for the ‘deductive’ part of the method. Deduction
concerns ‘correct inference’, and this in turn is based on the notion of truth (the
preservation of truth value). But what then is the foundation of logical truth? It
cannot rest on proof, on pain of a vicious circle, so it must be assumed. But any
assumption without a firm foundation, whether it be derived through intuition,
convention, meaning or whatever, is fallible.
In summary, mathematical truth and proof rest on deduction and logic. But logic
itself lacks certain foundations. It too rests on irreducible assumptions. Thus the
dependence on logical deduction increases the set of assumptions on which
mathematical truth rests, and these cannot be neutralized by the ‘if-thenist’ strategy.
A further presumption of the absolutist view is that mathematics is fundamentally
free from error. For inconsistency and absolutism are clearly incompatible. But this
cannot be demonstrated. Mathematics consists of theories (e.g., group theory,
category theory) which are studied within mathematical systems, based on sets of
assumptions (axioms). To establish that mathematical systems are safe (i.e., consistent),
for any but the simplest systems we are forced to expand the set of assumptions of
the system (Godel’s Second Incompleteness Theorem, 1931). We have therefore to
assume the consistency of a stronger system to demonstrate that of a weaker. We
cannot therefore know that any but the most trivial mathematical systems are secure,
and the possibility of error and inconsistency must always remain. Belief in the safety
of mathematics must be based either on empirical grounds (no contradictions have
yet been found in our current mathematical systems) or on faith, neither providing
the certain basis that absolutism requires.
Beyond this criticism, there are further problems attendant on the use of proof as
a basis for certainty in mathematics. Nothing but a fully formal deductive proof can
serve as a warrant for certainty in mathematics. But such proofs scarcely exist. Thus
absolutism requires the recasting of informal mathematics into formal deductive
systems, which introduces further assumptions. Each of the following assumptions is
a necessary condition for such certainty in mathematics. Each, it is argued, is an
unwarranted absolutist assumption.

Assumption A
The proofs that mathematicians publish as warrants for asserting theorems can, in
principle, be translated into fully rigorous formal proofs.
The informal proofs that mathematicians publish are commonly flawed, and are by
no means wholly reliable (Davis, 1972). Translating them into fully rigorous formal

16
 A Critique of Absolutist Philosophies

proofs is a major, non-mechanical task. It requires human ingenuity to bridge gaps and
to remedy errors. Since the total formalization of mathematics is unlikely to be carried
out, what is the value of the claim that informal proofs can be translated into formal
proofs ‘in principle’? It is an unfulfilled promise, rather than grounds for certainty. Total
rigor is an unattained ideal and not a practical reality. Therefore certainty cannot be
claimed for mathematical proofs, even if the preceding criticisms are discounted.

Assumption B

Rigorous formal proofs can be checked for correctness.

There are now humanly uncheckable informal proofs, such as the Appel-Haken
(1978) proof of the four colour theorem (Tymoczko, 1979). Translated into fully
rigorous formal proofs these will be much longer. If these cannot possibly be
surveyed by a mathematician, on what grounds can they be regarded as absolutely
correct? If such proofs are checked by a computer what guarantees can be given that
the software and hardware are designed absolutely flawlessly, and that the software
runs perfectly in practice? Given the complexity of hardware and software it seems
implausible that these can be checked by a single person. Furthermore, such checks
involve an empirical element (i.e., does it run according to design?). If the checking
of formal proofs cannot be carried out, or has an empirical element, then any claim
of absolute certainty must be relinquished (Tymoczko, 1979).

Assumption C
Mathematical theories can be validly translated into formal axiom sets.
The formalization of intuitive mathematical theories in the past hundred years
(e.g., mathematical logic, number theory, set theory, analysis) has led to unanticipated
deep problems, as the concepts and proofs come under ever more piercing scrutiny,
during attempts to explicate and reconstruct them. The satisfactory formalization of
the rest of mathematics cannot be assumed to be unproblematic. Until this
formalization is carried out it is not possible to assert with certainty that it can be
carried out validly. But until mathematics is formalized, its rigour, which is a
necessary condition for certainty, falls far short of the ideal.

Assumption D

The consistency of these representations (in assumption C) can be checked.

As we know from Godel’s Incompleteness Theorem, this adds significantly to the
burden of assumptions underpinning mathematical knowledge. Thus there are no
absolute guarantees of safety.

Each of these four assumptions indicates where further problems in establishing

certainty of mathematical knowledge may arise. These are not problems concerning

17
The Philosophy of Mathematics Education

the assumed truth of the basis of mathematical knowledge (i.e., the basic
assumptions). Rather these are problems in trying to transmit the assumed truth of
these assumptions to the rest of mathematical knowledge by means of deductive
proof, and in establishing the reliability of the method.

7. The Fallibilist View

The absolutist view of mathematical knowledge has been subject to a severe, and in my
view, irrefutable criticism.6 Its rejection leads to the acceptance of the opposing fallibilist
view of mathematical knowledge. This is the view that mathematical truth is fallible and
corrigible, and can never be regarded as beyond revision and correction. The fallibilist
thesis thus has two equivalent forms, one positive and one negative. The negative form
concerns the rejection of absolutism: mathematical knowledge is not absolute truth, and
does not have absolute validity. The positive form is that mathematical knowledge is
corrigible and perpetually open to revision. In this section I wish to demonstrate that
support for the fallibilist viewpoint, in one form or the other, is much broader than might
have been supposed. The following is a selection from the range of logicians,
mathematicians and philosophers who support this viewpoint:
In his paper ‘A renaissance of empiricism in the philosophy of mathematics’,
Lakatos quotes from the later works of Russell, Fraenkel, Carnap, Weyl, von
Neumann, Bernays, Church, Godel, Quine, Rosser, Curry, Mostowski and Kalmar (a
list that includes many of the key logicians of the twentieth century) to demonstrate
their common view concerning ‘the impossibility of complete certainty’ in
mathematics, and in many cases, their agreement that mathematical knowledge has
an empirical basis, entailing the rejection of absolutism. (Lakatos, 1978, page 25,
quotation from R.Carnap)

It is now apparent that the concept of a universally accepted, infallible body

of reasoning—the majestic mathematics of 1800 and the pride of man—is
a grand illusion. Uncertainty and doubt concerning the future of
mathematics have replaced the certainties and complacency of the
past…The present state of mathematics is a mockery of the hitherto deep-
rooted and widely reputed truth and logical perfection of mathematics.
(Kline, 1980, page 6)

There are no authoritative sources of knowledge, and no ‘source’ is

particularly reliable. Everything is welcome as a source of inspiration,
including ‘intuition’… But nothing is secure, and we are all fallible.
(Popper, 1979, page 134)
I should like to say that where surveyability is not present, i.e., where there
is room for a doubt whether what we have really is the result of this
substitution, the proof is destroyed. And not in some silly and unimportant
way that has nothing to do with the nature of proof.

18
 A Critique of Absolutist Philosophies

Or: logic as the foundation of mathematics does not work, and to show
this it is enough that the cogency of logical proof stands and falls with its
geometrical cogency….
The logical certainty of proofs—I want to say—does not extend beyond
their geometrical certainty.
(Wittgenstein, 1978, pages 174–5)
A Euclidean theory may be claimed to be true; a quasi-empirical theory—
at best—to be well-corroborated, but always conjectural. Also, in a
Euclidean theory the true basic statements at the ‘top’ of the deductive
system (usually called ‘axioms’) prove, as it were, the rest of the system; in a
quasi-empirical theory the (true) basic statements are explained by the rest
of the system…Mathematics is quasi-empirical
(Lakatos, 1978, pages 28–29 & 30)
Tautologies are necessarily true, but mathematics is not. We cannot tell
whether the axioms of arithmetic are consistent; and if they are not, any
particular theorem of arithmetic may be false. Therefore these theorems are
not tautologies. They are and must always remain tentative, while a
tautology is an incontrovertible truism…
[T]he mathematician feels compelled to accept mathematics as true, even
though he is today deprived of the belief in its logical necessity and
doomed to admit forever the conceivable possibility that its whole fabric
may suddenly collapse by revealing a decisive self-contradiction.
(Polanyi, 1958, pages 187 and 189)
The doctrine that mathematical knowledge is a priori mathematical apriorism
has been articulated many different ways during the course of reflection
about mathematics…I shall offer a picture of mathematical knowledge
which rejects mathematical apriorism…the alternative to mathematical
apriorism—mathematical empiricism—has never been given a detailed
articulation. I shall try to give the missing account.
(Kitcher, 1984, pages 3–4)
[Mathematical knowledge resembles empirical knowledge—that is, the criterion
of truth in mathematics just as much as in physics is success of our ideas in
practice, and that mathematical knowledge is corrigible and not absolute.
(Putnam, 1975, page 51)
It is reasonable to propose a new task for mathematical philosophy: not to
seek indubitable truth but to give an account of mathematical knowledge as
it really is—fallible, corrigible, tentative and evolving, as is every other kind
of human knowledge.
(Hersh, 1979, page 43)
Why not honestly admit mathematical fallibility, and try to defend the
dignity of fallible knowledge from cynical scepticism, rather than delude

19
The Philosophy of Mathematics Education

ourselves that we shall be able to mend invisibly the latest tear in the fabric
of our ‘ultimate’ intuitions.
(Lakatos, 1962, page 184)

8. Conclusion

The rejection of absolutism should not be seen as a banishment of mathematics from

the Garden of Eden, the realm of certainty and truth. The ‘loss of certainty’ (Kline,
1980) does not represent a loss of knowledge.
There is an illuminating analogy with developments in modern physics. General
Relativity Theory requires relinquishing absolute, universal frames of reference in
favour of a relativistic perspective. In Quantum Theory, Heisenberg’s Uncertainty
Principle means that the notions of precisely determined measurements of position
and momentum for particles also has had to be given up. But what we see here are
not the loss of knowledge of absolute frames and certainty. Rather we see the growth
of knowledge, bringing with it a realization of the limits of what can be known.
Relativity and Uncertainty in physics represent major advances in knowledge,
advances which take us to the limits of knowledge (for so long as the theories are
retained).
Likewise in mathematics, as our knowledge has become better founded and we
learn more about its basis, we have come to realize that the absolutist view is an
idealization, a myth. This represents an advance in knowledge, not a retreat from past
certainty. The absolutist Garden of Eden was nothing but a fool’s paradise.

Notes

1 In this chapter, for simplicity, the definition of truth in mathematics is assumed to be unproblematic
and unambiguous. Whilst justified as a simplifying assumption, since none of the arguments of the
chapter hinge on the ambiguity of this notion, the meaning of the concept of truth in mathematics
has changed over time. We can distinguish between three truth-related concepts used in
mathematics:
(a) There is the traditional view of mathematical truth, namely that a mathematical truth is a
general statement which not only correctly describes all its instances in the world (as does a
true empirical generalisation), but is necessarily true of its instances. Implicit in this view is the
assumption that mathematical theories have an intended interpretation, namely some
idealization of the world.
(b) There is the modern view of the truth of a mathematical statement relative to a background
mathematical theory: the statement is satisfied by some interpretation or model of the theory.
According to this (and the following) view, mathematics is open to multiple interpretations,
i.e., possible worlds. Truth consists merely in being true (i.e., satisfied, following Tarski, 1936) in
one of these possible worlds.
(c) There is the modern view of the logical truth or validity of a mathematical statement relative
to a background theory: the statement is satisfied by all interpretations or models of the theory.
Thus the statement is true in all of these possible worlds.
Truth in sense (c) can be established by deduction from the background theory as an axiom set. For
a given theory, truths in a sense (c) are a subset (usually a proper subset) of truths in sense (b).

20
 A Critique of Absolutist Philosophies

Incompleteness arises (as Godel, 1931, proved) in most mathematical theories as there are sentences
true in sense (b) (i.e., satisfiable) which are not true in sense (c).
Thus not only does the concept of truth have multiple meanings, but crucial mathematical
issues hinge upon this ambiguity. Beyond this, the modern mathematical view of truth differs
from the traditional mathematical view of truth (a), and the everyday sense of the term, which
resembles it. For in a naive sense truths are statements which accurately describe a state of
affairs—a relationship—in some realm of discourse. In this view, the terms which express the
truth name objects in the realm of discourse, and the statement as a whole describes a true state
of affairs, the relationship that holds between the denotations of the terms. This shows that the
concept of truth employed in mathematics no longer has the same meaning as either the
everyday, naive notion of truth, or its equivalent (a) as was used in mathematics, in the past
(Richards, 1980, 1989).
The consequence of this is that the traditional problem of establishing the indubitable
foundations of mathematical truth has changed, as the definition of truth employed has changed. In
particular to claim that a statement is true in sense (b) is much weaker than senses (a) or (c). ‘1+1=1’
is true in sense (b)(it is satisfied in Boolean algebra, but not in sense (a) which assumes the standard
Peano interpretation).
2 For the proof to be rigorous, a formal language L for Peano Arithmetic should be specified in full.
L is a first-order predicate calculus in universally quantified free variable form. The syntax of L will
specify as usual the terms and formulas of L, the formula ‘P(r)’ in the term ‘r’, and the result ‘P(t)’
of substituting the term ‘t’ for the occurrences of Y in ‘P(r)’ (sometimes written P(r)[r/t]). It should
also be mentioned that a modernised form of the Peano Axioms is adopted above (see, for example,
Bell and Machover, 1977), which is not literally that of Peano (Heijenoort, 1967).
3 Scholars believe that Euclid’s fifth postulate was not considered to be as self-evident as the others. It
is less terse, and more like a proposition (a theorem) than a postulate (it is the converse of
proposition I 17). Euclid does not use it until proposition I 29. For this reason, over the ages, many
attempts to prove the posulate were made including Sacchieri’s attempt to prove it by reductio ad
absurdam based on its denial (Eves, 1953).
4 It is worth remarking that the classical predicate calculus is translatable into intuitionist logic in a
constructive way that preserves deducibility (see Bell and Machover, 1977). This means that all the
theorems of classical mathematics expressible in the predicate calculus can be represented as
intuitionistic theorems. Thus classical mathematics cannot easily be claimed to be intuitionistically
unintelligible. (Note that the reverse translation procedure is intuitionistically unacceptable, since it
replaces ‘-P’ by ‘P’, and ‘-(x)-P’ by ‘(Ex)P’, reading-, (x), and (Ex) as ’not’, ‘for all x’ and ‘there exists
x’, respectively).
5 Some readers may feel that assertion requires justification. What valid warrant can there be for
mathematical knowledge other than demonstration or proof? Clearly it is necessary to find other
grounds for asserting that mathematical statements are true. The principal accounts of truth are the
correspondence theory of truth, the coherence theory of truth (Woozley, 1949), the pragmatic
theory of truth (Dewey, 1938) and truth as convention (Quine, 1936; Quinton, 1963). We can first
dismiss the coherence and pragmatic theories of truth as irrelevant here, since these do not claim
that truth can be warranted absolutely. The correspondence theory can be interpreted either
empirically or non-empirically, to say that basic mathematical truths describe true states of affairs
either in the world or in some abstract realm. But then the truths of mathematics are justified
empirically or intuitively, respectively, and neither grounds serve as warrants for certain
knowledge.
The conventional theory of truth asserts that basic mathematical statements are true by virtue of
the meanings of the terms therein. But the fact that the axioms express what we want or believe
terms to mean does not absolve us from having to assume them, even if we simply stipulate them
by fiat. Rather it is an admission that we simply have to assume certain basic propositions. Beyond
this, to say that complex axioms such as those of Zermelo-Fraenkel set theory are true by virtue of
the meanings of the constituent terms is not supportable. (Maddy, 1984, gives an account of set-
theoretic axioms in current use which by no stretch of the imagination are considered true). We
must regard these axioms as implicit definitions of their constituent terms, and it is evident we must
assume the axioms to proceed with set theory.

21
The Philosophy of Mathematics Education

6 The critique of absolutism can be used to criticize this chapter, as follows. If no knowledge,
including mathematics is certain how can the modestly founded assertions of this chapter be true?
Is not the assertion that there is no truth self-defeating?
The answer is that the assertions and arguments of this chapter do not pretend to be the truth,
but a plausible account. The grounds for accepting the truths of mathematics, imperfect as they are,
are far firmer than the arguments of the chapter. (The argument can be defended analogously to the
way Ayer, 1946, defends the Principle of Verification.)

22
 2

The Philosophy of Mathematics

Reconceptualized

1. The Scope of the Philosophy of Mathematics

In the previous chapter we entertained the hypothesis that mathematical knowledge

is a set of truths, in the form of a set of propositions with proofs, and that the
function of the philosophy of mathematics is to establish the certainty of this
knowledge. Having found that this hypothesis is untenable we are forced to
reconsider the nature of the philosophy of mathematics. What is the function and
scope of the philosophy of mathematics?

As the philosophy of law does not legislate, or the philosophy of science

devise or test scientific hypotheses—the philosophy of mathematics does
not add to the number of mathematical theorems and theories. It is not
mathematics. It is reflection upon mathematics, giving rise to its own
particular questions and answers.
(Korner, 1960, page 9)
The philosophy of mathematics begins when we ask for a general account
of mathematics, a synoptic vision of the discipline that reveals its essential
features and explains just how it is that human beings are able to do
mathematics.
(Tymoczko, 1986, page viii)

Priest (1973) boldly outlines the task as follows:

All the problems concerning the philosophy of mathematics can neatly be

summarized by the question:
Question 0. What is pure mathematics?…
Firstly, what is meant by ‘mathematics’? The only answer we can give
without begging the question is ‘That which is done and has been done for
the last four thousand years by mathematicians.’…Knowledge of the nature
of mathematics lies in an ability to do it.
[To answer question 0 we need to answer the following:]

23
The Philosophy of Mathematics Education

Question 1. Why are the truths of mathematics true?

Any reasonable answer must also permit reasonable answers to the following
questions
Question 1(a). Why is it that such truths appear necessary and inviolable,
and why are we unable to conceive of them being false?
Question 1(b). How is it we come to know such truths?
Question 1(c). Why is it that the truths of mathematics can be applied in
practical matters e.g., surveying, building bridges, sending rockets to the
moon, etc. In short, why are they useful?…
Now the naive answer to question 1 is that mathematical truths are so
because they are true of certain objects such as numbers, functions,
propositions, points, groups, models etc., i.e., these are what mathematics is
about.
Hence we must be able to answer:
Question 2. What exactly are the above objects, and in what sense do
they exist?…
Question 2 (cont.). And if they don’t exist, why is it we have such a
strong impression that they do?
(Priest, 1973, pages 115–117)

According to these views, the role of the philosophy of mathematics is to reflect on,
and give an account of the nature of mathematics. The key issue concerns how ‘giving
an account of’ mathematics is conceived. Absolutist philosophies of mathematics such
as logicism, formalism and intuitionism attempt to provide prescriptive accounts of the
nature of mathematics. Such accounts, as we have seen, are programmatic, legislating
how mathematics should be understood, rather than providing accurately descriptive
accounts of the nature of mathematics.Thus they are failing to account for mathematics
as it is, in the hope of fulfilling their vision of how it should be. But ‘to confuse
description and programme—to confuse ‘is’ with ‘ought to be’ or ‘should be’—is just
as harmful in the philosophy of mathematics as elsewhere.’
(Korner, 1960, page 12)
The inquiry can begin with the traditional questions of epistemology and ontology.
What is the nature and basis of mathematical knowledge? What is the nature of, and
how do we account for, the existence of mathematical objects (numbers, functions,
sets, etc.)?
However, the answers to these questions will not provide a descriptive account of
the nature of mathematics. For the narrow focus of these ‘internal’ questions
concerning the philosophy of mathematics fails to locate mathematics within the
broader context of human thought and history. Without such a context, according to
Lakatos, the philosophy of mathematics loses its content.
Under the present dominance of formalism (i.e., foundationism), one is
tempted to paraphrase Kant: the history of mathematics, lacking the
guidance of philosophy has become blind, while the philosophy of

24
 The Philosophy of Mathematics Reconceptualized

mathematics turning its back on the most intriguing phenomena in the

history of mathematics, has become empty.
(Lakatos, 1976, page 2)

Thus much more should fall within the scope of the philosophy of mathematics than
merely the justification of mathematical knowledge, provided through its
reconstruction by a foundationist programme. Mathematics is multi-faceted, and as
well as a body of prepositional knowledge, it can be described in terms of its
concepts, characteristics, history and practices. The philosophy of mathematics must
account for this complexity, and we also need to ask the following questions. What is
the purpose of mathematics? What is the role of human beings in mathematics? How
does the subjective knowledge of individuals become the objective knowledge of
mathematics? How has mathematical knowledge evolved? How does its history
illuminate the philosophy of mathematics? What is the relationship between
mathematics and the other areas of human knowledge and experience? Why have
the theories of pure mathematics proved to be so powerful and useful in their
applications to science and to practical problems?
These questions represent a broadening of the scope of the philosophy of
mathematics from the internal concerns of absolutism. Three issues may be selected
as being of particular importance, philosophically and educationally. Each of these
issues is expressed in terms of a dichotomy, and the absolutist and fallibilist
perspectives on the issue are contrasted. The three issues are as follows.
First of all, there is the contrast between knowledge as a finished product, largely
expressed as a body of propositions, and the activity of knowing or knowledge
getting. This latter is concerned with the genesis of knowledge, and with the
contribution of humans to its creation. As we have seen, absolutist views focus on the
former, that is finished or published knowledge, and its foundations and justification.
Absolutist views not only focus on knowledge as an objective product, they often
deny the philosophical legitimacy of considering the genesis of knowledge at all, and
consign this to psychology and the social sciences. One partial exception to this is
constructivism, which admits the knowing agent in a stylized form.
In contrast, fallibilist views of the nature of mathematics, by acknowledging the
role of error in mathematics cannot escape from considering theory replacement and
the growth of knowledge. Beyond this, such views must be concerned with the
human contexts of knowledge creation and the historical genesis of mathematics, if
they are to account adequately for mathematics, in all its fullness.
Because of the importance of the issue, it is worth adding a further and more
general argument for the necessity for considering the genesis of knowledge. This
argument is based on the reality of knowledge growth. As history illustrates,
knowledge is perpetually in a state of change in every discipline, including
mathematics. Epistemology is not accounting adequately for knowledge if it
concentrates only on a single static formulation, and ignores the dynamics of
knowledge growth. It is like reviewing a film on the basis of a detailed scrutiny of a
single key frame! Thus epistemology must concern itself with the basis of knowing,

25
The Philosophy of Mathematics Education

which underpins the dynamics of knowledge growth, as well as with the specific
body of knowledge accepted at any one time. Traditional philosophers such as Locke
and Kant admit the legitimacy and indeed the necessity of genetic considerations in
epistemology. So do an increasing number of modern philosophers, such as Dewey
(1950), Wittgenstein (1953), Ryle (1949), Lakatos (1970), Toulmin (1972), Polanyi
(1958), Kuhn (1970) and Hamlyn (1978).
Secondly, there is the distinction between mathematics as an isolated and discrete
discipline, which is strictly demarcated and separated from other realms of
knowledge, as opposed to a view of mathematics which is connected with, and
indissolubly a part of the whole fabric of human knowledge. Absolutist views of
mathematics accord it a unique status, it being (with logic) the only certain realm of
knowledge, which uniquely rests on rigorous proof. These conditions, together with
the associated internalist denial of the relevance of history or genetic or human
contexts, serve to demarcate mathematics as an isolated and discrete discipline.
Fallibilists include much more within the ambit of the philosophy of mathematics.
Since mathematics is seen as fallible, it cannot be categorically divorced from the
empirical (and hence fallible) knowledge of the physical and other sciences. Since
fallibilism attends to the genesis of mathematical knowledge as well as its product,
mathematics is seen as embedded in history and in human practice. Therefore
mathematics cannot be divorced from the humanities and the social sciences, or from
a consideration of human culture in general. Thus from a fallibilist perspective
mathematics is seen as connected with, and indissolubly a part of the whole fabric of
human knowledge.
The third distinction can be seen as a specialisation and further development of
the second. It distinguishes between views of mathematics as objective and value free,
being concerned only with its own inner logic, in contrast with mathematics seen as
an integral part of human culture, and thus as fully imbued with human values as
other realms of knowledge and endeavour. Absolutist views, with their internal
concerns, see mathematics as objective and absolutely free of moral and human
values. The fallibilist view, on the other hand, connects mathematics with the rest of
human knowledge through its historical and social origins. Hence it sees
mathematics as value-laden, imbued with moral and social values which play a
significant role in the development and applications of mathematics.
What has been proposed is that the proper concern of the philosophy of
mathematics should include external questions as to the historical origins and social
context of mathematics, in addition to the internal questions concerning knowledge,
existence, and their justification. For some years there has been a parallel debate over
an internalist-externalist dichotomy in the philosophy of science (Losee, 1987). As in
the philosophy of mathematics there has been a split between philosophers promoting
an internalist view in the philosophy of science (such as the logical empiricists and
Popper) and those espousing an externalist view. The latter include many of the most
influential recent philosophers of science, such as Feyerabend, Hanson, Kuhn, Lakatos,
Laudan and Toulmin. The contributions of these authors to the philosophy of science
is a powerful testimony to the necessity of considering ‘external’ questions in the

26
 The Philosophy of Mathematics Reconceptualized

philosophy of the sciences. However in the philosophy of science, even philosophers

espousing an internalist position, such as Popper, admit the importance of considering
the development of scientific knowledge for epistemology.

Criteria for an Adequate Philosophy of Mathematics

It has been argued that the role of the philosophy of mathematics is to account for
the nature of mathematics, where this task is conceived broadly to include ‘external’
issues such as the history, genesis and practice of mathematics, as well as ‘internal’
epistemological and ontological issues, such as the justification of mathematical
knowledge. These criteria can be stated more explicitly: a proposed philosophy of
mathematics should account for:
(i) Mathematical knowledge: its nature, justification and genesis,
(ii) The objects of mathematics: their nature and origins,
(iii) The applications of mathematics: its effectiveness in science, technology and
other realms.
(iv) Mathematical practice: the activities of mathematicians, both in the present
and the past.
It is proposed, therefore, to adopt these as adequacy criteria for any proposed
philosophy of mathematics. These criteria represent a reconceptualization of the role
of the philosophy of mathematics. However, this role, it is argued, represents the
proper task of the philosophy of mathematics, which was obscured by the mistaken
identification of the philosophy of mathematics with the study of the logical
foundations of mathematical knowledge.

2. A Further Examination of Philosophical Schools

The new criteria provide a means to assess the adequacy of schools of thought in the
philosophy of mathematics.

A. The Absolutist Schools

In the previous chapter we saw that the logicist, formalist and intuitionist schools are
absolutist. We have given an account of the failure of the programmes of these
schools, and indeed refuted in general the possibility of absolutism in the philosophy
of mathematics. On the basis of the above criteria we can further criticize these
schools for their inadequacy as philosophies of mathematics. Their task should have
included accounting for the nature of mathematics, including external social and
historical factors, such as the utility of mathematics, and its genesis. Because of their
narrow, exclusively internal preoccupations, these schools have made no contribution

27
The Philosophy of Mathematics Education

to a broadly conceived account of mathematics (with the possible exception of

intuitionism, see below). Thus, not only have they failed in their self-chosen
foundationist goals, but even had they succeeded they would remain inadequate
philosophies of mathematics, in terms of the criteria adopted. Furthermore, this
criticism is probably applicable to any absolutist philosophy with a foundationist
programme.

B. Progressive Absolutism

Although the various forms of absolutism are grouped and criticized together,
different forms of absolutism in mathematics can be distinguished. Drawing a parallel
with the philosophy of science, Confrey (1981) separates formal absolutist and
progressive absolutist philosophies of mathematics.1 The formal absolutist view of
mathematics is

the epitome of certainty, immutable truths, and irrefutable methods…

secure through the infallibility of its supreme method, deduction…
Concepts in mathematics do not develop, they are discovered…the previous
truths left unchanged by the discovery of a new truth… mathematics
proceeding by an accumulation of mathematical truths and as having an
inflexible, a priori structure.
(Confrey, 1981, pages 246–247)
This is contrasted with the progressive absolutist view of mathematics, which whilst
absolutist sees mathematics as resulting from human striving for truth, rather than its
attainment. According to this view

progress is a process of replacement of previous theories by superior

theories which account for all the previous data and more. Each progressive
theory approximates truth more and more precisely…progress consists of
discovering mathematical truths which are not consistent with a theory or
not accounted for in the theory, and then extending the theory to account
for this larger realm of mathematical phenomenon.
(Confrey, 1981, pages 247–248)

The key distinction is between static and dynamic absolutist conceptions of

mathematical knowledge and theories, with human activity contributing the
dynamic in progressive absolutism. Formalism and logicism are formal absolutist.
They accept the discovery and proof of new theorems within a formal mathematical
theory, building on its axioms. However, they do not treat the creation or change of
mathematical theories nor informal mathematics, let alone human agency. According
to such views, mathematics consists of nothing more than fixed, formal mathematical
theories.
In contrast, progressive absolutist philosophies:

28
 The Philosophy of Mathematics Reconceptualized

1 accommodate the creation and change of axiomatic theories;

2 acknowledge that more than purely formal mathematics exists, for
mathematical intuition is needed as the basis for theory creation; and hence
3 acknowledge human activity and its outcomes, in the creation of new
knowledge and theories.

Intuitionism (and constructivism, more generally) fit this description. For intuitionism
is foundationist and absolutist, seeking secure foundations for mathematical
knowledge through intuitionistic proofs and ‘ur-intuition’ (Kalmar, 1967). However,
intuitionism (1) acknowledges human mathematical activity as fundamental in the
construction of proofs or mathematical objects, the creation of new knowledge, and
(2) acknowledges that the axioms of intuitionistic mathematical theory (and logic) are
fundamentally incomplete, and need to be added to as more mathematical truth is
revealed informally or by intuition (Brouwer, 1927; Dummett, 1977).
In consequence, intuitionism, and progressive absolutist philosophies in general,
satisfy more of the adequacy criteria than formal absolutist philosophies, whilst
nevertheless remaining refuted overall. For they give some place, although restricted,
to the activities of mathematicians (criterion 4). They acknowledge human agency,
albeit in stylized form, in the domain of informal mathematics. This partial fulfilment
of the criteria deserves acknowledgment, for it means that not all absolutist
philosophies are on a par. It also turns out to be significant for education.

C. Platonism

Platonism is the view that the objects of mathematics have a real, objective existence
in some ideal realm. It originates with Plato, and can be discerned in the writings of
the logicists Frege and Russell, and includes Cantor, Bernays (1934), Hardy (1967)
and Godel (1964) among its distinguished supporters. Platonists maintain that the
objects and structures of mathematics have a real existence independent of humanity,
and that doing mathematics is the process of discovering their pre-existing
relationships. According to platonism mathematical knowledge consists of
descriptions of these objects and the relationships and structures connecting them.
Platonism evidently provides a solution to the problem of the objectivity of
mathematics. It accounts both for its truths and the existence of its objects, as well as
for the apparent autonomy of mathematics, which obeys its own inner laws and logic.
The problem of platonism, unlike that of the absolutist foundational school, is not
entirely one of failure, for it offers no foundationist programme to reconstruct and
safeguard mathematics. What is of more interest is to account for the fact that such an
implausible philosophy provides aid and comfort to successful mathematicians of the
stature of Cantor and Godel.
This interest notwithstanding, platonism suffers from two major weaknesses. First
of all, it is not able to offer an adequate account of how mathematicians gain access
to knowledge of the platonic realm. We can grant that platonism accounts for

29
The Philosophy of Mathematics Education

mathematical knowledge in the way that naive inductivist science accounted for its
knowledge. That is as being based on observations of the real world (an ideal world,
in the case of platonism), subsequently generalized. But if mathematics is the natural
history of the crystalline platonic universe, how is it that mathematicians gain
knowledge of it? It must surely be through intuition, or some such special mental
faculty, and no account of this is given. If access is through intuition, then a
reconciliation is needed between the facts that (i) different mathematicians’
intuitions vary, in keeping with the subjectivity of intuition, and (ii) platonist
intuition must be objective, and lead to agreement. Thus the platonist view is
inadequate without an account of human access to the realm of platonic objects
which overcomes these difficulties.
If, on the other hand, the platonist’s access to the world of mathematical objects
is not through intuition but through reason and logic, then further problems arise.
How does the platonist know that his or her reasoning is correct? Either another
form of intuition is needed, which allows the platonist to see which proofs correctly
describe mathematical reality, or the platonist is in the same boat as everybody else
with regard to proof. But in this second case, what is platonism but empty faith, since
it provides no insight into truth or existence?
The second flaw in the platonist account is that it is not able to offer an adequate
account of mathematics, neither internally nor externally. Internally, an important
part of mathematics is its constructive, computational side. This depends vitally on
the representation of dynamic mathematical processes, such as iteration, recursive
functions, proof theory, and so on. Platonism accounts only for the static set-theoretic
and structural aspects of mathematics. Thus it omits a central area of mathematics
from its account. Externally, platonism fails to account adequately for the utility of
mathematics, its relations with science, human activity or culture, and the genesis of
knowledge. For platonists to say that mathematics advances as it is progressively
uncovered, just as geography advanced with the voyages of the explorers, is not
enough. Nor does it suffice to say its utility stems from the fact that mathematics
describes the necessary structure of observable reality. For these explanations beg the
very questions they are meant to settle.
Since it fails on all the above counts, platonism is rejected as a philosophy of
mathematics.

D. Conventionalism

The conventionalist view of mathematics holds that mathematical knowledge and

truth are based on linguistic conventions. In particular, that the truths of logic and
mathematics are analytic, true by virtue of the meanings of the terms involved. A
moderate form of conventionalism, such as that of Quine (1936) or Hempel (1945),
uses linguistic convention as the source of basic mathematical truth on which the
edifice of mathematics is constructed. According to this view linguistic conventions
provide the basic, certain truths of mathematics and logic, and deductive logic

30
 The Philosophy of Mathematics Reconceptualized

(proofs) transmits this truth to the remainder of the body of mathematical

knowledge, thus establishing its certainty. This form of conventionalism is more or
less the same as ‘ifthenism’, discussed in Chapter 1 as a fall-back position for defeated
foundationists. This view remains absolutist, and as such is subject to the same
refutation.
The more interesting forms of conventionalism are not absolutist (and it is these
that I shall be referring to with the term ‘conventionalism’). Priest (1973) proposes
to revive conventionalism, but the best known proponent of this view is
Wittgenstein, who both laid the foundations of the moderate form by declaring the
truth of mathematics to be tautologous (Wittgenstein, 1922), before making his
extensive later contributions (Wittgenstein, 1953, 1978). Wittgenstein’s later
philosophy of mathematics is not clearly laid out because of his epigrammatic style,
in which he eschews systematic exposition, and because most of his contributions to
the philosophy of mathematics were published posthumously, in an unfinished state
(Wittgenstein, 1953, 1978).
Wittgenstein criticizes the foundationist schools, and dwells at length upon
knowing as a process in mathematics (Wittgenstein, 1953, 1978). In his
conventionalism, Wittgenstein claims that mathematics is a ‘motley’, a collection of
‘language games’, and that the notions of truth, falsity and proof depend upon our
accepting the conventional linguistic rules of these games; as the following
quotations indicate.

Of course, in one sense mathematics is a branch of knowledge—but still it

is also an activity. And ‘false moves’ can only exist as the exception. For if
what we now call by that name became the rule, the game in which they
were false moves would have been abrogated.
The word ‘agreement’ and the word ‘rule’ are related to one another,
they are cousins. If I teach anyone the use of the one word, he learns the
use of the other word with it.
‘So you are saying that human agreement decides what is true and what is
false?’—It is what human beings say that is true and false: and they agree in the
language they use. That is not agreement in opinions but in form of life.
(Wittgenstein, 1953, pages 227,86 and 88, respectively)
What is unshakably certain about what is proved? To accept a proposition as
unshakably certain—I want to say—means to use it as a grammatical rule:
this removes uncertainty from it.
(Wittgenstein, 1978, page 170)
These quotations illustrate Wittgenstein’s view that the uses of language (in various
language games or meaning contexts) involve the acceptance of rules, which are a
precondition, a sine qua non, for linguistic communication. The agreement he refers
to is the sharing of ‘a form of life’, a group socio-linguistic practice based on the
common following of rules, which is essential for any meaningful language use. Such
agreement is not merely the voluntary assent to a practice, such as with the

31
The Philosophy of Mathematics Education

conventions of bridge. Rather it is inbuilt in our communicative behaviour, which

presupposes a common underlying language usage and rule following.
Thus according to Wittgenstein’s conventionalist philosophy of mathematics the
‘truths’ of mathematics and logic depend for their acceptance on the linguistic rules
of use of terms and grammar, as well on the rules governing proofs. These underlying
rules confer certainty on the ‘truths’, for they cannot be false without breaking the
rules, which would be flying in the face of accepted use. Thus it is the linguistic rules
underlying the ‘truths’ of mathematics and logic that ensure that they cannot be
falsified.

I have not yet made the role of miscalculating clear. The role of the
proposition: ‘I must have miscalculated’. It is really the key to an
understanding of the ‘foundations’ of mathematics.
(Wittgenstein, 1978, p. 221)

What Wittgenstein is saying here is that if our results contradict the underlying rules
of use, then we reject the results, we do not question the underlying rules.
In summary, Wittgenstein proposes that the logical necessity of mathematical (and
logical) knowledge rests on linguistic conventions, embedded in our socio-linguistic
practices.2
Conventionalism might appear, on the basis of the account given, to be absolutist,
for it claims that mathematical axioms, for example, are absolutely true on the basis
of linguistic conventions. But locating the foundations of mathematical knowledge in
the rules governing natural language usage allows for the development of
mathematical knowledge, and indeed for changes in the nature of mathematical truth
and meaning, as its basis evolves. For language and its patterns of use develop all the
while organically, and its sets of conventions and rules change. This is especially true
of informal mathematical language, in which the rules governing the use of such
terms as ‘set’, ‘infinity’, ‘infinitesimal’ and ‘proof have changed dramatically in the
last hundred years, as a mathematical practice has developed. Likewise new
conventions have warranted new truths (such as Hamilton’s ‘ij =-ji’ and, in logic
‘1=2 implies 1=1’). Thus conventionalism is not absolutist, for it allows for the
dethronement and replacement of basic mathematical truths (such as ‘xy=yx’). This
form of conventionalism is therefore consistent with fallibilism.
The conventionalist philosophy of mathematics has been criticized by previous
authors on two grounds. First of all, it is claimed to be uninformative: ‘apart from
pointing out the essentially social nature of mathematics, conventionalism tells us
remarkably little’. (Machover, 1983, page 6). The force of this criticism is that to be
an adequate philosophy of mathematics, a much more elaborated version of
conventionalism is needed.
The second objection is due to Quine.

Briefly the point is that the logical truths, being infinite in number, must be
given by general conventions rather than singly; and logic is needed then to

32
 The Philosophy of Mathematics Reconceptualized

begin with, in the metatheory, in order to apply the general conventions to

individual cases’.
(Quine, 1966, page 108)

Thus according to Quine, our linguistic conventions must either include the infinite
number of truths of the form ‘(Sentence 1) and (Sentence 2) implies (Sentence 2)’,
or this single, general convention, in which case we need logic in the metalanguage
to derive all its instances.
But notice that the same objection applies to the possibility of grammatical
conventions in language. We either need to know of the infinite number of
grammatical instances of the form ‘(Subject) is a (Predicate)’, or we would need
metalinguistic rules of substitution to derive its instances from the general
grammatical convention. But we evidently do not need such additional rules to
speak, because the very scheme is a ‘production rule’. The sole function of such a
rule in a natural language is to generate instances. Likewise, logical schemes are rules
which guide the production of logical truths. Thus it is not the case that we need to
presuppose logic in a meta-language to derive instances from our logical scheme. It
is inappropriate to seek all the forms and distinctions of formal languages in natural
languages, which, for example, already differ in being their own meta-languages.
In fact, truths of the form ‘A&B implies B’ are not likely to depend on the above
sentential scheme, but on the rules governing the use of the word ‘and’. These rules
are likely to be semantic rules linking ‘and’ with ‘combine’, ‘join’, and ‘put together’,
that is with the conjunctive meaning of ‘and’. These semantic rules imply that the
consequences of ‘A&B’ are the consequnces of ‘A’ combined with those of ‘B’.
Quine’s objection is therefore dismissed in that it does not apply to natural
languages, and imposes an overly restrictive role on general conventions. On the
other hand he is right to say that we will not find all the truths of mathematics and
logic represented literally as linguistic rules and conventions.
Although Quine is critical of conventionalism in logic, he regards its potential as
a philosophy of mathematics quite differently.
For set theory the linguistic doctrine has seemed less empty; in set theory,
moreover, convention in quite the ordinary sense seems to be pretty much
what goes on. Conventionalism has a serious claim to attention in the
philosophy of mathematics, if only because of set theory.
(Quine, 1966, page 108)

Conventionalism offers the beginnings of a descriptive account of the nature of

mathematics, formulated in terms of its linguistic basis. It accommodates a fallibilist
view of mathematics, and may account for both the objectivity of mathematical
knowledge, through our necessary acceptance of linguistic rules, and for at least part
of its genesis, via the acquisition of language. Since language connects mathematics
with other areas of knowledge, conventionalism has the potential to account for the
applications of mathematics. Thus conventionalism is not refuted, and indeed may
satisfy many of the adequacy criteria proposed earlier. Conventionalism is discussed

33
The Philosophy of Mathematics Education

further in the next chapter, as one of several contributors to a proposed social

constructivist philosophy of mathematics.

E. Empiricism

The empiricist view of the nature of mathematics (‘naive empiricism’, to distinguish

it from Lakatos’ quasi-empiricism) holds that the truths of mathematics are empirical
generalisations. We can distinguish two empiricist theses: (i) the concepts of
mathematics have empirical origins, and (ii) the truths of mathematics have empirical
justification, that is, are derived from observations of the physical world.The first thesis
is unobjectionable, and is accepted by most philosophers of mathematics (given that
many concepts are not directly formed from observations but are defined in terms of
other concepts which lead, via definitional chains, to observational concepts). The
second thesis is rejected by all but empiricists, since it leads to some absurdities. The
initial objection is that most mathematical knowledge is accepted on theoretical, as
opposed to empirical grounds. Thus I know that 999,999+1=1,000,000 not through
having observed its truth in the world, but through my theoretical knowledge of
number and numeration.
Mill (1961) partly anticipates this objection, suggesting that the principles and
axioms of mathematics are induced from observations of the world, and that other
truths are derived from these by deduction. However, empiricism is open to a
number of further criticisms.
First of all, when our experience contradicts elementary mathematical truths, we do
not give them up (Davis and Hersh, 1980). Rather we assume that some error has crept
in to our reasoning, because there is shared agreement about mathematics, which
precludes the rejection of mathematical truths (Wittgenstein, 1978). Thus, ‘1+1=3’ is
necessarily false, not because one rabbit added to another does not give three rabbits, but
because by definition ‘1+1’ means ‘the successor of 1’ and ‘2’ is the successor of ‘1’.
Secondly, mathematics is largely abstract, and so many of its concepts do not have
their origins in observations of the world. Rather they are based on previously
formed concepts. Truths about such concepts, which form the bulk of mathematics,
cannot therefore be said to be induced from observations of the external world.
Finally, empiricism can be criticized for focusing almost exclusively on
foundationist issues, and failing to account adequately for the nature of mathematics.
This, as has been argued above, is the major purpose of the philosophy of
mathematics. On the basis of this criticism we can reject the naive empiricist view of
mathematics as inadequate.

3. Quasi-empiricism

Quasi-empiricism is the name given to the philosophy of mathematics developed by

Imre Lakatos (1976,1978). It is the view that mathematics is what mathematicians do

34
 The Philosophy of Mathematics Reconceptualized

and have done, with all the imperfections inherent in any human activity or creation.
Quasi-empiricism represents a ‘new direction in the philosophy of mathematics’
(Tymoczko, 1986), because of the primacy it accords to mathematical practice.
Supporters of this view include Davis (1972), Hallett (1979), Hersh (1979),
Tymoczko (1979) and at least in part, Putnam (1975). A preliminary sketch of the
quasi-empiricist view of mathematics is as follows.
Mathematics is a dialogue between people tackling mathematical problems.
Mathematicians are fallible and their products, including concepts and proofs, can
never be considered final or perfect, but may require renegotiation as standards of
rigour change, or as new challenges or meanings emerge. As a human activity,
mathematics cannot be viewed in isolation from its history and its applications in the
sciences and elsewhere. Quasi-empiricism represents ‘a renaissance of empiricism in
the recent philosophy of mathematics’ (Lakatos, 1967).

A. Exposition of Lakatos’ Quasi-empiricism

Five theses of quasi-empiricism can be identified, as follows.

1. Mathematical knowledge is fallible

In quasi-empiricism, the search for a basis for absolute certainty in mathematics is
rejected, and mathematical knowledge is acknowledged to be fallible, corrigible, and
without certain foundations. (See Lakatos quotations in Chapter 1.)

2. Mathematics is hypothetico-deductive
Mathematics is acknowledged to be a hypothetico-deductive system, like the widely
accepted conception of empirical science due to Popper (1959). As in science, the
emphasis in such a system is not on the transmission of truth from true premises to
conclusions (the absolutist view), but on the re-transmission of falsity from falsified
conclusions (‘falsifiers’) to hypothetical premises. Since axiomatic theories are
formalizations of previously existing informal mathematical theories, their potential
falsifiers are the informal theorems of the pre-existing theory, (in addition to formal
contradictions). The existence of such a (informal theorem) falsifier shows that the
axiomatization has not validly expressed the informal theory, i.e. its source (Lakatos, 1978).

3. History is central
The epistemological task of the philosophy of mathematics is not simply to answer
the question ‘how is (any) mathematical knowledge possible?’, but to account for the
actual mathematical knowledge that exists. Thus, the philosophy of mathematics is

35
The Philosophy of Mathematics Education

indissolubly linked with the history of mathematics, since the latter is the history of
the evolution of mathematical knowledge.

4. The primacy of informal mathematics is asserted

Informal mathematics is of paramount importance, both as a practice and as a
product. As a product, it is the source of all formal mathematics, since it is what is
formalized. It is also, as we have seen, the source of potential falsifiers of formal
mathematics. The importance of mathematical practice is that it is the ‘stuff’ of the
history of mathematics, and the quasi-empirical source of mathematics. It provides
individuals with the premises and conclusions of deductive mathematics (informal
axioms, definitions and conjectures), and the informal proofs through which the
premises and conclusions are connected.

5. A theory of knowledge creation is included

A central concern of the philosophy of mathematics is the logic of mathematical
discovery, or ‘heuristic’. This underlies ‘the autonomous dialectic of mathematics’
(Lakatos, 1976, page 146), the mechanism for the genesis of mathematical knowledge.
In this process the productions of individual mathematicians (usually a constellation of
definitions, conjectures and informal proofs) are exposed to criticism, and reformulated
in response to the criticism, in an iterated dialectical cycle. This process, following its
own autonomous logic, is necessary for new items (definitions, theorems, proofs) to
become incorporated into the body of accepted mathematical knowledge.

There is a simple pattern of mathematical discovery—or of the growth of

informal mathematical theories. It consists of the following stages:
1 Primitive conjecture.
2 Proof (a rough thought-experiment or argument, decomposing the
primitive conjecture into subconjectures or lemmas).
3 ‘Global’ counterexamples (counterexamples to the primitive conjecture)
emerge.
4 Proof re-examined: the ‘guilty lemma’to which the global counter-example
is a ‘local’ counterexample is spotted. This guilty lemma may have
previously remained ‘hidden’ or may have been misidentified. Now it is
made explicit, and built into the primitive conjecture as a condition. The
theorem—the improved conjecture—supersedes the primitive conjecture
with the new proof-generated concept as its paramount new feature.
These four stages constitute the essential kernel of proof analysis. But there
are some further standard stages which frequently occur:
5 Proofs of other theorems are examined to see if the newly found
lemma or the new proof-generated concept occurs in them: this

36
 The Philosophy of Mathematics Reconceptualized

concept may be found lying at cross-roads of different proofs, and thus

emerge as of basic importance.
6 The hitherto accepted consequences of the original and now refuted
conjecture are checked.
7 Counterexamples are turned into new examples—new fields of
inquiry open up.
(Lakatos, 1976, pages 127–128)
Mathematical activity is human activity. Certain aspects of this activity—as
of any human activity—can be studied by psychology, others by history.
Heuristic is not primarily interested in these aspects. But mathematical
activity produces mathematics. Mathematics, this product of human activity,
‘alienates itself from the human activity which has been producing it. It
becomes a living, growing organism, that acquires a certain autonomy from
the activity which has produced it; it develops its own autonomous laws of
growth, its own dialectic. The genuine creative mathematician is just a
personification, an incarnation of these laws which can only realise
themselves in human action. Their incarnation, however, is rarely perfect.
The activity of human mathematicians as it appears in history, is only a
fumbling realisation of the wonderful dialectic of mathematical ideas. But
any mathematician, if he has talent, spark, genius, communicates with, feels
the sweep of, and obeys this dialectic of ideas.
(Lakatos, 1976, page 146).
It can be seen that at the heart of Lakatos’ philosophy of mathematics is a theory of
the genesis of mathematical knowledge. This is a theory of mathematical practice,
and hence a theory of the history of mathematics. Lakatos is not offering a
psychological theory of mathematical creation or discovery, for he does not deal with
the origins of the axioms, definitions and conjectures in the mind of the individual.
His focus is instead on the process which transforms private creations into accepted
public mathematical knowledge, a process which centrally involves criticism and
reformulation. In this, his philosophy closely resembles the falsificationist philosophy
of science of Karl Popper, a debt that Lakatos readily acknowledges. For Popper
(1959) proposes a ‘logic of scientific discovery’, in which he argues that science
advances through a process of conjectures and refutations. The difference is that
Popper concerns himself only with rational reconstructions or idealizations of
theories, and denies the philosophical validity of the application of his model of
science to history. Lakatos on the other hand, refuses to separate the philosophical
theory of knowledge growth from its historical realization.
Despite his avoidance of the pitfalls of psychologism, Lakatos may be accused of
straying outside the boundaries of legitimate philosophical concern. For in contrast
with most epistemology in the English-speaking world, which focus exclusively on
objective knowledge or a single knowing subject, quasi-empiricism discusses knowing
or knowledge generation as part of a social process. However, in the philosophy of
mathematics, as we have seen, there is a dearth of theories offering an adequate

37
The Philosophy of Mathematics Education

account of mathematics. Thus the traditional limitations on what counts as legitimate

in philosophy may in fact be an obstacle to an adequate philosophy of mathematics.
We thus turn to an evaluation of Lakatos’ quasi-empiricist philosophy of
mathematics. However it must be remarked that Lakatos’ philosophy of mathematics
is far from a complete or fully worked out system. This is due to two factors. First of
all, his untimely death. Lakatos’ wrote only one slim volume and five papers on the
philosophy of mathematics, and a number of these were unfinished and published
posthumously. Secondly, his style of presentation in his major work was indirect,
utilizing a platonic dialogue to reconstruct an aspect of the history of mathematics.
Thus Lakatos’ has bequeathed us an exciting but incomplete philosophy of
mathematics, far from fully worked or elaborated. Thus the potential, as well as its
realization, must be born in mind in assessing his quasi-empiricism.

B. The Adequacy Criterion and Quasi-empiricism

To evaluate the quasi-empiricist philosophy we consider it in the light of the

adequacy criteria.
Quasi-empiricism offers a partial account of the nature of mathematical
knowledge, and its genesis and justification. In this Lakatos offers a more extensive
account than the other philosophies of mathematics we considered, far exceeding
them in scope. To the traditional stratum of formalized mathematical knowledge he
adds a new, lower stratum, namely that of informal mathematical knowledge. To this
extended system he adds a dynamic, which shows not only how knowledge in the
lower stratum develops, but also the relationship between the two strata. In particular,
how knowledge in the lower stratum is reflected upwards, by formalization, to form
idealized images at the upper level, which are seen as the indubitable truths of
mathematics. Lakatos accounts for the nature of mathematical knowledge as
hypothetico-deductive and quasi-empirical, building a striking analogy with Popper’s
(1979) philosophy of science. He accounts for errors in mathematical knowledge, and
provides an elaborate theory of the genesis of mathematical knowledge. This
potentially accounts for much of mathematical practice, and for its history.
Since Lakatos’ theory of the genesis of mathematical knowledge puts it on a par,
in many respects, with scientific knowledge, the success of the applications of
mathematics are potentially explicable by analogy with science and technology.
Explaining the success of applied mathematics would be a significant strength,
especially in the face of the neglect shown by other philosophies of mathematics
(Korner, 1960). Finally, a key strength of Lakatos’ philosophy of mathematics is that
it is not prescriptive, but descriptive, and attempts to describe mathematics as it is
and not as it ought to be practised.
In terms of the four adequacy criteria, quasi-empiricism partly satisfies those
concerning mathematical knowledge (i), applications (iii) and practice (iv).

38
 The Philosophy of Mathematics Reconceptualized

C. Weaknesses of Lakatos’ Quasi-empiricism

Quasi-empiricism can be criticized on a number of counts. First of all, there is no

account of mathematical certainty. Lakatos fails to explain why mathematical truth is
seen to be the most certain of all knowledge. Likewise, he fails to account for the
seeming certainty of deductive logic. Thus Lakatos fails to account for the apparent
certainty of mathematical knowledge, and the weight that is attached to its
justification.
Second, Lakatos gives no account of the nature of the objects of mathematics, or
of their origins. There is no indication in his account of the plausibility of platonism.
He does not even offer grounds for the objectivity of heuristic, the ‘autonomous
laws of growth’ of mathematical knowledge, assumed in his account.
Third, Lakatos does not account for either the nature or the success of the
applications of mathematics, or its effectiveness in science, technology, and in other
realms. It may be argued, as I have done, that there is a potential justification of the
utility of mathematics in quasi-empiricism. The fact remains it is neither-hinted at
nor given. Furthermore, Lakatos deals and refers exclusively to pure mathematics, so
an account of the nature of applied mathematics is missing from his philosophy.
Fourth, Lakatos does not sufficiently establish the legitimacy of bringing the
history of mathematics into the heart of his philosophy of mathematics. In this, he is
flying in the face of a philosophical tradition (although, as we have seen, an
increasing number of philosophers bring genetic issues into epistemology). His
approach is potentially fruitful, perhaps more so than previous philosophies of
mathematics. Nevertheless his historical approach needs explicit philosophical
justification (beyond the implicit justification that he is trying to account for
mathematics descriptively).
Fifth, there is the problem of the contrasting status of his philosophy and the
historical thesis. Lakatos fails to provide justification for introducing an empirical (i.e.
conjectural) historical thesis into an analytic philosophical approach, on an equal
footing with logical methodology. This, I believe, is due to his failure to distinguish
the general dialectical logic of mathematical discovery from a specific empirical
thesis concerning the stages of development and elaboration of mathematical
knowledge. The general logic of mathematical discovery, like Popper’s (1968) logic
of scientific discovery, is hypothetico-deductive. It involves the general dialectical
method of conjectures (including proofs) and refutations. It is a purely logical
methodology, describing the general form and conditions for the improvement, and
ultimately, the acceptance of mathematical knowledge.
In contrast, Lakatos’ empirical hypothesis concerns the actual historical stages
through which mathematical knowledge passes during its development from initial
conjecture to accepted mathematical knowledge (the seven stage heuristic quoted
above). This is an empirical conjecture, and does not constitute a necessary condition
for the validity of the general quasi-empiricist philosophy of mathematics. Thus the
details of this specific empirical thesis may be modified, without calling into question
the underlying (and general) philosophy.3 In fact the philosophy and the historical

39
The Philosophy of Mathematics Education

thesis are logically independent, in that the rejection of one has no logical
implications for the other. Lakatos seems to be unaware of this distinction.
Sixth, Lakatos’ quasi-empiricist philosophy of mathematics provides grounds that
are necessary hut not sufficient for establishing mathematical knowledge. Examples
can be found of mathematical knowledge that after development and reformulation,
following the general pattern of Lakatos’ heuristic, is still not incorporated into the
body of accepted mathematical knowledge. Consider, as a fictional counterexample,
the idiosyncratic mathematics that may be developed by a group of mystics, sharing
a set of conventions and norms, including the basis of their critical methodology,
which is peculiar to themselves. The fact that this group’s mathematical creations
survive their process of proofs and refutations does not give them general acceptance.
To rule out such examples, quasi-empiricism requires the assumption of a shared
basis for its critical methodology, if there is to be universal agreement on its
outcomes. In effect, this is the assumption of the use of a standard logic, and of its
validity.
Lastly, there is no systematic exposition of quasi-empiricism, putting forward its
theses in detail, and anticipating and rebutting objections to it. Lakatos’ publications
on the philosophy of mathematics comprise reconstructed historical case studies and
polemical writings.
Overall, it can be seen that the major defects of quasi-empiricism are sins of
omission, rather than of commission. The above critique, admittedly from a
sympathetic viewpoint, discloses no fundamental flaw or defect. Rather it indicates a
needed research programme, namely to develop quasi-empiricism systematically and
fill the gaps.

D. Quasi-empiricism and the Philosophy of Mathematics

Quasi-empiricism offers a partial account of the nature of mathematical knowledge,

and its genesis and justification. In this Lakatos offers potentially a far more extensive
account than the other philosophies of mathematics which have been considered. In
large part, this depends on Lakatos’ reconceptualization and redefinition of the
philosophy of mathematics. He has called into question the dominant orthodoxy in
the philosophy of mathematics with regard to both its foundationism and absolutism
(the arguments in the previous chapter are unlikely to have been formulated, but for
Lakatos). Thus he has freed the philosophy of mathematics to reconsider its function
in the terms described in this chapter, as well as to question the hitherto
unchallenged status of mathematical truth. Although these are critical (i.e. essentially
negative) achievements of quasi-empiricism their importance cannot be
overestimated.
In positive terms quasi-empiricism has the potential to offer solutions to many of
the new problems Lakatos posed for the philosophy of mathematics. In the next
chapter I shall attempt to build a novel social constructivist philosophy of
mathematics on the basis of Lakatos’ quasi-empiricism.

40
 The Philosophy of Mathematics Reconceptualized

Notes

1 ‘Formal absolutism’ is my term, in place of Confrey’s ‘absolutism’, to avoid ambiguity. Confrey also
has a third category of ‘conceptual change’ theories, corresponding to fallibilism and drawing heavily
on Lakatos (1976).
2 Wittgenstein’s conventionalism is more radical than the account given in the chapter, although the
selective account presented there can be considered independently of some of his other views.
Dummett (1959) terms it a ‘full blooded conventionalism’, because all of the ‘truths’ of mathematics
and logic, not just the axioms, are direct expressions of linguistic conventions. Consequently,
Wittgenstein denies that mathematics has any logical foundations, it rests instead on the rules of
actual practice, both linguistic and mathematical. He claims that every time we accept a new
theorem, we accept a new rule of language. He also holds a very strict constructivist view of
mathematics known as ‘strict finitism’, and thus rejects even more of classical mathematics than the
intuitionists. Dummett (1959) criticizes some of these views.
3 Considered as an empirical hypothesis about the development of mathematics, Lakatos’ thesis has its
limitations. It is based on a single case study from nineteenth century mathematics, the Euler
conjecture (perhaps two, counting uniform continuity). In other domains, such as number theory,
conjectures concerning well defined concepts (not needing redefinition, ‘monster barring’, etc) may
be proposed in final form, simply needing proofs (for example, Ramanujan’s conjectures proved by
Hardy and Littlewood). In yet further domains, such as axiomatic set theory, it may not be possible
to distinguish sharply between substance (concepts) and form (proofs), as Lakatos does in his case
studies. See my review of Lakatos (1976) in Mathematical Reviews for further elaboration of this point.

41
 3

Social Constructivism as a Philosophy

of Mathematics

1. Social Constructivism1

In this chapter I propose a new philosophy of mathematics called ‘social

constructivism’. Naturally, as it concerns a novel philosophy of mathematics, this
chapter is more tentative than those preceding it, which were largely concerned with
the exposition of well-established ideas. On the other hand, not too much novelty
should be claimed, since social constructivism is largely an elaboration and synthesis
of pre-existing views of mathematics, notably those of conventionalism and quasi-
empiricism.2
Social constructivism views mathematics as a social construction. It draws on
conventionalism, in accepting that human language, rules and agreement play a key
role in establishing and justifying the truths of mathematics. It takes from quasi-
empiricism its fallibilist epistemology, including the view that mathematical
knowledge and concepts develop and change. It also adopts Lakatos’ philosophical
thesis that mathematical knowledge grows through conjectures and refutations,
utilizing a logic of mathematical discovery. Social constructivism is a descriptive as
opposed to a prescriptive philosophy of mathematics, aiming to account for the nature
of mathematics understood broadly, as in the adequacy criteria.
The grounds for describing mathematical knowledge as a social construction and
for adopting this name are threefold:

(i) The basis of mathematical knowledge is linguistic knowledge, conventions

and rules, and language is a social construction,
(ii) Interpersonal social processes are required to turn an individual’s subjective
mathematical knowledge, after publication, into accepted objective
mathematical knowledge,
(iii) Objectivity itself will be understood to be social.

42
 Social Constructivism

A. Overview of Social Constructivism

Like quasi-empiricism, a central focus of social constructivism is the genesis of

mathematical knowledge, rather than just its justification. Newly generated
mathematical knowledge can be either subjective or objective knowledge, and a
unique feature of social constructivism is that it considers both these forms of
knowledge, and links them in a creative cycle. It is not uncommon to see subjective
knowledge and objective knowledge treated together in philosophy, as in Popper
(1979). What is less common is for their links to be treated, since this admits the
genesis of knowledge into philosophy.
Social constructivism links subjective and objective knowledge in a cycle in
which each contributes to the renewal of the other. In this cycle, the path followed
by new mathematical knowledge is from subjective knowledge (the personal
creation of an individual), via publication to objective knowledge (by intersubjective
scrutiny, reformulation and acceptance). Objective knowledge is internalized and
reconstructed by individuals, during the learning of mathematics, to become the
individuals’ subjective knowledge. Using this knowledge, individuals create and
publish new mathematical knowledge, thereby completing the cycle. 3 Thus
subjective and objective knowledge of mathematics each contributes to the creation
and re-creation of the other. The assumptions underpinning the social constructivist
account of knowledge creation are as follows.

1. An individual possesses subjective knowledge of mathematics

A key distinction is that between subjective and objective knowledge. The
mathematical thought of an individual (both the process and its product, mathematical
knowledge) is subjective thought. This is largely learned (i.e. reconstructed objective)
knowledge, but, subject to certain powerful constraints, the process of re-creation
results in unique subjective representations of mathematical knowledge. Furthermore,
individuals use this knowledge to construct their own, unique mathematical
productions, the creation of new subjective mathematical knowledge.

2. Publication is necessary (but not sufficient) for subjective knowledge to become objective
mathematical knowledge
When an individual’s subjective mathematical knowledge production enters the
public domain through publication, it is eligible to become objective knowledge.
This will depend on its acceptance, but first it must be physically represented (in
print, electronically, in writing, or as the spoken word). (Here knowledge is
understood to include not only statements, but also their justification, typically in the
form of informal proofs).

43
The Philosophy of Mathematics Education

3. Through Lakatos’ heuristic published knowledge becomes objective knowledge of

mathematics
Published mathematics is subject to scrutiny and criticism by others, following
Lakatos’ (1976) heuristic, which may result in its reformulation and acceptance as
objective (i.e., socially accepted) knowledge of mathematics. The successful
application of this heuristic is sufficient for acceptance as (tentative) objective
mathematical knowledge, although the knowledge always remains open to challenge.

4. This heuristic depends on objective criteria

During the genesis of mathematical knowledge, objective criteria play an essential
part (Lakatos’ autonomous logic of mathematical discovery, understood
philosophically, not historically). These criteria are used in the critical scrutiny of
mathematical knowledge, and include shared ideas of valid inference and other basic
methodological assumptions.

5. The objective criteria for criticizing published mathematical knowledge are based on
objective knowledge of language, as well as mathematics
The criteria depend to a large extent on shared mathematical knowledge, but
ultimately they rest on common knowledge of language, that is, on linguistic
conventions (the conventionalist view of the basis of knowledge). These too are
socially accepted, and hence objective. Thus both published mathematical knowledge
and the lingusitic conventions on which its justification rests are objective knowledge.

6. Subjective knowledge of mathematics is largely internalized, reconstructed objective

knowledge
A key stage in the cycle of mathematical creation is the internalization, that is the
inner subjective representation, of objective mathematical and linguistic knowledge.
Through the learning of language and mathematics inner representations of this
knowledge, including the corresponding rules, constraints and criteria are
constructed. These permit both subjective mathematical creation, and participation
in the process of criticizing and reformulating proposed (i.e., public) mathematical
knowledge.

7. Individual contributions can add to, restructure or reproduce mathematical knowledge

On the basis of their subjective knowledge of mathematics individuals make potential
contributions to the pool of objective knowledge. These can add to, restructure, or
simply reproduce existing knowledge of mathematics (subject to the heuristic).

44
 Social Constructivism

Additions can be new conjectures or proofs, which may include new concepts or
definitions. They can also be new applications of existing mathematics. Restructuring
contributions may be new concepts or theorems that generalize or otherwise link two
or more previously existing parts of mathematical knowledge. Contributions that
reproduce existing mathematics are typically textbooks or advanced expositions.

B. Immediate Problems of Social Constructivism

Two immediate problem areas arise from this brief account. First of all, there is the
identification of objectivity with the social or socially accepted. To identify the
immutable and enduring objectivity of the objects and truths of mathematics with
something as mutable and arbitrary as socially accepted knowledge does, initially,
seem problematic. However we have already established that all mathematical
knowledge is fallible and mutable. Thus many of the traditional attributes of
objectivity, such as its enduring and immutable nature, are already dismissed. With
them go many of the traditional arguments for objectivity as a super-human ideal.
Following Bloor (1984) we shall adopt a necessary condition for objectivity, social
acceptance, to be its sufficient condition as well. It remains to show that this
identification preserves the properties that we expect of objectivity.
Secondly, there is the problem of the proximity of social constructivism to
sociological or other empirical accounts of mathematics. Since it is quasi-empirical,
and has the task of accounting for the nature of mathematics including mathematical
practice, in a fully descriptive fashion, the boundary between mathematics and other
disciplines is weakened. By removing traditional philosophical barriers these
consequences bring the philosophy of mathematics closer to the history and sociology
of mathematics (and psychology too, concerning subjective knowledge). Thus there is
the danger of social constructivism straying into the provinces of history, sociology or
psychology.We saw that Lakatos (1976) conflates his theory of the historical evolution
of mathematical knowledge with his philosophical account of the genesis of
mathematical knowledge. Thus there is a real danger of conflating empirical with
philosophical accounts of mathematics, which social constructivism must avoid.

2. Objective and Subjective Knowledge

A. The Nature of Objective and Subjective Knowledge

Before proceeding further with the exposition and development of social

constructivism it is necessary to establish some philosophical preliminaries. A key
distinction that is employed is that between subjective and objective knowledge. This
is clarified by a consideration of Popper’s (1979) definition of three distinct worlds,
and the associated types of knowledge.

45
The Philosophy of Mathematics Education

We can call the physical world ‘world 1’, the world of our conscious
experiences ‘world 2’, and the world of the logical contents of books,
libraries, computer memories, and suchlike ‘world 3’.
(Popper, 1979, p. 74)

Subjective knowledge is world 2 knowledge, objective knowledge is world 3

knowledge, and according to Popper includes products of the human mind, such as
published theories, discussions of such theories, related problems, proofs; and is
human-made and changing.
I shall use the term ‘objective knowledge’, in a way that differs from Popper, to
refer to all knowledge that is intersubjective and social. I wish to count all that
Popper does as objective knowlege, including mathematical theories, axioms,
conjectures, proofs, both formal and informal. One difference is that I also want to
include additional ‘products of the human mind’ as objective knowledge, notably the
shared (but possible implicit) conventions and rules of language usage. Thus I am
referring to publicly shared, intersubjective knowledge as objective, even if it is
implicit knowledge, which has not been fully articulated. This extension is very likely
unacceptable to Popper.
In fact, I wish to adopt the social theory of objectivity proposed by Bloor.

Here is the theory: it is that objectivity is social. What I mean by saying that
objectivity is social is that the impersonal and stable character that attaches
to some of our beliefs, and the sense of reality that attaches to their
reference, derives from these beliefs being social institutions.
I am taking it that a belief that is objective is one that does not belong
to any individual. It does not fluctuate like a subjective state or a personal
preference. It is not mine or yours, but can be shared. It has an external
thing-like aspect to it.
(Bloor, 1984, page 229).

Bloor argues that Popper’s world 3 can defensibly and fruitfully be identified with
the social world. He also argues that not only is the three-fold structure of Popper’s
theory preserved under this transformation, but so are the connections between the
three worlds. Naturally, the social interpretation does not preserve the meaning that
Popper attaches to objectivity, who regards the logical character of theories, proofs
and arguments sufficient to guarantee objectivity in an idealistic sense. Despite this,
the social view is able to account for most, if not all, features of objectivity: the
autonomy of objective knowledge, its external thing-like character (presumably the
original meaning of ‘object’-ivity), and its independence from any knowing subject’s
subjective knowledge. For the social view sees objective knowledge, like culture,
developing autonomously in keeping with tacitly accepted rules, and not subject to
the arbitrary dictates of individuals. Since objective knowledge and rules exist
outside individuals (in the community), they seem to have an object-like and
independent existence.

46
 Social Constructivism

Thus it can be seen that the social view accounts for many of the necessary
characteristics of objectivity. Beyond this, it is worth remarking that Bloor’s social view
of objectivity explains and accounts for objectivity. In contrast traditional views
(including Popper’s) elaborate on, or at best define objectivity (intensively or extensively),
but never account for or explain objectivity. For the autonomous, independent existence
of objective knowledge is traditionally shown to be necessary, without any explanation
of what objectivity is, or how objective knowledge can emerge from subjective human
knowledge. In contrast, the social view of objectivity is able to offer an account of the
basis and nature of objectivity and objective knowledge.
One immediate problem the social view must face is that of accounting for the
necessity of logical and mathematical truth. The answer given by Bloor (1983,1984),
and adopted here, is that this necessity (understood in a fallibilist sense) rests on
linguistic conventions and rules, as Wittgenstein proposes. This is the full
conventionalist account of the basis of logic and mathematical knowledge.

B. The Role of Objective Knowledge in Mathematics

Having clarified the sense in which objectivity is understood to be social, it is worth

reiterating the social constructivist account of objective mathematical knowledge.
According to social constructivism, published mathematics, that is mathematics that is
represented symbolically in the public domain, has the potential to become objective
knowledge. The application of Lakatos’ logic of mathematical discovery to this
published mathematics is the process that leads to social acceptance, and thus to
objectivity. Once mathematical axioms, theories, conjectures, and proofs are
formulated and presented publicly, even if only in conversation, the autonomous (i.e.,
socially accepted) heuristic begins to work. Both the process and its product are
objective, being socially accepted. Likewise, both the implicit and explicit
conventions and rules of language and logic on which this heuristic rests are
objective, also being socially accepted. It is these conventions and rules which it is
claimed, following conventionalism, underpin mathematical knowledge (including
logic). For they provide the basis of logical and mathematical definitions, as well as
the basis for the rules and axioms of logic and mathematics.

C. The Role of Subjective Knowledge in Mathematics

Given the centrality of the role of objective knowledge, I wish to argue that the role of
subjective mathematical knowledge must also be acknowledged, or else the overall
account of mathematics will be incomplete. For subjective knowledge is needed to
account for the origins of new mathematical knowledge, as well as, according to the
theory proposed, the re-creation and sustainment of existing knowledge. Since objective
knowledge is social, and not a self-subsistent entity existing in some ideal realm, then
like all aspects of culture this knowledge must be reproduced and transmitted from

47
The Philosophy of Mathematics Education

generation to generation (admittedly with the aid of artefacts, such as text books).
According to the social constructivist account, subjective knowledge is what sustains
and renews objective knowledge, whether it be of mathematics, logic or language.Thus
subjective knowledge plays a central part in the proposed philosophy of mathematics.
Having said this, it must be acknowledged that the treatment of subjective as well
as objective knowledge, in the proposed theory, is at odds with much modern
thought in philosophy, and in the philosophy of mathematics, as we have seen
(barring intuitionism, which we have rejected). For example, Popper (1959) has
distinguished very carefully between the ‘Context of discovery’ and the ‘context of
justification’ in science. He regards the latter context as being subject to logical
analysis, and thus being the proper concern of philosophy. The former context,
however, concerns empirical matters, and therefore is the proper concern of
psychology, and not of logic or philosophy.
Anti-psychologism, the view that subjective knowledge—or at least its
psychological aspects—is unfitted for philosophical treatment, rests on the following
argument. Philosophy consists of logical analysis, including methodological problems
such as the general conditions for the possibility of knowledge. Such inquiry is a
priori, and is wholly independent of any particular empirical knowledge. Subjective
issues are of necessity psychological issues, since they of necessity refer to the
contents of individual minds. But such matters, and psychology in general, are
empirical. Therefore, because of this category difference (the a priori versus the
empirical realm) subjective knowledge cannot be the concern of philosophy.4
This argument is rejected here on two grounds. First of all, a powerful critique of
absolutism, and hence of the possibility of certain a priori knowledge has been mounted
(Chapter 1). On this basis all so-called a priori knowledge, including logic and
mathematics, depends for its justification on quasi-empirical grounds. But this
effectively destroys the unique categorical distinction between a priori knowledge and
empirical knowledge. Thus this distinction cannot be used to deny the applicability of
the a priori philosophical methods of objective knowledge to subjective knowledge, on
the grounds that the latter is empirically tainted. For now we see that all knowledge,
including objective knowledge, is empirically (or rather quasi-empirically) tainted.
The second argument, which is independent of the first, is as follows. In
discussing subjective knowledge it is not proposed to discuss the specific contents of
individuals’ minds, nor specific empirical psychological theories of the mind under
the guise of philosophy. Rather the intention is to discuss the possibility of subjective
knowledge in general, and what can be concluded about its possible nature on the
basis of logical reasoning alone (given a number of theoretical assumptions). This is a
legitimate philosophical activity, just as the philosophy of science can legitimately
reflect on an empirical realm, namely science, without therefore being an empirical
realm itself. Thus subjective knowledge is a proper matter for philosophical inquiry.
Indeed, in discussing belief or the knowing subject, this is precisely what
epistemologists such as Sheffler (1965), Woozley (1949), Chisholm (1966) and indeed
Popper (1979), are considering. Further back, epistemology has traditionally
considered subjective knowledge, at least from the time of Descartes (and probably

48
 Social Constructivism

further back to Plato), through the British empiricists Locke, Berkeley and Hume,
via Kant to the present day. Thus subjective knowledge is a legitimate area of
philosophical enquiry, based on a substantial philosophical tradition.
Although the claim that the consideration of subjective knowledge is
psychologistic is thus refuted, it is acknowledged that there are real dangers and
legitimate concerns arising from the philosophical treatment of subjective
knowledge. For it makes it easier to commit the error of using psychologistic
reasoning in philosophy, that is reasoning based on psychological belief of necessity as
opposed to logical argument. Furthermore, the distinction between subjective and
objective knowledge is a vital one to maintain, both for social constructivism, and for
philosophy in general. These are two genuinely distinct domains of knowledge.
For these reasons, in the explication of the social constructivist philosophy of
mathematics, the domains of objective and subjective knowledge will be treated
separately. The objective aspect of this philosophy is independent of the subjective
aspect in terms of its justification. So the reader wary of psychologism can follow the
objective aspect of social constructivism without qualms (concerning this issue, at least).

3. Social Constructivism: Objective Knowledge

In providing a social constructivist account of objective knowledge in mathematics

we need to establish a number of claims. We need to justify the account of objective
mathematical knowledge by demonstrating both the objectivity of what is referred
to, as well as the fact that it is indeed warranted knowledge. Having established these
minimal conditions for an account of objective mathematical knowledge, we next
need to establish that social constructivism provides an adequate philosophical
account of mathematics. This involves satisfying the adequacy criteria for the
philosophy of mathematics formulated in the last chapter.

A. Objectivity in Mathematics

In Chapter 1, on the basis of a powerful critique of absolutism, we accepted the

fallibility of mathematical knowledge. Whilst this is a central assumption of social
constructivism, the fact remains that the objectivity of both mathematical knowledge
and the objects of mathematics is a widely accepted feature of mathematics, and must
be accounted for by any philosophy of mathematics. It has been established that
objectivity is understood to reside in public, intersubjective agreement, that is, it is
social. Thus the objectivity of mathematics means that both the knowledge and
objects of mathematics have an autonomous existence upon which there is
intersubjective agreement, and which is independent of the subjective knowledge of
any individual. We therefore need to establish the shared basis for this knowledge,
which allows public access to it, and guarantees intersubjective agreement on it.
Subsequently, we extend the discussion to the objectivity of the ontology of

49
The Philosophy of Mathematics Education

mathematics, that is the basis for the autonomous existence of the objects of
mathematics. We consider first the substratum which provides the foundation for
objectivity in mathematics, namely language.

The linguistic basis of objectivity in mathematics

The claim is that the objectivity of mathematical knowledge is based on shared
knowledge of natural language. Such an account has already been sketched in the
treatment of conventionalism in the previous chapter. There Wittgenstein’s account
of the lingusitic basis for mathematics and logic was presented and judged promising.
Here we develop the argument further. It will be argued that acquiring competence
in natural language necessarily involves the acquisition of a large, implicit, body of
knowledge. Part of this knowledge is elementary knowledge of mathematics and
logical reasoning, and their applications. Linguistic communication requires that the
rules and conventions of this language, which embody its meaning, are presupposed.
These shared presuppositions, without which communication breaks down, are the
basis for the objectivity of mathematical knowledge (and objects). This is the essence
of the argument.
This is not argued on the basis of psychological or empirical facts, but on logical
and philosophical grounds. For it is a truism that any logical system of knowledge, be
it deductive or definitional, depends ultimately on a set of primitive propositions or
terms. For objective mathematical knowledge these primitive propositions and terms
are to be found in the objective knowledge of natural language.
To flesh out the bare bones of this argument, we notice first that traditionally,
objective knowledge is identified with a set of propositions or statements, (or the
contents thereof), that is as a linguistically expressed body of knowledge. In previous
chapters it has been argued that knowledge includes in addition to prepositional
knowledge such things as processes and procedures. However, these too can be
represented as propositions. Thus the fundamental assumption that knowledge has a
lingusitic basis is not in question. This means that the understanding of such
knowledge depends essentially on linguistic competence, as does most of human
cognitive and social activity.
Linguistic competence consists of the ability to communicate linguistically. This in
turn depends on the shared use of grammatical forms, of the relationships between
terms, and of the applicability of terms and descriptions to situations, including shared
meanings of terms, at least in publicly observable, behavioural uses. It also depends on
the ability to interrelate social contexts and certain forms of discourse. In brief, linguistic
competence depends on the common following of rules, in conformity with public
usage. As we have seen in the discussion of Wittgenstein’s conventionalism, the
participation in our shared social ‘language games’ requires the acceptance and following
of language rules and conventions. Through being woven into the whole complex of
human action and communication (Wittgenstein’s ‘forms of life’), these rules acquire a
necessity, that cannot be questioned without threatening to unravel the whole enterprise.

50
 Social Constructivism

We cannot question the fact that ‘A and B’ entails ‘A’ or that 1+1=2 without
withdrawing some of the possibility of communication. We can only get around this
temporarily, by circumscribing a small domain of language use, and exposing and
questioning some of the rules governing its use.We may ‘freeze’ and thus suspend some
of these rules to dissect them. But in our other language games, including our meta-
language, these rules remain in force. And when our inquiry moves on, the rules
become reanimated, and reassume their living certitude. Like a boat in mid voyage, we
may tentatively remove a plank from the hull and question its role. But unless we
reinstate it before we continue our inspection, the whole enterprise may founder.
This is the general argument for the necessity of the rules associated with language
use. These rules codify the shared linguistic behaviour which allows the possibility of
communication. In detail, these rules depend on the particular terms and rules of
mathematics and logic embedded in our language. We consider these next.
Our natural language contains informal mathematics as a subset, including such
terms as ‘square’, ‘circle’, ‘shape’, ‘zero’, ‘one’, ‘two’, ‘number’, ‘add’, ‘less’, ‘greater’,
‘equals’, ‘set’, ‘element’, ‘infinite’ and so on. Some of these terms are directly
applicable to the shared world of our experience, and natural language includes rules
and conventions on how to apply these terms. In this sense, these terms resemble
those of science, for their basic terms are learned together. Such terms allow us to
describe events and objects in the world by classification and quantification. The
intended interpretations of informal mathematics, such as these, are implicit in the
semantics of natural language (which often provides multiple meanings for these
terms). In addition, the inter-relations between terms are established by linguistic
conventions and rules. Thus, for example, ‘one is less than two’ and ‘an infinite set
has more than two elements’ are both warranted on the basis of the semantic rules of
language. As was stated, the elementary applications of mathematics are also built into
the rules of linguistic use. The presence of these two types of rule, those concerning
the interconnections of terms and their intended applications in the world, account
for much of the implicit mathematical knowledge we unconsciously acquire with
linguistic competence.
This account is grossly oversimplified in one sense. For it appears to assume a
single external world. In fact, there are many overlapping domains of linguistic
discourse, many language games, each with their own shared worlds of reference.
Some relate to what is socially accepted by the majority as objective reality, others
less so, and some are wholly fictional or mythological. Each contains an informal
theory, a set of relationships between the entities that inhabit them. What they all
share is social agreement on the rules relating to discourse about them.
Many of our linguistic utterances, whatever ‘language game’ we are engaged in, are
laden with mathematical concepts, or highly ‘mathematized’ (Davis and Hersh, 1986).
For an example of the embeddedness of mathematics in every day language use,
consider the Zen koan ‘What is the sound of one hand clapping?’.This is based on the
linguistic knowledge that it takes two hands to clap, one is half of two, but half the
number of hands does not give half the amount of sound (I focus here on
mathematical content, and not on the purpose of the koan which is through cognitive

51
The Philosophy of Mathematics Education

challenge to induce satori). Overall, I wish to claim that natural language such as
English (and Japanese, apparently), and even more so informal mathematical language,
is rich with implicit mathematical rules, meanings and conventions. These rules, such
as ‘two is the successor of one’, necessitate the acceptance of truths, such as ‘1+1=2’.

The linguistic basis of logic

The same can be said for logic in language. Our use of the key logical terms ‘not’,
‘and’, ‘or’, ‘implies’, ‘if, and only if’, ‘entails’, ‘there exists’, ‘for all’, ‘is a’, and so on,
strictly follows linguistic rules. (We ignore inconsistent colloquial variations such as
‘not not=not’, which are rejected by mathematics and logic.) These rules fix as true
basic statements such as ‘if A, then A or B’, and rules of inference such as ‘A’ and ‘A
implies B’ together entail ‘B’. For these rules reflect the use of these terms, and
hence (according to Wittgenstein) their meaning. The rules and conventions of logic
underpin more than just the ‘truths’ of logic. As we have seen, they also underpin
logical relationships, including implication and contradiction. Thus reasoning, and
indeed, the whole basis of rational argument rests on the shared rules of language.
The rather more abstract and powerful forms of logic used in mathematics also
rest on the logic embedded in natural language use. However, the rules and meanings
of mathematical logic represent a formalized and refined version of this logic. They
make up a tighter set of language games overlapping with those of natural language
logic.

The linguistic basis accommodates conceptual change

It has been argued, on conventionalist grounds, that everyday mathematical
knowledge is linguistic knowledge, and that it derives its security, indeed its apparent
necessity, from the regularity and agreed use of language. But while linguistic
convention provides everyday mathematical knowledge with its secure foundation,
so too it provides the grounds for change in mathematics, as linguistic conventions
and usage develop over time.
Consider the most unassailable of arithmetical truths. Since time immemorial it
has been inconceivable to question the elementary fact ‘1+1=2’ (but see Restivo,
1984, on ‘2+2=4’). However, since the time of George Boole we can assert the
contradictory fact ‘1+1=1’. It can be rejoined that this is only because Boole has
invented a formal system which assigns different meanings to the symbols. This is
true, but the fact remains that ‘1+1=1’ is no longer false, and that ‘1+1=2’ is no
longer absolutely true. It is true given certain presuppositions (admittedly embedded
in our natural language) which, when conflict arises, need to be made explicit. Prior
to Boolean algebra the questioning of ‘1+1=2’ was simply not coherently possible.
The real change, therefore, lies behind the scenes. It resides in the fact that we can
suspend our everyday rules for portions of language, and consider the consequences
of hypothetical conventions, i.e. those contrary or distinct from those embedded in

52
 Social Constructivism

natural language usage. This is the change that led Russell to claim that Boole
originated pure mathematics. Does this mean that the unique meanings of
mathematics have been lost? On the contrary, it means that we have added new,
more abstract language games, to those associated with the mathematical part of
natural language.
This notion of a range of language games encompassing the mathematical part of
natural language allows a possible objection to be addressed. This concerns the claim
that since the basis of mathematical and logical knowledge is inherent in natural
language usage, then all of mathematical knowledge must be inherent in natural
language. But this is patently false, the only legitimate conclusion from these
premises about the sum of all mathematical knowledge is that its basis, and not the
whole itself, is inherent in language usage. Given this basis, more and more new
language games embodying mathematical meaning and knowledge can be (and are)
developed, without necessitating the corresponding enlargement of the linguistic
basis. For specialist formal and informal mathematical discourses may be enlarged,
resting on the same natural language basis.
The mathematical knowledge embedded in language usage provides a basis for
informal (and ultimately formal) mathematical knowledge. The meanings and rules
embodied in this knowledge can be described in terms of a series of language games.
These games provide the basis of further, more refined language games, which abstract,
refine, extend and develop their rules and meanings. Thus a loose hierarchy can be
posited, with the mathematical knowledge embedded in natural language making up
the base. On this is built a series of language games embodying informal and ultimately
formal mathematical knowledge. In the upper reaches of the hierarchy informal
mathematical systems become formalized into axiomatized theories. At this level the
rules of the games or systems become almost completely explicit. In this way the
knowledge of mathematics implicit in language provides a basis for all mathematical
knowledge.Truths embodied in and vouchsafed by linguistic usage are reflected up the
hierarchy to justify elementary assumptions adopted in mathematics. The same is true
for the assumptions and rules of logic. In the next section we will explore the role of
such assumptions, in the justification of mathematical knowledge.
In this section we have seen that linguistic convention and usage provides
mathematical knowledge with its secure foundations. Similarly, it provides grounds
for change in mathematics, as linguistic conventions and usage develop over time.
Mathematics, like every other realm of knowledge depends essentially on tacit
linguistic assumptions. Fallibilism forces us to admit their presence, as well as their
changing nature, over the passage of time.

B. The Conventionalist Warrant for Mathematical Knowledge

According to the social constructivist view, mathematical knowledge is fallible, in

that it is open to revision, and it is objective in that it is socially accepted and
publicly available for scrutiny. Valid mathematical knowledge is knowledge which is

53
The Philosophy of Mathematics Education

accepted on the basis of there being a public justification of the knowledge (a

published proof) which has survived (or been reformulated in the light of) public
scrutiny and criticism.5
The justification for a particular item of mathematical knowledge consists of a
valid informal or formal deductive proof. The analysis of a proof justifying an item of
knowledge must consider two aspects: the explicit starting assumptions, and the
sequence of steps leading from them to the conclusion. We consider first the starting
assumptions. These consist of (i) hypothetical statements or assumed axioms (e.g., the
continuum hypothesis), (ii) definitions (e.g., Peano’s inductive definition of ‘+’), (iii)
previously accepted mathematical knowledge, typically previously established
theorems, (iv) the ‘truths’ of accepted informal mathematical knowledge, which are
embedded in mathematical language, or their formalization (e.g., the Peano Axioms),
or (v) logical axioms. Of these types, (iii) is reducible to others (via proofs). The
remaining assumptions are either hypothetical assumptions (case i and case iv in
some instances), or rest on the conventions and rules of mathematical language.
Definitions of type (ii) are conventions by fiat, they are simply laid down as such. The
remaining two types of assumptions are either informal mathematical rules, or their
formalization (case iv), or logical axioms (case v). The justification for these two types
of assumptions is conventionalist, and is offered below.
Secondly, a mathematical proof consists of a finite sequence of steps leading from
the initial assumptions of the proof, to its conclusion. The key feature of such a
sequence is the means of proceeding from one step to the next, that is the justification
for concluding a step from its predecessors. The justification for such a step consists of
(i) the use of a logical rule of inference (e.g. the rule Modus Ponens), (ii) the use of a
mathematical principle of inference (e.g. the Pigeon Hole Principle), (iii) the
introduction of a new assumption (this is like the cases treated in the preceding
paragraph), (iv) the claim that the step is justified by an elementary combination of the
previous types of steps, and (v) by analogy with a similar proof given elsewhere.
Assuming that any claims under cases (iv) and (v) are verified, this leaves (i) and (ii) to
consider. These depend on the assumption of a rule or principle of mathematics or
logic. These will either be reducible to simple assumptions (as is the Pigeon Hole
Principle) or are themselves basic mathematical and logical rules. Such rules do not in
principle differ from the basic mathematical and logical assumptions discussed above.
In fact assumptions and rules are inter-translatable, so rules can be replaced by
assumptions in sentence form, although at least one logical rule of inference is
necessary. If, for simplicity, we thus dispose of the mathematical rules (replacing them
by assumptions in prepositional form), the assumptions on which the inferential steps
in a mathematical proof are based reduce simply to some basic logical rules of
inference. These rules of inference will be justified by the conventionalist argument.
We have seen that the warrant for asserting mathematical knowledge consists of
a mathematical proof (of one step only, in the case of a basic assumption). The basis
for such a warrant resides in a number of basic assumptions (excluding genuinely
hypothetical axioms, such as ‘V=L’ of Godel, 1940, or the axioms of Tensor
Theory). These basic assumptions consist of informal mathematical ‘truths’, and

54
 Social Constructivism

logical axioms and rules of inference. These were justified, in the previous section,
as linguistic conventions, which are a part of the rules of meaning and use inherent
in our grasp of language. Thus, it is argued, the whole body of mathematical
knowledge is warranted by proofs, whose basis and security rests on linguistic
knowledge and rules.6

C. The Objects of Mathematics

The objectivity of mathematical knowledge is social, based on the acceptance of

linguistic rules, which are necessary for communication as we know it. Social
acceptance also provides the basis for the independent existence of the objects of
mathematics. For embedded in the rules and truths of mathematics is the assumption,
the assertion even, that the concepts and objects of mathematics have an objective
existence.
In natural language, each set of language games may be regarded as a discourse,
including a set of linguistic entities, and rules and truths, together making up a naive
theory. Associated with such a discourse and its function is a semantic realm, a
universe of discourse. This is an informally delineated set of entities, with certain
properties and relationships as specified by the associated naive theory. Thus the
existence of a shared set of language games entails a realm of entities with an
existence independent of any individual. In particular, a mathematical theory or
discourse brings with it a commitment to the objective existence of a set of entities.

Classical mathematics, as the example of primes greater than a million

clearly illustrates, is up to its neck in commitments to an ontology of
abstract entities. Thus it is that the great medieval controversy over
universals has flared up anew in the modern philosophy of mathematics.
The issue is clearer now than of old, because we now have a more explicit
standard whereby to decide what ontology a given theory or form of
discourse is committed to: a theory is committed to those, and only those
entities to which the bound variables of the theory must be capable of
referring in order that the affirmations made in the theory be true.
(Quine, 1948, pages 13–14)

Objective mathematical definitions and truths specify the rules and properties
determining the objects of mathematics. This confers on them as much of an
objective existence as that of any social concepts. Just as universal linguistic terms,
such as ‘noun’, ‘sentence’, or ‘translation’ have a social existence, so too the terms
and objects of mathematics have the properties of autonomous, self-subsistent
objects. The objects of mathematics inherit a fixity (i.e., a stability of definition) from
the objectivity of mathematical knowledge, entailing in turn their own permanency
and objective existence. Their objectivity is the ontological commitment that
inevitably accompanies the acceptance of certain forms of discourse.

55
The Philosophy of Mathematics Education

Of course, this is not the end of the matter, for discourses commit us to all
manner of entities, from chairs, tables and cars, to ghosts, angels and souls. It cannot
be claimed that these are all on a par. But likewise, the objects of mathematics vary
from the relatively concrete, embedded in the natural language descriptions of the
sensible world, to the abstract theoretical entities of mathematics e.g., the least
inaccessible cardinal (Jech, 1971), many steps removed from this basis. However, most
of the objects of mathematics have more reality than the objects in some discourses,
such as the fantasy creatures of Tolkien’s (1954) Middle Earth. For they are the result
of social negotiation, not just the product of a single individual’s imagination.
Many of the elementary terms and concepts of mathematics have concrete
applications and examples in the world. For they are part of a language developed to
describe the physical (and social) world. Thus such terms as ‘one’, ‘two’, ‘ten’, ‘line’,
‘angle’, ‘square’, ‘triangle’, and so on, describe properties of objects or sets of objects,
in the world. Other terms such as ‘add’, ‘subtract’, ‘divide’, ‘measure’, ‘rotate’, and so
on, describe actions that can be performed on concrete objects. The denotations of
these terms, gain ‘objecthood’ by their concrete applications in objective reality.7 Yet
more terms, such as ‘equation’, ‘identity’, and ‘inequality’ refer to linguistic entities.
Each of these sets of terms describes aspects of objective reality, whether external or
linguistic, and thus provides a concrete basis for a ‘mathematical reality’. On this basis
further mathematical ter ms, such as ‘number’, ‘operation’, ‘shape’, and
‘transformation’, are defined, one level removed from concrete reference. At higher
and higher levels, further mathematical terms, increasingly abstract, apply to those
below them. Thus through such an hierarchy virtually all mathematical terms have
definitions and denote objects at lower levels. These denotations behave exactly like
objectively existing, autonomous objects. Thus the objects of mathematics are
objective in the same way as the knowledge of mathematics. They are public
linguistic objects, some concrete but most abstract.
An example is provided by algorithms. These denote precisely specified sequences
of actions, procedures which are as concrete as the terms they operate on. They
establish connections between the objects they operate on, and their products. They
are a part of the rich structure that interconnects, and thus helps to implicitly define,
the terms, and hence the objects of mathematics.
This account may seem to fall short of providing all that is required for objective
existence. However, the analogy between the above conceptual hierarchy of mathematics
and an empirical scientific theory should be noted. For although defined analogously,
the theoretical entities of theoretical science are understood to have an autonomous
existence. Hempel (1952) likens a scientific theory to a net. Knots represent terms, and
threads represent sentences of the theory (definitions, theoretical statements, or
interpretative links) which both bind together the net and anchor it to the bedrock of
observation. The theoretical terms of science, such as ‘neutrino’, ‘gravitational field’,
‘quark’, ‘strangeness’ and ‘big bang’ correspond to the abstract entities of mathematics,
in the analogy. The difference being that only concrete mathematical terms have an
empirical reference, whereas the theoretical terms of science are taken to denote the
physical entities whose empirical existence is posited by the current theory.

56
 Social Constructivism

Both these types of entities exist in the ‘world 3’ of objective knowledge.

Whether all such objects, especially those of mathematics ‘really’ exist or not, is the
fundamental question of ontology, and is the subject of the traditional debate
between realism and nominalism (see, for example, Putnam 1972). The social
constructivist view is that the objects of mathematics are social constructs or cultural
artefacts. They exist objectively in the sense that they are public and there is
intersubjective agreeement about their properties and existence. The social
constructivist view is that mathematical entities have no more permanent and
enduring self-subsistence than any other universal concepts such as truth, beauty,
justice, good, evil, or even such obvious constructs as ‘money’ or ‘value’. Thus if all
humans and their products ceased to exist, then so too would the concepts of truth,
money and the objects of mathematics. Social constructivism therefore involves the
rejection of platonism.8

D. The Genesis of Mathematical Knowledge

In accepting that mathematics is a social construct, it is implied that objective

mathematical knowledge is the product of human beings. To defend this thesis, we
need to be able to account for the addition of the mathematical creations of
individuals (or groups) to the body of accepted mathematical knowledge. But the
growth of mathematical knowledge is not exclusively incremental. Thus we must also
account for the way that as a result of new contributions, the body of existing
mathematical knowledge develops and changes.
Although he does not explicitly address both of these issues, we have seen that
Lakatos’ quasi-empiricism offers a potentially fruitful account of the genesis of
mathematical knowledge, and we shall build on his account.
According to the usage adopted, the mathematical thought of an individual is
subjective thought. In order for it to become objective thought it must be linguistically
represented, typically in written form. The key act which transforms this published
subjective thought into objective thought is social acceptance, following critical public
scrutiny. Then it can be said to be a contribution to mathematical knowledge, even if,
like Fermat’s famous conjecture written in his copy of Diophantus, it is not scrutinized
in the author’s lifetime. Objectivity is conferred on the mathematical thought through
its social acceptance, following publication. Here no restriction to written publication
is intended.Thus communicating mathematical thought via a lecture to colleagues also
constitutes publication, and can equally be a contribution to objective thought,
providing it is socially accepted.
A crucial feature in the genesis of mathematical knowledge is transformation from
publicly represented (subjective) knowledge in mathematics to objective, that is
socially accepted mathematical knowledge. This transformation depends on surviving
a process of public scrutiny and criticism. During this process, which is Lakatos’
autonomous logic of mathematical discovery, objective criteria play an essential part.
They are used to judge the correctness of inferences, the consistency of assumptions,

57
The Philosophy of Mathematics Education

the consequences of definitions, the validity of formalizations in expressing informal

notions, and so on. The shared criteria used in such a process of critical scrutiny
include the ideas of logic and correct inference and other basic methodological
notions and procedures, which depend to a large extent on shared mathematical and
logical knowledge.
The fact that objective criteria exist, however, does not mean that all criticism is
rational.9 However, this account represents a discussion of the philosophical features
of objective knowledge growth, and not the empirical factors that may arise in
practice. The account is based on that of Lakatos, although elaborated in several
respects. The original insight as to the essential role of public criticism in the growth
of knowledge, as Lakatos acknowledges, is that of Popper (1959).

The varieties of mathematical creation

What has not yet been accounted for is how some additions to knowledge are
incremental, whereas others result in a restructuring or reformulation of existing
knowledge. Like science, mathematics is acknowledged to be hypothetico-deductive.
Thus mathematicians work within an established mathematical theory. Much of this
work consists of the development of new consequences of existing aspects of the
theory, or the application of existing methods from within the theory to a range of
problems. When fruitful, such work results in incremental additions to the body of
mathematical knowledge.
Mathematicians also utilize the concepts and methods from one mathematical
theory in another, or manage to establish links between two previously separate
theories. Such work causes new structural links to be formed between separate parts
of mathematics. This constitutes a restructuring of mathematics, which can be quite
considerable if under the influence of the new links the two theories are reworked,
reformulated and drawn closer together. Finally, work in some theories, often
directed at the solution of some problem, may generate a new mathematical theory.
This may simply be an additional theory, or it may subsume previous theories in a
larger, more general theory. The move towards increasing abstraction and generality,
such as in this case, is a major factor in the restructuring of mathematical knowledge.
For increasingly general theories are more widely applicable, and several more
specialized, pre-existing theories may fall within their more general structual
patterns. An example is provided by Cantor’s theory of sets, which initially seemed
very specialized and recondite. Since its introduction, because of its wide generality,
it has come to encompass most other mathematical theories, and provide them with
reformulated and unified foundations.
This account of the genesis of mathematical knowledge provides an idealized
account of the mechanism underlying the historical development of mathematics. At
any given time, the existing body of objective mathematical knowledge is being
reformulated and developed as the result of new contributions, which may either
restructure existing knowledge, or simply add to it.

58
 Social Constructivism

E. The Applicability of Mathematics

For adequacy, social constructivism must account for the ‘unreasonable effectiveness
of mathematics in science’ (Wigner, 1960). It is able to account for the applicability
of mathematics on two grounds: (1) mathematics is founded on our empirical natural
language; and (2) the quasi-empiricism of mathematics means that it is not so very
different from empirical science anyway.
First of all, we have already argued that mathematical knowledge rests on the rules
and conventions of natural language. We have seen that there is a rich mathematical
vocabulary directly applicable to the world of our experience, and natural language
includes rules and conventions on how to apply these terms. Many of these belong
both to mathematics and to science, and allow us to use classification and
quantification in describing events and objects in the world (via conjectured
explanations). Everyday and scientific uses of natural language are a key feature of its
role, and in such uses the embedded mathematical concepts play an essential part.
Thus the linguistic basis for mathematics, as well as the other functions language
performs for mathematics, provides interpretative links with real world phenomena.
In this way its linguistic roots provide mathematics with applications.
Secondly, we have accepted Lakatos’ argument that mathematics is a quasi-
empirical hypothetico-deductive system. In acknowledging this, we are admitting a
much closer link between mathematics and empirical science than the traditional
absolutist philosphies allow. This is reflected in the close resemblance between
mathematical theories and scientific theories, which we have observed. Both types of
theory contain relatively concrete exemplifiable or observational terms, and
theoretical terms, interconnected by a ‘net’ of links and relationships. Quine (1960)
even sees them both as interwoven in a single, connected fabric. In view of this
striking structural analogy, it is not surprising that some of the general structures and
methods of mathematics are imported into physical theories. Indeed, much of
empirical theory is wholly expressed in the language of mathematics. Likewise, it is
not surprising that many scientific problems, formulated in the language of
mathematics, become the stimulus for mathematical creation. The need for ever
better models of the world, as science advances, provides mathematics growth points
for development. The consequent cross-fertilization and interpenetration of science
and mathematics is a fact that the absolutist philosophical separation between a priori
and empirical knowledge has masked and mystified. In its origins and throughout its
development, mathematics has maintained contact with the physical world by
modelling it, often in conjunction with empirical science. In addition, the forces that
lead to generalization and integration in mathematical knowledge, described above,
ensure that the contact and influence of the empirical world on mathematics are not
merely marginal. The applicable theories within mathematics are subsumed into
more general theories, as mathematics is restructured and remade. By these means,
the applicability of mathematics extends to the central abstract theories of
mathematics, and not merely those on its periphery.
Overall, the applicability of mathematical knowledge is sustained by the close

59
The Philosophy of Mathematics Education

relationships between mathematics and science both as bodies of knowledge and as

fields of enquiry, sharing methods and problems. Mathematics and science are both
social constructs, and like all human knowledge they are connected by a shared
function, the explanation of human experience in the context of a physical (and a
social) world.

4. A Critical Examination of the Proposals

The social constructivist account of objective mathematical knowledge potentially

satisfies the adequacy criteria for the philosophy of mathematics, since it treats
knowledge, ontology, applications and practice. However, a number of telling
criticisms can be directed at the account, and these need to be anticipated and
answered.

A. Mathematics is arbitrary and relative

First of all, there is the problem of the relativism of mathematical knowledge and
truth. If, as is claimed, mathematical truth rests on social conventions, then it is both
arbitrary and relative. It is arbitrary since it rests on arbitrary beliefs, practices and
conventions. It is relative since it rests on the beliefs of one group of humans.
Consequently there is no need for other groups of humans, let alone other intelligent
creatures in the universe, to accept the necessity of mathematical knowledge, which
only holds relative to a particular culture at a particular period.
To answer this, I wish to question two presuppositions. These are first, the notion
that linguistic and mathematical conventions are arbitrary and mutable; and second,
the misconception that mathematical and logical knowledge are necessary and
immutable (although this was largely refuted in Chapter 1).

Arbitrariness
The arbitrariness of mathematics, in the account given, arises from the fact that
mathematical knowledge is founded on linguistic conventions and rules. There is no
necessity behind these rules, and they could have developed differently. This is
undeniable. But the fact remains that language operates within the tight constraints
imposed by physical reality and interpersonal communication. The conventions of
language can be formulated differently, but the purpose of language in providing a
functioning social description of the world remains constant. The shared rules and
conventions of language are a part of a naive empirical theory of the physical world
and social life. Thus, although every symbol in natural language is arbitrary, as the
choice of any sign must be, the relationship between reality and the overall model of
it, provided by language is not.

60
 Social Constructivism

Although such modelling may not be the whole function of language, it provides
crucial external anchors for language that keep it functioning viably. To maintain this
viability, some of the logical rules of language appear to be necessary. For example,
White (1982) argues that the principle of contradiction is necessary for any assertion
to be made by means of language. For without the principle in operation there would
be simultaneous assertion and denial. But assertions are ruled out by denial. Now in
some language uses the principle might be relaxed for certain purposes, such as
describing a deity. However it is difficult to argue that a language could function
viably without some such rules. Thus although much of the formulation of language
rules and conventions may be arbitrary in detail, the need for viability reduces much
of the scope of the arbitrariness of language to inessential details. For example, the
differences between natural languages indicate areas of arbitrariness in their formulation.

Relativism
The definition of objectivity adopted opens social constructivism to the charge of
relativism. That is, it is just the knowledge of a particular group at a particular time.
This is true, but there are two mitigating circumstances which remove much of the
force of this criticism. As we have seen, mathematics through language must provide
a viable description of aspects of empirical and social reality. Thus the relativism of
mathematics is reduced by its anchoring via these applications. In other words, both
mathematics and language are highly constrained by the need to describe, quantify
and predict events in the physical and human worlds effectively. In addition,
mathematics is constrained by its growth and development through the inner logic
of conjectures, proofs and refutations, described above. Thus mathematics not only
has its feet rooted in reality, but its upper parts have to survive the rigorous public
procedures of justification and criticism, based on the thorough-going application of
a small number of principles. Thus mathematical knowledge is relativistic knowledge
in that its objectivity is based in social agreement. But its relativism does not make it
equal or interchangeable with other social belief systems, unless they satisfy these
same two criteria.
Critics of the possibility of relativism in mathematics claim that alternative
mathematics or logic is inconceivable, which confirms the necessity and unique
status of mathematics and logic. This raises the question: what would an alternative
mathematics (or logic) look like? Bloor (1976) asks this question, and illustrates his
answer with alternative notions of number, calculus, and so on from the history of
mathematics. A critic’s reply to this might be that although our conceptions have
evolved and changed throughout history, they were just steps on the path to the
necessary modern notions. If the questionable teleological aspect of this claim is
ignored, then it is necessary to exhibit simultaneous competing alternatives for
mathematics, to answer the criticism. However, a further question needs to be asked:
how different does an alternative mathematics need to be to count as alternative (and
hence to refute the uniqueness claim)?

61
The Philosophy of Mathematics Education

The answer I propose is that an alternative mathematics (or logic) should be based
on concepts defined differently, with different means of establishing truths, and
which results in a very different body of truths. Furthermore, if the alternative is to
be taken seriously, there should be a respectable body of mathematicians who adhere
to the alternative, and who reject standard mathematics. This, in my view, is an
adequately strong characterization of an alternative form of academic (as opposed to
culturally embedded) mathematics. Strong as it is, it is not difficult to satisfy this
requirement. Intuitionist mathematics fits the requirement perfectly. Intuitionist
concepts, from the logical connectives ‘not’, ‘there exists’, to the concepts of ‘set’,
‘spread’ and the ‘continuum’ are quite different in meaning and in logical and
mathematical outcomes, from the corresponding classical concepts, where they exist.
Intuitionist axioms and principles of proof are also different, with the rejection of the
classical Law of the Excluded Middle, ‘-P↔P’, and ‘-(x)-A ↔(Ex)A’. Intuitionist
mathematics has its own body of truths including the countability of the continuum,
the Fan Theorem and the Bar Theorem, which do not appear in classical
mathematics, as well as rejecting the bulk of classical mathematics. Finally, since the
time of Brouwer, intuitionism has always had a cadre of respected adherent
mathematicians, committed to intuitionism (or constructivism) and who reject
classical mathematics (e.g. A.Heyting, H.Weyl, E.Bishop, A.Troelstra). Thus there is an
alternative mathematics, which includes an alternative logic.
This century there has been an explosion of other alternative or ‘deviant’ logics,
including many-valued logics, Boolean-valued logic, modal logic, deontic logic and
quantum logic. These show that further alternatives to logic are not only possible, but
exist. (However these deviant logics may not satisfy the last criterion given above, i.e.
the adherence of a group of mathematicians, who reject classical logic).
The example of intuitionism shows that classical mathematics is neither necessary
nor unique, for an alternative is not only possible, but it exists. It also shows that
there are alternatives to classical logic. The example also demonstrates the relativism
of mathematics, subject to the constraints discussed above, since there are two
mathematical communities (classical and intuitionist) with their own, opposing
notions and standards of mathematical truth and proof. In previous chapters the
absolutist view of mathematics as a body of immutable and necessary truth was
refuted, and a fallibilist view argued in its place. This weakened the claim of necessity
for mathematics. This has now been supplemented with an example of a genuine
alternative, dispelling any possible claims of uniqueness or necessity for mathematics.

B. Social Constructivism Fails to Specify any Social Group

The account of social constructivism given refers to ‘social acceptance’, ‘social

construction’ and ‘objectivity as social’. However it fails to specify in any way which
social groups are involved, and for the term social to have meaning, it must refer to
a specific group. There are also hidden secondary problems such as how does one tell
when something is accepted by the mathematical community? What happens when

62
 Social Constructivism

there is conflict in this community? Does this mean that a new piece of mathematics
can hover on the boundary between subjective and objective knowledge?
To answer the main point first: it would be inappropriate in a philosophical account
to specify any social groups or social dynamics, even as they impinge upon the
acceptance of objective knowledge. For this is the business of history and sociology,
and in particular, the history of mathematics and the sociology of its knowledge. The
claim that there is a social mechanism involved in objectivity and in the acceptance
of mathematical knowledge, and a conceptual analysis and elaboration of it remains
within the province of philosophy. The importation of concepts from history and
sociology to develop this theory, valuable as this might be, takes the discussion
outside the philosophy of mathematics. Thus this is not a valid criticism.
The secondary criticisms do hold some problems for social constructivism. If
there is simultaneous social acceptance of different sets of mathematical knowledge,
as was explored in section A above, then they both constitute objective mathematical
knowledge.
The transition of mathematical knowledge from subjective to objective
knowledge may benefit from further clarification. It needs to be made clear that
there is an intermediate state, which is neither. Subjective mathematical knowledge
resides in the mind of an individual, possibly supported by external representation.
For individuals developing subjective knowledge often do so with the aid of visual,
oral or other representations. Already such representations mean that there is a public
aspect supporting the individual’s subjective knowledge. When fully represented in
the public domain, it is no longer subjective knowledge as such, although the
originating individual may have corresponding subjective knowledge. Public
representations of knowledge are just that. They are not subjective knowledge, and
need not be (or represent, to be precise) objective knowledge either. However, they
have the potential to lead to the latter, when they are socially accepted.
Strictly speaking, public representations of knowledge are not knowledge at all,
for they consist only of symbols, and meanings and assertions have to be projected
into them by understanding subjects. Whereas knowledge is meaningful. This is
consistent with the view adopted in communications theory, that signals have to be
coded, transmitted and then decoded. During the transmission phase, that is when
coded, signals have no meaning. This has to be constructed during decoding.
It is convenient to adopt the current (but strictly speaking false) usage of
identifying public representations of objective knowledge (coded signals) with the
knowledge itself, and speaking as if the representation embodied information and
meaning. Such an attribution of meaning only works if it is assumed that the
appropriate community share the decoding knowledge. In the case of mathematical
knowledge this consists of knowledge of natural language and additional knowledge
of mathematics.
These then are some to the essential presuppositions about the social groups on
which social constructivism depends.

63
The Philosophy of Mathematics Education

C. Social Constructivism Assumes a Unique Natural Language

Social constructivism employs a conventionalist justification for mathematical

knowledge. This assumes that mathematical knowledge rests on a unique natural
language, contradicting the fact that over 700 different natural languages are known,
many with very different bases from English.
Whilst it can be said that mathematical concepts and truths do depend on certain
structural features of English, these are found equally in European and some other
languages, but not necessarily in all natural languages. This has two major
consequences, neither of which is critical for social constructivism. First, if
mathematics were based on languages with significantly different logic and structural
features, then an alternative (i.e., different) mathematics could result. This is not a
problem for social constructivism. Second, native language speakers whose language
differs significantly from English, French, etc., in logic and structural features either
have to acquire a second language, or restructure their understanding of their own, in
order to learn academic Western mathematics. This again seems plausible, and in fact
there is some evidence to support this.10 In fact such evidence of cultural relativism
strengthens rather than weakens the case in favour of social constructivism.

D. Previously Raised Objections

1. Social acceptance is not the same as objectivity.

An account has been given of objective mathematical knowledge, but objectivity has
been re-interpreted to mean socially accepted, in the manner of Bloor (1984). Thus
it is true to say that objectivity (understood socially) is used to mean something
different. In defence of the social interpretation the following has been argued. First,
the important properties of objectivity, such as impersonality and verifiability, are
preserved. Second, objective existence in mathematics means consistently postulable.
The immense ontological consequences of this definition for mathematics distort the
meaning of ‘objectivity’ far beyond the sense of ‘existing like an object’. Third, the
social interpretation uniquely provides an explanation of the nature of objectivity in
mathematics.

2. Social constructivism attends insufficiently to the warranting of mathematical knowledge.

It is true that the account given focuses on the genesis of mathematical knowledge,
but it does not neglect to account for the justification of mathematical knowledge,
although in doing so it challenges absolutist accounts. Mathematical knowledge is
justified as hypothetico-deductive knowledge, which, in the case of derived
knowledge, involves proofs. Some elementary truths and the basis of logic and
proof are justified in ter ms of natural language, using the arguments of
conventionalism. A problem with this latter justification is that natural language

64
 Social Constructivism

does not literally contain all the basic truths and rules of mathematics and logic.
Rather it embodies the basic meanings, rules and conventions, which in refined and
elaborated form, provide the elementary truths and rules of mathematics and logic.
The account offered is superior in scope to that of traditional philosophies of
mathematics, because it provides an objective basis, warranting these elementary
assumptions. At best, other philosophies offer intuition (intuitionism, formalism,
platonism) or induction (empiricism), for these assumptions, if they offer any basis
at all.

3. Social constructivism conflates the contexts of discovery and justification, and commits the
error of psychologism.
By challenging the widespread assumption that the business of philosophy is only
with the context of justification, and not that of discovery, social constructivism may
seem to lay itself open to this charge. The account given acknowledges the
importance of these concepts, and distinguishes carefully between the two contexts,
as well as between the different proper concerns of philosophy, history, psychology
and sociology. However it is argued that on adequacy grounds the philosophy of
mathematics must account for the development and genesis of mathematical
knowledge, albeit from a philosophical perspective, as is accepted analogously in the
philosophy of science. It is also argued that subjective knowledge is a legitimate area
of philosophical enquiry, and need not lead to psychologism. Subjective thought and
knowledge must be included in a social constructivist account because it is the fount
of new mathematical knowledge. Naturally it must be treated philosophically, and
not psychologically, to avoid psychologism.

Notes

1 Other authors refer to mathematics as a social construct, notably Sal Restive (1985, 1988), in
developing a sociology of mathematical knowledge. Although approaching mathematics from
another perspective, he offers a range of insights compatible with the philosophy proposed here. This
is considered further in chapter 5.
2 Ideas from constructivist epistemology and learning theory, due to Glasersfeld, Piaget and others,
have also contributed to social constructivism.
3 Knowledge reproducing cycles occur also in sociology and education, but are concerned with the
genesis and reproduction of knowledge, not its justification (see Chapters 11 and 12). Such a cycle
is relatively novel in philosophical studies of mathematics, because it treats the genesis and social
origins of knowledge, as well as its justification. The approach adopted may be seen as part of a new
naturalistic approach to the philosophy of mathematics, typified by Kitcher (1984), and other
authors.
4 ‘Twentieth-century epistemology has been characterized by an attitude of explicit distaste for
theories of knowledge which describe the psychological capacities and activities of the subject. This
attitude has fostered an apsychologistic approach to knowledge…present in the writings of Russell,
Moore, Ayer, C.I.Lewis, R.Chisholm, R.Firth, W.Sellars and K.Lehrer, and is presupposed by the
discussions of science offered by Carnap, Hempel and Nagel.’ (Kitcher, 1984, page 14)
5 Notice the analogy with the ‘replicability’ criterion for experimentation in science, which demands

65
The Philosophy of Mathematics Education

that results, for acceptance, should not be peculiar to a unique scientist, but should be replicable.
Likewise a mathematical proof not only has to be surveyable to others, but this survey must result
each time in acceptance.
6 One possible objection to this account is that there seems to be a gap between the implicit ‘truths’
embedded in natural language, and the more abstract and sophisticated logical and mathematical
assumptions required to warrant mathematical knowledge. The answer to this is that linguistic
competence is not defined by a static level of performance and knowledge. Linguistic competence
presupposes competence in performing certain functions in certain social situations. The mastery of
these different linguistic functions, which representing the mastery of different ‘language games’ in
different social contexts, represents different types of linguistic competence. This range of contexts
brings with it a range of ever sharper linguistic conventions and rules, and those interested in
mathematical knowledge will of necessity have mastered a range of sophisticated mathematical and
logical language games.
In addition, the relatively elementary notions of rationality (implication, and contradiction)
underpin the more refined notions of logic. This ensures coherence and means that the more
refined notions are extrapolations of, and not discontinuous with, the simpler notions. Thus I
am claiming that the ‘gap’ is self closing. Those with knowledge enough to participate in the
language games of warranting mathematical knowledge will have extended their linguistic rules
and truths to the required point. Should they fall short, they will have their usage, and rules,
extended.
7 By objective reality I am referring to the socially agreed features of the external world. I
acknowledge the basic ontological assumption implicit in using this phrase, that is the assumption
of the existence of a physical world (Popper’ s world 1). I also acknowledge the existence of
human beings, as the basis for a social view of objective knowledge. However, I do not concede
that this commits me to ontological assumptions about the particular, conventionally labelled
objects in the world, beyond the fact that there is social agreement on their existence and
objectivity.
8 In case it seems a weakness of the social constructivist view that it denies the absolute existence of
mathematical objects, it is worth noting some of consequences of the mathematical notion of
existence. The criterion for existence in mathematics is consistency. If a mathematical theory is
consistent, then there is a set of objects (a model) providing denotations for all its terms and
satisfying all the conditions of the theory. That is, these objects have mathematical existence. Thus, for
example, provided the theories of Peano Arithmetic, the Real Numbers and Zermelo-Fraenkel (ZF)
Set Theory are consistent (which is accepted), there are models satisfying them. Thus the Natural
Numbers, Real Numbers and the universe of sets can be said to exist, which already allows for an
unbelievable richness of entities. But worse is to come. The Generalized Lowenheim-Skolem
Theorem (Bell and Slomson, 1971) establishes that these theories have models of every infinite
cardinality. Thus countable models of both Real Number Theory and ZF set theory exist, as do
models of the natural numbers of every (infinite) size. These all exist in world 3. Social
constructivism denies the existence this unpredictable and undreamed of multitude of mathematical
entities. In contrast, the outcome of the platonist view of existence in mathematics would be to
populate the universe far more densely than merely putting infinitely many angels to dance on the
head of every pin. The question is, what does it mean to say that the objects in the platonist
ontology actually exist? Surely this is using the term ‘exist’ in a novel way. The social constructivist
response to these ontological problems is basically to adopt the traditional conceptualist solution, but
in a new social guise (Quine, 1948).
9 Russell’s (1902) criticism of Frege’s system was rational, i.e. logical, being based on logical features
of the system (namely its inconsistency). Kronecker’s criticism of Cantor’s set theory evidently was
not purely rational or logical, since it had strongly moral and religious undertones.
The Kuhn-Popper debate in the philosophy of science hinged on the issue of rational
versus irrational criticism of scientific theories, which is analogous to the point made here
concerning mathematics. Popper’s position is prescriptive, he posits falsification as a rational
criterion for the rejection of a scientific theory. Kuhn, on the other hand, proposes a more
descriptive philosophy of science, which while treating the growth of objective knowledge
acknowledges that rational features are neither necessary nor sufficient to account for theory
acceptance or rejection.

66
 Social Constructivism

10 Alan Bishop has presented evidence that in some of the 350 languages to be found in Papua
and New Guinea the conceptual structure is markedly different from that presupposed by
elementary mathematics. For example in one language the same word is used to denote ‘up’,
‘top’, ‘surface’ and ‘area’, indicating a conceptual structure very different from English and
comparable languages.

67
 4

Social Constructivism and

Subjective Knowledge

1. Prologue

This chapter faces a difficult task: that of showing the relationship between subjective
and objective knowledge of mathematics in social constructivism. The task is difficult
for a number of reasons. It skirts the edge of psychologism, and it needs to conjoin
two different languages, theories and modes of thought that apply to two different
realms, the subjective and the objective. Beyond this, the epistemology underpinning
social constructivism is quite slippery to grasp, since it is claimed that there is no
realm where a determinate entity ‘knowledge’ basks in tranquillity. Knowledge,
perhaps analogous to consciousness, is seen as an immensely complex and ultimately
irreducible process of humankind dependent on the contributions of a myriad of
centres of activity, but also transcending them. Science fiction authors (Stapledon,
1937) and mystical philosophers (Chardin, 1966) have groped for a vision of how the
consciousnesses of individual human beings can meld into a greater whole. But these
provide too simplistic a vision to account for knowledge and culture as dynamic,
cooperative dances uniting millions of thinking and acting but separate human
beings. The seduction of idealism is great: to say that knowledge exists somewhere in
an ultimate form, possibly growing and changing, but that all our representations of
knowledge are but imperfect reflections. The pull to view human knowledge
attempts as parts of a convergent sequence that tends to a limit in another realm, is
almost irresistible.
Once these simplifying myths are rejected, as they are by social constructivism,
there is the complex task of accounting for knowledge. It is social, but where is the
social? Is it a moving dance, a cloud of pirouetting butterflies, which when caught is
no more? Books do not contain knowledge, according to this account. They may
contain sequences of symbols, carefully and intentionally arranged, but they do not
contain meaning. This has to be created by the reader, although books may guide the
reader to create new meanings. This is subjective knowledge, the unique creation of
each individual. Yet by some miracle of interaction, the way human beings use this
knowlege in their transactions fits together.
The concept of the individual, the knowing subject, in Western thought is

68
 Social Constructivism and Subjective Knowledge

another problem. Since Locke, or earlier, the subject is a tabula rasa, and gradually
knowledge is inscribed on its blank page by experience and education. But the form
and content of mind cannot be separated like this, and there is no universal form of
knowledge that can be written in our minds. The view that follows is that
knowledge has to be created anew in the mind of every human being, and solely in
response to their active efforts to know. Consequently, objective knowledge is all the
while being born anew. Thus knowledge is more like a human body, with its every
cell being replaced cyclically, or like the river: never the same twice! This is why I
called the epistemology involved slippery!
These are some of the problems that the present chapter raises and tries to tackle.

2. The Genesis of Subjective Knowledge1

The fundamental problem to be accounted for in the growth of subjective

knowledge concerns the acquisition of knowledge of the external world. How can
an individual acquire knowledge of the external world by means of sensory organs
alone? The external world includes other human beings, so acquisition of knowledge
of the external world includes knowledge of human beings, their actions, and
ultimately their speech. Only when we have accounted for the acquisition of speech
can we begin to consider how the substantive structures of mathematics are acquired.
We begin, therefore, by inquiring how any subjective knowledge is acquired.

A. The Construction of Subjective Knowledge

How does the individual acquire knowledge of the external world? Human beings
have incoming sense impressions of the world, as well as being able to act physically
on the world, and thus in some measure are able to control aspects of the
environment. Clearly subjective knowledge is acquired on the basis of interaction
with the external world, both through incoming sense data and through direct
actions. What is also clear is that these interactions are necessary but not sufficient for
the acquisition of knowlege of the external world. For the sense data are particulars.
Whereas our knowledge is evidently general, since it includes general concepts
(universals), and it allows for anticipation and the prediction of regularities in our
experience. Therefore some further mechanism is required to account for the
generation of general knowledge of the world of our experience, on the basis of
particular items of information or experiences.
This is precisely the problem that the philosophy of science faces, but expressed at
the subjective level. Namely, how can we account for (and justify) theoretical
scientific knowledge on the basis of observations and experiment alone? The
solution proposed is the same. The development of subjective knowledge, like that of
science, is hypothetico-deductive. The answer proposed is that the minds of
individuals are active, conjecturing and predicting patterns in the flow of experience,

69
The Philosophy of Mathematics Education

and thus building theories of the nature of the world, although these may be
unconsciously made theories. These conjectures or theories serve as guides for action,
and when they prove inadequate, as inevitably they do, they are elaborated or
replaced by new theories that overcome the inadequacy or failing of the previous
theory. Thus our subjective knowledge of the external world consists of conjectures,
which are continually used, tested and replaced when falsified.
Thus the account of the formation of subjective knowledge is a recursive one.
Our knowledge of the world of our experience consists of private conjectures or
theories, which order the world of our experience. These theories are based on two
factors. First, our immediate experiences of the world, including interactions with it,
as perceived and filtered through our theories. Second, our previously existing
theories. Thus the formation of our subjective theories is recursive in that it depends
essentially on these theories, albeit in an earlier state.
This account mirrors that of Popper (1959), but at the level of subjective as
opposed to objective knowledge.2 However, it is clear that Popper intends his account
of science to apply only to objective knowledge, and furthermore, he has nothing to
say on the genesis of scientific theories. As a purely subjective view of knowledge, this
view is elaborated by Glasersfeld (1983, 1984, 1989) as ‘radical constructivism’.
‘The world we live in’ can be understood also as the world of our
experience, the world as we see, hear and feel it. This world does not consist
of ‘objective facts’ or ‘things-in-themselves’ but of such invariants and
constancies as we are able to compute on the basis of our individual
experience. To adopt this reading, however, is tantamount to adopting a
radically different scenario for the activity of knowing. From an explorer
who is condemned to seek ‘structural properties’ of an inaccessible reality,
the experiencing organism now turns into a builder of cognitive structures
intended to solve such problems as the organism perceives or conceives.
Fifty years ago, Piaget characterised this scenario as one could wish:
‘Intelligence organises the world by organising itself’ (Piaget, 1937). What
determines the value of the conceptual structures is their experimental
adequacy, their goodness of fit with experience, their viability as means for
the solving of problems, among which is, of course, the never-ending
problem of consistent organisation that we call understanding.
The world we live in, from the vantage point of this new perspective, is
always and necessarily the world as we conceptualize it. ‘Facts’, as Vico saw
long ago, are made by us and our way of experiencing, rather than given by
an independently existing objective world. But that does not mean that we
can make them as we like. They are viable facts as long as they do not clash
with experience, as long as they remain tenable in the sense that they
continue to do what we expect them to do.
(Glasersfeld, 1983, p. 50–51)
Constructivism is a theory of knowledge with roots in philosophy,
psychology and cybernetics. It asserts two main principles…(a) knowledge

70
 Social Constructivism and Subjective Knowledge

is not passively received but actively built up by the cognizing subject; (b)
the function of cognition is adaptive and serves the organization of the
experiential world, not the discovery of ontological reality.
(Glasersfeld, 1989, page 162)

This view accounts for the development of subjective knowledge of the external
world. It explains how an individual constructs subjective knowledge, notably a
theoretical model of a portion of the external world which fits that portion, and how
this knowledge or model develops, improving the fit. It does this without
presupposing that we construct true knowledge matching3 the given portion of the
world, which would contradict much modern thought, especially in the philosophy
of science. Thus the theory provides an account of how external reality serves as a
constraint in the construction of subjective knowledge, a constraint that ensures the
continued viability of the knowledge. What the theory does not yet do, is to account
for the possibility of communication and agreement between individuals. For the
sole constraint of fitting the external world does not of itself prevent individuals from
having wholly different, incompatible even, subjective models of the world.
Such differences would seem inescapable. However, this is not the case. Suitably
elaborated, the social constructivist view also provides an account of the development
of knowledge of the world of people and social interaction, and the acquisition of
language. The very mechanism which improves the fit of subjective knowledge with
the world also accounts for the fit with the social world, including patterns of
linguistic use and behaviour. Indeed, the experiential world of the cognizing subject
which Glasersfeld refers to, does not differentiate between physical or social reality.
Thus the generation and adaptation of personal theories on the basis of sense data and
interactions equally applies to the social world, as the following account shows.
Individuals, from the moment of birth, receive sense impressions from, and
interact with, the external and social worlds. They also formulate subjective theories
to account for, and hence guide, their interactions with these realms. These theories
are continually tested through interaction with the environment, animate and
inanimate. Part of this mental activity relates to other persons and speech. Heard
speech leads to theories concerning word (and sentence) meaning and use. As these
theories are conjectured, they are tested out through actions and utterances. The
patterns of responses of other individuals (chiefly the mother or guardian, initially)
lead to the correction of usage. This leads to the generation of an ever growing set of
personal rules of language use. These rules are part of a subjective theory, (or family
of theories) of language use. But the growth of this theory is not monotonic. The
correction of use leads to the abandonment of aspects of it, the adaptation of the
theory and hence the refinement of use. This subjective knowledge of language is
likely to be more procedural than prepositional knowledge. That is, it will be more
a matter of ‘knowing how’ than ‘knowing that’ (Ryle, 1949).
The acquisition of language involves the exchange of utterances with other
individuals in shared social and physical contexts. Such interaction provides
encounters with rule governed linguistic behaviour. In other words, it represents the

71
The Philosophy of Mathematics Education

confrontation with, and accommodation to, socially accepted or objective features of

language. The acquisition of linguistic competence results from a prolonged period
of social interaction. During this period, by dint of repeated utterances and
correction, individuals construct subjective theories or personal representations of
the rules and conventions underpinning shared language use. The viability of these
theories is a function of their mode of development. Quine refers to the ‘objective
pull’, which brings about adequate levels of agreement between individuals
utterances and behaviour:
Society, acting solely on overt manifestations, has been able to train the
individual to say the socially proper thing in response even to socially
undetectable stimulations.
(Quine, 1960, pages 5–6)
Halliday (1978) describes linguistic competency in terms of mastery of three inter-
locking systems, namely the forms, the meanings and the (social) functions of
language. Of these, language forms and functions are publicly manifested systems,
which thus lend themselves to correction and agreement. Whilst the system of
meanings is private, the other systems ensure that where they impact on public
behaviour, there is a pull towards agreement.
The uniformity that unites us in communication and belief is a uniformity
of resultant patterns overlying a chaotic subjective diversity of connections
between words and experience. Uniformity comes where it matters
socially...Different persons growing up in the same language are like
different bushes trimmed and trained to take the shape of identical
elephants. The anatomical details of twigs and branches will fulfil the
elephantine form differently from bush to bush, but the overall outward
results are alike.
(Quine, 1960, page 8)
What has been provided is an account of how individuals acquire (or rather
construct) subjective knowledge, including knowledge of language. The two key
features of the account are as follows. First of all, there is the active construction of
knowledge, typically concepts and hypotheses, on the basis of experiences and
previous knowledge. These provide a basis for understanding and serve the purpose
of guiding future actions. Secondly, there is the essential role played by experience
and interaction with the physical and social worlds, in both the physical actions and
speech modes. This experience constitutes the intended use of the knowledge, but it
provides the conflicts between intended and perceived outcomes which lead to the
restructuring of knowledge, to improve its fit with experience. The shaping effect of
experience, to use Quine’s metaphor, must not be underestimated. For this is where
the full impact of human culture occurs, and where the rules and conventions of
language use are constructed by individuals, with the extensive functional outcomes
manifested around us in human society.
Bauersfeld describes this theory as the triadic nature of human knowledge:

72
 Social Constructivism and Subjective Knowledge

the subjective structures of knowledge, therefore, are subjective

constructions functioning as viable models, which have been formed
through adaptations to the resistance of ‘the world’ and through
negotiations in social interactions’
(Grouws et al., 1988, page 39)

The theory has a number of implications for communication, for the representation
of information, and for the basis and location of objective knowledge. With regard to
communication, the theor y imposes severe limits on the possibility of
communicating meanings by linguistic or other means. Since the subjective
meanings of individuals are uniquely constructed (with certain constraints
accommodated in view of their genesis), it is clear that communication cannot be
correctly described as the transfer of meanings. Signals can be transmitted and
received, but it is impossible to match the meanings that the sender and recipient of
the signals attach to them, or even talk of such a match. However, the ways in which
linguistic competence is acquired, mean that a fit between sender and receiver
meanings can be achieved and sustained, as evidenced by satisfactory participation in
shared language games. This view of communication is fully consistent with the
Communication Theory of Shannon (cited in Glasersfeld, 1989).
However, there is something I want to call objective, which enters into
communication. This is not the informational content of messages, but the pre-
existing norms, rules and conventions of linguistic behaviour that every speaker meets
(in some form) when entering into a linguistic community. These, in Wittgenstein’s
(1953) term, are a ‘form of life’, the enacted rules of linguistic behaviour shared (at
least in approximation) by speakers. These rules, represent the constraints of the world
of interpersonal communication, which permit the possibility of a fit between
senders’ and receivers’ meanings. Such a fit will depend on the extent to which the
actors are drawn from communities which share the same norms of linguistic
competence, as well as on the success of the individuals in reconstructing these norms
for themselves.These norms or rules are objective, in the sense that they are social, and
transcend individuals. However, at any one time, they are located in the regularities of
the linguistic behaviour of the group, sustained by the subjective representations of
them, in the minds of the individual group members.
A further consequence of the view of subjective knowledge growth concerns the
extent to which meanings are inherent in symbolic representations of information,
such as a book or a mathematical proof. According to the view proposed, such
meanings are the constructions of the reader. (This view is in essence, Derrida’s
deconstructive approach to textual meanings; Anderson et al., 1986). The linguistic
rules, conventions and norms reconstructed by a reader during their acquisition of
language constrain the reader to a possible interpretation whose consequences fit
with those of other readers. In other words, there is no meaning per se in books and
proofs. The meanings have to be created by readers, or rather, constructed on the
basis of their existing subjective meanings. Within a given linguistic community, the
readers’ private meaning structures are constructed to fit the constraints of publicly

73
The Philosophy of Mathematics Education

manifested linguistic rules. Thus it is the fit between the readers’ subjective theories
of language, brought about by a common context of acquisition, including shared
constraints, rather than an inherent property of text that brings about a fit between
interpretations. However, the social agreement within a community as to how a
symbolism is to be decoded constrains individuals’ meaning constuctions, giving the
sense that there is informational content in the text itself.
This is consistent with the account given of objective knowledge in the previous
chapter. For it was stated that public representations of subjective knowledge are just
that. Knowledge, truth and meaning cannot be attributed to sets of marks or symbols.
Only the assignment of meanings to a set of marks, or a symbol system, which
ultimately has to be done by an individual, results in the knowledge or meaning of
a published document. As in communications theory, decoding is essential if meaning
is to be attributed to a set of broadcast codes.
The social constructivist account of subjective knowledge is also consistent with
the conventionalist account of the basis of mathematical, logical and linguistic
knowledge given in previous chapters. For according to the constructivist view, the
growth of the subjective knowledge of an individual is shaped by interactions with
others (and the world). This shaping takes place throughout a linguistic community,
so that the constraints accommodated by all of its individuals allow shared
participation in language games and activities. These constraints are the objective,
publicly manifested rules and conventions of language. On the basis of these
constraints individuals construct their own subjective rules and conventions of
language. It can be said that these ‘fit’ (but do not necessarily match) since they allow
for shared purposes and interchange which satisfy the participants to any degree of
refinement desired.
One problem that arises from the account of the constructivist epistemology that
has been given is that it seems to necessitate cumbersome circumlocutions.
Knowledge is no longer ‘acquired’, ‘learned’, or ‘transmitted’, but ‘constructed’ or
‘reconstructed’ as the creative subjective response of an individual to certain stimuli,
based on the individual’s pre-existing knowledge which has been shaped to
accommodate rules and constraints inferred (or rather induced) from interactions
with others. Whilst the latter account is the accurate one from the constructivist
viewpoint, it is convenient to retain the former usage on the understanding that it is
merely a façon de parler, and an abbreviation for the latter.
An analogy for such usage is provided by the language of analysis in mathematics.
To say that a function f(x) (defined on the reals) approaches infinity is acceptable,
provided this is understood to have the following more refined meaning (provided in
the nineteenth century). Namely, that for every real number r there is another s such
if x>s, then f(x)>r. This reformulation no longer says that the function literally
approaches infinity, but that for every finite value, there is some point such that
thereafter all values of the function exceed it. The two meanings expressed are quite
different, but the convention is adopted that the first denotes the second. The
rationale for this is that an abbreviated and historically prior mode of speech is
retained, which in all contexts can be replaced by a more precise definition.

74
 Social Constructivism and Subjective Knowledge

Likewise, we may retain the use of terminology of knowledge transmission in

situations where there is no danger of confusion, on the understanding that it has
constructivist meaning, which can be unpacked when needed.
In summary, it has been argued that: (a) subjective knowledge is not passively
received but actively built up by the cognizing subject, and that the function of
cognition is adaptive and serves the organization of the experiential world of the
individual (Glasersfeld, 1989). (b) This process accounts for subjective knowledge of
the world and language (including mathematics), (c) Objective constraints, both
physical and social, have a shaping effect on subjective knowledge, which allows for
a ‘fit’ between aspects of subjective knowledge and the external world, including
social and physical features, and other individuals’ knowledge, (d) Meanings can only
be attributed by individuals, and are not intrinsic to any symbolic system.

B. The Construction of Mathematical Knowledge

It has been argued that linguistic knowledge provides the foundation (genetic and
justificatory) for objective mathematical knowledge, both in defending the
conventionalist thesis, and subsequently as part of the social constructivist philosophy
of mathematics. What is proposed here is the parallel but distinct claim that linguistic
knowledge also provides the foundation, both genetic and justificatory, for the
subjective knowledge of mathematics. In a previous section we saw how social (i.e.
objective) rules of language, logic, etc., circumscribe the acceptance of published
mathematical creations, allowing them to become part of the body of objective
mathematical knowledge. Thus we were concerned with the subjective origins of
objective knowledge. In this section the focus is on the genesis of subjective
mathematical knowledge, and it will be argued that the origins of this knowledge lie
firmly rooted in linguistic knowledge and competence.
Mathematical knowledge begins, it can be said, with the acquisition of linguistic
knowledge. Natural language includes the basis of mathematics through its register of
elementary mathematical terms, through everyday knowledge of the uses and inter-
connections of these terms, and through the rules and conventions which provide
the foundation for logic and logical truth. Thus the foundation of mathematical
knowledge, both genetic and justificatory, is acquired with language. For both the
genetic basis of mathematical concepts and propositions, and the justificatory
foundation of prepositional mathematical knowledge, are found in this linguistic
knowledge. In addition, the structure of subjective mathematical knowledge,
particularly its conceptual structure, results from its acquisition through language.
One of the characteristics of mathematical knowledge is its stratified and
hierarchical nature, particularly among terms and concepts. This is a logical property
of mathematical knowledge, which is manifested both in public expositions of
objective mathematical knowledge and, as will be claimed here, in subjective
mathematical knowledge. We consider first the hierarchical nature of objective
mathematical knowledge.

75
The Philosophy of Mathematics Education

It is acknowledged that concepts and terms, both in science and mathematics, can
be divided into those that are defined and those taken as primitive and undefined, in
any theory (see, for example, Popper, 1979; Hempel, 1966; Barker, 1964). The defined
terms are defined using other terms. Ultimately, after a finite number of defining links,
chains of definition can be chased back to primitive terms, or else the definitions
would be based on, and lead to, an infinite regress4. On the basis of the division of terms
into primitive and defined, a simple inductive definition of the level of every term
within an hierarchical structure can be given. Assuming that each concept is named by
a term, this provides an hierarchy of both terms and concepts. Let the terms of level 1
be the primitive terms of the theory. Assuming that the terms of level n are defined, we
define the terms of level n+1 to be those whose definitions include terms of level n, but
none of any higher level (although terms of lower level may be included).This definition
unambiguously assigns each term of an objective mathematical theory to a level, and
hence determines an hierarchy of terms and concepts (relative to a given theory).5
In the domain of subjective knowledge, we can, at least theoretically, divide
concepts similarly, into primative observational concepts, and abstract concepts defined
in terms of other concepts. Given such a division, an hierarchical structure may be
imposed on the terms and concepts of a subjective mathematical theory precisely as
above. Indeed Skemp (1971) offers an analysis of this sort. He terms observational and
defined concepts primary and secondary concepts, respectively. He bases the notion of
conceptual hierarchy upon this distinction in much the same way as above, without
assigning numbers to levels. His proposals are based on a logical analysis of the nature
of concepts, and their relationships. Thus the notion of a conceptual hierarchy can be
utilized in a philosophical theory of subjective knowledge without introducing any
empirical conjecture concerning the nature of concepts.
To illustrate the hierarchical nature of subjective mathematical knowledge, consider
the following sample contents, which exemplify its linguistic origins. At the lowest
level of the hierarchy are basic terms with direct empirical applications, such as ‘line’,
‘triangle’, ‘cube’, ‘one’, and ‘nine’. At higher levels there are terms defined by means of
those at lower levels, such as ‘shape’, ‘number’, ‘addition’ and ‘collection’. At higher
levels still, there are yet more abstract concepts such as ‘function’, ‘set’, ‘number system’,
based on those at lower levels, and so on. In this way, the concepts of mathematics are
stratified into a hierarchy of many levels. Concepts on succeeding levels are defined
implicitly or explicitly in terms of those on lower levels. An implicit definition may
take the following form: numbers consist of ‘one’, ‘two’, ‘three’, and other objects with
the same properties as these. ‘Shape’ applies to circles, squares, triangles, and other
objects of similar type. Thus new concepts are defined in terms of the implicit
properties of a finite set of exemplars, whose membership implicitly includes
(explicitly includes, under the new concept) further exemplars of the properties.
It is not the intention to claim here that there is a uniquely defined hierarchy of
concepts in either objective or subjective mathematical knowledge. Nor is it claimed
than an individual will have but one conceptual hierarchy. Different individuals may
construct distinct hierarchies for themselves depending on their unique situations,
learning histories, and for particular learning contexts. We saw in the previous section

76
 Social Constructivism and Subjective Knowledge

that different individuals’ use of the same terms in ways that conform to the social
rules of use does not mean that the terms denote identical concepts or meanings
(such an assertion would be unverifiable, except negatively). Similarly, such
conformity does not mean that individuals’ conceptual structures are isomorphic,
with corresponding connections. All that can be claimed is that the subjective
conceptual knowledge of mathematics of an individual is ordered hierarchically.
It is conjectured that the generation of a hierarchy of increasingly abstract concepts
reflects a particular tendency in the genesis of human mathematical knowledge.
Namely, to generalize and abstract the shared structural features of previously existing
knowledge in the formation of new concepts and knowledge. We conjecture the
existence of some such mechanism to account for the genesis of abstract concepts and
knowledge (as was noted above). At each succeeding level of the conceptual hierarchy
described, we see the results of this process. That is the appearance of new concepts
implicitly defined in terms of a finite set of lower level terms or concepts.
This abstractive, vertical process contrasts with a second mode of mathematical
knowledge generation: the refinement, elaboration or combination of existing
knowledge, without necessarily moving to a higher level of abstraction. Thus the
genesis of mathematical knowledge and ideas within individual minds is conjectured
to involve both vertical and horizontal processes, relative to an individual’s
conceptual hierarchy. These directions are analogous with those involved in inductive
and deductive processes, respectively. We discuss both these modes of knowledge
generation in turn, beginning with that described as vertical.
Before continuing with the exposition of the mechanisms underpinning the
genesis of mathematical knowledge, a methodological remark is called for. It should
be noted that the conjectures concerning the vertical and horizontal modes of
thought in the genesis of subjective mathematical knowledge are inessential for social
constructivism. It has been argued that some (mental) mechanism is needed to
account for the generation of abstract knowledge from particular and concrete
experience. This is central to social constructivism. But as a philosophy of
mathematics it is not necessary to analyse this mechanism further, or to conjecture its
properties. Thus the rejection of the following exploration of this mechanism need
not entail the rejection of the social constructivist philosophy of mathematics.
The vertical processes of subjective knowledge generation involve generalization,
abstraction and reification, and include concept formation.Typically, this process involves
the transformation of properties, constructions, or collections of constructions into
objects. Thus, for example, we can rationally reconstruct the creation of the number
concept, beginning with ordination, to illustrate this process.The ordinal number ‘5’ is
associated with the 5th member of a counting sequence, ranging over 5 objects. This
becomes abstracted from the particular order of counting, and a generalization ‘5’, is
applied as an adjective to the whole collection of 5 objects.The adjective ‘5’ (applicable
to a set), is reified into an object, ‘5’, which is a noun, the name of a thing-in-itself.
Later, the collection of such numbers is reified into the set ‘number’.Thus we see how
a path can be constructed from a concrete operation (using the ordinal number ‘5’),
through the processes of abstraction and reification, which ultimately leads (via the

77
The Philosophy of Mathematics Education

cardinal number ‘5’) to the abstract concept of ‘number’.This account is not offered as
a psychological hypothesis, but as a theoretical reconstruction of the genesis of subjective
mathematical knowledge by abstraction.
What is proposed is that by a vertical process of abstraction or concept formation,
a collection of objects or constructions at lower, pre-existing levels of a personal
concept hierarchy become ‘reified’ into an object-like concept, or noun-like term.
Skemp refers to this ‘detachability’, or ‘the ability to isolate concepts from any of the
examples which give rise to them’ (Skemp, 1971, page 28) as an essential part of the
process of abstraction in concept formation. Such a newly defined concept applies to
those lower level concepts whose properties it abstracts, but it has a generality that
goes beyond them. The term ‘reification’ is applied because such a newly formed’
concept acquires an integrity and the properties of a primitive mathematical object,
which means that it can be treated as a unity, and at a subsequent stage it too can be
abstracted from, in an iteration of the process.6
The increasing complexity of subjective mathematical knowledge can also be
attributed to horizontal processes of concept and property elaboration and clarification.
This horizontal process of object formation in mathematics is that described by Lakatos
(1976), in his reconstruction of the evolution of the Euler formula and its justification.
Namely, the reformulation (and ‘stretching’) of mathematical concepts or definitions
to achieve consistency and coherence in their relationships within a broader context.
This is essentially a process of elaboration and refinement, unlike the vertical process
which lies behind ‘objectification’ or ‘reification’.
Thus far, the account given has dwelt on the genesis and structure of the
conceptual and terminological part of subjective mathematics. There is also the
genesis of the propositions, relationships and conjectures of subjective mathematical
knowledge to be considered. But this can be accommodated analogously. We have
already discussed how the elementary truths of mathematics and logic are acquired
during the learning of mathematical language. As new concepts are developed by
individuals, following the hierarchical pattern described above, their definitions,
properties and relationships underpin new mathematical propositions, which must be
acquired with them, to permit their uses. New items of prepositional knowledge are
developed by the two modes of genesis described above, namely by informal
inductive and deductive processes. Intuition being the name given to the facility of
perceiving (i.e., conjecturing with belief) such propositions and relationships
between mathematical concepts on the basis of their meaning and properties, prior
to the production of warrants for justifying them. Overall, we see, therefore, that the
general features of the account of the genesis of mathematical concepts also holds for
prepositional mathematical knowledge. That is we posit analogous inductive and
deductive processes, albeit informal, to account for this genesis.
In summary, this section has dealt with the genesis of the concepts and
propositions of subjective mathematical knowledge. The account given of this genesis
involves four claims. First of all, the concepts and propositions of mathematics
originate and are rooted in those of natural language, and are acquired (constructed)
alongside linguistic competence. Secondly, that they can be divided into primitive

78
 Social Constructivism and Subjective Knowledge

and derived concepts and propositions. The concepts can be divided into those based
on observation and direct sensory experience, and those defined linguistically by
means of other terms and concepts, or abstracted from them. Likewise, the
propositions consist of those acquired linguistically, and those derived from pre-
existing mathematical propositions, although this distinction is not claimed to be
clear cut. Thirdly, the division of concepts, coupled with the order of their definition,
results in a subjective (and personal) hierarchical structure of concepts (with which
the propositions are associated, according to their constituent concepts). Fourthly, the
genesis of the concepts and propositions of subjective mathematics utilizes both
vertical and horizontal processes of concept and proposition derivation, which take
the form of inductive and deductive reasoning.
These claims comprise the social constructivist account of the genesis of
subjective mathematical knowledge. However, in providing the accounts, examples
have been given, especially concerning the third and fourth of these claims, which
may have the status of empirical conjectures. The hierarchical nature of subjective
mathematical knowledge can be accepted, without relying on such empirical
conjectures. Likewise the existence of the horizontal process of subjective concept
refinement or prepositional deduction, by analogy with Lakatos’ logic of
mathematical discovery, can be accepted in principle. This leaves only the vertical
processes of abstraction, reification or induction to account for, without assuming
empirical grounds. But some such procedure is necessary, if subjective knowledge is
to be constructed by individuals on the basis of primitive concepts derived from
sense impressions and interactions, or elementary mathematical propositions
embedded in language use, as we have assumed. For it is clear that relatively abstract
knowledge must be constructed from relatively concrete knowledge, to account for
the increasing abstraction of the subjective knowledge of mathematics. Hence, as
with the horizontal process, the existence of this vertical process is needed in
principle, irrespective of the fact that some of the details included in the account
might be construed as empirical conjectures. For this reason, these details were
characterized as inessential to the central thesis of social constructivism.

C. Subjective Belief in the Existence of Mathematical Objects

The account given above of the development of individuals’ knowledge of the

external world is that it is a free construction of the individual subject to the
constraints of the physical and social worlds. The individual directly experiences
these worlds and has his or her conjectural maps of these worlds confirmed as viable
or demonstrated to be inadequate on the basis of the responses to their actions. The
consequence of this is that the individual constructs personal representations of these
worlds, which are unique and idiosyncratic to that individual, but whose
consequences fit with what is socially accepted. Such a fit is due to the shared
external constraints which all individuals accommodate (more or less), and in
particular, the constraint of viable negotiation of meanings and purposes in social

79
The Philosophy of Mathematics Education

intercourse. Thus, according to this account, individuals’ construct their own

subjective knowledge and concepts of the external and social worlds, as well as that
of mathematics, so that they fit with what is socially accepted.
These self-constructed worlds represent reality to the individuals who have made
them, be it physical or social reality. Since the same mechanism lies behind the
construction of mathematics as the other representations, it is not surprising that it
too seems to have a measure of independent existence. For the objects of
mathematics have objectivity, in that they are socially accepted constructs. Other
socially constructed concepts are known to have a powerful impact upon our lives,
such as ‘money’, ‘time’ (o’clock), ‘the North Pole’, ‘the equator’, ‘England’, ‘gender’,
‘justice’ and ‘truth’. Each of these is, undeniably a social construct. Yet each of these
concepts has as tangible an impact as many concretely existing objects.
Consider ‘money’. This represents an organizing concept in modern social life of
great power, and more to the point, of undeniable existence. Yet it is clearly an
abstract human-made symbol of conventional, quantified value, as opposed to some
aspect of the physical world. Let us explore ‘money’ further. What is it that gives
money its existence? There are two features on which its ontological status is based.
First, it is socially accepted, which gives it objectivity. Second, it is represented by
tokens, which means that it has tangible concrete reference.
Now consider the analogy with the objects of mathematics. These have
objectivity, being socially accepted. In addition, the primitive concepts of
mathematics, such as ‘square’ and ‘7’, have concrete examples in our perceptions of
the physical world. So far, the analogy is good. The defined concepts of mathematics
do not fit so well with the analogy, for they may only have concrete applications
indirectly, via chains of definition. Although there is an analogy between these
abstract objects of mathematics and the more abstracted applications of money
(budgeting, financial forecasting, etc.), this is stretching it a bit far. What can be said
is that the analogy between ‘money’ and mathematical objects lends some
plausibility to subjective belief in the latter objects. They are both objective social
constructions and have concrete manifestations.
Of course mathematics has a further feature supporting this belief. This is the
necessary relationships between its objects, due to their strict logical relationships in
deductive systems. Logical necessity attaches to the objects of mathematics through
their defining relationships, their inter-relationships and their relationships with
mathematical knowledge. This lends necessity to the objects of mathematics (a
feature that money lacks).
In a nutshell, the argument is this. If an individual’s knowledge of the real world,
including its conventional components, is a mental construct constrained by social
acceptance, then belief in such constructs evidently can be as strong as beliefs in
anything. Subjective knowledge of mathematics, and acquaintance with its concepts
and objects is also a mental construct. But like other socially determined constructs,
it has an external objectivity arising from its social acceptance. The objects of
mathematics also have (i) concrete exemplifications, either directly (for the primitive
mathematical concepts), or indirectly (for the defined mathematical concepts); and

80
 Social Constructivism and Subjective Knowledge

(ii) logical necessity, through their logical foundations and deductive structure. These
properties are what give rise to a belief in the objective existence of mathematics
and its objects.
Traditionally, knowledge has been divided into the real and the ideal. It is common
to accept the reality of the external world and our scientific knowledge of it (scientific
realism). It is also common to accept the ideal existence of (objective) mathematics
and mathematical objects (idealism or platonism). This dichotomy places physical and
scientific objects in one realm (Popper’s world 1) and mathematical objects in another
(subjective knowledge of them in world 2, objective knowledge in world 3). Thus it
places mathematical and physical objects in different categories. The social
constructivist thesis is that we have no direct access to world 1, and that physical and
scientific objects are only accessible when represented by constructs in world 3
(objective concepts) or in world 2 (subjective concepts). Thus our knowledge of
physical and mathematical objects has the same status, contrary to traditional views.
The difference resides only in the nature of the constraints physical reality imposes on
scientific concepts, through the means of verification adopted for the two types of
knowledge (scientific or mathematical). The similarity, including the social basis of the
objectivity of both types of knowledge, accounts for the subjective belief in the
existence of mathematical objects (almost) just as for theoretical physical objects.7

3. Relating Objective and Subjective Knowledge of Mathematics

The relationship between subjective and objective knowledge of mathematics is

central to the social constructivist philosophy of mathematics. According to this
philosophy, these realms are mutually dependent, and serve to recreate each other.
First of all, objective mathematical knowledge is reconstructed as subjective
knowledge by the individual, through interactions with teachers and other persons,
and by interpreting texts and other inanimate sources. As has been stressed,
interactions with other persons (and the environment), especially through negative
feedback, provides the means for developing a fit between an individuals subjective
knowledge of mathematics, and socially accepted, objective mathematics. The term
‘reconstruction’, as applied to the subjective representation of mathematical
knowledge, must not be taken to imply that this representation matches objective
mathematical knowledge. As has been said, it is rather that the subjective knowledge
‘fits’, to a greater or lesser extent, socially accepted knowledge of mathematics (in
one or more of its manifestations).
Secondly, subjective mathematical knowledge has an impact on objective
knowledge in two ways. The route through which individuals’ mathematical
creations become a part of objective mathematical knowledge, provided they survive
criticism, has been described. This represents the avenue by means of which new
creations (including the restructuring of pre-existing mathematics) are added to the
body of objective mathematical knowledge. It also represents the way in which
existing mathematical theories are reformulated, inter-related or unified. Thus it

81
The Philosophy of Mathematics Education

includes creation not only at the edges of mathematical knowledge, but also
throughout the body of mathematical knowledge. This is the way that subjective
knowledge of mathematics explicitly contributes to the creation of objective
mathematical knowledge.8 However, there is also a more far-reaching but implicit way
in which subjective mathematical knowledge contributes to objective mathematical
knowledge.
The social constructivist view is that objective knowledge of mathematics is
social, and is not contained in texts or other recorded materials, nor in some ideal
realm. Objective knowledge of mathematics resides in the shared rules, conventions,
understandings and meanings of the individual members of society, and in their
interactions (and consequently, their social institutions). Thus objective knowledge of
mathematics is continually recreated and renewed by the growth of subjective
knowledge of mathematics, in the minds of countless individuals. This provides the
substratum which supports objective knowledge, for it is through subjective
representations that the social, the rules and conventions of language and human
interaction, is sustained. These mutually observed rules, in their turn, legitimate
certain formulations of mathematics as accepted objective mathematical knowledge.
Thus objective knowledge of mathematics survives through a social group enduring
and reproducing itself. Through passing on their subjective knowledge of
mathematics, including their knowledge of the meaning to be attributed to the
symbolism in published mathematical texts, objective knowledge of mathematics
passes from one generation to the next.
This process of transmission does not merely account for the genesis of
mathematical knowledge. It is also the means by which both the justificatory canons
for mathematical knowledge, and the warrants justifying mathematical knowledge
itself are sustained. Kitcher (1984) likewise claims that the basis for the justification of
objective mathematical knowledge is passed on in this way, from one generation of
mathematicians to the next, starting with empirically warranted knowledge.
As a rational reconstruction of mathematical history to warrant mathematical
knowledge, Kitcher’s account has some plausibility. Like Kitcher, social
constructivism sees as primary the social community whose acceptance confers
objectivity on mathematical knowledge. However, unlike Kitcher, social
constructivism sees the social as sustaining the full rational justification for objective
mathematical knowledge, without the need for historical support for this
justification. According to social con¬ structivism, the social community which
sustains mathematics endures smoothly over history, with all its functions intact, just
as a biological organism smoothly survives the death and replacement of its cells.
These functions include all that is needed for warranting mathematical knowledge.
It should be made clear that the claim that objective mathematical knowledge is
sustained by the subjective knowledge of members and society does not imply the
reducibility of the objective to the subjective. Objective knowledge of mathematics
depends upon social institutions, including established ‘forms of life’ and patterns of
social interaction. These are sustained, admittedly, by subjective knowledge and
individual patterns of behaviour as is the social phenomenon of language. But this no

82
 Social Constructivism and Subjective Knowledge

more implies the reducibility of the objective to the subjective, than materialism
implies that thought can be reduced to, and explained in terms of physics. The sum
of all subjective knowledge is not objective knowledge. Subjective knowledge is
essentially private, whereas objective knowledge is public and social. Thus although
objective knowledge of mathematics rests on the substratum of subjective
knowledge, which continually recreates it, it is not reducible to subjective
knowledge.
As a thought experiment, imagine that all social institutions and personal
interactions ceased to exist. Although this would leave subjective knowledge of
mathematics intact, it would destroy objective mathematics. Not necessarily
immediately, but certainly within one lifetime. For without social interaction there
could be no acquisition of natural language, on which mathematics rests. Without
interaction and the negotiation of meanings to ensure a continued fit, individual’s
subjective knowledge would begin to develop idiosyncratically, to grow apart,
unchecked. The objective knowledge of mathematics, and all the implicit knowledge
sustaining it, such as the justificatory canons, would cease to be passed on. Naturally
no new mathematics could be socially accepted either. Thus the death of the social
would spell the death of objective mathematics, irrespective of the survival of
subjective knowledge.
The converse also holds true. If, as another thought experiment we imagine that
all subjective knowledge of mathematics ceased to exist, then so too would objective
knowledge of mathematics cease to exist. For no individual could legitimately assent
to any symbolic representations as embodying acceptable mathematics, being
deprived of the basis for such assent. Therefore there could be no acceptance of
mathematics by any social group. This establishes the converse relationship, namely
that the existence of subjective knowledge is necessary for there to be objective
knowledge of mathematics.
Of course it is hard to follow through all the consequences of the second thought
experiment, because of the impossibility of separating out an individual’s subjective
knowledge of language and mathematics. Knowledge of language depends heavily
on the conceptual tools for classifying, categorising and quantifying our experience
and for framing logical utterances. But according to social constructivism these form
the basis for mathematical knowledge. If we delete these from subjective knowledge
in the thought experiment, then virtually all knowledge of language and its
conceptual hierarchy, would collapse. If we leave this informal knowledge and only
debate explicit knowledge of mathematics (that learned as mathematics and not as
language), then subjective knowledge of mathematics could be rebuilt, for we would
have left its foundations intact.9
In summary, the social constructivist thesis is that objective knowledge of
mathematics exists in and through the social world of human actions, interactions
and rules, supported by individuals’ subjective knowledge of mathematics (and
language and social life), which need constant re-creation. Thus subjective
knowledge re-creates objective knowledge, without the latter being reducible to the

83
The Philosophy of Mathematics Education

former. Such a view of knowledge is endorsed by a number of authors. Paul Cobb,

argues from a radical constructivist perspective that:
the view that cultural knowledge in general and mathematics in particular
can be taken as solid bedrock upon which to anchor analyses of learning
and teaching is also questioned. Instead it is argued that cultural knowledge
(including mathematics) is continually recreated through the coordinated
actions of the members of a community.
(Cobb, 1988, page 13)
Paulo Freire has elaborated an epistemology and philosophy of education that places
individual consciousness, in the context of the social, at the heart of objective
knowledge. He ‘recognize(s) the indisputable unity between subjectivity and
objectivity in the act of knowing.’ (Freire, 1972b, page 31) Freire argues, as we have
done, that objective knowledge is continually created and re-created as people reflect
and act on the world.
Even the received view of epistemology (see, for example, Sheffler, 1965) can be
interpreted as logically founding objective knowledge on subjective knowledge. For
this view defines knowledge (rather more narrowly than it has been used above) as
justified true belief. Belief includes what has been termed subjective knowledge, in
this chapter. In mathematics, justified true belief can be interpreted as consisting of
assertions that have a justification necessitating their acceptance (in short, a proof).
According to the social constructivist philosophy, such mathematical statements are
socially accepted, on the basis of their justification, and thus constitute objective
mathematical knowledge. Thus, in the terms of this chapter, ‘knowledge is justified
true belief’ translates into ‘objective knowlege of mathematics is socially acccepted
subjective knowledge, expressed in the form of linguistic assertions’. According to
this translation, objective knowledge of mathematics depends logically on subjective
knowledge, because of the order of definition.
The social constructivist view of mathematics places subjective and objective
knowledge in mutually supportive and dependent positions. Subjective knowledge
leads to the creation of mathematical knowledge, via the medium of social
interaction and acceptance. It also sustains and re-creates objective knowledge, which
rests on the subjective knowledge of individuals. Representations of objective
knowledge are what allow the genesis and re-creation of subjective knowledge. So
we have a creative cycle, with subjective knowledge creating objective knowledge,
which in turn leads to the creation of subjective knowledge. Figure 4.1 shows the
links between the private realm of subjective knowledge and the social realm of
objective knowledge each sustaining the creation of the other. Each must be publicly
represented for this purpose. Thereupon there is an interactive social negotiation
process leading to the reformulation of the knowledge and its incorporation into the
other realm as new knowledge.
Of course there are powerful constraints at work throughout this creative cycle.
These are the physical and social worlds, and in particular the linguistic and other
rules embodied in social forms of life.

84
 Social Constructivism and Subjective Knowledge

Figure 4.1: The Relationship between Objective and Subjective Knowledge of

Mathematics

4. Criticism of Social Constructivism

The account of social constructivism has brought together three philosophical

perspectives as a basis for a unified philosophy of mathematics. These are quasi-
empiricism, conventionalism and radical constructivism. As a consquence, two types of
criticism can be directed at social constructivism. First of all, there is criticism directed
at the assumptions and philosophical stance adopted by one of these tributary
philosophies. For example, there are the problems in accounting for logical necessity
in mathematics from the social perspective of conventionalism. An attempt has been
made to anticipate and answer such criticisms. To those who reject the conventionalist
assumptions such arguments will not be convincing. However, no further criticisms of
this type, that is, directed at the tributary philosophies, will be addressed. For in
addition to having been treated above, they can also be found in the relevant literature.

85
The Philosophy of Mathematics Education

Secondly, there are criticisms that can be directed at the novel synthesis that is
provided by social constructivism. Two criticisms in this category will be considered.
These concern the use of empirical assumptions in social constructivism, and the
tension between the subsumed theor ies of conventionalism and radical
constructivism.
Throughout the exposition, it has been argued that the genesis as well as the
justification of knowledge is the proper concern of the philosophy of mathematics.
Consequently both these contexts have been discussed, but an attempt has been
made to distinguish between them carefully, and to avoid or to demarcate carefully
any empirical assumptions, especially concerning the genesis of mathematical
knowledge. This was a criticism directed at Lakatos in Chapter 2, that in his account
of the conditions of the genesis of mathematical knowledge, he introduced an
historical (i.e. empirical) conjecture. It may be felt that the social constructivist
account errs similarly. However, I believe that this would indicate that a clearer
exposition of social constructivism is needed, rather than necessitating a rejection of
the entire philosophy.
A more substantial criticism arises from a possible tension, or even inconsistency,
between the subsumed theories of conventionalism and radical constructivism. For
the former gives primacy to the social, comprising accepted rules and conventions
underpinning the use of language and objective knowledge of mathematics. This
reflects a ‘form of life’, constituting accepted social and verbal behaviour patterns.
The latter, gives primacy to the knowing subject, an unreachable monad
constructing hypo-thetical world-pictures to represent exper iences of an
unknowable reality. To an adherent of one but not the other philosophy, their
conjunction in social constructivism may seem to be an unholy alliance, for neither
of the two foci is given precedence. Rather each is the centre of a separate realm.
The knowing subject is at the centre of the private realm of individuals and
subjective knowledge. This realm assumes a real but unknowable world, as well as the
knowing subject. But this realm is not enough to account for objective knowledge,
let alone for humanity. For humankind is a social animal, and depends essentially on
interchange and language. The social realm takes this as its basis, including social
institutions and social agreements (albeit tacit). This realm assumes the existence of
social groups of human beings. However, this perspective seems weak in terms of the
interior life and consciousness it ascribes to individuals.
Thus although the primacy of focus of each of conventionalism and radical
constructivism is sacrificed in social constructivism, their conjunction in it serves to
compensate for their individual weaknesses, yet this conjunction raises the question
as to their mutual consistency. In answer it can be said that they treat different
domains, and both involve social negotiation at their boundaries (as Figure 4.1
illustrates). Thus inconsistency seems unlikely, for it could only come about from
their straying over the interface of social interaction, into each other’s domains.
The separateness of the private and social realms, together with their separate
theoretical accounts, has another consequence. It means that the parts of the social
constructivist account could be modified (e.g., the account of subjective knowledge)

86
 Social Constructivism and Subjective Knowledge

without changing the whole philosophy. This suggests that the philosophy lacks a
single overarching principle. However, there are unifying concepts (or metaphors)
which unite the private and social realms, namely construction and negotiation. For
both subjective and objective knowledge are deemed to be human constructions,
built up from pre-existing knowledge components. The second unifying concept is
that of social negotiation. This not only plays a central role in the shaping of
subjective and objective knowledge. It also plays a key role in the justification of
mathematical knowledge according to social constructivism, from the quasi-
empiricist component.

Notes

1 Reference is made to subjective knowledge. Philosophically this usage is problematic, because knowlege
has been defined as justified belief, where the justification is understood to be objective. Thus
knowledge is socially verifiable and socially accepted belief. But subjective belief which is not
publicly represented cannot be socially accepted, and hence cannot be knowledge. To follow this
strict usage, all references to subjective knowledge should be replaced by references to subjective
belief. However, there is a well known precedent for continuing with the current usage. Individuals
are frequently described as possessing procedural knowledge (Sheffler, 1965) or knowledge as ‘know
how’ (Ryle, 1949). Thus a precedent exists for not observing the strict usage described above. I shall
continue to refer to ‘subjective knowledge’, in full consciousness of the possible transgression
involved, but without assuming that subjective knowledge is objectively justified.
2 There is also a powerful analogy with Kuhn’s (1970) theory of the structure of scientific revolutions.
For in the proposed account we have an alternation between viable subjective theories used to guide
action, which parallels Kuhn’s periods of ‘normal science’, followed by conflicts between the
predictions of the subjective theories and observations, paralleling Kuhn’s periods of ‘revolutionary
science’ (as well as Popper’s falsification of theories), when the old subjective theories are rejected,
and new theories are built from them and accepted, completing the cycle (see Ernest, 1990).
3 I follow Glasersfeld’s (1984) use of the terms ‘fit’ and ‘match’ which make an important distinction.
They both refer to the relationship between a representation and what it represents. Thus a map
‘matches’ the geographical region it represents because there is a morphism, a structure preserving
relation, between them. In this way ‘matching’ resembles the correspondence theory of truth. In
contrast, a key ‘fits’ a lock when it works and turns the lock. Such a ‘fit’ does not represent the
structure of the lock, but merely a functional relationship between the two. When a theory, account
or representation ‘fits’ it means that it satisfies a basic set of constraints, but that beyond that it can
have any form. ‘Fitting’ resembles the pragmatic conception of truth. The crucial difference is that
theories that ‘fit’ a portion of the world do not tell us about the structure of the world, whereas any
that ‘match’ would do so.
4 If this assertion is considered to be controversial, we simply exclude from the discussion any terms
which do not fit the description. Subsequent discussion of hierarchies therefore only refer to those
terms included in them, and not any excluded circularly defined terms.
5 This hierarchy can also be used as a basis for imposing a hierarchical structure on the sentences and
formulas of the theory. We simply define the level of a sentence or formula to be the maximum level
of the terms occurring in it.
6 Such an account has previously been offered to explain the genesis of set-theoretic objects from
mathematical constructions, within an objective philosophy of mathematics, by Machover (1983).
The purpose being to found abstract, classical mathematics on the more intuitive (and presumably
safer) realm of constructive mathematics, analogous to the formalist strategy. However, by proposing
that the objects of mathematics are interpreted as ‘reified constructions’, Machover is opening the
door for a genetic epistemology in mathematics, as is proposed by Piaget (1972) and Kitcher (1984),
among others, albeit from a different viewpoint.
It is also worth mentioning that the vertical process of object formation or reification is part of
set theory, both formal and informal, in the standard objective mathematical theory. Typically, such a

87
The Philosophy of Mathematics Education

process is permitted by a comprehension axiom, which allows the collection of objects satisfying a
defined mathematical property to become a new mathematical object, a set:
For all P, there exists p, such that p={x/P(x)}) (Where ‘P(x)’ is a first-order defined property
with free variable ‘x’, and ‘p’ is the set comprising the extension of ‘P(x)’).
Of course such an unrestricted comprehension principle leads to contradictions (Russell, 1902).
Indeed many of the innovations in logic and set theory in the early part of this century (Russell’s
theory of types, Zermelo-Fraenkel and Godel-Bernays-von Neumann set theories) were expressly
motivated to permit the safe use of some form of the comprehension principle.
Thus the process of reification, i.e., the elevation of (the extension of) a property to objecthood,
is an established principle in objective mathematics. Since, according to social constructivism,
objective mathematics is a socially accepted reflection of subjective mathematics, this adds plausibility
to the assumption of the vertical reifying process in the genesis of subjective mathematical
knowledge.
7 There is a further explanation for the subjective belief in the objects of mathematics based on the
propensity to objectify and reify the concepts of mathematics into objects, discussed above. Once
objectified, such mentally constructed mathematical objects can be (nearly) as potent as physical
objects. Indeed a platonist view of mathematical existence is often held by practising mathematicians
(such as Frege, Hardy, Godel, Thom) whose extensive mathematical activities by constantly using
their mathematical concepts and objects reinforce their subjective ‘solidity’ or ‘object’-like qualities.
Thus the reification of mathematical objects, coupled with other features, such as a propensity for
belief in socially accepted constructs, gives them an apparent existence, accounting for some
mathematicians’ platonism.
8 Subjective knowledge of mathematics (syntactical knowledge of mathematics, Schwab, 1975) also
plays a part in the process of acceptance of new mathematical knowledge, via the process of criticism.
9 Popper (1979, pages 107–108) uses this second interpretation of the thought experiment as a basis
for an argument that books contain objective knowledge. For with our capacity to learn from books
intact, Popper argues that we can re-acquire objective knowledge from libraries. However, the social
constructivist view is that only the explicit, more advanced part of subjective (and objective)
knowledge is eradicated in this interpretation of the thought experiment. The reconstruction of
these forms of knowledge with the aid of libraries shows that the foundations of this knowledge has
survived. (However, this conclusion does not contradict Popper’s, since his sense of ‘objectivity’ does
not include implicit, socially accepted linguistic knowledge).

88
 5

The Parallels of Social Constructivism

1. Introduction

Today’s burgeoning intellectual climate views all human knowledge as a social

construction. Mathematics, the last bastion of certainty in knowledge, has been trying
to resist this current of thought. But as we have seen, more and more authors are
joining the stream and viewing mathematics as a social construction. Of course social
constructivism, the particular philosophy expounded above, is just one account of
mathematics from this perspective. Not many alternatives are yet formulated in the
philosophy of mathematics, but beyond, parallel views of mathematics and knowledge
abound in other disciplines. This chapter explores some of these parallels, showing
how overwhelming the intellectual current is becoming.

2. Philosophical Parallels

A. Sceptical Philosophy

The most central claim of social constructivism is that no certain knowledge is

possible, and in particular no certain knowledge of mathematics is possible.
Concerning empirical knowledge, this thesis is one that is subscribed to by many
philosophers. These include continental sceptics beginning with Descartes; British
empiricists such as Hume; American pragmatists such as James and Dewey; some
modern American philosophers such as Goodman, Putnam, Quine and Rorty; and
modern philosophers of science, including Popper, Kuhn, Feyerabend, Lakatos and
Laudan.
Throughout a number of these strands of thought there is agreement that
empirical knowledge of the world is a human construction. Beyond those cited, this
view is shared by Kant and his followers, who see any knowledge of the world as
shaped by innate mental categories of thought.
Scepticism concerning absolute empirical knowledge has grown to be the
dominant view. However, until Lakatos (1962) the full extension of scepticism to
mathematics was not made. Since then, it has gained partial acceptance, whilst

89
The Philosophy of Mathematics Education

remaining controversial. Social constructivism is an attempt to extend Lakatos’

sceptical approach systematically to a philosophy of mathematics.
However, social constructivism does not entail the fullest form of scepticism, such
as cartesian doubt. For it accepts the existence of the physical world (whilst denying
certain knowledge of it) and accepts the existence of language and the social group.
Both the physical and the social worlds play an essential part in the social constructivist
epistemology. As a commentator on Wittgenstein says: ‘Doubt presupposes mastery of
a language-game.’ Kenny (1973, page 206) Social constructivism is sceptical about the
possibility of any certain knowledge, particularly in mathematics, but it is not sceptical
about the necessary pre-conditions for knowledge.

B. The Private Language Problem

One challenge for the social constructivist account of subjective knowledge is the
‘private language’ problem. If an individual’s concepts are personal constructions,
how are they able to communicate using a shared language? Why should different
mathematicians understand the same thing by a concept or proposition, when their
meanings are personally unique? May not each have a private language, to refer to
his or her own private meanings?
Social constructivism overcomes this problem through the interpersonal
negotiation of meanings to achieve a ‘fit’. Support for this position, if not the precise
form of argument, is widespread. Wittgenstein (1953) first answered the problem,
arguing that private languages cannot exist. A number of philosophers commenting
on his work, such as Kenny (1973) and Bloor (1983), support the rejection of private
languages, as do others including Ayer (1956) and Quine (1960). With regard to
mathematics, the private language problem is also considered soluble, for example by
Tymoczko (1985) and Lerman (1989), both arguing from a position close to social
constructivism.
The solution of the private language problem by social constructivism reflects a
substantial body of philosophical opinion. Generally, it is argued that the shared rules
and ‘objective pull’ of inter-personal language use makes it public, consistent with
social constructivism.

C. Knowing and the Development of Knowledge

The social constructivist philosophy of mathematics treats knowledge as the result of

a process of coming to know, including the social processes leading to the
justification of mathematical knowledge. Thus it attaches great weight to knowing
and the development of knowledge, in addition to its product, knowledge. This
emphasis, although far from universal, is to be found in the works of a number of
philosophers, including Dewey (1950), Polanyi (1958), Rorty (1979), Toulmin (1972),
Wittgenstein (1953) and Haack (1979).

90
 The Parallels of Social Constructivism

Other authors have looked to an evolutionary model to account for the growth
and development of knowledge. This includes the genetic epistemology of Piaget
(1972, 1977), and the evolutionary epistemologies of Popper (1979), Toulmin (1972)
and Lorenz (1977).
The majority of modern philosophers of science view it as a growing and
developing body of knowledge either detached from history (Popper, 1979) or
embedded in human history (Kuhn, Feyerabend, Lakatos, Toulmin and Laudan).
Educational thinkers have also stressed the processes and means of knowledge
acquisition, as a basis for the curriculum, including, most notably, Schwab (1975) and
Bruner (1960).
The process of coming to know relates to practical knowledge and the
applications of knowledge. Ryle (1949) established that practical knowledge
(‘knowing how’) belongs to epistemology as well as declarative knowledge
(‘knowing that’). Sneed (1971) proposes a model of scientific knowledge which
incorporates the range of intended applications (models) as well as the core theory.
This model has been extended to mathematics by Jahnke (Steiner, 1987). Such
approaches admitting practical knowledge or its applications into the traditional
domain of knowledge thus parallel aspects of the social constructivist proposals.
The social constructivist account of the nature and genesis of subjective
knowledge of mathematics is to a large extent based on the radical constructivism of
Glasersfeld (1984, 1989). This has parallels in the thought of Kant, and even more so,
Vico, as well as with the American pragmatists and modern philosophers of science
cited above.
Thus there is a growing current of thought in modern philosophy which gives a
central place in epistemology to considerations of the human activity of knowing
and the evolution of knowledge, as in social constructivism.

D. The Divisions of Knowledge

A key tenet of social constructivism, following Lakatos, is that mathematical

knowledge is quasi-empirical. This leads to the rejection of the categorical
distinction between a priori knowledge of mathematics, and empirical knowledge.
Other philosophers have also rejected this distinction, most notably Duhem and
Quine (1951), who hold that because the assertions of mathematics and science are
all part of a continuous body of knowledge, the distinction between them is one of
degree, and not of kind or category. White (1950) and Wittgenstein (1953) also reject
the absoluteness of this distinction, and a growing number of other philosophers also
reject the water-tight division between knowledge and its empirical applications
(Ryle, 1949; Sneed, 1971; Jahnke).
A further parallel is found in ‘post-structuralist’ and ‘post-modernist’
philosophers, such as Foucault (1972) and Lyotard (1984), who take the existence of
human culture as their starting point. Foucault claims that the divisions of knowledge
are modern constructs, defined from certain social perspectives. Throughout history,

91
The Philosophy of Mathematics Education

he argues, the different disciplines have changed. Their objects, concepts, accepted
rules of thought and aims have evolved and changed, even amounting, in extreme
cases, to discontinuities. Knowledge, in his view, is but one component of ‘discursive
practice’, which includes language, social context and social relations. In evidence, he
documents how certain socially privileged groups, such as doctors and lawyers, have
established discourses creating new objects of thought, grouping together hitherto
unconnected phenomena defined as delinquent behaviour or crime. Elsewhere,
Foucault (1981) shows how a new area of knowledge, the discourse of human
sexuality, was defined by church and state, to serve their own interests.
Lyotard (1984) considers all human knowledge to consist of narratives, whether
literary or scientific. Each disciplined narrative has its own legitimation criteria,
which are internal, and which develop to overcome or engulf contradictions. He
describes how mathematics overcame crises in its axiomatic foundations due to
Godel’s Theorem by incorporating meta-mathematics into an enlarged research
paradigm. He also claims that continuous differentiable functions are losing their pre-
eminence as paradigms of knowledge and prediction, as mathematics incorporates
undecidability, incompleteness, Catastrophe theory and chaos. Thus a static system of
logic and rationality does not underpin mathematics, or any discipline. Rather they
rest on narratives and language games, which shift with organic changes of culture.
These thinkers exemplify a move to view the traditional objective criteria of
knowledge and truth within the disciplines as internal myths, which attempt to deny
the social basis of all knowing. This new intellectual tradition affirms that all human
knowledge is interconnected through a shared cultural substratum, as social
constructivism asserts.
Another post-structuralist is Derrida, who as well as supporting this view, argues
for the ‘deconstructive’ reading of texts:
In writing, the text is set free from the writer. It is released to the public
who find meaning in it as they read it. These readings are the product of
circumstance. The same holds true even for philosophy. There can be no
way of fixing readings…
Anderson et al. (1986, page 124)
This offers a parallel to the social constructivist thesis that mathematical texts are
empty of meaning. Meanings must be constructed for them by individuals or groups
on the basis of their knowledge (and context).

E. The Philosophy of Mathematics

Various modern philosophers of mathematics have views consistent with some if not
all of the theses of social constructivism. Here we draw together some of the points
of contact between them and social constructivism.
Some philosophers emphasize the significance of the history and empirical
aspects of mathematics for philosophy. Kitcher (1984) erects a system basing the

92
 The Parallels of Social Constructivism

justification of mathematical knowledge on its empirical basis, with the justification

transmitted from generation to generation by the mathematical community.1 An
empirical or quasi-empirical justification of mathematical knowledge, drawing on
mathematical practice, is also adopted by N.D.Goodman, Wang, P.Davis, Hersh,
Wilder, Grabiner, Tymoczko (all in Tymoczko, 1986), Tymoczko (1986a), Stolzenberg
(1984), MacLane (1981), McCleary and McKinney (1986), and Davis (1974). Thus a
move away from the traditional aprioristic view or justification of mathematics, as
advocated by social constructivism, is widespread.
A number of other contributory theses of social constructivism are espoused by
philosophers of mathematics. The conventionalist viewpoint is implicit in several of
these authors’ work. Those who make it explicit include Stolzenberg (1984), as well
as Bloor, Quine and Wittgenstein, cited above, and others mentioned in Chapter 2.
In addition, the thesis that the objects of mathematics are reified constructions is
proposed by both Davis (1974) and Machover (1983).
Beyond such piecemeal comparisons, two philosophers who have anticipated
much of the social constructivist philosophy of mathematics are Bloor (1973, 1976,
1978, 1983, 1984) and Tymoczko (1985, 1986, 1986a). Both argue that objectivity in
mathematics can best be understood in terms of social acceptance, and draw upon
the seminal work of Wittgenstein and Lakatos.
Although no new paradigm is yet fully accepted, social constructivism sits
comfortably in a growing quasi-empir icist tradition. Beyond this, a few
contemporary philosophers are beginning to propose approaches to the philosophy
of mathematics similar to and coherent with social constructivism.

3. Sociological Perspectives of Mathematics

A. Cultural and Historical Approaches

Several authors offer historico-cultural accounts of the nature of mathematics,

treating the relationships between the social and cultural groups involved in
mathematics, and the genesis and nature of mathematical knowledge. These include
Crowe (1975), Mehrtens (1976), Restivo (1985), Richards (1980, 1989), Szabo
(1967), Wilder (1974, 1981) and Lakatos (1976). These authors have offered theories
of the development of mathematical knowledge, which relate it to its social,
historical and cultural context. In particular, they theorize how the micro social
context (i.e. interactions in small groups), in the case of Szabo and Lakatos, or the
macro social context, in the case of Crowe, Mehrtens, Restivo, Richards and Wilder,
influence the development of mathematical knowledge.
Studies of the micro social context concern negotiation within groups of individuals,
leading to the acceptance of logical argument or mathematical knowledge, as well as
concepts. Such theories reflect the quasi-empiricist account of the acceptance of
knowledge, on the empirical level. Lakatos (1976) offers an account of this type with
his conjectured 7 stage pattern of mathematical discovery. As an empirical conjecture

93
The Philosophy of Mathematics Education

this fits here, because it represents an historical parallel with aspects of quasi-empiricism
and social constructivism, at the micro social level. Szabo (1967) argues that the
deductive logic of Euclid derives from pre-socratic dialectics, with conversation serving
as the model. Again, this fits with the social constructivist account.
Studies of the macro social context offer theories of the structural patterns, social
relationships or ‘laws’ in the development of mathematical knowledge in history and
culture. Many of these are social constructionist accounts, consistent with
conventionalism, and hence social constructivism, albeit in a different realm. In this
bracket can be included a new breed of histories of mathematics acknowledging its
fallibility (Kline, 1980) and its multi-cultural social construction (Joseph, 1990).
Historical and cultural studies of mathematics with a bearing on the philosophy
of mathematics draw strength and inspiration from the comparable ‘externalist’
approaches to the philosophy of science, such as those of Kuhn (1970) and Toulmin
(1972). Such historical approaches, as well as the philosophy of science, provide
parallels and support for social constructivism. Likewise, when the social constructi¬
vist account is supplemented with empirical hypotheses, a theory of the history of
mathematics results, as in the quasi-empiricism of Lakatos (1976).

B. The Sociology of Knowledge

A number of sociologial theses offer parallels to social constructivism.

Knowledge as a social construction

First of all, there is the ‘social constructionist’ thesis that all knowledge is a social
construction. There is a tradition in the sociology of knowledge supporting and
elaborating this thesis, including such theorists as Marx, Mannheim, Durkheim,
Mead, Schutz, Berger and Luckman, and Barnes (although the first few named in this
list assert that some knowledge, notably mathematics, can be free from social bias).
This is the dominant view in the sociology of knowledge, contrasting with the main
traditions in philosophy which claim that there is certain knowledge of the world
from observations (empiricism) or through abstract thought (idealism).
In the sociology of knowledge there are variations in the degree of relativism
ascribed to knowledge. In the extreme case, all human knowledge is seen as relative
to social groups and their interests, and physical reality itself is regarded as a social
construction. More moderate positions regard knowledge (and not reality) as a social
construction, and accept an enduring world as a constraint on the possible forms of
knowledge. For example, Restivo (1988a) argues that although the new sociology of
science regards knowledge as a social construction, it is better aligned with realism
than with simple relativism, with which it has no necessary connection. Such
positions are parallel to social constructivism in the assumptions they adopt, although
they remain sociological as opposed to philosophical theories. Their existence

94
 The Parallels of Social Constructivism

suggests the potential fruitfulness of a sociological version of social constructivism, to

account for the social structures and development of mathematics.

The ‘strong programme’ in the sociology of knowledge

Bloor (1976) has laid down criteria (the tenets of the ‘strong programme’) that a
sociology of knowledge should satisfy if it is to provide a sociologically acceptable
account of knowledge. Briefly, these require that for adequacy a theory of
knowledge should account for: (i) the social genesis of knowledge; (ii and iii) both
true and false knowledge and beliefs symmetrically; and (iv) itself (reflexivity).
Although designed for the sociology of knowledge, it is interesting to apply these
criteria to social constructivism. Re (i): the account given evidently accounts for the
social genesis of mathematical knowledge. Re (ii): it can be said that social
constructivism accounts for beliefs and knowledge without regard to their truth or
falsity. For the generation of knowledge by the hypothetico-deductive method has
no implications concerning its truth. Social constructivism accounts both for the
adoption of new, and for the rejection of old beliefs and knowledge when falsified,
or for other reasons, denied acceptance. Like sociological accounts of knowledge,
social constructivism is symmetrical in these explanations, in terms of social
acceptance, and not in terms of a ‘match’ with a transcendent reality.
Re (iv): Although social constructivism is primarily a philosophy of mathematics, it
can be extended to account for itself, at least in part. For it is based on a number of
epistemological and ontological assumptions, from which conclusions are inferred. As
such it has similar status to that which it ascribes to mathematics, namely an
hypothetico-deductive theory, except for differences in subject matter and rigour.
Both start with a set of plausible but conjectural assumptions (albeit concerning
different realms of knowledge), from which consequences are inferred. In addition,
any justification for social constructivism must reside in its social acceptance, directly
paralleling its account of mathematics. Finally, social constructivism rejects the
analytic-empir ical distinction, and views all knowledge as inter-related.
Consequently, it is legitimately applicable throughout realms of human knowledge,
including to itself. Thus social constructivism may be said to be reflexive, since a
parallel account can be applied to itself.
Overall, social constructivism largely satisfies the criteria of the ‘strong
programme’. This contrasts with absolutist philosophies, which treat truth quite
differently from falsehood, failing to satisfy (ii) and (iii), as well as being unable to
satisfy (iv). Whilst in terms of traditional philosophy, this is of limited significance, it
suggests that a sociological parallel will satisfy the criteria, auguring well for an over-
arching social constructivist theory.

Knowledge is value laden

Third, there is the value laden-ness of knowledge. Values are the basis for choice, and

95
The Philosophy of Mathematics Education

so to be value-laden is to represent the preference or interest of a social group.Values

can be manifested explicitly, as in a conscious act of choice, or tacitly, as in unconscious
compliance or acceptance. For example, Polanyi (1958) argues that many of the shared
values of the scientific community, such as the endorsement of the scientific
consensus, are tacit. However, the traditional view of science and scientific knowledge
is that it is logical, rational, objective, and hence value-free.2 Both social constructivism
and the sociology of knowledge reject this view, for different reasons. The sociology
of knowledge asserts that all knowledge is value-laden, because it is the product of
social groups, and embodies their purposes and interests.
Social constructivism denies that mathematical knowledge is value-free. First,
because it rejects the categorical distinction between mathematics and science, and as
is increasingly accepted by philosophers of science, science is value-laden. Second,
because it posits a shared linguistic basis for all knowledge, which since it serves all
human purposes, it is imbued with human values. The mathematical use of language,
formal and informal, attempts to eradicate values, by adhering to objective logical
rules for the definition and justification of mathematical knowledge. However, the
use of the hypothetico-deductive (i.e. axiomatic) method means that values are
involved in the choice of hypotheses (and definitions). Beyond this, there are values
implicit in logic and the scientific method.
Although mathematics was thought to epitomize value-free objectivity, alongside
the sociology of knowledge social constructivism rejects this belief, arguing that
objectivity itself is social, and that consequently mathematical knowledge is laden
with human and cultural values.

The reification of concepts

Fourth, there is the reification of concepts, in which they become autonomous,
objective things-in-themselves. In sociology, this mechanism was first proposed by
Marx, by analogy with the genesis of religion.

…the productions of the human brain appear as independent beings

endowed with life, and entering into a relation both with one another and
with the human race.
(Marx, 1967, page 72)

He argues that the form of products becomes reified and fetishized into an abstract
thing: money, value or commodity (Lefebvre, 1972). Subsequent theorists in this
tradition of thought, such as Lukacs, have extended the range of operation of
reification to a much broader range of concepts.
Evidently the social constructivist thesis concerning the reification of newly
defined concepts has a strong parallel in Marxist sociology. This parallel has been
extended to mathematics by Davis (1974) and others such as Sohn-Rethel, as
Restivo (1985) reports.

96
 The Parallels of Social Constructivism

C. The Sociology of Mathematics

The sociology of mathematics is a substantial field of study concerning the social

development and organization of mathematics, as the survey in Restivo (1985)
indicates. In contrast with the philosophy of mathematics, it is concerned to offer
empirical theories of the growth, development and organisation of mathematical
knowledge. To achieve this, it tries to account for mathematics and mathematical
knowledge as a social construction 3 (unlike the traditional philosophies of
mathematics). Consequently, the social constructivist philosophy of mathematics
offers a parallel with sociologial accounts, but whereas the former is concerned with
a logical and conceptual analysis of the conditions of knowledge, the latter is
concerned with the social determinants of the actual body of knowledge.
One of the aims of social constructivism is to offer a descriptive philosophy of
mathematics, as opposed to the prescription of the traditional philosophies. Thus
parallel accounts of mathematics from sociological, as well as historical and
psychological perspectives should be possible. Therefore this section offers a
preliminary sociology of mathematical knowledge.

A social constructivist sociology of mathematics

From a sociological perspective, mathematics is the name given to the activities of,
and knowledge produced by a social group of persons called mathematicians. When
linked to social history by a definition like this, the term ‘mathematics’ has an
organic, changing denotation, as does the set of mathematicians.

‘[M]athematics’ by 1960, consisted of various subgroups working, to some

extent, within different cognitive and technical norms, on different orders
of phenomena and different types of problems. What had changed, with
some exceptions—such as computing—was the relative numerical strength
and status within the overall discipline of groups carrying particular norms.
(Cooper, 1985, page 7)

Subjects (e.g. mathematics) will be regarded not as monoliths, that is as

groups of individuals sharing a consensus both on cognitive norms and on
perceived interests, but rather as constantly shifting coalitions of individuals
and variously sized groups whose members may have, at any specific
moment, different and possibly conflicting missions and interests. These
groups may, nevertheless, in some arenas, all successfully claim allegiance to
a common name, such as ‘mathematics’.
(Cooper, 1985, page 10)

These complexities form a backdrop of the brief, conjectural sociological account of

mathematics that follows, in line with social constructivism.

97
The Philosophy of Mathematics Education

(i) Mathematicians. At any one time, the nature of mathematics is determined primarily
by a fuzzy set of persons: mathematicians. The set is partially ordered by the relations
of power and status. The set and the relations on it are continually changing, and thus
mathematics is continuously evolving. The set of mathematicians has different
strengths of membership (which could in theory be quantified from 0 to 1). This
includes ‘strong’ members (institutionally powerful or active research mathematicians)
and ‘weak’ members (teachers of mathematics). The ‘weakest’ members could simply
be numerate citizens. The notion of a fuzzy set usefully models the varying strengths
of individuals’ contribution to the institution of mathematics. Mathematical
knowledge is legitimated through acceptance by the ‘strongest’ members of the set. In
practice the set of mathematicians is made up of many sub-sets pursuing research in
sub-fields, each with a similar sub-structure, but loosely interconnected through
various social institutions (journals, conferences, universities, funding agencies).

(ii) Joining the set. Membership of the set of mathematicians results from an extended
period of training (to acquire the necessary knowledge and skills) followed by
participation in the institutions of mathematics, and presumably the adoption of (at
least some) of the values of the mathematics community (Davis and Hersh, 1980;
Tymoczko, 1985). The training requires interaction with other mathematicians, and
with information technology artefacts (books, papers, software, etc.). Over a period
of time this results in personal knowledge of mathematics. To the extent that it exists,
the shared knowledge of mathematics results from this period of training in which
students are indoctrinated with a ‘standard’ body of mathematical knowledge. This is
achieved through common learning experiences and the use of key texts, which have
included Euclid, Van der Waerden, Bourbaki, Birkhoff and MacLane, and Rudin, in
the past. Many, probably most students fall away during this process. Those that
remain have successfully learned part of the official body of mathematical knowledge
and have been ‘socialized’ into mathematics. This is a necessary, but not sufficient
condition for entry into the set of mathematicians (with a membership value
significantly greater than 0). The ‘standard’ body of knowledge will have a shared
basis, but will vary according to which subfields the mathematician contributes.

(iii) Mathematicians’ culture. Mathematicians form a community with a mathematical

culture, with sets of concepts and prior knowledge, methods, problems, criteria of
truth and validity, methodology and rules, and values, which are shared to a varying
degree. A number of authors have explored the culture and values of mathematics,
including Bishop (1988), Davis and Hersh (1980) and Wilder (1974, 1981). Here we
will undertake a more limited inquiry, restricted to the different realms of discourse
and knowledge of mathematicians, and their associated values. The analysis given here
is three-fold, proposing that mathematicians operate with knowledge on the three
levels of the syntax, semantics and pragmatics of mathematics. This is based on the
classificatory system of Charles Morris (1945) who distinguishes these three levels in
language use. In his sytem the syntax, semantics and pragmatics of a language refer to
the formal rule system (grammar and proof), the system of meanings and interpretations,

98
 The Parallels of Social Constructivism

and the nexus of human rules, purposes and decisions concerning language use,
respectively. In constructing this system, Morris added to the formal logical levels of
syntax and semantics a further level of pragmatics, inspired by pragmatism.
There is also a parallel with the three interlocking systems of language
distinguished by Halliday (1978), consisting the forms, meanings and functions of
language. In the sociology of mathematics, Restivo (1985) distinguishes the syntactical
and semantic properties of an object (following Hofstadter), paralleling the syntax-
semantics distinction. Hersh (1988) makes an analogous distinction between the
‘front’ and ‘back’ of mathematics. Restivo (1988) also distinguishes between ‘social’
and ‘technical’ talk of mathematics, paralleling the distinction between the third level
of pragmatic considerations and the first two levels taken as one, respectively. Thus
precursors of these three levels, in various forms, are to be found in the literature.
The three levels of mathematical discourse proposed are as follows. First of all,
there is the level of syntax or formal mathematics. This consists of rigorous
formulations of mathematics, consisting of the formal statement and proof of results,
comprising such things as axioms, definitions, lemmas, theorems and proofs, in pure
mathematics, and problems, boundary conditions and values, theorems, methods,
derivations, models, predictions and results in applied mathematics. This level includes
the mathematics in articles and papers accepted for conferences and journal
publication, and constitutes what is accepted as official mathematics. It is considered
to be objective and impersonal, the so-called ‘real’ mathematics. This is the level of
high status knowledge in mathematics, what Hersh (1988) terms ‘the front’ of
mathematics. This level is not that of total rigour, which would require exclusive use
of one of the logical calculi, but of what passes in the profession for acceptable rigour.
Secondly, there is the level of informal or semantic mathematics. This includes
heuristic formulations of problems, informal or unverified conjectures, proof
attempts, historical and informal discussion. This is the level of unofficial
mathematics, concerned with meanings, relationships and heuristics. Mathematicians
refer to remarks on this level as ‘motivation’ or ‘background’. It consists of subjective
and personal mathematics. It is considered to be low status knowledge in
mathematics, what Hersh (1988) terms ‘the back’ of mathematics.
Third, there is the level of pragmatic or professional knowledge of mathematics
and the professional mathematical community. It concerns the institutions of
mathematics, including the conferences, places of work, journals, libraries, prizes,
grants, and so on. It also concerns the professional lives of mathematicians, their
specialisms, publications, position, status and power in the community, their work
places and so on. This is not considered to be mathematical knowledge at all. The
knowledge has no official status in mathematics, since it does not concern the
cognitive content of mathematics, although aspects of it are reflected in journal
announcements. This is the level of ‘social talk’ of mathematics (Restivo, 1988).
These three levels are the different domains of practice within which
mathematicians operate. As languages and domains of discourse they form a
hierarchy, from the more narrow, specialized and precise (the level of syntax), to the
more inclusive, expressive and vague (the level of pragmatics). The more expressive

99
The Philosophy of Mathematics Education

systems can refer to the contents of the less expressive systems, but the relation is
asymmetric.
The hierarchy also embodies some of the values of mathematicians. Namely, the
more formal, abstract and impersonal that the mathematical knowledge is, the more
highly it is valued.The more heuristic, concrete and personal mathematical knowledge
is, the less it is prized. Restivo (1985) argues that the development of abstract mathematics
follows from the economic and social separation of the ‘hand’ and ‘brain’. For abstract
mathematics is far removed from practical concerns. Since the ‘brain’ is associated with
wealth and power in society, this division may be said to lead to the above values.
The values described above lead to the identification of mathematics with its
formal representations (on the syntactical level). This is an identification which is
made both by mathematicians, and philosophers of mathematics (at least those
endorsing the absolutist philosophies). The valuing of abstraction in mathematics may
also partly explain why mathematics is objectified. For the values emphasize the pure
forms and rules of mathematics, facilitating their objectification and reification, as
Davis (1974) suggests.4 This valuation allows the objectified concepts and rules of
mathematics to be depersonalized and reformulated with little concerns of ownership,
unlike literary creations. Such changes are subject to strict and general mathematical
rules and values, which are a part of the mathematical culture. This has the result of
offsetting some of the effects of sectional interests exercised by those with power in
the community of mathematicians. However, this in no way threatens the status of the
most powerful mathematicians. For the objective rules of acceptable knowledge serve
to legitimate the position of the elite in the mathematical community.
Restivo (1988) distinguishes between ‘technical’ and ‘social’ talk of mathematics,
as we saw, and argues that unless the latter is included, mathematics as a social
construction cannot be understood. Technical talk is identified here with the first and
second levels (the levels of syntax and semantics), and social talk is identified with the
third level (that of pragmatics and professional concerns).
Denied access to this last level, no sociology of mathematics is possible, including
a social constructionist sociology of mathematics. However, social constructivism as a
philosophy of mathematics does not need access to this level, although it requires the
existence of the social and language, in general. An innovation of social
constructivism is the acceptance of the second level (semantics) as central to the
philosophy of mathematics, following Lakatos. For traditional philosophies of
mathematics focus on the first level alone.
Sociologically, the three levels may be regarded as distinct but inter-related
discursive practices, after Foucault. For each has its own symbol systems, knowledge
base, social context and associated power relationships, although they may be hidden.
For example, at the level of syntax, there are rigorous rules concerning acceptable
forms, which are strictly maintained by the mathematics establishment (although
they change over time). This can be seen as the exercise of power by a social group.
In contrast, the absolutist mathematician’s view is that nothing but logical reasoning
and rational decision-making is relevant to this level. Thus a full sociological
understanding of mathematics requires an understanding of each of these discursive

100
 The Parallels of Social Constructivism

practices, as well as their complex inter-relationships. Making these three levels

explicit, as above, is a first step towards this understanding.

D. Sociological Parallels of Social Constructivism

The above suggests that social constructivism may offer a potentially fruitful parallel
sociological account of mathematics. Such a parallel, highly compatible with social
constructivism is already partly developed by Restivo (1984, 1985, 1988) and others.
Although sociological parallels do not add weight to social constructivism in purely
philosophical terms, they offer the prospect of an interdisciplinary social constructivist
theory, offering a broader account of mathematics than a philosophy alone. Mathematics
is a single phenomenon, and a single account applicable to each of the perspectives of
philosophy, history, sociology and psychology is desirable, since it reflects the unity of
mathematics. If successful, such an account would have the characteristics of unity,
simplicity and generality, which are good grounds for theory choice.5

4. Psychological Parallels

A. Constructivism in Psychology

Constructivism in psychology can be understood in narrow and broad senses.6 The

narrow sense is the psychological theory of Piaget and his school. Piaget’s
epistemological starting point resembles that of social constructivism in its treatment
of subjectve knowledge. His epistemological assumptions are developed into the
philosophy of radical constructivism by von Glasersfeld, as we have seen. However,
Piaget’s psychological theory goes far beyond its epistemological starting point. Fully
articulated, it is a specific empirical theory of conceptual development, with
particular concepts and stages. It also assumes the narrow Bourbakiste structural view
of mathematics, which is not compatible with social constructivism.
The Bourbaki group have been developing and publishing a unified axiomatic
formulation of pure mathematics for about fifty years in Elements de Mathematique
(see for example, Kneebone, 1963). Their formulation is structuralist, based on
axiomatic set theory in which three ‘mother-structures’ are defined: algebraic,
topological and ordinal, providing the foundation for pure mathematics. As a view of
mathematics, the Bourbaki programme may be criticised as narrow. First, because it
excludes constructive mathematical processes, and second, because it represents
mathematics as fixed and static. Thus it reflects the state of mathematics during a
single era (mid twentieth-century), although this is denied in Bourbaki (1948). It is
incompatible with social constructivism because of this narrowness, and because it is
a foundationist programme, and hence is implicitly absolutist.
However, the Bourbaki programme is not a philosophy of mathematics, and does
not need to defend itself against this criticism. For it can be seen merely as a

101
The Philosophy of Mathematics Education

programme, albeit ambitious, to reformulate the structural part of mathematics. But

Piaget views Bourbaki as revealing the nature of mathematics. Thus this criticism can
be validly directed at Piaget’s implicit philosophy of mathematics, rendering the
details of his psychological theory incompatible with social constructivism. For he
takes the three ‘mother-structures’ of Bourbaki as a priori, and assumes that they are
an integral part of the psychological development of individuals. This is evidently an
error, due to a misinterpretation of the significance of Bourbaki.
Other aspects of Piaget’s theory do offer a parallel to social constructivism. For
example, the notion of ‘reflective abstraction’, which allows mental operations to
become objects of thought in their own right, accommodates the social
constructivist thesis of mathematical objects as reifications. However, much of
Piaget’s developmental psychology, such as his stage theory, goes beyond any parallel
with social constructivism, and is extensively criticized on both psychological
(Bryant, 1974; Brown and Desforges, 1979; Donaldson, 1978), and mathematical
grounds (Freudenthal, 1973).
The broad sense of constructivism in psychology is what Glasersfeld (1989) refers
to as ‘trivial constructivism’, based on the principle that knowledge is not passively
received but actively built up by the cognizing subject. This broad sense encompasses
many different psychological theories including the personal construct theory of
Kelly (1955), the information processing theory of Rumelhart and Norman (1978),
the schema theory of Skemp (1979) and others, the social theory of mind of
Vygotsky (1962), as well as the basis of the constructivism of Piaget and his followers.
This list indicates some of the diversity of thought that falls under the broad sense of
constructivism. What these authors share is a belief that the acquisition and develop¬
ment of knowledge by individuals involves the construction of mental structures
(concepts and schemas), on the basis of experience and reflection, both on
experience and on mental structures and operations. Many, but not all psychologists
in this group accept that knowledge grows through the twin processes of assimilation
and accommodation, first formulated by Piaget.
On the basis of their epistemological assumptions alone, both the broad and
narrow senses of constructivism offer a psychological parallel to social constructivism.
The auxiliary hypotheses of individual constructivist psychologies, such as Piaget’s,
may be incompatible with the social constructivist philosophy of mathematics. But
the potential for a psychological theory of mathematics learning paralleling social
constructivism clearly exists.
A number of researchers are developing a constructivist theory of mathematics
learning, including Paul Cobb, Ernst von Glasersfeld and Les Steffe (see, for example,
Cobb and Steffe, 1983; Glasersfeld, 1987; Steffe, Glasersfeld, Richards and Cobb,
1983). As they appear to have rejected many of the problematic aspects of Piaget’s
work, such as his stages, much of their theory can be seen as parallel to social
constructivism, on the psychological plane. However it is not clear that all of their
auxiliary assumptions, such as those involved in accounting for young children’s
number acquisition, are fully compatible with social constructivism.
No attempt will be made to develop a psychological parallel to social

102
 The Parallels of Social Constructivism

constructivism here, although in the next sections we consider briefly some of the
key components of such a theory.

B. Knowledge Growth in Psychology

Following Piaget, schema theorists such as Rumelhart and Norman (1978), Skemp
(1979) and others, have accepted the model of knowledge growth utilizing the twin
processes of assimilation and accommodation. These offer parallels to the social
constructivist accounts of subjective and objective knowledge growth. For knowledge,
according to this account, is hypothetico-deductive. Theoretical models or systems are
conjectured, and then have their consequences inferred.This can include the applications
of known procedures or methods, as well as the elaboration, application, working out
of consequences, or interpretation of new facts within a mathematical theory or
framework. In subjective terms, this amounts to elaborating and enriching existing
theories and structures. In terms of objective knowledge, it consists of reformulating
existing knowledge or developing the consequences of accepted axiom systems or
other mathematical theories. Overall, this corresponds to the psychological process of
assimilation, in which experiences are interpreted in terms of, and incorporated into
an existing schema. It also corresponds to Kuhn’s (1970) concept of normal science, in
which new knowledge is elaborated within an existing paradigm, which, in the case of
mathematics, includes applying known (paradigmatic) procedures or proof methods to
new problems, or working out new consequences of an established theory.
The comparison between assimilation, on the psychological plane, and Kuhn’s
notion of normal science, in philosophy, depends on the analogy between mental
schemas and scientfic theories. Both schemas (Skemp, 1971; Resnick and Ford, 1981)
and theories (Hempel, 1952; Quine 1960) can be described as interconnected
structures of concepts and propositions, linked by their relationships. This analogy has
been pointed out explicitly by Gregory (in Miller, 1983), Salner (1986), Skemp
(1979) and Ernest (1990), who analyzes the parallel further.
The comparison may be extended to schema accommodation and revolutionary
change in theories. In mathematics, novel developments may exceed the limits of
‘normal’ mathematical theory development, described above. Dramatic new methods
can be constructed and applied, new axiom systems or mathematical theories
developed, and old theories can be restructured or unified by novel concepts or
approaches. Such periods of change can occur at both the subjective and objective
knowledge levels. It corresponds directly to the psychological process of
accommodation, in which schemas are restructured. It also corresponds to Kuhn’s
concept of revolutionary science, when existing theories and paradigms are
challenged and replaced.
Piaget introduced the concept of cognitive conflict or cognitive dissonance (which
will not be distinguished here). In the social constructivist account of mathematics, this
has a parallel with the emergence of a formal inconsistency, or a conflict between a
formal axiom system and the informal mathematical system that is its source (Lakatos,

103
The Philosophy of Mathematics Education

1978a). This is analogous to cognitive conflict, which occurs when there is conflict
between two schemas, due to inconsistency or conflicting outcomes. In psychology, this
is resolved through the accommodation of one or both of the schemas. Likewise in
mathematics, or in science, this stimulates revolutionary developments of new theories.
Overall, there is a striking analogy between theory growth and conflict in the social
constructivist philosophy of mathematics and schema theory in psychology, and
underlying it, between theories and schemas. Unlike the situation in the philosophy of
mathematics, schema theory, as sketched above, represents the received view in
psychology, lending support to a psychological parallel for social constructivism.

C. Reification and Concept Formation

The social constructivist philosophy of mathematics distinguishes two modes of

concept development, vertical processes of concept formation, involving the
reification of concepts into objects, and horizontal processes. These can be elaborated
as part of a psychological parallel to social constructivism.
We may conjecture that psychological concept formation involves both vertical
and horizontal processes. The vertical processes include the standard processes of
concept formation, namely the generalization and abstraction of shared features of
earlier formed concepts to form new concepts. Beyond this, we conjecture the
existence of a psychological mechanism or, tendency which transforms mental
procedures or processes into objects. This mechanism changes a property, a
construction, a process, or an incomplete collection into a mental object, a complete
thing-in-itself. What is represented as a process, a verb or an adjective becomes
represented as a noun. This is ‘reification’ or ‘objectification’. Psychologically, much
concept formation has this character. Even in the act of coordinating different
perceptions of an external object, in sensory concept formation, we reify the set of
perceptions into the concept, an enduring object-representation in a schema.
There is some parallel between this conjectured ‘Vertical’ mechanism and Piaget’s
notion of reflective abstraction, the process whereby an individual’s operations, both
physical and mental, become represented cognitively as concepts. Thus reflective
abstraction includes concept reification, although the former is a broader notion.
A number of other researchers have proposed psychological theories dealing
specifically with concept reification (Skemp, 1971). Dubinsky (1988, 1989) includes
‘encapsulation’ as part of his explication of the notion of reflective abstraction.
Encapsulation converts a subjective mathematical process into an object, by seeing it
as a total entity. Sfard (1987, 1989) has been testing a theory of mathematical concept
development, in which operational concepts are transformed into structural concepts,
by a process of reification. Both these researchers have empirical data consistent with
the hypothesis that a process of encapsulation or reification occurs in vertical concept
formation. Thus there is evidence for a psychological process of vertical concept
formation, parallel to the social constructivist account, and accounting for subjective
belief in platonism.

104
 The Parallels of Social Constructivism

D. Individualism in Subjective Knowledge

A central feature of the social constructivist theory is that subjective knowledge

comprises idiosyncratic personal meanings, concepts and knowledge structures.
These are subject to the constraints imposed by the external and social worlds, but
this leaves room for considerable variation. A psychological version of this thesis
would predict that significant variations in concepts and knowledge should occur
between individuals, both within a single culture, and even more so in inter-cultural
comparisons. This hypothesis seems to be confirmed empirically, although there is, of
course, the methodological problem of comparing private meanings. Any evidence
about individuals’ personal meanings and knowledge must be based on inference and
conjecture, for subjective knowledge is, by definition, unavailable for public scrutiny.
A number of different psychological approaches provide evidence of the
uniqueness of individuals’ subjective knowledge. First of all, there is research on
errors in mathematics learning (Ashlock, 1976; Erlwanger, 1973; Ginsburg, 1977).
From the patterns observed, it is clear that many errors are systematic and not
random. The range of errors observed in learners suggests that they are not taught,
and that learners construct their own idiosyncratic concepts and procedures.
Secondly, researchers are finding that ‘alternative conceptions’ (i.e., idiosyncratic
personal concepts) are also very widespread in science (Abimbola, 1988; Driver,
1983; Pfundt and Duit, 1988). Thirdly, researchers have tried to represent learners’
cognitive structures in mathematics, using a variety of data-gathering methods. Their
findings have included spontaneous (i.e., untaught) sequences of procedures in
learning arithmetic (Steffe et al., 1983; Bergeron et al., 1986), and unpredictable
growth in the links in personal concept hierarchies (Denvir and Brown, 1986).
These approaches illustrate the broad base of empirical and theoretical support for
a psychological version of social constructivism. Individuals do seem to construct
unique personal meanings and conceptual structures. There are, however, patterns to
be found in these constructions across individuals (Bergeron et al., 1986), presumably
reflecting the similar mental mechanism generating subjective knowledge, and the
comparable experiences and social contexts of individuals.

E. Social Negotiation as a Shaper of Thought

A central thesis of social constructivism is that the unique subjective meanings and
theories constructed by individuals are developed to ‘fit’ the social and physical
worlds. The main agency for this is interaction, and in the acquisition of language,
social inter-action. This results in the negotiation of meanings, that is the correction
of verbal behaviour and the changing of underlying meanings to improve ‘fit’.
Briefly put, this is the conjectured process by means of which the partial inner
representation of public knowledge is achieved.
This thesis is close to the social theory of mind of Vygotsky (1962) and his
followers. Vygotsky’s theory entails that for the individual, thought and language

105
The Philosophy of Mathematics Education

develop together, that conceptual evolution depends on language experience, and of

particular relevance to social constructivism, that higher mental processes have their
origin in interactive social processes (Wertsch, 1985).

Vygotsky’s point is not that there are hidden cognitive structures awaiting
release through social interaction. His point is the radical one that they are
formed through social interaction. Development is not the process of the
hidden becoming public, but on the contrary, of the public and inter-
subjective becoming private.
(Williams, 1989, page 113)

Thus Vygotsky’s social theory of mind offers a strong parallel with social
constructivism, one that can also be found elsewhere in psychology, such as Mead’s
(1934) symbolic interactionism. A further development in this direction is the
Activity Theory of Leont’ev (1978), with perceives psychological motives and
functioning as inseparable from the socio-political context. Possibly less radical is the
move to see knowing as bound up with its context in ‘Situated cognition’ (Lave,
1988; Brown et al., 1989), although Walkerdine (1988, 1989) proposes a fully social
constructionist psychology of mathematics. Social constructionism as a movement in
psychology is gaining in force, as Harre (1989) reports, and is replacing the
traditional developmental or behaviourist paradigms of psychology with that of social
negotiation. Harre goes so far as to propose that inner concepts such as self-identity
are linguistic-related social constructions.

F. Psychological Parallels

A number of psychological parallels of social constructivism have been explored,

including ‘constructivism’ and ‘social constructionism’. Many reflect a dominant
view in psychology, contrasting with the controversial position of social
constructivism in the philosophy of mathematics. Thus it seems likely that a
psychological version of social constructivism, enriched with appropriate empirical
hypotheses, could offer a fruitful account of the psychology of mathematics.

5. Conclusion: a Global Theory of Mathematics

Social constructivism is a philosophy of mathematics, concerned with the possibility,

conditions and logic of mathematical knowledge. As such, its acceptability depends on
philosophical criteria. It has been shown to have more in common with some other
branches of philosophy, than with traditional philosophies of mathematics, for it
inescapably raises issues pertaining to empirical knowledge, and to the social and
psychological domains. Despite raising such issues, no empirical assumptions concerning
the actual history, sociology or psychology of mathematics have been made.

106
 The Parallels of Social Constructivism

Due to the multidisciplinary nature of the issues raised, there is also the prospect
of a unified social constructivist account of mathematics. The aim of this section is to
propose an overall social constructivist theory of mathematics, incorporating its
philosophy, history, sociology and psychology. These are distinct disciplines, with
different questions, methodologies and data. What is proposed is an overarching
social constructivist meta-theory of mathematics, to provide schematic explanations
treating the issues and processes in each of these fields, to be developed to suit the
characteristics and constraints of that field. This would result in parallel social
constructivist accounts of:
1 the history of mathematics: its development at different times and in different
cultures;
2 the sociology of mathematics; mathematics as a living social construction,
with its own values, institutions, and relationship with society in the large;
3 the psychology of mathematics: how individuals learn, use and create
mathematics.
The goal of providing such a meta-theory of mathematics is ambitious, but legitimate.
Theoretical physics is currently seeking to unify its various theories into a grand theory.
In the past century other great strides have been made to unify and link sciences.There
have been ambitious schemes to document a shared methodology and foundations,
such as the International Encyclopedia of Unified Science.The history of mathematics
likewise provides many examples of theoretical unification.What is claimed here is that
this is also a desirable goal for the philosophy of mathematics.
There are a number of reasons why such a project is worthwhile. First of all, as
mathematics is a single discipline and social institution, it is appropriate to coordinate
different perspectives of it, for the unity of mathematics should transcend the divisions
between disciplines. A meta-theory which reflects this unity gains in plausibility, and
reflects the characteristics of a good theory, namely agreement with the data,
conceptual integration and unity, simplicity, generality and, it is to be hoped, fertility.
Secondly, beyond this general argument is the fact of the strong parallels between
the social constructivist philosophy and the history, sociology and psychology of
mathematics demonstrated above.These are not coincidental, but arise from genuinely
interdisciplinary issues inherent in the nature of mathematics as a social institution.
Thirdly, in exploring these parallels one factor has recurred, the greater
acceptability of the parallel theses in general philosophy, sociology, psychology and
the history of mathematics, than in the philosophy of mathematics. In these fields,
many of these theses are close to the received view or a major school of thought. In
particular, social constructionist views in sociology and psychology have a great deal
of support. This contrasts strongly with the position in the philosophy of
mathematics, where absolutist philosophies have dominated until very recently. Thus
the call for a social constructivist meta-theory of mathematics is stronger from the
surrounding fields than from the traditional philosophy of mathematics.
Fourthly, one of the theses of social constructivism is that there is no absolute
dichotomy between mathematical and empirical knowledge. This suggests the

107
The Philosophy of Mathematics Education

possibility of a greater rapprochement than hitherto, between the logical concerns of

philosophy, and the empirical theories of history, sociology and psychology. An
overarching social constructivist meta-theory of mathematics would offer such a
rapprochement. Such a theory is therefore proposed, in the spirit of developing the
self-consistent (i.e., reflexive) application of social constructivism.7

Notes

1 There is an interesting analogy between deductive proof and Kitcher’s (1984) justification of
mathematical knowledge, which might be its source. Just substitute axioms for his basis and inference
as the means of deriving each stage from the next.
2 An example of this ‘standard’ view is that of Scheffler: ‘science is a systematic public enterprise,
controlled by logic and empirical fact whose purpose is to formulate the truth about the empirical
world’. (Brown et al., 1981, page 253)
3 A leading exponent of social constructionism as a sociology of mathematics is Sal Restivo (1984,
1985, 1988). (In addition Restivo, 1984, offers valuable insights into social constructivism as a
philosophy of mathematics). David Bloor (1976, 1983) has made major contributions to both the
sociology and philosophy of mathematics as a social construction.
4 Restivo (1985, page 192) also suggests, following Struik, that the separation of form from content in
the objectification of mathematical knowledge is a product of the prevailing social conditions. The
argument is that idealism results from, and provides a solution to problems in social outlook, during
periods of social decline, such as the disintegration of the western Roman empire, and the
enfeeblement of empire. Similarly Koestler (1964, page 57) suggests that Plato’s idealism was a
response to the decline of Greece. An interesting analogy might be drawn with the development of
the rigid philosophy of logical positivism in post-Great war Austria and Germany.
5 The strength of the sociological parallels might be used to direct a charge of sociologism against
social constructivism, claiming that it is a sociological theory of mathematical knowledge, which
although avoiding overtly empirical matters, remains essentially sociological. My response is that the
primary focus is on the general conditions and justification of mathematical knowledge, which is the
proper concern of the philosophy of mathematics.
6 ‘Constuctivism’ has many meanings. Below two senses of ‘constructivism’ are distinguished in
psychology. In the philosophy of mathematics, ‘constructivism’ encompasses intuitionism and similar
schools of thought. The psychological and philosophical senses are quite distinct (Lerman, 1989).
‘Social constructivism’ introduces another sense into the philosophy of mathematics. Social
constructivism is also applied in the sociology of mathematics, by Restivo. ‘Constructivism’ also
denotes a movement in the history of modern art, with proponents such as Gabo, Pevsner and Tatlin.
The account of judicial reasoning of Ronald Dworkin (in his 1977 book ‘Taking Rights Seriously’)
is termed ‘constructivist’, according to the Fontana Dictionary of Modern Thought.
Doubtless further ‘constructivist’ schools of thought exist in other disciplines. What they seem to
share is the metaphor of construction: the product involved is built up by a synthetic process from
previously constructed components.
7 An outcome of the social constructivist meta-theory of mathematics might be to demystify the
philosophy of mathematics. For if the meta-theory is possible, then the strict demarcation of the
disciplines may be seen as the reification, mystification and even the fetishization of philosophy and
mathematics. The force with which the inviolability of the boundaries has been asserted (e.g. by
logical positivists and empiricists) resembles a social taboo. It is surely in the interests of knowledge
to offer a rational challenge to such a taboo, even if it is against the interests of the professionals who
have created the mystique.

108