Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT. The aim of this paper is to elucidate the concepts of society. social graup,
power, class, structure, and change. This clarification is performed essential1y with the
help of the concepts of relation (in particular equivalence relation). set (in particular
equivalence c1ass), funetion, and matrix. In addition to that analysis, a framework for
sociological hypotheses and theories is evolved, reminiscent of certain ideas in mathe~
matical ecology. Thus social structure is defined as a matrix exhibiting the distribution
of the total population of a cornmunity among the various social groups resulting from
the partitions of the society induced by so many social equivalence relations. And
social change is defined as a redistribution of the relative populations oC those social
groups. The advantages and Iimitations of this approach are discussed. Also. a number
of methodological remarks are made.
INTROOUCTlON ANO STATEMENT OF THE PROBLEM
The social sciences are centrally concemed with social groups, structures,
and changes. It is hoped that the corresponding concepts are made precise
in the course of becoming part and parcel of sociological theories. This
hope is only partially justilied: systematization, though necessary for
elucidation, is not sufficient unless the theories in question are formulated
in a full and aceurate way. But this degree of relinement is attained only
rarely: the social scientist is usually in a hurry to account for some social
facts rather ilian to make a contribution to the foundations, methodology
or philosophy of his subject.
The elucidation of key sociological concepts may be regarded as one
of the two tasks of the foundations of sociology - ilie other being the
axiomatic reconstruction of basic sociological theories. That discipline
may indeed be construed as the set of conceptual analyses and syntheses
aiming at clarifying and interrelating (systematizing) the generic, hence
basic, ideas of social science. The emphasis here is on the generc in contrast to the specific - e.g. on social groups rather than income groups or
educational groups; on social structure in general instead of, say, kinship
or class structure; and on arbitrary social changes rather ilian, for instance, urbanization or poltica! revolution.
The foundations of sociology should supply frameworks within which
J..ellf/ellnttIK6hler (etls.). Developmel/ts in the Methodology 01 Social Science. 175-215 .4.11 Rights Rtserved
Copyright
1974 by Mario Bunge.
176
MARIO
BUNGE
the relation of being greater than: they are 'interna!' relations in the sense
that, if a and b are members of S and R is in R, and furthermore a holds
R to b, tben this 'makes a difference' to a or b. The environment E is
shared and transformed by all the members of j. Even if a member of S
owns no par! of E and takes no part in production, it has access to E and
in turo does its share in the modification (mainly degradation) of E. Note
finally Ihat, in Ihe simplesl case, a single organism (p = 1) transforms a
single element of E (i.e. q= 1) into anolher part of E. In this particular
case, Ihen, Ihe extension of the transformation relalion is incJuded in Ihe
carlesian product of S by E 2 Sometimes (e.g. coral reefs, beehives, industrial societies) it lakes a number p of organisms acling upon a number
q of environmental nnils lo process the laller. Whenee the exponents p
and q in clause (iv).
Since we shall be concerned with human communities, we mayas well
define this latter concept as a certain specification of the generic concept
of a community. We adopt
the generic socio10gical concepts can get cJeansed and interrelated. Such
frameworks may not only cJarify sorne ofthe existing constructs, but may
also help to build substantive hypotheses or theories. However, tbe formulation of particular sociological hypotheses and theories - e.g. on the
relation of participation to social cohesiveness - lies aboye the foundational level: it is tbe concem of substantive tbeoretical work. Our own
work, concemed as it is with tbe concept of social structure ratber tban
witb social structure, will belong to the foundations of sociology.
In this paper we wish to elucidate five basic generic sociological concepts: those of society, social group, social power, social structure, and
social change. To this end we shall avail ourselves of some extremely
simple matbematical concepts and we shall adapt some ideas of mathematical ecology.
1. SOCIETY AND PRODUCTION
1.1. Society
177
I,
.)
ing il), and hold social relations among themselves. Work, an ingredient
of human communities, does nol suffiee to ebaraclerize them: Ihe animal
communities that work (e.g. antbills) do not qualify as human communitieso Therefore any definilion of man as 'the ... ing animal' (e.g. Ihe 1001
178
MARIO BUNGE
an artifact;
sometbing not found in a state of nature. Solitary meditation in a monastery does not qualify. Of course it may happen that a community subsists for sorne time
179
00
But even so it wiII have to engage in sorne \York or other and, in any case,
it won't be self supporting. hence lasting. In other words. it is an unspoken axiom that the primary manpower of every society. referred to in
clause (vi). be nonempty. Note fnrther that our concept of primary work
does not coincide with the standard concept of prirnary industry: the
primary producers referred to above may be workingmen or farmers.
Nor does our notion of an artifact coincide with the usual one. Thus
In other words. the active part of a society is formed by all those persons
- farmers, industrial workers, and housewives - who transform raw material into artifacts. and tbose whose job it is to act upon other people.
. It is not implied that prirnary work be manual: it may consist in controlling automated machinery. Nor is it necessary !hat secondary work
consist in the direct control of other men: the invisible decision maker
in a buge industrial concero does a secondary job and so does the civil
servant. Finally it is not necessary that the memberships of the two manpowers be disjoint: in the family business P and Q may overlap or even
coincide.
181
MARIO BUNGE
hand and secondary workers on the other, and relations among secondary
workers. Figure 1 illustrates tbis triehotomy. To simplify the formal
definitions we shall restrict ourselves to ternary and quaternary work
relalions. We propose
180
Manager
N
Foreman
1.2. Production
The preceding definitions will aIlow us to clarify three notions that are
central to Marxian economie and soeial thought but seem to have remained heretofore at an intuitive stage: namely those of material foree
of produetion, social relation ofproduetion, and superstrueture. We take
the former to eonsist of the environment together with the primary work
in a community, and its produetion. Sinee doing productive work of any
kind involves knowing how to do it (wilh greater or lesser awareness and
e/lidency), speeifying the set W of work relalions involves describing lhe
technology available to the soeiety of interest. Thus we have
:-~
Fig. 1.
Let f/' = (S, R, E, T, W, M) represent a soeiety with primary manpower P and secondary manpower Q. Then
(i) for any X and y in P : x and y hold a primary (or coproduction)
relationship between each olher if there is at least one member K of R
such that, for some members W, W k and W m of W,
Kxy iff ((Wxea and WkYe'a') or W".xyea for somee, e'EEandsome
DEFINITION 1.7.
Workingmen
a' EAc::A);
(ii) for any x in Q and any y in P: x andyhold amixedsocialrelation
oJ production if there is at least one member L of M such that, for sorne
Q,
"
W,in W,
Lxy and W,yea for some eEE and sorne oEA, cA;
(iii) for any x and y in Q : x and y hold a secondary relation oJ
produclion iff lhere is al least one relalion N in M such that, for sorne W.
in W,
Nxy and there is a z in P such that (Lxz or Lyz) and W.zea for sorne
eEE and some aEA.cA.
That is, for a social relation of production to hold between two persons,
it is necessary and suffieient that either of lhe following situations obtain:
182
MARIO BUNCE
(a) bolh individuals take parl in Ihe produetion of lhe soeiely as primary
produeers; (b) one of Ihe individuals produces while !he olher manages
the former; (e) the two iodividuals eomaoage at leasl ooe primary produeer. Thus while all lhe social relations of produetion are 'based 00'
work, as !he metaphor has it, not every kiod of work gives rise lo a social
relatioo ofproduetion. Forexample, makiog a toy for one's own ehild does
not establish a social relatioo of produetion.
Finally we come lo the Ihird notion we intended to elarify in this subseetion: that of superstructure. In Platonic terms, the latter is the set of
beliefs and rules as well as the fund of knowhow and knowledge of a
society. Since in our ootology ideas do not fiy around but are thought
out by people, we pro pose
"
.1
183
DEFINITION
"
MARIO BUNGE
184
185
We assume that there are societies in which sorne individuals exert power
of a eertain kind over other individuals, and Ihat this relationship seIs the
relata apart in different social elasses, such as Ihe landed aristocracy and
the serfs. Moreover we shall single out for analysis one particular kind of
power, namely economic power, and make Ihe methodological assumption that the strength or intensity of power is a numerical function. But
before introducing this notion it will be convenient to clarify the qualitative notion of influenee. To lhis end we shall take it for granted
Ihat the notions of human activity and communication have been
elucidated elsewhere. We need Ihe following combination of these two
coneepts:
DEFINmON 2.2. Let x and y be members of a society represented by
.'7' = <S, R, E, T, W, M). Then x influences y iff the activity of y while
communicating with x differs from whal y performs when no such communication flow exists.
Obviously, two persons may influence one another with unequal in
tensity. We need therefore a comparative and, better, a quantitative coneept of influenee. We offer the following:
DEFlNITION 2.3. Let x and y be members of a society represented by
.'7'= <S, R, E, T, W, M) and call N(y) the number ofactions y performs,
and N(y I x) tbe number of actions y performs nnder the inflnence of x.
Then the strength oi rhe influence x exerts over y is the function
1:S x S -> Q+ such that 1(x, y)
IN(y x) - N(y)1
Max{N(y), N(y Ix)}'
187
MARIO BUNGE
managed by the powerful party; afortiori it is immaterial whether property is private or pub lic. (These finer distinctions, though important, are
irrelevant to the question of effective economic control: the management
of a state enterprise may exert a s!ronger economic power than the owner
of a private concern.) Finally, the fourth clause is necessary as well: the
first three conditions alone are also satisfied by the pair consisting of a
manager and his docile wife.
Having secured a qualitative ootion of economc power, we can now
proceed to building a quantitative concept. We shall assume that economic power can be measured by the number of persons under control. More
precisely, we lay down
186
Not only people but also things can be subjected to control. We take it
that controlling a thing is working on or with it, or having somebody else
do the work for ns. More formally, we propose
DEFlNlTlON 2.5. Let x be a member of a society represented by [1' =
=<S,R,E,T, W,M) and call F={<E, W"A,) 1 <;;<;;r} me material
resources of S (Definition 1.6). Then
(i) the person x has direct control of the ming yEEu A iff there is a
W, in W such that Wxya for some aEA or Wxey for some eEE;
(ii) the person exer!s indirect control over the tlllng yEEu A iff mere
is at least one other person ZE S such that x controls z and in turn Z has
direct control of y;
(iii) the person x mobilizes (owns or manages) part of the material
resources F of S iff x has direct or indirect control of sorne of the things
that make up F.
Now, control can be ofvarious kinds: physical, psychological, or social.
We are interested in economic control, a kind of social control. We define
it as follows:
P(x)
N(x) .
N-l
Clearly, the values ofthe functionP: S ... Q+ are comprised between Oand
1. Any specific quantitative concept ofpower, such as that of economic or
political power, is obtained by suitably qualifying the words 'control'
and 'power' in the preceding definition.
The following derivative concepts suggest themselves:
DEFlNITION 2.8. Let x and y be members of a society represented by
[I'=<S, R, E, T, W, M) and call P(x) and P(y) respectively the powers
they wield. Then
(i) x is powerless in S iff P (x)", O;
(ji) x is more powerful than y in S iff P(xP(y);
(iii) x is the most powerful member of S iff, for an y in S,
188
MARIO BUNGE
x, yeS, d(x, y)
IP(x) - P(y)l,
IN (x) - N (y)1
N-1
where N(x) and N(y) are the numbers ofindividuals conlrolled economically by x and y respectively, and N is the population (cardinality) of S.
We can now exactify two further notions of interest:
DEFINITION 2.11. Let G"=(S, d) be an economic space. Then
(i) S is a class society iff, for sorne x, yeS, d(x, y);"O;
189
(ii) S is a e/assless society iff S is not a elass society (i.e. if, for all x,
P(x) =
I P(x I y) = eS
I
eS
I y)
T()
Y
T (x
Example: A slave holder who owns lOO slaves exerts a total power of
200/3, twice the power exerted by the manager who employs the same
number of workers.
The preceding interpersonal power measure suggests defining a social
separalitm function:
Two individuals are socially separated just in case either overpowers the
other. But this is clearIy inadequate, because two iudividuals may hold a
dependence relationship and yet be socially distant, like a given landlord
190
MARIO
BUNGE
and the serfs of another landlord. Besides, the above social separation
function u is not even a qnasimetric, which is mathematically annoying.
2.3. Social Class
In the last subsection we defined !he notion of a class society but not that
of class. We could try and use !he notion of economic distance to define
at least the concept of same elass, namely as follows;
Let C=(S, d) be an economic space. Then, for any persons x andy in C,
(i) x is classwi,. equivalent to y, i.e. x~,y, ilf x and y wield!he same
power, i.e. if they are not economically distant ;
,y ilf d (x, y) = O;
[x]
~,relation
elass of x in S;
= {y eS IY ~~}.
191
Now, there are two concepts of family; the anthropological and the
sociological ones, the former based on kinship relations only, the latter
adding a social componen!. In sociology a famUy is a kinship group of
people living together and sharing their possessions or at least sorne of
them. Call ~ I !he social equivalence relation of belonging lo Ihe sarne
family in !he sociological sense. The quotient of S by ~ I is then the
family ofF of all !he families in S. It is this set ofF that will be categorized
into social classes. A social class will be dellned as the collection of famiIies containing individuals who wield an economic power comprised between given bounds. More formally, we make
DEFINITION 2.12. Let .9'= (S, R, E, T, W, M) represent a society with
the family partition ofF = SI ~ l ' and let P be the economic power function
introduced by Definition 2.6. Then if a and b are preassigned reals, an
economic class in S is defined as the interval
I
tm
2 middle fams.; {GeofF I (3y)(yeG & -h '" P(y) '" rt)}
10 lower fams.; {He ofF I (3z) (zeH &P(z) = O)}
I top family;
MARIO BUNGE
We have built a concept of social cIass coinciding with that of an econoroic cIass bu! to be kept distinet froro Ibat of an occupational group.
The notion of equal social cIass follows irnmediately:
192
[x] = {YEff 1 y ~A
It is easily seen that ~ o is an equivalence relation. Consequently the
quotient set of ff by ~ o' i.e. ff1~ o = (SI ~ f)1 ~ o, wiII be Ibe class partition of the family partition of S. To emphasize we make
193
P(uP(v).
This dominance relation helps to define a hierarchical society:
DEFINITION 2.17. Let S be a society. Then S is a hierarehical society iff
(9" >- >- >is a simple order (i.e. if, for any Si' SjE9" eilber Si>- >-S} or
SJ>- >-S,).
The notion of economic distance, defined previously for individuals (Definition 2.8), carries over to classes. In fact we can
, stipulate that Ibe distance between two classes equals the smallest distance be!ween elements
of those cIasses. That is, we propase
DEFINITION 2.18. Let S be a society with family partition ff and cIass
structure 9" and let S, and SJ be members of 9,. The function
a:9o x9,-+R+ such that a(Si' SJ)=Min{d(u, v)[ for all
UEXES,
195
MARIO BUNGE
heterogeneity and, in particular, of elass heterogeneity. The simplest measure of the c1ass diversity of a society is, of coursc, just tbe number of
social elasses of the society, i.e. the number of members of the elass partition !!Y',. However, this simple index is clumsy because it does not indicate
how the total population is distributed among the classes. These two
independent variables, the number of elasses and their relative abundance,
194
D(S)=(N-1)-Z
D (S) oc
d (i, j).
, j e S
This diversity index is zero for a elassless soeiety and it increases with
both the social distanees and the abundance or population of eaeh elass.
The maximum value oceurs for d(i,j)=8lj' where is fue Kronecker
delta. In this case, hardly a realistie one exeept for a two tier society,
d(i,j)=tN(N-l).
,jES
To form an exaet definition we must take into aeeount that fue individual of interest is now a whole soeiety. Hence we must take the colleetion .11 of human socieles and define D on this set. We adopt
IN(i)-N(j)I,
index of cIass heterogeneity, and moreover ane in tune with our prcvious
work.
The coneept of eeonomic distance (Definition 2.7) allows us to build
a measure of the overall elass diversity of a community S, namely
L
,l eS
n=3 and m=5.) Since d(i, j)=ml(N-I) if i is upper elass andj lower
elass but zero otherwise, fue value ml(N -1) occurs n times in the summation. But nm=N-I, whenee the sum equals m, and we have
D (S)
m
= -N-l
- = 8 (upper, lower).
Example. Greek elassical period : m= 1, D(S)= l /N. Late Roman Empire: m= 100, D(S)=mIN= 1001N, i.e. 100 times fue economic heterogeneity of fue previous case.
DEFINlTION 2.19. Let .11 be fue set of human eommunities. Tbe c/ass
diversity (heterogeneity) measure is fue funetion D:.II -+ Q+ sueh that
foranySE.II,D(S)=(N-1)-'
L
l.j eS
d(i,j),
Fig. 2.
196
MARIO BUNGE
~., witb k between 1 and n, and the corresponding partitions fili'k=S/ ~k'
That is, we must consider the whole
D(S)
nm(l + 2)
(N -1)'
197
m +ml
~-,...
N-l
N-l
mI
Fig. 3.
3. SO C IALSTRUCTUR E
The family and class partitions of society are just two among many possible. (They may not even be universal, as anthropology and prehistory
tel! us.) A social equivalence relation other tban those of being in the
same family and being in the same cIass wil! generate a difrerent partition
of the same society. Hence one and the same society S can be partitioned
in n different ways, where n, a finite natural number, is the nunlber of
equivalence relations defined in S. If we wish to attain a satisfactory
picture of a society we must take into account all n equivalence relations
Every element S" of this matrix is a set eitber empty or nonvoid and it
represents tbe " social group resulting from the k" partition (or aspect)
of S. We compress tbe foregoing in the following
3.1. Let S be a society and fiIi's the col!ection of al! n partitions
of S, each induced on S by a social equivalence relation. Then
(i) every element S" of fiIi', EfiIi's is called a social eeU of S and
() the entire matrix :E= 11 S" 11 is called tbe social eomb of S.
DEFINmON
MARIO BUNGE
198
THEOREM 3.1. Let 1:= 11 S" 11 be the social comb of a society S. Then for
each fixed (aspect) k between l and n,
m
(a)
U Sik = S
199
N,. (1)
N 2.(I)
= 1
and
m
(b)
n S,,=0 .
i= 1
In other words, each relation of social equivalenee effeets the partition of a soeiety into social eeUs that are pairwise disjoint and'
jointIy exhaustive, Sinee these various partitions are mutuaUy compatible, one and the same individual may belong to different social eeUs
Slk and Sm. provided k"" n (Le. as long as different aspeets of soeiety are
studied),
The concept of social ce11 exactifies that of social group, and the coneept
of social comb elucida tes that of soCial slruclure, The two are exaet coneepts witbout being quantitative, If we wish to obtain an equaUy aeeurate
but far more detailed pieture of soeiety we must take a further step and
measure at least the size of each social ceU in a community. And as a
second step we should inquire into tbe variation of such partial abundances in tbe eourse of time. That is, we should study tbe distribution of
the total population of a society among the various ceUs of its social
comb, as weU as the ehange in sueh a distribution. This we proceed to
do next.
3.2. Densilies o[ Social eells
Let S be a human community with social comb 1:= II S"II. CaU N(I) the
total population of S at time 1, This population is distributed among the
various cells Si.' Can N" (1) the population of ceU Si. at time l. These
m x n numbers are subject to the sale condition tbat, for every fixed partition [1Jke[1JS, they add up to N. Dividing now each Nik(l) by the total
population N(I) al the same time, we obtain the population density of
the i lb social cell in the le lh social comb (respect). Call D the function that
carries Sik in lO N,,(I)/N(I) al time 1, This funclion D associates aU Ihe
possible parlitions of a society to the relative abundance of every one of
the resulting ceUs. Hence it deserves being caUed Ihe social struelure
= N for every
S;
S;
n.
(S,I)>->(I/N(I) II N,,(t)1I
is called the social Slrueture funclion, and tbe value of D at (S, 1) the
social slruelure o[ S al time t.
Imaginary example. Let S be formed right now (1=0) by 100 persons,
10 of whom are upper elass, none middle elass, and 90 lower elass. Besides, there are in S 70 illiterates and 30 literates. Finally, 50 members of
S vote white, 25 red, and 25 black. Make the pretence that these three
aspeets are all those that are relevant to the social strueture of S. Then the
social structure of S at the present time is given by the density matrix
MARIO BUNGE
for an eventual change in that entry. Change is, precisely, what will occupy our altention for the remainder of this papero
with, since changes in one column are bound to be correlated with changes
in other columns. Thus we are led to propose
4. SOCIAL CHANGE
DEFINITION 4.1. Let S be a society wilh social comb II S"II and the corresponding social structure D = IID'kll, where D is the m x n density matrix.
Then the nef sfrucfural change of S between the times f 1 and f 2 is them x n
matrix
200
Lf (S; f 2 , f 1 ) = D (S, f 2 )
- 111
Lf (S; T, O) = - 1O11
+1
In tbis example a single aspect (column) of lhe social structure was
taken into account. But the whole social comb IIS'kl1should be reckoned
D (S" f 1 )
This measure of social change may Dot be as perspicuous as this other ane:
'. (S'
a,k
'2'
() _
1
On lhe other hand the computation of the extent of social change is easier
with Definition 4.1. This is an nstance of the general maxim lhat simplicity cannot be achieved in every respect al the same time.
That which effects the change from D (S, f 1 ) to D (S, f 2 ) is a certan
matrix E(S; f" (2) characteristic of both the society S and the time interval [t" f 2 ]. In the prevous example, a possible operator ofths knd is
E(S; O, T) =
1
1
- 110 O
- 1 1O O
It is easi1y seen that several other operators do the same trick, .e. that the
The net structural change Lf (S; O, T) between the instants Oand Tmay be
regarded to be D(S, T)-D(S, O), i.e.
201
E(S; O, T) =
O O 1
O O 1
1 1 O
effects the same change. This is just as well both mathematically and
sociologically. The first because the problem we are dealing with, namely
that of tracing lhe evolution of a society between two given social structures, is an inverse problem, and it is typical of inverse problems not to
have unique solutons. (The corresponding direct problem is, of course:
Given a social structure and an evoluton operator, find the resulting
203
MARIO BUNGE
new structure.) It is also sociologically reasonable, for there are alternative ways of producing a given structural change in a society.
The preceding considerations are sununed up in the fo11owing convention and consequences of it.
century, according to fue speed of social change and the fineness of our
analysis.) The two matrices are related, according to Definition 4.2, by
the evolution matrix E(S; 1,1+ 1) in the following way:
DEFINlTlON 4.2. Let D(S, 1, ) and D(S, (2 ) represent the social structures
of a society S at times 1, and 12 respectively. Then any matrix E(S; 1" 12 )
such that
D (S, 12 ) = E(S; 1 12) D (S, 1, )
If E itself does not change in the course of time, then the social structure
acquired by the society p time units later is obtained by iterating E p
times, where p is a number such that p = T, T being the total time lapse
considered. The condition of invariance of E is satisfied when the society
changes with a constant mode of change, such as gradual industrialization, or literacy. In general terms we have
202
"
I] D (S, 1, )
I ~J I ~ill
and sought the coefficients
eik
D(S, t
+ 1) =
E(S; 1, I + 1) D(S, 1)
+ p) =
Set 12 =1, + 1 in
D(S, I
The preceding framework will enable us to characterize with sorne precision sorne of fue main types of structural change studied by historians.
Call D (S, 1) the social structure of S at time I and D (S, 1+ 1) the value
of Done time unit later. (This unit is conventional: it can be a week or a
:<{. i:<{.
MARIO BUNGE
204
N,,(T)
IE- bIl = O.
RecaU now Corollary 4.1 and note Ihat, even if there is a net structural
change over every single time unt (i.e. if E""I), there may be some p for
which E' =1. In tbis case, after p untary changes the society resumes its
initial social structure. The whole process constitutes a cycle with period
p. Example :
O 0 ... 0
E= O 0 ... 1 O
1 O... O O
205
DEFINITION
iff S ehanges irreversibly in lhose respeets and if the net social ehange in
those respecls is very large, i.e. if some eells D'k ehange appreciably.
DEFlNITlON 4.7. A soeiety S experiences a total revolution iff S ehanges
irreversibly in every respeet and to a large extent, i.e. if every eeU D 'k
changes appreciably.
At tirst blush the irreversibility condition we have plaeed on revolutionary
ehanges to qualify as such is ineonsistent with the faet that there are
drastic eyelie changes (Definition 4.5). There is no sueh ineonsistency: a
genuine social revolution, in contrast to a strictly politieal revolution or
sorne other partial upheaval, affeets irreversibly a number of eolumns in
the social combo Thus the English and Freneh restorations of the monarchy in the 17th and 19lh centuries respeetively did not undo the profound social and cultural changes brought about by the eorresponding
revolutions.
N(O)'D(O)
II!~II
N(T)'D(T)
1 1~011
(Peaceful)
206
207
MARIO BUNGE
But if the upper class is physicaIly liquidated and the lower class deci-
altogether others are only greatly reduced and others greatly increased.
Example:
N,
I ~III
N(T)'D(T) =
N(O) D(O)
(Violenl)
I ~II
are only interested in the density matrix.) Each type of change can be
represented by an evolution matrix of a peculiar formo In the example
aboye we may set
I ~ ~II
since
I ~ 9/~01 1
N(T)'D(T) =
since
a peacefullevelling revolution by E=
p
O 0 ...
O 0 ... 0
1 1...1
and
a violent levelling revolution by
E=
,
aN 2 + bN 3
N (O)
.
I ~ 9/~011'11!~11=I180111
O O O
O a O
O O b
fore knowledge of the final social structure is insufficient to retrieve (retrodict) the initial structure.
and
E, =
E=
Ep
N2
N3
The resulting social structure is the same Ihrough either method. Indeed,
in both cases the final density matrix is
D(T) =
O 0 ... 0
O 0 ... 0
.. .....
O O. .J
The en tries of the social structure matrix D= IIDik ll at time tare the instantaneous relative populations of the social ceIls Sik of a society S. If
we collect the whole lot of values of the density NiklN of every such ceIl
over a period T, we obtain the history of that ceIl during that periodo And
if we do the same for every cel! in the comb of S we obtain the social
history of S. More explicitIy, we have
DEFINITION 4.8. Let S be a human society with instantaneous social structure D (S, t). Then
(i) Ihehistoryofthecell Sik throughout the period [O, T] is the sequence
IIHikll
209
MARIO BUNGE
lhe introduction of certain ideas that might help formulate specific sociological theories - e.g. by specifying the population dcnsity matrix D and
lhe social evolution operator E. A framework has thus been produced
that might aceommodate a number of eompeting sociological theories. For
a charaeterization of the framework concept see Appendix V.
Because of its eminentJy analytic rather than substantive charaeter,
this paper contains more definitions lhan postula tes and theorems. Neverlheless we have tried to make adequate rather than arbitrary definitions
and, in the process, we have been forced to make some assumptions - for
example, that the population density matrix D is an adequate representative of the social structure. We have also made a few assumptions of a
simplifying character. One of lhem is lhat social structure and social
change are independent of lhe total population. (An exception was the
trealment ofrevolutions in Section 4.3.) This is only approximately true :
we know that below a eertain population threshold some social partitions
(e.g. division of labor) do not come into existence, and that there are
critical masses as weU, beyond which certain functions (e.g. participation
in management and politics) become impossible.
Another pretence, of a different kind, is that the social comb does not
change: only the occupation numbers N ik of the corresponding ceUs Sik
may change, in particular from zero or into nought. In other words, we
have regarded the social comb as a permanent set of cubes, sorne of them
empty, through which people pass. To iIIustrate:
208
orems.
cepts are so basie lhat they are usuaUy taken for granted and therefore
are not always as elear as they should. A byproduct of our effort has been
..
That is, our study has not dealt with the fascinating question ofthe very
origin and extinction of the cells of a social eomb. The reason is plain:
such an alternative study involves conjecturing specific dynamieal laws
about the mechanisms or 'forces' (biological, psyehologieal, eeonomical,
political, or cultural) lhat bring abont or dissolve the various social partitions (aspeets) of society. Sueh a study of the emergence and extinetion of
the various social categories belongs to social science proper rather than
210
MAl<.lO BUNGE
to its foundations and is therefore way beyond the scope of this papero
However, we can be sure that any such study of origins and decays can
use sorne of the concepts analyzed in this papero
Foundations and Philosophy of Science Unil
McGill Universi/y
NOTE
211
contribute to sketching the meaning of the formulas. For example, suppose we tried to represent the phenomenon of social parasitism by means
of the Lotka-Volterra equations
dH
- = (all
dt
+ a12P) H,
dP
-
dt
(a21
+ a22H ) P .
* This paper was written at the University of Aarhus while tbe author held a John
Simon Guggenheim Foundation fellowship. <O Copyright 1974 by Mario Bunge.
APPENDIXES
1. SEMANTIC ASSUMPTIONS
2.
SET MATRICES
DEFINITION
.,
(iv) thenuIlmatrixO is the matrix every entry ofwhich is the empty set;
(v) theunit matrixIis the diagonal p xp matrix, with p=Max{m, n},
sueh thatl,.=o .. U, where O'k is the Kroneeker delta and Uthe largest of
the sets being eonsidered.
Because of the multiplieation law (iii) and of the eommutativity of set
212
MARIO BUNGE
intersection, it follows, with Ihe he1p of (V), Ihat lA =AI=A for every A in
.
A = IIA
A 2l
A 12
A 22
A1311
A 23
we set
~.
U 11 11
I= 0 U 11
11 11 U
with
U::oA I"
value. Besides, set matrices in general have no in verses under multiplication. (The exception is constituted by certain matrices sorne of whose
elements are unit matrices.) There is nothing wrong with Ihe lack of an
inverse, either as a matter of principIe or as a practical malter. Singular
matrices, though disagreeable, are known to physicists and economists.
D (S, t) are reliable, we must iofer that the mode of evolution itself did
in fact evolve - e.g. that, after a certain point, industrialization led to
neglecting agriculture, which in turo caused famine, which originated
riots, which ended up in a civil war.
Our formulas for social change in Sections 4.1 and 4.2 constitule a
framework for social projection (prediction or retrodiction), hence for
social history of a !dnd. On Ihe other hand they give hardly any insight
into social change. Indeed the evolution matrix E does not exhibit any
change mechanism: it just summarizes the operation of Ihe social mechanisms of structura! change. In olher words, we have supplied a nonexplanatory black box. But this box is ample enough to contain any
explanatory mechanism and moreover it provides a framework for any
such mechanismic hypolheses.
4.
213
'1
1.2.1. Deterministic
1.2.2. Stochastic
t)
()
9 (N Ik , Pr N'k)' t
214
THE CON'
MARIO BUNGE
2.1. Static
2.1.!. Deterministic N ik = (Nj ,)
2.1.2. Stochastic
Pr(Nik N m,) = P"m,
2.2. Dynamic
2.2.2. Stochastic
215
garded as a collection __
el objects, of the referents of
the discipline. For example, our DefinitlOn 1.1 of a cornmunity as the list
'6'= (S, R, E, T) is a mode! object of a community. Such a relational
system constitutes a conceptual model of a cornmunity. In order to obtain a definite theory of communities one should put forth a number of
hypotheses characterizing and inter-relatng the various coordinates of'6'.
As long as this transformation of a framework into a theory is not performed, the derived concepts wil! be defined in terms of more or less hazy
notions.
... -dN"
2.2.1. D etermUllstlc
dI
~TRUCTURE
(Nik' Nmil' t )
dPr (N", t)
dt
= g (N", Pr(Nm,), t)