SOCIOTECHNICAL SYSTEMS

ROBERT COOPER MICHAEL FOSTER1

School of Business Studies, University oj Liverpool, Human Resources Centre, Tavistock Institute of Human
Liverpool, England Relations, London, England

M
ANY years ago, the philosopher Alfred openness to environmental influences and of de-
North Whitehead (1925) applied his pendence on the environment for needed energy
philosophy of organism to the modern resources. A generalized, theoretical exposition of
factory in the following words: the systems approach is to be found in the General
System Theory of the biologist Ludwig von Bertal-
A factory, with its machinery, its community ol operatives,
its social service to the general population, its dependence anffy (1968).
upon organizing and designing genius, its potentialities as While increasing attention is being paid by social
a source of wealth to the holders of its stock is an organ- scientists to organization-environment relationships,
ism exhibiting a variety of vivid values. What we want to the bulk of work in organizational studies is
train is the habit of apprehending such an organism in its
focused on internal organizational processes, that is,
completeness [p. 249].
at the S0 o level of analysis. Probably the most
Without excusing the conceptual lag, we can at extensive body of theoretical and empirical work
least claim that this way of thinking about organi- exemplifying the systems perspective at this level
zations is at last being honored by social scientists. is the sociotechnical systems theory developed over
Thus, current perspectives view organizations as the past 20 years by social scientists at the Tavi-
dynamic complex structures in symbiotic relation- stock Institute of Human Relations, London, Eng-
ship with their environments. The organization- land. The concept of the sociotechnicEil system is
environment matrix constitutes one total system, in based on the simple fact that any production system
which understanding of the behavior of any part requires both a technology—machinery, plant lay-
depends on the extent to which it is seen as being out, raw materials—and a work-relationship struc-
integral with other system parts. Let 5 denote a ture that relates the human operators both to the
system or system part, and the subscripts o and e technology and to each other. The technology
denote organization and environment, respectively makes demands and places limits on the type of
work structure possible, while the work structure
itself has social and psychological properties that
generate their own unique requirements with re-
S0 o refers to processes within the organization, the gard to the task to be done.
area of internal interdependencies and relationships;
S0 e and Se „, to exchanges and reciprocations be- SOCIOTECHNICAL FRAMEWORK
tween the organization and its environment; and Figure 1 indicates how the human and technical
Sc Q, to relationships between different parts of the elements interface with each other in a typical in-
environment. Each system is a legitimate area of dustrial production system. The basic elements
study in its own right, while at the same time ex- are operations performed by individuals and ma-
erting influence on other systems. They possess the chines on material. Operations then combine into
characteristics—common to all living systems—of sequences of operations. Operation sequences in-
1
In addition to the published Tavistock Institute work teract at the individual and machine level to pro-
on sociotechnical systems that we discuss in this article, duce the conventional man-machine chart. First-
we have drawn on ideas contained in several unpublished order sociotechnical units are represented in (a) an
internal papers by former Institute members. In this activity, that is, a completed sequence of operations
connection, we would especially like to recognize the work by a person (e.g., in assembling an apparatus) a.nd
of P. G. Hcrbsl.
Requests for reprints should be sent to Robert Cooper,
(b) the set of operations performed by a machine
School of Business Studies, University of Liverpool, P.O. with a given setup (e.g., trimming metal sheets of a
Box 147, Liverpool, England. given size and cutting two slots). A completed se-
467
468 AMKUICAN PSYCHOLOGIST

quence of operations on a set of materials leading to systems in this way, there is a danger of assuming
an intermediate product defines a production or comparability of function between men and ma-
service process (e.g., welding, inspecting, or stor- chines, that is, of equating human functions with
ing). Interaction of the first-order units is re- mechanical ones. While such a view may be valid
flected in (a) the allocation of men to machines, in at the physiological level, it is completely inappro-
which the primary aim is to minimize idle man- priate to an understanding of psychological proc-
machine time and satisfy the requirements of pro- esses. Modern psychological research presents a
duction schedules and process sequences; (b) allo- picture of man as a proacting, self-realizing organ-
cation of processes to machines to determine a par- ism capable of considerable achievements given the
ticular machine setup; and (c) allocation of men right environmental conditions, but who becomes
to processes which is determined by initial job se- refractory when placed in a dependency relation-
lection or subsequent training. Second-order socio- ship and when forced to function at a fraction of
technical units are defined by (a) sets of activities his capacity. Therefore, in considering the "fit"
which are grouped together—usually by a job de- between man and machine, it is more constructive
scription—and which define jobs; (b) a total set of to think of how they can complement each other.
operations performed by a machine and whose func- The principle of man-machine complementarity rec-
tion defines the type of machine (e.g., forming, join- ognizes that men are better than machines at doing
ing, materials moving, etc.); and (c) the sequence certain things while machines have the advantage
of processes performed on a set of raw materials over men on other points. Men, for example, are
leading to a given end product and which defines a more flexible than machines, but machines tend to
manufacturing process, usually represented by the be more consistent in their behavior. H the task
flow process chart. First- and second-order socio- environment is predictable and demands consistent
technical units are essentially relationships between performance, the machine is superior to the man;
an individual and his tools or machinery. For this but if the task demands are unpredictable, then the
reason, we prefer to describe such units as "psycho- man, backed up with the proper tools, is needed.
technical units." Tn our view, the distinction be- Machines themselves do occasionally break down,
tween psycholechnical and sociotcchnical units has and then they necessarily depend on human inter-
considerable practical value as many production vention. Where the machine is critical to overall
technologies specify a work-relationship structure system functioning, then it should be designed so
that is based on the individual worker and not on that man can serve as a manual back-up in the
the social group. event of breakdown. Man performs optimally
Total system structure comprises: (a) the social when he knows he can control environmental con-
system of the factory, that is, the total set of indi- tingencies; when he lacks such control, his perform-
viduals' activities together with their interrelation- ance becomes correspondingly degraded. A further
ships; (b) the plant layout of the factory, that is, aspect of man-machine complementarity is that men
the total set of machines and their displacement; will not assume responsibility for machine operation
and (c) the manufacturing system of the factory, —or indeed any other kind of operation—unless
that is, the total set of manufacturing processes and their responsibilities arc clearly specified and unless
their interrelationships. Interaction of these sys- they are provided with the means for exercising
tems is represented by (a) the manufacturing proc- effective control over their areas of responsibility.
ess together with the plant layout which determine But no matter how well we relate man to machine,
the possible routing and materials handling system,] unless he is properly motivated he will not function
and (b) the manufacturing process which deter- very effectively as a complement to it. This moti-
mines the plant layout and together determine the vation must derive from the task itself. The role
routing and materials handling system. Both the of motivation as the mainspring of man-machine
plant layout and the handling system determine to performance has been well depicted by Jordan
a large extent the social system. The interaction of (1968):
social and technical systems constitutes the socio-
Unless a task represents a challenge to the human operator
technical system. he will not use his flexibility or his judgement, he will not
Figure 1 is, of course, a type of mapping familiar Icarn nor will he assume responsibility, nor will he serve
to industrial engineers. Tn portraying production efficiently as a manual back-up. By designing man-machine
 SOCIOTKCHNICAL SYSTUMS 469

Elemenls Operations by Operations by Operations on

individuals machines materials

Element Operation sequences Sequences of Operation sequences

sequences shown by operation machine shown by process
charts operations charts

Man-
machine charts

First-order Activities of Machine operates A specific production

sociotechnical individuals with given setup or service process
units

Allocation of ' Allocation of Allocation of processes

men to machines men to processes to machines

Second-order Jobs carried out Machine type Manufacturing process

sociolec/mical by persons for a specific product
mills shown by flow process
chart

System Social system Plant layout Manufacturing system

structure

Routing and materials

•handling system

The sociotechnical system

Fio. 1. The sociotechnical framework of a typical industrial production system (from Hcrbst, 1959).

systems for man to do least we also eliminate all challenge logical factors will become salient in a production
from the job. We must clarify to ourselves what it is system. The level of mechanization—the extent to
makes a job a challenge to man, and build in those chal-
lenges in every task, and activity, and responsibility which
which the machine simulates human skills—is prob-
we assign to the human operator. Otherwise man will not ably the single most important property of technol-
complement the machine but will begin to function like a ogy affecting the variables of the social system. In
machine [pp. 206-2071. mass production, for example, the human operator
Any environment can be analyzed in terms of is linked to the machine in what is essentially a
those features which make particular behaviors mechanical role. But in many automated flow
possible (supports) and those which preclude or technologies, the operator not only functions au-
limit particular behaviors (constraints). The sup- tonomously of the technology but has some measure
ports and constraints inherent in the technical en- of control over it. The technology may also deter-
vironment determine what particular social-psycho- mine the spatial and temporal character of the
470 AM KIUCAN I 'SYCHOLOGIST

social system. Jn automobile assembly, the frac- to be located in the middle of (he chain so that the
tionated nature of the production process differenti- later tasks are more dependent than the earlier
ates operators largely on a spatial basis; chassis ones. In the divergent dependence pattern (A to
assembly is spatially differentiated from transmis- B, C, D), A is the key task since it has to be
sion assembly. In contrast, in flow technologies, completed before any of the subsequent tasks can
temporal factors often are more important than be started. In the convergent pattern, ]) is directly
spatial layout. In electric power generation, the dependent on Tasks A, B, and C. Knowledge of the
transformation from coal to steam to electricity task dependence patterns enables the identification
occurs over a very short period so that the different of work regions which are most likely to disrupt
aspects of the technology are temporally highly the system as well as regions which are most vul-
integrated. Workers performing the different op- nerable to disruption. For example, the disruption
erations on the same shift constitute a more articu- potential would be high in a divergent dependence
lated system than do the workers who perform the pattern while convergent patterns would be more
same operation on different shifts. Slow through- vulnerable to disruption.
put appears to accentuate the territorial and tech- Figure 2 illustrates the task dependence structure
nological differences in Ihc sociotechnical system, of an extraction cycle (covering three shifts) in coal
while swift throughput accentuates time-related mining. It particularly shows a convergent depen-
differences (Miller, 1959). The physical nature of dence pattern of work relationships with a strong
the throughput can also be an important technology- vulnerability to disruption in the final shift of the
related factor affecting the social variables. Rice extraction process.
(19S8) has described, for example, how variation Goals of course may be either mutual or inde-
in the tensile strength of cotton creates problems of pendent. The existence of a mutual or common
a social-psychological nature for the operators of goal increases the need for, and the likelihood of,
weaving machinery. reciprocal supporting behaviors in the group. In
Such features of the technical system influence work regions susceptible to disruption, the existence
social-psychological variables by means of the of common goals may serve to reduce disruption
work-relationship structure which relates the system possibilities and to counter their effects.
operators to the technology and to each other. The When sociotechnical systems become increasingly
most important dimensions of the work structure differentiated along the work structure dimensions,
are (a) role differentiation, (/;) task dependence, there tends to be an increase in system segmenta-
and (c) goal dependence. Role differentiation re- tion. Increased segmentation in turn requires an
fers to the degree of task variety in a given job and,
in the case of group tasks, to differences between
the tasks carried out by group members. Jobs of
limited complexity offer little challenge and thus
affect individual motivation, while greater differen- Sequential
tiation of roles in the group as a whole increases the dependence
group's flexibility of task operation but may limit
the development of cohesivencss.
Task dependence may be determined by the
nature of the task on which two or more people
Disturbance
work together or to a common limited pool of ser-
channel
vices, materials, or tools in the case of people work-
ing alone. The simplest form of task dependence
is the chain pattern where A transmits to B, B to
C, C to D, and so on. The centrality of a task—
that is, the number of tasks directly dependent on
it—decreases as one moves to later stages of the
chain. In certain chains, centrality is less impor-
tant than the location of "pace-setter" tasks. Thus, Fro. 2. Task dependence structure in the longwall mefhod
in automated steel tube mills, "pace-setters" tend of coal extraction (from Herbst, 1'JS9).
 SOCIOTECHNICAL SYSTKMS 471

external system of coordination and control, thus production process did not permit the carrying on
reducing the autonomy of the sociotechnical system. of later tasks while the hold-ups were being dealt
with, the system was necessarily highly sensitive to
IMPLICATIONS FOR TASK ORGANIZATION disruption both at the production and social-psy-
Sociotechnical theory dates formally from an in- chological levels. And since control was no longer
vestigation carried out about 20 years ago by Trist functionally vested in the working group, the miners
and Bamforth (1951) of the Tavistock Institute of experienced a sense of impotence and frustration in
Human Relations into the effects of technological face of the complex and inflexible technical system.
change in a British coal mine. The prechangc Consequently they developed various defensive
method of coal-getting involved small groups of maneuvers: a norm of low productivity as a means
miners—usually two or three—working closely to- of reducing disturbances which were beyond their
gether. Such groups were almost wholly autono- control; the creation of small informal groups whose
mous in that control over the task was internal to major obligations were to themselves and not to the
the group—miners even chose their own workmates cycle system as a whole; individual competition for
—and the only contact the group had with external the more workable parts of the coal face and for
colliery management was in contracting to work a special favors from co-workers whose own work
particular wall of the coal face. Within these small could help or hinder one's particular task area.
undifferentiated groups, each coal miner was called Physical withdrawal in the form of absenteeism
on to execute a variety of tasks, often substituting also became widespread.
for his mate. In addition, group members experi- In a later series of studies in other coal fields
enced the satisfaction of completing a whole task in- (Trist, Higgin, Murray, & Pollock, 1963), the
asmuch as they completed the entire cycle of op- Tavistock researchers were able to compare the effi-
erations necessary to hew a given face. In socio- ciency of the conventional longwall method of ex-
technical terms, the social-psychological and the traction with another method, the composite long-
task requirements were congruent. The techno- wall method, which, while still employing the new
logical innovation consisted of substituting mechan- technology described above, utilized some of the
ical coal-cutters and conveyors for the old hand-got social-psychological features of the older hand-got
methods, thus transforming the production tech- methods. In the composite method, men arriving
nology to a type characteristic of mass production for i:i new shift take up the cycle point left by the
methods. Instead of working a series of short previous shift. When their main task is completed,
faces, a costly method of coal extraction, mechani- they then carry on with the next task, whether or
zation made it possible to work a single long wall. not this happens to be part of the current cycle or
The new longwall method demanded a different begins another cycle; that is, unlike the men in
form of individual and group working than prevailed conventional longwalling, they act somewhat inde-
with the hand-got method. The production unit pendently of the strict cycle process determined by
was organized around the cycle group of 40 men the technology. The composite method also pro-
who had to extract about 200 tons of coal per cycle. vides a greater variety of skills for the individual
A cycle extended over 24 hours, made up of three worker, thus making for greater individual satisfac-
shifts of 7-£ hours each. The allocation of workmen tion. Composite teams are also self-selected and are
to each shift was approximately 10 to the first paid on a common paynote in which all group mem-
("cutting") shift, 10 to the second ("ripping") bers share equally. Tables 1 and 2 show the rela-
shift, and 20 to the third ("rilling") shift. Within tive efficiency levels of teams working the same
each of the shifts, individuals were restricted to technology and coal seams but under different socio-
narrowly defined work roles in contrast to the task technical systems.
variety inherent in the hand-got method. However, Rice's (1958) study of the reorganization of the
the real lack of integration between the social and work-relationship structure in the weaving room of
technical systems lay in the high degree of inter- an Indian textile mill again reveals the failure of
dependence between the tasks throughout the entire management to recognize the importance of task-
cycle. Operational problems experienced at one related social variables in the production system.
stage of the process were carried forward to later Although machinery layout and allocation of work
stages, and because the inflexible nature of the loads had been carefully studied by engineers, the
472 AMKIUCAN PSYCHOLOGIST

TAHUC I tems. The lack of integration between the social

PRODUCTIVITY INDEXES KOI; Two D I F K E K E N T PRODUCTION and technical aspects of such production systems is
SYSTKMS IN COAL MINING usually reflected in their low levels of employee
morale and cohesion (absenteeism and labor turn-
Longwall
Index
over). In other words, in addition to the goods or
iual Comp( lite services dictated by its primary task, a sociotech-
nical unit also generates social-psychological out-
J'rorluctivity (% of coal face comes. Figure 3 details the system outcomes ac-
potential) 78 95
State of cycle progress
cording to an input-conversion-output model
Tu advance 22 adapted from Stogdill (1959). The human inputs
Normal 73 are subjected to the mediating effects of the tech-
Behind 5 nology via the work-relationship structure to create
the outputs of productivity, morale, and cohesion.
A state of balance-imbalance characterizes the rela-
mill still failed to attain satisfactory performance tionship between the output variables such that, for
levels. The nature of the production process de- example, if productivity is too strongly pursued,
manded the utmost coordination between the weav- either through technological or management pres-
ing room jobs, but the work groups were initially sures, then morale and cohesion will suffer. As far
organized on the basis of role similarity and not as the general economics of automobile manufacture
according to task interdependence. Not surpris- are concerned, this imbalance may be quite accepta-
ingly, this form of grouping mitigated against con- ble inasmuch as productivity gains far outweigh
tinuity of production. Rice redesigned the groups the costs incurred by employee apathy and absence.
by bringing together all workers whose tasks were However, even within the restraints imposed by eco-
interdependent. As a result of the reorganization, nomic considerations, there is probably always some
productivity rose from an average of 80% to an latitude for improving the level of sociotechnical
average of 95%, with damage dropping from 52% integration and thus the state of balance between
to 20%. In the rest of the weaving room, where the outputs.
no changes were made, productivity dropped to All production systems are designed with the aim
70% and finally rose to 80%, while damage re- of minimizing the degree of variance from goal at-
mained at an average of 31 %. tainment. In the design of sociotechnical units,
goal variance is minimized by: (a) the joint opti-
CONCEPT or ORGANIZATIONAL CHOICE mization of the social and technical systems, and
From the point of view of sociotechnical design, (b) providing conditions for the variance to be
the foregoing studies indicate that there is often an controlled within the unit. In practice, this means
element of choice in designing effective work sys- that the unit, whether an individual or a group, has
tems and that this choice must take into account to be given "responsible autonomy" so that it be-
the mutual dependence of the social and technical
systems. However, the degree of such mutual de- TAKLK 2
pendence differs between production systems. In
ABSENCE KATES (% OF POSSIBLE SHUTS) FOR
the coal and textile situations, the degree of inter- Two DIFFERENT PRODUCTION SYSTEMS
dependence was very high, productivity being a IN COAJ. MINING
direct function of the integration (or lack of it)
Longwall
between the social and technical aspects of the sys- Reasons for absence
tem. In other production systems, the degree of Conventional Composite

mutual dependence between the social and technical No reason given 4.3 .4
aspects may appear to be less marked. For exam- Sickness and other 8.9 4.6
ple, automobile assembly lines enable high levels of Accident 6.8 3.2
production to be attained although they would 8.2
Total 20.0
hardly be described as balanced sociotechnical sys-
 SOCIOTECHNICAL SYSTEMS 473

Unman inputs Technical characteristics System outputs

Individual performances Level of mechanization Productivity

Psychological requirements > Spatio-temporal dis- Morale
placement of technology
Social interactions Nature of processed Cohesion
material

Work relationship structure

FIG. 3. An input-conversion-output model of a sociotechnical system.

comes capable of self-regulation.2 Since the major by a process of successive decomposition arrives
functions of conventional management are assumed finally at a stage where a myriad of product parts is
by the unit itself, the control and regulative func- covered by a myriad of isolated man-machine op-
tion of management changes to one of support, par- erations. At each design level, there is a blue-
ticularly in mediating the relationships which the print which serves to specify the next lower level
sociotechnical unit has with its task environment. until every movement of operator and machine is
The design of autonomous units always proceeds rigidly specified. As this design philosophy makes
from the minimal set of conditions required to no provision for self-regulation, a superstructure of
create self-regulating production units. These mini- supervision, inspection, scheduling, and so on is
mal conditions are obtained from an analysis of the required in order to control the unwanted variances
required or existing work-relationship structure. that occur at each level. This additional structure
In the longwall method of coal extraction, for ex- creates the possibility of further variance.
ample, an important minimal condition was that Quite apart from the value that sociotechnical
operators should be allowed to work with some thinking has for the design of viable man-machine
independence of the inexorable cycle process im- systems, there are additional reasons, at present
posed by the technology. In the textile mill, a only incipiently recognized, that commend its utili-
minimal condition concerned the groupings of op- zation on a wider scale. The developing applica-
erators according to task interdependencies. After tion of advanced forms of technology both in fac-
identification of the minimal conditions, the non- tory and office means that organizations will be-
critical variables are left free in order to allow the come increasingly structured around their work
system some degree of potential variability for cop- technology, and this will require us to think much
ing with variance from both internal and external more seriously about the relationship between man
sources. Tn fact, cyberneticians acknowledge this and machine. As technology becomes more com-
principle as a basic requirement for the self-main- plex, so does human nature. The affluence of ad-
tenance of any system. Ashby (1956) has dubbed vanced industrial countries has brought about a
it "the law of Requisite Variety": only variety can change in people's value preoccupations. We have
destroy variety. The innovatory nature of design become less overly concerned with the themes of
by minimal specification may perhaps be better working life—adequacy and security of income be-
appreciated if it is compared with traditional ing foremost—that dominated our thinking up until
philosophies of work organization which are based the recent past. Instead, our aspirations are mov-
on the assumption that work is best performed un- ing more and more in the direction of greater indi-
der maximally specified regimes. The traditional vidual autonomy and expression, and it will not be
approach proceeds from the ultimate product, and long before this movement, sustained by educa-
tional and social changes, will begin to make itself
2
The concept of "responsible autonomy" is, of course, a felt in industry. Such developments will precipi-
basic principle of Theory Y (McGregor, 1960). It has
been most thoroughly expatiated at the work group level
tate the need to complement technology with psy-
by Herhst (1962). chology.
474 AMEKICAN PSYCHOLOGIST

KKKKRKNCES Rrci:, A. K. Productivity and social organization: The

Ahmedabad experiment. London: Tavistock Publications,
AsiniV, W. K. An introduction to cybernetics. New York:
1958.
Wiley, 19S6.
STOGDILL, R. M. Individual behavior and group achieve-
HKRHST, P. G. Task slructurc and work relation.',. (Docu- ment. New York: Oxford University Press, 1959.
ment No. 528) London: Tavislock Institute of Human TRIST, E. L., & BAMFOHTJI, K. W. Some social and psycho-
Relations, 19S9. (Mimeo)
logical consequences of the longwall method of coal-
HiautsT, P. G. Autonomous group functioning: An explor- getting. Human Relations, 1951, 4, 3-38.
ation in behaviour theory anil measurement-. London: TRTST, K. L., HrcorN, G. W., MUKHAY, H., & POLLOCK, A. U.
Tavistock Publications, 1062. Organizational choice: Capabilities of groups at the coal
JORDAN, N. Themes in speculative psychology. New York: face under changing technologies. London: Tavislock
Barnes & Noble, 1968. ['ublications, 1963.
McGKKOOK, D. The human side of enterprise. New York: VON BKRTAT,ANI''|''Y, L. General system theory: Founda-
McGraw-Hill, 1960. tions, development, applications. New York: Brazillcr,
MILT.ER, K. J. Technology, territory, and time: The in- 1968.
ternal differentiation of complex production systems. WiiiTEHKAD, A. N. Science and the modern world. New
Human Relations, 1959, 12, 243-272. York: Macmillan, 1925.