The Notion of Closure in the Social Sciences: an Overview

Although a polemical subject, the application of the concept of closed systems in the social sciences has been seen as a way to limit and even rule out the influence of certain variables in the study of social objects, which in turns seems to allow the elicitation of clear-cut causal relationships. Those causal relationships can then be generalised in the form of scientific laws and theories, which could then be used as a mean to explain and predict social phenomena.

Closed systems have been used with great success in natural sciences such as physics or chemistry: famous early examples of their application are for instance Newton’s double prism experience, which greatly improved our understanding of the nature of light, or Lavoisier’s mercury calx experiment, which led to the discovery of oxygen and led the path to modern chemistry. Closed systems have made possible the manipulation of certain parameters in the framework of an experimental setting, which in turns allowed researchers to push further their understanding of their objects of study. This control of an artificial system’s conditions, although unquestionably successful in the establishment of causal relationships within the realm of the “hard” sciences, can be quite problematic when adapted to knowledge construction in the field of the social sciences. Indeed, the early years of the development of social science in Europe saw the introduction of the idea that the concept of closed systems could be applied to the study of social phenomena. However, the particular nature of the social world makes the adaptation of this concept to the study of social relations and processes quite delicate.

First, we could start by considering the social world as being an inherently open system. We could refer to the definition of open systems proposed by Mark Smith (Smith, 1998, p. 348), by saying that the complex, interdependent relationships between the entities that compose the social world (as well as the influence of external factors on their internal properties) make it very difficult (not to say impossible) to apprehend society as a whole and in all its complexity: It would therefore be a mistake to think that social phenomena could be predicted. Moreover, several issues arise when we look closer at the three ways in which closed systems have been applied in the social sciences by the means of statistical, theoretical or empirical closure, to which we are now turning to.

Experimental closure is obtained in disciplines such as psychology by isolating the subjects of the experiment from the external world and by submitting them to a particular set of empirical conditions, or stimuli. The subjects’ reactions are then analysed in order to infer causal relationships between the stimuli and the subjects’ responses. This type of closure is questionable as one could argue that the very fact of isolating the subjects by placing them in a controlled, artificial setting would have in itself quite an influence on the subjects of the experiment: what can appear as an almost literary way to achieve closure could also be seen as an interference to the normal behaviour of the subjects. In addition, ethical issues can be entailed by the use of human subjects for psychological or sociological experiments; a spectacular illustration of this problem is the outcomes of the infamous Stanford Prison experiment, in which Philip Zimbardo studied the effects of captivity by reproducing artificially a particular view of the conditions that can be found in prison. An interesting critical review of Zimbardo’s finding, showing some of the problems entailed by experimental closure has been made by Erich Fromm questioning Zimbardo’s experimental settings and the impact of his personal values and expectations on his findings and conclusions (Fromm, 1973).

Another method to obtain closure is achieved for instance in economics by making a model of a particular object of study based on a particular set of theories or assumptions about the system (such as, for instance, the economy) and by using this model to make predictions about the future state of this object of study. The validity of a model can then be verified according to the accuracy of its predictions. Although such models can allow us to make useful prediction regarding, for instance the effects of the increase of the money supply on inflation (Smith, 1998, p. 43), they are only a partial reflection of a much more complex system. Therefore, the validity of such a method of closure, called theoretical closure, is also disputable. Indeed, we could consider that the particular theories on which are based those models make a certain number of original assumptions that may be able to predict a particular facet of the object of study, but which have in the same time a great influence on the general validity of the model’s prediction. This can be illustrated by the fact stated by Smith (Smith, 1998, p.43) that certain econometric models would be more successful for predicting inflation while other would be more accurate on unemployment levels. Moreover, the process of interpretation and simplification that takes place when the original object of study is “transformed” into a simpler model can also be seen as a reason to be critical towards the use of the results of such modeling methods in order to explain actual causal relationships in the real world.

Statistical closure involves the use of mathematics in the treatment of empirical statistical data in a way that allows the identification of correlations between particular variables. Particular sampling techniques, as well as the use of statistical mathematical tools to validate the results provided by the analysis of the collected statistical data are advocated as means to prove the validity of the causal relationships, or correlations identified between variables as an explanation of social phenomena. However, the definition of the variables or concepts being studied can be seen as problematic for the validity of such results: Indeed, as Smith argue, some concepts, such as for instance homelessness, are greatly subjective and can be refereed to as “fuzzy concepts” (Smith, 1998). This particular problem is overcome by social scientists by crafting a definition of such concepts on the basis of a consensus between the participants of the statistical study. However, one could argue that such a strategy would explain causal relationship between concepts that do not have any empirical reality, but are constructions of the human imagination.

Finally, we could conclude this discussion by considering one of the main differences in the application of closed system in the hard sciences when compared to social sciences. Within the realm of natural sciences, researchers are relatively detached from their objects of study (even though this assumption can itself be subject to debate). As social researchers are themselves part of the object they are studying (such as the society, or as human beings studying other human beings), detachment towards the object of study becomes quite problematic. Indeed, the meanings and values entangled in the concepts that are manipulated in the context of the practice of the social sciences have a profound impact on the way things are perceived by the researchers, and can deeply influence their findings and conclusions. Therefore, one can argue that when attempting to simulate a closed system for the study of any social phenomenon, researchers could be themselves in some way part of the system they attempt to create, which raises the questions of the need for a strong effort towards reflexivity, and of the consideration of the symbolic meanings of the concepts manipulated in such a framework.

References

Fromm, E. (1973) Fromm… on Zimbardo’s Prison Experiment (Extract from The Anatomy of Human Destructiveness, Fawcett Books, 1973).

Smith, M.J. (1998) Social Science in Question: Towards a Postdisciplinary Framework.

Related posts:

1. Applying social scientific concepts to domain definition: a short overview
2. What is Falsificationism?
3. Tweeting Bourdieu: Social software and social capital
4. Research proposal : A Foucauldian analysis of the evolution of the discourse about software development methodologies
5. Software as Discourse