by When people hear the word “complexity,” they respond in different ways. Some think “complicated” or “messy,” not being able to see the forest for the trees. Others think of a clutter of matter going this way and that with no chance to get a purchase on its behavior, to take hold of the “blooming, buzzing confusion” (James 1890, 462). Others think “chaos,” in the traditional sense, something unrestrained and uncontrollable, a realm of unpredictability and uncertainty that doesn’t yield to human understanding. None of these interpretations does justice to the tractable, understandable, evolved, and dynamic complexity that contemporary sciences say aptly characterizes our world.’ Neither its complications nor its chaotic dynamics should scare away the curious, nor drive them
Topics:
Lars Pålsson Syll considers the following as important: Theory of Science & Methodology
This could be interesting, too:
Lars Pålsson Syll writes Kausalitet — en crash course
Lars Pålsson Syll writes Randomization and causal claims
Lars Pålsson Syll writes Race and sex as causes
Lars Pålsson Syll writes Randomization — a philosophical device gone astray
When people hear the word “complexity,” they respond in different ways. Some think “complicated” or “messy,” not being able to see the forest for the trees. Others think of a clutter of matter going this way and that with no chance to get a purchase on its behavior, to take hold of the “blooming, buzzing confusion” (James 1890, 462). Others think “chaos,” in the traditional sense, something unrestrained and uncontrollable, a realm of unpredictability and uncertainty that doesn’t yield to human understanding. None of these interpretations does justice to the tractable, understandable, evolved, and dynamic complexity that contemporary sciences say aptly characterizes our world.’ Neither its complications nor its chaotic dynamics should scare away the curious, nor drive them to replace a clear-eyed investigation of the nuanced beauty of complexity with the austere, clean lines of the simple and timeless.
The world is indeed complex; so, too, should be our representations and analyses of it. Yet science has traditionally sought to reduce the “blooming, buzzing confusion” to simple, universal, and timeless underlying laws to explain what there is and how it behaves.
As a philosopher of science, it is interesting to note that many economists and other social scientists appeal to a requirement that explanations, in order to be considered scientific, must be capable of “reducing an individual case to a general law.” As a fundamental principle, a general law is often invoked as “if A, then B,” and if one can demonstrate in individual cases that if A and B are present, then one has ‘explained’ B.
However, this positivist-inductive view of science is fundamentally untenable.
According to a positivist-inductive view of science, the knowledge possessed by science constitutes proven knowledge. Starting with entirely unbiased observations, an ‘impartial scientific observer’ can formulate observational statements from which scientific theories and laws can be derived. Using the principle of induction, it becomes possible to formulate universal statements in the form of laws and theories that refer to occurrences of properties that hold always and everywhere. Based on these laws and theories, science can derive various consequences with which one can explain and predict what happens. Through logical deduction, statements can be derived from other statements. The logic of research follows the schema of observation — induction — deduction.
In more uncomplicated cases, scientists conduct experiments to justify the inductions with which they establish their scientific theories and laws. As Francis Bacon vividly put it, experimentation involves “putting nature on the rack” and forcing it to answer our questions. With the help of a set of statements that accurately describe the circumstances surrounding the experiment — initial conditions — and the scientific laws, scientists can deduce statements that can explain or predict the phenomenon under investigation.
As a result of the well-known problems with the hypothetico-deductive method, more moderate empiricists have reasoned that since there is usually no logical procedure for discovering a law or theory, one simply starts with laws and theories from which a series of statements that serve as explanations or predictions are deduced. Instead of investigating how scientific laws and theories are arrived at, the focus is on explaining what a scientific explanation and prediction are, the role theories and models play in them, and how they can be evaluated.
In the positivist (hypothetico-deductive, deductive-nomological) model of explanation, explanation refers to the subordination or derivation of specific phenomena from universal regularities. To explain a phenomenon (explanandum) is the same as deducing a description of it from a set of premises and universal laws of the form “If A, then B” (explanans). Explanation simply involves being able to place something under a specific regularity, and this approach is sometimes referred to as the “covering law model”. However, theories should not be used to explain specific individual phenomena but to explain the universal regularities that are part of a hypothetico-deductive explanation. [But there are problems with this view even in natural science. Many of the laws of natural science do not really say what things do but rather what they tend to do. This is largely due to the fact that the laws describe the behaviour of different parts rather than the entire phenomenon itself (except possibly in experimental situations). And many of the laws of natural science actually apply to fictional entities rather than real ones. Often, this is a consequence of the use of mathematics within the respective science and leads to the fact that its laws can only be exemplified in models (and not in reality).] The positivist model of explanation also exists in a weaker variant known as the probabilistic explanation, according to which explaining essentially means showing that the probability of an event B is very high if event A occurs. This variant dominates in the social sciences. From a methodological standpoint, this probabilistic relativization of the positivist explanatory approach does not make a significant difference.
One consequence of accepting the hypothetico-deductive model of explanation is often the acceptance of the so-called symmetry thesis. According to this thesis, the only difference between prediction and explanation is that in the former, the explanans is assumed to be known, and a prediction is attempted, while in the latter, the explanandum is assumed to be known, and the goal is to find initial conditions and laws from which the observed phenomenon can be derived.
However, a problem with the symmetry thesis is that it does not consider that causes can be confused with correlations. The fact that storks appear at the same time as human babies does not explain the origin of children.
The symmetry thesis also fails to acknowledge that causes can be sufficient but not necessary. The fact that a cancer patient gets run over does not make cancer the cause of death. Cancer could have been the actual explanation for the individual’s death. Even if we could construct a medical law – in accordance with the deductive model – stating that individuals with a specific type of cancer will die from that cancer, the law does not explain this individual’s death. Therefore, the thesis is simply incorrect.
Finding a pattern is not the same as explaining something. To receive the answer that the bus is usually late when asking why it is delayed does not constitute an acceptable explanation. Ontology and natural necessity must be part of a relevant answer, at least if one seeks something more than “constant conjunctions of events” in an explanation.
The original idea behind the positivist model of explanation was to provide a complete clarification of what an explanation is and to show that an explanation that did not meet its requirements was actually a pseudo-explanation. It aimed to provide a method for testing explanations and demonstrate that explanations in accordance with the model were the goal of science. Clearly, all these claims can be legitimately questioned.
An important reason why this model has gained traction in science is that it seemed to offer a way to explain things without needing to use ‘metaphysical’ notions of causality. Many scientists see causality as a problematic concept that should be avoided if possible. Simple observable variables should suffice. The problem is that specifying these variables and their possible correlations does not explain anything at all. The fact that union representatives often wear grey suits and employer representatives wear pinstriped suits does not explain why youth unemployment in Sweden is so high today. What is missing in these “explanations” is the necessary adequacy, relevance, and causal depth without which science risks becoming empty science fiction and a mere playground for models.
Many social scientists seem to be convinced that in order for research to be considered science, it must apply some variant of the hypothetico-deductive method. From the complex array of facts and events in reality, one should extract a few common lawful correlations that can serve as explanations. Within certain fields of social science, this endeavour to reduce explanations of social phenomena to a few general principles or laws has been a significant driving force. By employing a few general assumptions, the aim is to explain the nature of an entire macrophenomenon we call society. Unfortunately, there are no truly sustainable arguments provided as to why the fact that a theory can explain different phenomena in a unified manner would be a decisive reason to accept or prefer it. Uniformity and adequacy are not synonymous.