The notion of causality – and the many longstanding debates about it – has interested me for some time. It is a subject that regularly resurfaces in my study of knowledge, and one that I continue to write about in my notebooks. In truth, it is no small or trivial matter – debates about the nature of causality have existed since at least Aristotle. I would describe it as a concept of such importance that any subtle shift or advancement in the science of cause and effect would impact a significant number of general academic and scientific fields. In physics, we still search for a clear definition of it, while the notion of causality is also very important to epidemiology and other diverse places like psychology and cognitive linguistics. In philosophy and social science, the notion of causation remains a site of great contention, and the implications of these debates are arguably immense.
I was first introduced to the notion of causation in physics. Considering the subject has its presence in my daily studies as a theoretical physics student, wider contention about its definition or even its actuality has offered a site of both intrigue and concern. In physics, causality is an important concept – think, for example, of the Penrose Interpretation and his arguments about the causes of wave function collapse. Or, as another example, think of the paradox of the arrow of time and the search for non-local causality. Or, finally, think of Special Relativity, where it is not possible for an effect to occur from a cause that is not in the back light cone of that event. But these are just a few examples, with the point to be emphasised that causality is important in a diversity of places within the physical sciences, from things like causal calculus to such areas as mathematical biology to my area of interest in particle physics.
So why is it interesting to think of issues pertaining to causation? I suppose it begins by acknowledging that it is not just in the natural sciences where the idea of causation has a presence. Causation – or causal reason – is arguably a fundamental tool of human cognition. There is also an argument to be made, along quite substantial lines, that cause and effect is a fundamentally important concept for the study of the social world. I won’t write too much about my own position on the matter. What I will say is that, from my current vantage, I am of the mind that causality is a key aspect of physical reality. I speak of such an opinion from the basis of my studies in maths and physics. I am also of the mind that causality and causal reasoning is a key aspect of human experience, and I speak of this opinion on the basis of a lot of extracurricular study from statistics to psychology, history, anthropology and other areas like epistemology. I think there is also a sample basis to suggest that causation is integral to the very existence and possibility of moral philosophy and social rationality, should one wish to expand on the notion that far.
Having said all that, the concept is very difficult to capture in definitive way. There is some cool maths (think of causal systems, for example) and science that help to enable an elaboration; but the problem, traditionally, concerns the question of what causality actually means and how we effectively arrive at capturing it in complex cases where an assessment of root cause is not immediately accessible. It is easier to think about causation within the natural sciences. It becomes much more difficult when the study involves conscious subjects, and then even more difficult when investigating the development of some complex social phenomenon. In some of my extracurricular studies – like when I wrote on the subject of social pathology and human development, for example – I always had some idea of causality in mind. It operated as a sort of rational tension, if you will, expressed simply as the idea that there is causation at work in the development of x social phenomenon (the limits of investigation were defined by the notion of feedback loops). But how to do justice to it? How to drill down deep enough to elaborate a sufficient framework that might capture cause and effect, whether scientifically or when it comes to the study of human civilisation? On what scientific and empirical-philosophical foundation might one rely to expand its application, from physics to the engineering to the sociology of health?
Reflecting on these questions offers some hint as to why I am excited to read a new book byimmediately jumped to the top of my extracurricular reading list. From what I understand, the author’s argue toward a similar position that I have been tentatively sketching: that causal reasoning not only plays an integral role in the practice of modern science, but, as a cognitive tool, it has even also played an important role in human evolution.
For decades the idea that “correlation is not causation” or that “correlation does not imply causation” has been an established position in many of areas of science and philosophy. It is a mantra in statistics. And it has arguably also had significant effects on the shape of the contemporary human social world, especially pertaining to a theory of value (for better or for worse). One of the ideas, or basic tenants, is that if two things correlate this does not necessarily imply that one causes the other. There could be a third variable which, for example, causes the correlation. In that “correlation does not necessarily imply causation” is a logically valid statement, one of the more popular and mainstream examples “of a correlation being clear but causation being in doubt concerned smoking and lung cancer in the 1950s”.
In the natural sciences, as the author’s point out (from I have read about the book), the concept of causality still lives but it has become “taboo”. In areas of philosophy and social science, on the other hand, I have already mentioned that from my understanding causality is a markedly contentious concept that draws significant divides. The debates also seem to become very muddled, to the point where established post-modern philosophies even outright reject the idea of causal roots and objective knowledge for the sake of subjectivism. In a sense, it would seem a particular matter of “correlation does not equal causation” taken to the extreme of “there can be no clear sense of causation”, which, at least to my mind, is then a short step from a post-truth worldview. On the other side of the divide, efforts appear to be in search of a rigorous framework that can do justice to the idea of root causes in a social, economic and environmental sense, maintaining the idea of objective knowledge and rationality. I suspect these efforts also tie-in, or may tie-in, to systems thinking as well as other theories.
In general, I suppose one point I am making is that these matters are interesting to think about not only within a natural scientific context – why does scientific knowledge work? – but also within a social scientific context – how can we explain, or identify, the cause of a certain systemic trend?
Of all the psychology and broader social science I have read, it seems to me that often researchers struggle (when considering a wide range of matters) with the lack of some concise or fundamental definition of causation. One interesting trend that I have noticed is often in how, in order to fundamentally explain some phenomenon within the human and social world (i.e., root cause), this requires some frame of analysis or set of tools that can drill to some idea of causal roots; but, in lacking a fundamental and clear definition – the science of causality, if you will – the inclination is for that frame to become increasingly philosophically and hyper-speculative in effort to “explain away” a justification for the identification of that causal root. I notice this in a lot of literature – its like there is a sense of some causal root, but it’s not always clear how to arrive at it or capture it in a rigorous and evidenced-based way. I could be totally wrong, but in some ways I see the problem as the social sciences not yet being methodologically mature enough to realise a systematic study of fundamental root causes when it comes to explaining complex social phenomena. There are tools like causal chain analysis, or conceptual definitions like feedback loops in systems theory, which are helpful and begin to hint at the steps forward or how we might begin to elaborate on an effective framework. But it also remains that the study of causation can be so convoluted when concerning complex systems or phenomena. When thinking of causality in the context of social science, one is also dealing with the human and social world – a world of incredible bias and complex motivation in which subjectivity is also a key factor. It makes for a messy form of research, and is why I alluded earlier that it is easier to think of causality in the natural sciences than in the social sciences.
There is a simple example from psychology that elaborates on this concern: when a person notices a certain negative tendency in thought, or a certain behavioural pattern, or a certain anxiety, sometimes causality can be very easy to discern in relation to their neurosis. That person had x experience that left them feeling z, and therefore developed y psychological tendency or emotional memory. It is simple, like a child being bitten by a dog – even innocently and not overly violently causing serious harm – resulting in their developing an overly irrational fear of dogs. We have a causal relation of some description. In other instances, it could take many years of psychological care and study a few times a week, and maybe then the individual may or may not discover or scratch the surface of some clear causal relation that can reveal new perspectives about their behavioural patterns.
Unlike simple cases of cause and effect, many times it doesn’t seem like such aspects of human experience and neurosis can be derived so cleanly. If they could be derived so cleanly, humanism would be a science. In other words, it seems incredibly difficult to differentiate between correlation and causation within the context of social and human study, and thus it would seem equally difficult to navigate toward some concise explanation in terms root cause. Even though the above is an example on the level of the individual subject, it serves as a wider analogy. There are serious contemporary issues – from economics and the study of systemic global trends to culture, social psychology, geopolitical conflicts, environmental health and so on, where understanding causal roots is of fundamental importance but difficult to model. Social and systemic patterns and effects can easily be perceived as a correlative web, rather than a complex and integral chain or sequence with some causal root (or roots). Just think of the study of human history, which is notoriously subjective. The problem here is that there often seems to be many dimensions to a social phenomenon or a particular sociohistorical development. Complex systems can have many interrelations, and it doesn’t always seem possible to drill down to the level of root cause without simultaneously drilling into each of its dimensional factors.
I suppose what I am saying, to add to the above, is that causal chains are not always so simple and linear in terms of drawing a straight lined connection from one cause to an effect. In non-closed or non-simple systems – that is, in complex systems – there can be any number of factors that influence an outcome, or that influence multiple effects which then cause other effects. It’s not necessarily disorder, but the fact that in some places in the causal chain more than others, there are an increased number of factors where a simple cause-effect might deviate to multiple layers of cause and effect. Moreover, in many social cases – that is, the study of human behaviour in relation to macro trends – to obtain an increasingly accurate picture would seem to entail the inclusion of more and more variables or factors in the causal chain for that system or picture. More simply, it is the need to account for how an effect might influence a later cause, which can then create feedback in the advancing chain and produce other causes which lead to other effects which can then produce other branching causes. Another way to put it: in complex social systems, an important point that we must consider with great urgency, I think, is how to account for what I would argue is the more realistic picture of layer upon layer of instances in which causes produce effects and in which those effects may then produce other causes in a complicated, multidimensional and interactive network of events that produce some outcome.
What I am suggesting is the urgent need to account for some fundamental concept of causality within the context of complexity, should one agree with the suggestion that the most accurate model of human society is as fundamentally complex or even chaotic. Another issue is the question whether pure statistical analysis can offer explanation, especially in complex cases. As I mentioned earlier, causation and explanation of causation don’t appear to be the same thing. Instead, similar to causal chain analysis – if not the same in idea – I sometimes find it helpful to think of the explanation of causality as requiring something more multidimensional, interactive and deviant in the analysis of complex sequence. This is very abstract, but it is interesting to ponder causal chains in which any one event, or fact, can again also have dimensions of its own. I have sometimes used phrases like “the multidimensionality of phenomena” or “analysis by interdisciplinary coherence” in relation to principles of cause-effect chains. The main idea takes note from Richard Feynman and others, particularly when it comes to thinking of the study of causality in the natural sciences as a simultaneous study of each dimension of that phenomenon or system or event. I mention Feynman because in one of his lectures he hints at arriving at cause and effect – or knowledge – by way of contribution from each scientific field. In other words, each field deepens our understanding of the subject of concern. He calls it the “hierarchy of concepts”, and while very informal, it always stuck with me when thinking about epistemology. I also think there is something important that can be taken here in the sense of the interdisciplinarity of causal explanation.
Think, again, of the study of human behaviour. There are several dimensions to human behaviour: emotional, psychological, biochemical, cognitive, social (and historical), and so on. An explanation of an individual’s behaviour would likely include many of these dimensions insofar that an explanation would require one to drill down into each of these contributing factors, such as when studying violence for instance, and how each may contribute to layers of some cause-effect chain. Or, to offer another example, think of climate change, beginning with the greenhouse gas effect. Picturing a causal chain, we can also include systemic factors: economic, industrial, political, psychological or whatever could be mapped in the chain which themselves lead to indirect causes and then various impacts, and so on. With climate change, we know and understand the cause; but in terms of explanation, in the causal chain, a more complete picture would seem to begin from the root that is the production of greenhouse gases also factoring economy, governance, and knowledge then leading to industry and sector specific causes to immediate causes (i.e., deforestation) and their interrelated nodes to environmental impacts, cultural and social patterns, and so on. It’s a complicated picture that seems to require a simultaneously drilling down into each cause, indirect cause, and their impacts in addition to how they may all interrelate and produces their own spiraling effects.
In thinking in the way described above, the closest analogy or metaphor I can arrive at when referring to this notion of a pattern from within a “hierarchy of concepts”, particularly when studying the causality of a complex phenomenon with a mind toward explanation – it is like solving several simultaneous differential equations.
In closing, it will be interesting to learn what
*Edited on 15/07/18 for syntax and clarity. A link was also included to an unedited version of the first chapter of Pearl’s book.