Why Causality?
The Ideological Roots of Our Desire for Agency
The topic of causality seems omnipresent across academic disciplines, and is salient in science and philosophy alike. Illustrating the recent interest in the topic, the 2021 Nobel prize in economics was jointly awarded to Joshua Angrist and Guido Imbens “for their methodological contributions to the analysis of causal relationships”. However, while contributions like theirs advance causal inference from a technical perspective, rarely in science do we interrogate why we prioritize causal understanding of phenomena, and what we mean by a “cause”. Still, with the advancement of computational techniques that are more conducive to prediction (e.g., neural networks) than causal inference, coupled with pressure on science and research to leverage these novelties, it is important to understand why we consider causality to be “more rigorous” in the first place.
The goal of this article is to explore the ideological roots of why we revere causal understanding of phenomena, and what we mean by a cause. This discussion will, I hope, lay the groundwork for a follow-up essay that explores alternatives to causal inference that achieve underlying objectives that are similar to those which our current approach to causality strives towards.
Why Understand “Why”?
Despite being the means by which we understand “why”, causality itself seems to have an unclear “why”. That is, causality is simply accepted as the supreme approach to understanding phenomena. On the surface, the answer to “why causality” is deceptively banal— establishing a causal relationship implies that manipulating A leads to an actual change in B. This is distinct from a predictive or correlative relationship, in which A may be associated with B, but does not change B. In a discussion of causality that is iconic in the field of epistemology, philosopher David Hume evoked a rooster at dawn. Clearly, the presence of the rooster here does not cause the sun to rise, although if the rooster is crowing, it is likely dawn, indicating a correlation. Yet interestingly, such an observation is seemingly of less importance in the domain of what we consider “rigorous inquiry” than one in which the presence of a rooster could truly manipulate the sunrise.
Although the prior justification of the importance of causality might appear self-evident, the focus on phenomena over which we may possess causal agency is apparently a more recent trend. Moreover, the impulse to not merely understand the world, but to effect change and possess such agency is far from arbitrary. In fact, “science” itself has arguably not always valorized human agency to the extent it currently does. For example, during times where the locus of science was found outside the West, scientific inquiry was ostensibly more focused on areas less conducive to causal inference, such as arithmetic, geometry, and astronomy. The significant contributions to astronomy in India long before the advent of contemporary science indicate a desire to explore and understand the vastness of the cosmos, including phenomena far beyond the scope of what we could control, especially during the time of their initial exploration. Moreover, the peculiar absence of astronomy as a foundational science alongside biology, chemistry, and physics is indicative of our anthropocentric view of contemporary science, which highlights phenomena over which we (i.e., humans) possess agency. Hence, it is worth considering how we arrived at the point in which science is characterized by understanding the influence of humans or other agents over nature.
Causality as a Means of Controlling Nature
Given the historical trajectory of science towards being a discipline whose goal is to effect change, the role of causal inference is to identify the levers of control that manipulate nature. From an ideological perspective, it seems non-arbitrary that we should seek to not merely evolve with nature, but control the evolutionary process and arguably dominate nature through identifying the aforementioned levers of manipulation.
The observation that science can be seen as a tool for controlling nature seems curiously analogous to, and perhaps shares ideological roots with the motivation behind the accumulation of resources presented by Max Weber in The Protestant Ethic and the Spirit of Capitalism. In his treatise, Weber argues that our desire to not only materially survive (i.e., evolve), but thrive (i.e., control our destiny) via the accumulation of wealth is driven by ideological factors — specifically, the Calvinist ethic of industriousness, individualism, and self-sufficiency. This ethic, taken to its extreme, suggests that we humans are in total control over our well-being and destiny, and should thus be accountable for our outcomes, whether we are rich or poor, healthy or sick.
The Protestant ethic not only emphasizes the individual, but it instills Godly aspirations into them. A unique feature of Christianity is that its role model of Jesus Christ is simultaneously God and man, blurring the line between earthly and divine. Moreover, God created man in His image, inviting man to aspire to Godly levels of power. Hence, extending the Protestant ethic beyond the self, one may seek to not only control and take accountability for one’s own life, but to “play God” and do so for the world at large. The desire to influence the state of the world seems to indeed be the underlying thrust behind the valorization of causality, and relatedly, the implicit superiority of contemporary “science” amongst its competing modes of inquiry (e.g., philosophy, spirituality), which aim to “merely” understand the world.
Beyond the specific area of causality, modern science has seemingly been influenced by the aforementioned Calvinistic ideals to a surprising extent. Rather than having an antagonistic relationship, the areas of science and religion are peculiarly intertwined. One example is the scientific advancements aimed towards the development of arms, which have been partly motivated by colonialism and imperialism, and which constitute a project to control the world that has been justified “in the name of God”. Even militarism that has had the goal of peace, such as the development of the atomic bomb, was tested in a site named “trinity” by none other than renown scientist Robert J. Oppenheimer.
The influence of a certain Calvinist, millenarian ideal over the pursuit of agency does not stop at the development of weaponry. In addition, the quest for “Artificial Intelligence” (AI) is arguably rooted in the transcendence of the body towards a divination of the self through immortalization of the mind. The famous Cartesian statement “I think therefore I am” characterizes mental or spiritual existence as a transcendent phenomenon, beyond the bondage of physical life (Descartes was indeed a religious Christian). Descartes’s refuge of “pure thought” is indeed salient in AI, as the very discipline tacitly promotes the utility of “intelligence” divorced from emotion, pain, joy, or other “biases” associated more with the mortal state than what one may consider a “divine” one. Similarly, the concept of afterlife or reincarnation in Abrahamic and non-Abrahamic religions alike points to a desire to transcend corporal existence. Hence, Artificial intelligence allows us to capture this superphysical self. Moreover, the impulse to strive for this transcendence as mere mortals is potentially also driven by the quest for “Godliness” that is salient in Calvinism.
Returning to the field of astronomy, the ascent into space (aka, “the heavens”) is arguably driven in part by a Calvinist ethic as well. In another biblical reference, one of NASA’s early proposals for space travel was named Project Adam. Indeed, our pride in early space travel emphasized human agency by highlighting the milestone of having humans in space, rather than merely observing space through rockets, which would undoubtedly have been a simpler, less expensive, and less risky endeavor. This conquest of the cosmos (depicted aptly in the planting of the American flag on the Moon) again has millenarian overtones, in that it can be seen as a blurring of the lines between God and man. Through this endeavor, and even contemporary projects like SpaceX, man subconsciously attempts to see himself as simultaneously earthly and heavenly, emulating the role model of Jesus Christ, who himself experienced an ascension. In this light, it is perhaps unsurprising that the Space Force endeavor launched by the Trump administration was headed by none other than Mike Pence.
Taking a more contemporary and concrete example over our use of science as a tool for human control of nature, one might observe our reaction to COVID-19. Irrespective of one’s opinion on lab leak theories that suggest that humans intentionally created such a virus, it seems intuitive that the spread of such a virus is accelerated by scientific advancements that have created a more interconnected world, to which travel and urban concentration of the populous are integral. However, as an antidote to the problems amplified by the aforementioned science, we point to science itself. The slogan “trust science” is, in this sense, overly broad — even if one supports the effectiveness of vaccines, one cannot reduce the entire enterprise of science down to this specific byproduct, while dismissing the role of a broader conception of science in the genesis of such problems. While we may even acknowledge that, on a subconscious level, we are “solving the problems of science with science”, the ideological dimension of this endeavor is reflected in our immediate assumption that by “playing God” and controlling the aforementioned cycles rather than merely evolving with and adjusting to nature, we are necessarily doing good.
Even if one has a favorable attitude towards “control through science” based on the specific example of COVID vaccines, a potentially conflicting example might be that of climate change. Although many are indeed concerned about this issue, we are seemingly assuaged by the subconscious faith that humans will scientifically innovate themselves out of this problem. Again, despite the very problem being a byproduct of science, our preferred solutions tend towards those that involve advancements in science itself, such as autonomous vehicles and the extraction of carbon dioxide from the air (often led by solitary messianic figures such as Elon Musk) rather than the “boring”, low-hanging fruit solutions such as developing public transportation. As with coronavirus, climate change can thus be viewed as a cycle of problems created by the same ideological desire for control that attempts to rectify it. Again, the ideological dimension of this cycle is reflected in our unquestioning leap of faith that compels us to conclude that we will make sure this cycle is a net positive, despite such calculations being neither arbitrary nor objective.
Driven by the underlying ideology of intervening in nature, our interest in causality seems motivated by a desire to identify the levers of manipulation, so that we can influence such change most effectively. In order to achieve this, we must nevertheless precisely identify these levers of change — or, in econometric terms, achieve “identification”. The next section discusses the ideological dimensions behind our characterization of these cleanly-identified “causes”.
What Do We Mean By a “Why”?
Moving in a more epistemological direction, one might consider enlightenment philosopher David Hume’s discussion of the challenges in rigorously identifying a cause. While our conception of causality requires event A to be a necessary prerequisite of event B in order to be considered its cause, Hume argued that the only way we can be certain of this is through inductive leap of faith. Put simply, our conclusion that B occurs because of A is rooted in past observation, although we have no guarantee that this phenomenon will persist into the future. Moreover, it is unclear whether an underlying cause is required for phenomena to occur at all, and, even if so, whether it is possible to precisely identify such a cause.
Although causality has often been deemed the “holy grail” of science (another religious overtone), it seems to shrug off Hume’s challenges, assuming that (1) the existence of causal mechanisms can be verified through observational approaches and (2) that, with tools that are sufficiently sophisticated, these mechanisms can be broken down and isolated. While often practical in abstracting phenomena, these assumptions are again not as arbitrary as we would like to think. Moreover, a strict adherence to these assumptions is not always productive to our quest for rigorous knowledge, even within an agency-motivated framework of inquiry.
The Limits of Causal Induction
The idea that causality can be established through induction (i.e., observation of the past to understand the future) is a necessary but insufficient condition for the identification of a causal agent. That is, a cause must first exist before we attempt to precisely identify the cause itself. Curiously, we seem to ideologically lean towards the desire to have an underlying cause that can explain phenomena. This desire is illustrated by the popular expression that “everything happens for a reason”. This phrase suggests that, even if we cannot precisely identify a cause (a challenge that will be discussed later in this essay), we insist that there is such a cause. Hence, before even considering the possibility of phenomena without a cause, it is important to acknowledge the inborn aversion within our ideological landscape towards such a notion.
As with our desire for agency, there are interesting parallels between our desire for causal identification in science and religion alike. Interestingly, the tension between these two disciplines is not in whether a precise cause exists, but in whether such a cause can be inferred through induction, returning to Hume’s characterization of causality. Although science attempts to identify such a cause through empirical observation and induction, religion resorts to the divine, often citing “God” as a catch-all source of agency behind worldly phenomena. Despite having different approaches to identifying a cause, it is perhaps worth observing that the impulse to explain phenomena through a known agent is something that transcends science itself. The blind spot in our ideology is thus seemingly an ability to conceive of phenomena without a cause.
To explain how phenomena might lack a well-defined cause as an alternative to both science and religion, we can fortunately find inspiration in these disciplines themselves. For example, even the within sciences, there is some flexibility in the insistence on a precisely-defined cause that is identified through induction. As a starting point, we can look to Galileo’s experiment on the Leaning Tower of Pisa, which provided evidence that the cause of the rate of free fall of an object was gravity, rather than the mass of the object. While this conclusion requires the inductive step (i.e., that this phenomenon will persist into the future) and is not set in stone like a mathematical axiom, the cause (i.e., gravity) seems fairly well-accepted.
However, if we transition from physics to chemistry, we seem to observe more complex phenomena involving “interactions”. For example, the combination of two hydrogen atoms with one oxygen atom creates a wet molecule, despite its constituent elements not being wet. Hence, the emergence of a chemical property such as wetness does not demand a well-defined cause, beyond an ad-hoc explanation related to the way in which these specific atoms bond. Moving from chemistry to biology, the theory of evolution is arguably even more lax with regard to the need for a causal agent. Although this theory outlines a mechanism by which the evolutionary process occurs —i.e., those who are best suited to their environment are more likely to pass on their genes — the theory does not even attempt to identify well-defined levers by which species evolve in the precise manner they do. Instead, the discipline of biology seemingly submits to nature and describes its progress more abstractly.
All of the aforementioned examples fall well within the range of respected, rigorous “science”, and yet there appears to be variance in their insistence on a precise underlying cause across these frameworks of inquiry. Similarly, while religions may attribute creation to divine causes, they tend to have some conception of free will. Without diving into this thorny subject, suffice it to say that the acceptance of human agency over creation allows for some randomness in the causes of the creation, thus relaxing the strict requirement of a divine source. Hence, even ideologies that strive to explain the world through cleanly-identified causal mechanisms admit the limitations of such an approach. Similarly, the phenomena that science tries to explain may vary in the extent to which they truly have a precise cause.
The Limits of Causal Identification
Even assuming that a causal mechanism exists, it is questionable whether it is possible to identify. Yet, economics has been flooded with new methods, such as causal forests, which typically employ statistical techniques to empirically isolate a cause by controlling for observable variables and their non-linear combinations. Still, it is unclear whether, even with these approaches, a single, precise cause can be discerned.
Taking an intuitive example of the challenges in isolating a cause, one may consider the subject of coronavirus deaths. Considering the measurement of “death rate” from the virus, it is truly difficult to discern what constitutes a fatality from the virus itself. Although it is commonly known that patients with co-morbidities were far more likely to perish after catching the virus, the percentage of patients having the virus as the sole identifiable cause of mortality was about 5%. Hence, it is quite challenging to discern the virus from other agents as a cause of death.
In addition to being difficult to isolate, one might question whether a singular cause can be discerned for every phenomenon, even assuming that a cause exists. The assumption that a single cause can be isolated rests on the notion that phenomena can be broken down cleanly into independent, orthogonal axioms. Going back to the example of the wetness of hydrogen and oxygen combined in a two-to-one ratio, the emergence of wetness cannot be broken down into atoms, which can be considered the “axioms” of chemistry. Meanwhile, a blindly-executed experiment in search of causality might add one atom of oxygen to two atoms of hydrogen, produce a wet substance, and conclude that hydrogen is a cause of wetness. Yet, we understand even within the well-accepted framework of chemistry that the wetness is not caused by a specific atom, but that it is an emergent property resulting from the interaction between these atoms. Still, the ideal “science” seems to prefer the isolation of hydrogen or oxygen as an agent, potentially impeding advancements in the understanding of interactive phenomena.
The aforementioned framework for characterizing a cause rests on the assumption that phenomena can be “broken down” into independent axioms. Returning to the comparison between different sciences, even these suggest that there are varying levels of tolerance under which one can isolate such a cause —that is, while pure mathematics rests on such an assumption in order to develop abstract representations of the world, biology accepts the randomness of causes of evolution in order to deal with the nuances of empirical reality. Consequently, it may be worth having differing levels of tolerance towards identifying causal agents, depending on the complexity of the phenomenon being studied. However, science as a discipline seems to lack guidance on the degree to which such dependencies should be tolerated, and seems to view a more “broken down” conception of phenomena as more rigorous.
The preference for “breaking down” phenomena is yet another non-arbitrary ideology driving our conception of causality. Even the expression — “break it down” — seems to have uniquely positive connotations in the Anglosphere, as opposed to say the Francophone world in which “décomposer un phénomène” would be taken as vulgar, and arguably dismissing of the nuance of the context. Taking another intercultural example, while religions often employ God as a metaphor for the ultimate Truth with a capital “T”, they differ in their representations of this Truth. The preference for a “broken down” Truth is again potentially in line with the Christian ethic, which breaks down God himself into three constituent elements — the Father, the Son, and the Holy Spirit. In the Hegelian dialectical tradition of analyzing concepts “in contradiction”, one might juxtapose the aforementioned understanding of the Truth with that of other religions. In Islam, for example, one must bear witness that there is no God but God. This attestation is more than a mere pledge of allegiance, but a philosophical statement that there is “One Truth” that transcends anything that can be broken down. Meanwhile, the conception of Truth in a polytheistic worldview allows for multiple, parallel Truths, through concepts such as Satya.
Regardless of the exact ideological roots of this tendency for a “broken down” understanding of causes, modern inquiry seems to emphasize the precise identification of a well-defined cause, rather than the acceptance of complex phenomena that are either interactive, or have multiple causes. While often useful, the emphasis on this tendency may discourage exploration of such complex phenomena, which may not necessarily fit into the framework of an agency-based method of inquiry. Hence, it may be worth interrogating not only why we insist on a “why”, but the extent to which we demand a precise characterization of this “why”.
How Should We Change How We Seek “Why”?
Ultimately, the prior comments are not meant to critique our quest for why or its characterization, nor do they intend to undermine any of its underlying ideologies. In fact, these ideologies have likely been an inspiration behind advancements in our understanding of the world in many cases. Indeed, understanding the levers of manipulation, even whilst accepting philosophical assumptions such as the power of human agency, induction, and the ability to somewhat “break down” a phenomenon, can be of significant and practical value. In the example of COVID, although it is difficult to precisely identify the virus as the sole, reducible “why” behind a mortality, there are indeed prescriptive implications regarding the deleterious nature of the virus, even if these prescriptions are open to individual interpretation. At the same time, the quest for agency and control may lead humans down unproductive paths, as we arguably see in our desire to “innovate” ourselves out of challenges like climate change. While there is no denying that the role of human agency is integral to our understanding of the world, it seems that the obsession with causality forces us into a role of almost Godly omnipotence, whereas a better solution at times may be acceptance and fostering robustness to phenomena outside our control, or even with potentially “no cause”.
Given the concerns and advantages corresponding to the nature of our quest for why, one may ask — should we always seek a “cause” for a phenomenon? How precisely broken-down and identified must this “cause” be? When might it be insightful to merely observe or predict a phenomenon without necessarily specifying its exact underlying cause, and when might such analysis be appropriate within the scope of “science”? I intend to address some of these questions in a future essay, having understood some of the benefits and biases associated with the search for causality as we currently understand it.
