Select Page


These articles on causation are best read in the sequence that they appear in the menu.


There are many concepts deeply embedded in our intuitive understanding of the world which, when subjected to scrutiny, defy simple analysis. We can include here our notions of space, time, number, consciousness, and causation.

It is tempting to put seemingly intractable topics like these to one side, to ignore them, or to try and dissolve them into irrelevance. If we have, as a human race, no definitive answers to questions about what all these things are and what they do then we put them in the ‘too hard’ basket – right? But sweeping problems under the carpet (or into the basket) does not make them go away. science has made enormous advances as it has attempted to understand and explain what we mean by these ideas and how they impact our lives. The fact that we cannot make immediate sense of them does not mean that they are inconsequential metaphysical ghosts, or that studying them is futile: in making the attempt to understand, we may make unexpected incidental discoveries.


The Volcano Stromboli off the coast of Sicily
We once explained eruption as ’caused by’ supernatural agencies. Today we are inclined to say that it is due to ‘natural causes’
Courtesy of Wikipedia Commons
S.W. Denglar. Accessed 13 April 2017

If we ask the simple question ‘How does the universe work – what makes it tick?‘, somewhere in our answer will be the idea of causation. Causation underlies all our discourse about what is going on around us at all scales. Insofar as science tries to explain the order of the cosmos then causation is at its core. Anyone who is curious about the natural world must, at some time or another, have wondered about that elusive concept ’cause’.

This series of articles on causation became necessary, not just because of the intrinsic interest of the topic, but because causation is frequently called on to do work in the philosophical debate about reductionism and attempts to resolve today’s two competing scientific world views.

In the early 21st century many people in the Western world have adopted an understanding of the physical world based on the best possible explanations available to science. However, the world-view presented to us by science, its mataphysical assumptions about reality, are by no means settled. The once prevailing view of a unity of science grounded in mathematics and physics is under challenge with a new paradigm slowly emerging that looks beyond the reality of matter, energy, and fundamental particles to a reality that includes biological and social reality, meaning, and purpose. Among the topics that are part of this reconfiguraion of scientific reality are teleology, laws of nature and, underlying all this, our understanding of causation.

Most of our reasoning about what goes on around us in the world, our reasoning about ‘matters of fact’ is reasoning about causes. Causation is a foundational intellectual tool so, as scientists, we need at least some grasp of current thinking about its nature and perhaps, along the way, this will sharpen our thinking about natural processes and the use of causal inference.

Just to get warmed up for the subsequent articles consider whether you think the following are causal in any way: brain states, consciousness, equations, the plan of an aircraft, money, marriage, digestion . . .

no, dont tune out, think about it . . .


Intuitively, wouldn’t you expect the universe to be random and chaotic? Why should there be order rather than chaos?

There must be order because we know that not just anything happens from one moment to the next – that we can predict what will or might happen next to at least some degree – and that is why we can operate in the world.

And then, if there is order rather than primordial chaos then there must be a reason for that order, some kind of explanation. However, finding explanations for the order of the universe is not straightforward.

We see order all around us and it seems to take different forms – in the structure of physical objects, in the process by which the way they are formed, their arrangement in space, and the way they interact.

How did this order arise, and why is the order we see around us the way that it is? Was order imposed on the world from outside in some way or did it arise from within? How adequately does science deal with these questions?




Time-lapse photograph of an 8-ball break on a pool table
Courtesy Wikimedia Commons



Understanding the order of the world gives us the power of prediction, it relieves us of the uncertainty about why things are the way they are, and what might happen next. Humans are rational animals so the discovery of order puts us on a path that is intimately associated with reasons, explanations, understanding, purpose, and therefore meaning.

Our distant ancestors assumed that order was imposed on the world in the way that law and order are imposed on human societies, that is, by a mindful and law-giving agent like God. Today we attribute the order of the universe to its inner workings, to its predictability as determined by the laws of nature. The first well-recorded shift in thinking from ‘laws of God’ to ‘laws of nature’ was made by Western pre-Socratic and classical philosophers who tried to find naturalistic explanations for substance and change and their investigation of the lawfulness of nature. In so-doing they laid the foundations for the methods used by modern science. And from science came not only the universal laws of physics but the documentation of many other natural patterns, principles, and regularities, including those that we are still uncovering today. Here, then, is a general all-purpose definition of science: it is the study that investigates and explains the order that we observe in the universe.


Principle 1 – If there is order then it is possible to predict. Science is the study that investigates the order of the universe.

The nature of order

Defining science as the investigation of order in the world is not saying very much. Is it possible to refine further what we mean by order. Perhaps if this were done well we could even arrive at a distinction between science and non-science? How does order do its work, and are there different kinds of order? One way of clasifying order is in terms of the order we see in static structure.


Order and pattern in nature
Note the bilateral symmetry and functionally adapted (purpose-driven) structure of a male cicada Thopha saccata
Photo: Roger Spencer – Dolphin Point, New South Wales – 15 Dec. 2013

Static structure

Perhaps the most obvious evidence of order comes in the structures we see around us, the architecture of substance, not forgetting the structure we see at different scales using the technology of microscopes, telescopes, particle accelerators etc. Science has revealed order in matter in a way that our ancestors could not have imagined: the world of subatomic particles; the way substance is organized into elements of gradually increasing atomic complexity (the periodic table); the orderly way that these elements are combined to form chemical compounds; the orderly forms we see in animals and plants; the arrangement of geological strata one on top of another, and so on.

Ancient philosophers struggled to find the essence, the ultimate components of the world out of which everything else must be constructed, concluding that the fundamental building blocks of the universe were earth, air, fire, and water. Out of the science we inherited from these people has now come an answer to their questions about the essence of things. We can define gold scientifically, in a precise way, as the element with atomic number 79 where once there was only an indefinite set of properties associated with gold. We now classify the living world based on the structure of DNA rather than, as once, a superficial arrangement according to similarities and differences visible to the naked eye. The ancients understood the miraculous integration of structure and function to be seen in living organisms but knew nothing of heredity. Today we can explain the way biological forms (species) persist from generation to generation by passing on information about form and function encoded in the molecules of DNA.

Dynamic change

Then there is the order that we encounter in change: what happens over time, the order or predictability that we see in the dynamic relations between the physical objects of the universe. The most pervasive sense of order in our daily lives probably comes from the exertion of our human agency, the way we have reasons (causes) for everything we think and do.

If we are to manage our daily lives then we must categorize and describe the objects that are around us, this we do by naming the structures around us. But we also need to be aware of (and classify) the way these objects behave in relation to one-another and ourselves. If we had no idea of what was likely to happen next then we would be unable to act at all, we would be paralysed. We notice that many dynamic processes occur in an orderly way, that a preceding situation is succeded by another situation in a more or less predictable manner. We notice that plants grow well after rain, I sleep when I’m tired, and unsupported objects falls to the ground.

Cause & effect

In the temporal flux of the world some configurations of objects (causes) seem to give rise to other configurations (effects) in a relationship of antecedent and consequent. We express this orderliness by saying that effects are generated by causes, and that the effects then become causes themselves. We use this relationship to make inferences about the behaviour of the world that go beyond our immediate senses into the future and back into the past. We can predict the likelihood of some future events and sometimes explain what probably happened in the past to give rise to the present situation.

The relationship of cause and effect we call causation (or causality) and because causation is orderly we can make sense of change. But what exactly is the nature of a cause. What are we asserting when we say that C causes E? It is not that both C and E happened, or even that C was close to E in space and time. We are tempted to say that C made E happen, or something similar, but that is hardly more informative than saying that C caused E. Effects depend on causes in some way. What is this dependence? Is it a ‘necessary connection’, as philosopher Hume expressed the problem, and what is the nature of this necessary connection? Is it a physical force or something to do with the orderliness of the world as revealed by science and, if it is part of the order of the world then how exactly is this order imposed?


Causes precede effects so the cause-effect relation proceeds in time, and because it is orderly it means that we can make sense of change. Our intuitions about causation help us to deal with potential uncertainty by helping us to answer the question ‘Given a particular state of affairs, what is likely to happen next?’ In everyday conversation we use the word ’cause’ in the sense of ‘because’, that is, as a vehicle for explanation. The word and idea ’cause’ is present in so many contexts that we are unlikely to find a simple and precise definition so we must start out with broad generality and we have already discovered enough about causation to do this.

Causation is the name we give to the order that we see in movement and change: it is when one kind of thing is associated with or follows another kind of thing, and it is the foundation on which we build our explanations and predictions. Our understanding of causation allows us to intervene in and manipulate causal processes, that is, to manage the world around us. The idea of cause feeds into scientific explanation as our most effective predictive tool. No wonder 18th century philosopher David Hume described causation as ‘the cement of the universe‘.


Principle 2 – We use causation as a way to understand, describe, and explain the orderly nature of change.

The ordering principle

We have identified two kinds of order, the passive order evident in spatial structure, the way matter is organized. Then there is the order that occurs in the process of temporal change: given circumstance X, circumstance Y will follow.
If the central project of science is to explain the order of the cosmos then causation must be a key ingredient of scientific explanations that account for spatiotemporal change.

But we still have not given an account of how order itself arises. What is the principle that ensures that there is order at all? Among the candidates traditionally offered here we can include: the laws of nature, general principles and reguarities of science, physical constants, and natural selection. We will look at these in more detail later.

A dilemma

Perhaps causation is the orderly process itself, not something over and above it. When, for example, we give a full scientific account of an orderly process, say the clashing of billiard balls, we do not then, to complete the account, announce something extra, the cause. The scientific description is a sufficient explanation because it reveals the ‘order’ of what happened, its ’cause’. ‘Cause’, on this account, is not something in addition to what happened. So do we need the idea of cause in science? If indeed we can roughly equate ‘order’ and ’cause’ then we can avoid the generality of cause-talk and instead of asking ‘What was the cause of the billiard balls moving in this way?‘ we ask ‘What was the ordering process operating such that the billiard balls moved in this way?’.

By now you will probably be experiencing some of the intellectual unease associated with causation, the confusion of concept and object, the relation between causation, science, and semanics, the relation between causation and the laws of nature and more.

Here we want to take an overview of the subject with the intention of seeing how causation is associated with other topics investigated on this web site. The task is to set out the playing field and to put some order into our understanding of this key process that itself seems to put order into the world. As elsewhere this will be done through a series of challengeable principles that will be used to construct a summary of the findings. The principles and summary can then be checked against your own views.



Metaphysics & science

We want to answer two kinds of questions, one is philosophical and the other scientific. Firstly, ‘What do we mean by ‘causation’ and ’cause’?‘ and, secondly, ‘What are causes in the actual world?’. The first is an abstract question that requires some form of conceptual analysis (metaphysics), the second is a question about concrete scientific objects (ontology) and it requires an empirical investigation. These two approaches require further explanation so we will consider each in turn.

Metaphysics: what do we mean by causation?

To answer to this particular question we need to produce a description or definition of causation that is both informative and acceptable to both philosophers and scientists. The definition given in Principle 1 would be too general to serve this purpose. An informative account would need to explain causation in new terms – it would clarify what we mean by causation by explaining or ‘reducing’ it to other more familiar terms. In a reductive metaphysical account there can be no causal concepts in either the expression that is being defined (definiendum) or the definition itself (definiens), it must be reduced to non-causal facts and relations. For this reason metaphysical accounts of causation are also known as reductive accounts. A reductive account of causation aims to provide necessary and sufficient conditions for a causal relation to hold presented in terms that are both objective (they can be explained in physical terms and the laws of nature) and non-causal.

The best-known example of a reductionist theory is that of philosopher David Hume (1711-1776) whose Regularity Theory claimed that there is no such thing as causal nature, causal power, or causal necessity. We cannot draw any conclusions about what will happen next by experiencing one event in isolation. Causes do not make or force their effects, we only attribute this feature to the world because we see events occurring together in regular association (a later perspectival argument was that we project our human agency onto the world)[4]). Causation is a real, mind-independent feature of the world but the necessary connection we associate with causation does not lie in the objects and their relations but in our minds. For Hume the properties of causation, defined reductively, were temporal priority (a cause precedes its effect) and spatial contiguity (cause and effect are closely associated in both space and time) but, above all, constant conjunction (similar causes and effects occur regularly together). C causes E iff all Cs are followed by Es. Hume was an empiricist believing that only by experience can we establish the causal nature of the world and he was trying to account for our sense of necessary connection. His views are discussed in the next article under Regularity Theeory.


What if this metaphysical goal of an informative account of causation cannot be achieved? There is of course the possibility that causation simply cannot be reduced or explained in any simpler terms, that it is a fundamental and irreducible feature of existence like the colour red or concept of number. This view is referred to philosophically as primitivism and it must be admitted that since metaphysicians still disagree about the true nature of causation then primitivism is a possibility, perhaps we just cannot explain it in non-causal terms?


But perhaps we are dealing here with the opposite of primitivism. Our problem is not that causation is conceptually simple and indivisible, the difficulty is that it is a notion of such wide application and generality that we cannot find any conceptual purchase, the generality makes it hard to establish clear meaning.

Cognitive illusion

We must also take seriously the possibility that we can learn more about causation from experimental psychology than from philosophy, prompted in part by German philosopher Immanuel Kant‘s (1724-1804) view that causation is as an innate feature of our cognition (see later) or because regularity leads to a kind of causal anticipation (Hume, although the Humean theory seems to depend itself on causal assumptions).


There are many theories of cause (predominantly regularity, counterfactual dependence, and the transfer of physical factors) but each suffers counter-examples. Perhaps this is a consequence of the poor classification categories that we use for causation. Sometimes single words denote objects that might be better expressed using different words, as when we decide that one species of plant is better recognized as two. Maybe if we want clarity we should jettison its current usage by dividing it up into more meaningful categories, that is, we could use several different words to categorize the phenomena we currently place under the one word ‘causation’?

On the other hand the desire to provide a single informative summation of causation, like a definition of its essence, is doomed to failure. Some words (like the word ‘game’) lack a single essence, they share a family resemblance of characteristics so there is semantic connection but no neat necessary and sufficient definition.

At times it seems that a cause can be everything and nothing. Yet another approach is to regard ’cause’ as a highly diverse and heterogeneous word, a word under which we fit innumerable different things, such that cause-talk though of general utility, is best reconfigured into different language if we require any precision (see eliminativim below). The word ’cause’ is not that common in either everyday or scientific discourse and that could be because we are subconsciously aware of its lack of specificity. We might, for example, divide causes into ‘movers‘ (~ objects), ‘matterers‘ (~ facts), and ‘makers-happen‘ (events).[5] Rather than using a general verb like ’cause’ we use slightly more precise generalizing verbs like ‘influence’, ‘determine’, or ‘produce’ or we can nail what is going on using verbs like ‘chop’, ‘kick’, or ‘smash’.


Alternatively it may be that causation is an altogether unnecessary or superfluous concept. Perhaps ‘causation’ does not refer to anything real in the world. If this is so then all causation talk could be expunged from our language without loss, a view known as eliminativism. This view was held at one time by English philosopher Bertrand Russell who declared ‘Causality … is a relic of a bygone age, surviving like the monarchy, only because it is erroneously supposed to do no harm‘ (1912).

Principle 3 – Causation may be considered an idea so simple that it cannot be further defined (primitivism) or so general in meaning as to lack conceptual structure (polysemy) or so diverse as to warrant fragmentation into more meaningful categories (pluralism) or to be a superfluous concept lacking meaning altogether (eliminativism).

Difference-making & production

One pluralistic view divides causes into difference-makers (indirect cause) where change is usually related to circumstantial or contextual factors. Sometimes there are causes that produce effects even if they make no difference to outcomes, as with a single bullet when a firing squad uses live ammunition. But we have vaccinations that prevent illness without production so we can include here: counterfactuals ‘If not X then not Y’ (‘If I had not been born I would not have had the car accident’) included here would be presence and absence information (‘The accident only happened because the dog ran across the road; if the dog was absent it would not have happened’), omissions (‘The plant died because I forgot to water it; if the plant was watered it would have lived’). Such debates are largely topics for philosophers. Then there are the cause producers (direct cause) that is transitive, intrinsic, and local. This is the process and mechanistic causation studied by scientists. In many cases both apply.

Science: what is causation, what are causes in reality?

There is still no agreement as to whether facts about our experience can address Hume’s particular concerns about causation. As mentioned earlier necessitarians simply disagreed with Hume. When I cut a potato I am aware of a causal process that is directly evident to my senses. Empirical accounts of causation are assumed to describe what is actually going on when something ’causes’ something else, they are non-reductive because the phenomena do not need to be reduced to other terms, a position sometimes called causal realism. The task of science is to then establish empirically justified theories and hypotheses about causal mechanisms (the process or path by which an outcome is reached). We are so familiar with the idea of a cause that it might surprize scientists to know that this assignment, the scientific definition of cause, is still a work in progress.

Scientifically only things that do things can be causal, that is things that make something happen. Reductive theories include redundant causation includes omissions and absences, preemption, overdetermination, and threats to transitivity like manipulation and intervention.


There are a number of different non-reductive theories which treat causes as involving an exchange of force, energy, momentum, mass or some other ‘conserved quantity’. A major proponent of this approach is Wesley Salmon who regarded causal processes, causal interactions, and causal laws as providing the mechanisms by which the world works. Recent philosophical reductionist theories include: Process Theory, Path Analysis, Mark Theory, Conserved Quantity Theory.

One strong case has been developed by Australian philosophers Heathcote and Armstrong based on Armstrong’s theory of laws of nature, a law being a relation of necessity (N) between universals (general properties like mass (m), force (f), and rate of acceleration (a), where f = ma), so causation is simply the instantiation of a law so defined. So all Cs are followed by Es when C and E are universals as N(C,E).
The strength of causation therefore boils down to the strength of scientific explanation as based in scientific axioms a view often associated with the work of the logical positivists and logical empiricists of the period c. 1920-1950.

Much of the debate about causation boils down to differences of opinion between various schools of regularists and necessitarians.

Principle 4 – Causation can be approached by reductive conceptual analysis (mostly regularist metaphysical theories) or non-reductive analysis (mostly necessitarian empirical theories).

How to proceed?

Causation is a broad concept with the word ’cause’ used in many different and sometimes contrasting or contradictory senses. We have many different intuitions about the nature of causation. Where to start? To provide a classical and satisfactory account of a concept we must provide the necessary and sufficient conditions for its application. So do we require a stipulative definition? If we carry out a conceptual analysis what would this entail and how do we begin a metaphysical reduction? What is the scientific way to establish what we mean by ’cause’?

Stipulative definitions work well when specialist technical circumstances benefit from the clarity of precise definition but they do not resolve ambiguous or difficult situations. We cannot provide a stipulative definition of something as general as causation because the sheer diversity of opinion will regard any restrictive definition as a matter of personal preference. That is, as it were, cheating, and therefore justifiably challenged. And because of its generality it is difficult to say anything informative. Conceptual analysis can take many forms, the way the concept of causation is woven into other concepts, used in technical scientific or other ways, its role both within language and our cognition. Besides, it turns out that many concepts are not of the strict classical kind. In fact the majority of concepts have no such conditions but merely share a family resemblance. The concept ‘game’ is usually cited as an example, it being impossible to define ‘game’ in a classical way. In the absence of a classical definition of causation can we can safely assume that it is of the latter kind?

I have tackled the topic under the three general headings of language, cognition, and conceptual analysis even though these categories themselves might not seem appropriate to the task (theories):

1. To examine the various ways in which we use the word ’cause’ in our everyday and technical language, including what is sometimes called the folk concept of cause
2. To assess the relationship between our psychological intuitions about causation and the causation we assume is operating in the external world. That is, to compare our manifest image and scientific image of causation
3. To examine the range of theories of causation


In examining the relationship between science and language Steven Pinker in ‘The Stuff of Thought’ (2008) discusses the way that we embed the key scientific concepts of space, time, matter, and causality in everyday language. Nouns express matter as stuff and things extended along one or more dimensions. Verbs express causality as agents acting on something. Verb tenses express time as activities and events along a single dimension. Prepositions express space as places and objects in spatial relationships (on, under, to, from etc.). This language of intuitive physics may not agree with the findings of modern physics but, like all metaphor, ‘it helps us to reason, quantify experience, and create a causal framework for events in a way that allows us to assign responsibility‘. When looked at this way language is a toolbox that conveniently and immediately transfers life’s most obscure, abstract, and profound mysteries into a world that is factual, knowable, willable, and communicable: the logic of familiar concrete situations can be used as the metaphorical logic for mapping and making inferences about more abstract things like time, quantity, state, change, action, cause, purpose, means, categories, and so on. Metaphor allows us to understand one mental domain in terms of another.

Polysemy & the grammar of cause

Grammatically ‘causation’ is a synonym of ‘causality’ the latter seemingly more closely related to the idea of determinism. The adjective ‘causal’ captures much of the semantic variety of ’cause’ as referring to relationships like laws, regularities, correlations, properties, or patterns of functional dependence. Though ’cause’ as a transitive verb implies an action (to drink something is to cause it to be drunk, and so on), but also serves as an object that might be an abstract concept or relation between events (it was the smoke that caused the cancer). When we say ‘Smoking is a cause of cancer’ alternatively we could say ‘Smokers often die from cancer’.[3] That is, we can re-frame the statement to omit cause altogether. We are inclined to think of a cause like an object or force in situations where this is clearly not so. Could this always be the case – should we take an eliminativist stand?

Like most difficult concepts causation is tied up in the tangle of words and their meanings, encompassing a wide range of phenomena and ideas. We can extract from the brief history of the subject given above some key words and concepts that are part of the semantic cloud enveloping the word ’cause’: agency, determinism, explanation, intervention, law, manipulation, mechanism, necessity, order, probability, process, and regularity. In common usage the meanings of the words ‘reason’, ‘explanation’, and ‘cause’ are interwoven and of such impenetrable and slippery generality that it is hardly surprising that, even after 2000 years of study, we still have no neat necessary and sufficient formulation of what we mean by ’cause’ and ‘causation’. The word ‘causation’ does not seem closely tied to a particular relation but more to a property of some relationships and not others. Perhaps we can only get a purchase on the concept by developing a pluralist account that recognizes different kinds of causation.

Most of our causal claims are made at a macro and highly general scale (‘increased violence is a consequence of (caused by) immigration’).

The role of verbs as causal agents

Causation is deeply and pervasively embedded in language through our use of verbs. Verbs in general can be used to generate causal statements. Transitive verbs enshrine the idea of causation. To make something is to cause it to be made; to hurt someone is to cause them pain. Simile declares itself with the word ‘like’ but metaphor does not give itself away, it claims identity and, if we are not careful, we simply take it at face value, we take it for reality. In the philosophical literature it is conventional to treat causes as events but we do not do this in conversation as can be seen from the examples given above. When we speak of a cause, the causal agent can take many forms … it may be an object, relation, fact, event, thought, action, trope, or state of affairs – are some or all of these metaphor? But then objects can be causes too. We say ‘The stone broke the window’. Causality is central to every aspect of our existence, a critical intellectual tool that is harnessed by almost all forms of explanation and therefore most of the sentences we speak. Above all we see it in science as our most effective mode of explaining the world.

Causal ubiquity

Consider the following ‘X caused Y’ statements:

1. Discuss the causes of the French Revolution
2. What caused the red light to change from red to green?
3. Smoking causes lung cancer
4. Smoke causes lung cancer
5. The stone smashed the window
6. TV violence causes criminal behaviour
7. Forgetting to water the garden today caused the hydrangeas to die
8. Telling lies causes your nose to grow
9. Happy cows are productive cows
10. ‘You make me sick!’

Causal language in science

The word ’cause’ is of such generality that it can be informatively replaced by verbs expressing the specificity of the particular situation. Rather than ‘I caused the glass to break’ we say something like ‘I dropped the glass’. This is especially evident in science where th e word cause is rather scarce. Rather than saying ‘The chemical caused the litmus paper to turn blue’ we might say ‘The litmus paper turned blue’ or ‘The chemical gave an alkaline reaction’ are somesuch. In many contextx the use of the word ’cause’ can seem rather archaic.

Cause is ubiquitous, the concept of cause serving as a general abstraction of the many different specific and general kinds of cause expressed largely through mechanistic verbs. Indeed, unless we insist on ’cause’ used in a restrictive or technical sense, a cause can be just about anything the concept dissolving into formless semantic generality. To remove the sloppiness of everyday communication researches tend to regard a cause is an event not an object. An event is something that happens, a change. This restricts the possibilities, so instead of saying ‘The stone caused the window to break’ we make cause abstract such that ‘Hitting of the window by the stone caused the window to break’would say generally treated as an event this can easily obscure the generality with which we are working. It seems that anything may be regarded as a cause including abstract, imaginary, absent, or non-existent objects – as when the unicorn I imagine in my dreams causes me to wake up at night, or my forgetting to water the pot plants causes them to die, or the distance from home to work causes me to catch a train.

((We tend to use the word ‘event’ to indicate a particular change, a singular event, although that singular event can have generality or complexity as when we speak of a ‘swimming event’. When the event or change is strongly predictable we tend to speak of it as a ’cause’.))

Cause & explanation

In common usage ’cause’ is often used as a synonym for ‘explanation’. ‘Cause’ means ‘because’ and it is a characteristic of all answers to such questions that they express order in some form. If there can be an answer then there is a reason, and if there is a reason then that reason expresses orderliness in some comprehensible and communicable form. Further, if there is order then something must account for that order itself. Of its vary nature order is not random and chaotic, it expresses something that is determined in some way. Can we make any sense of the semantic cloud that envelopes ’cause’, ‘explanation’, ‘reason’, ‘because’, and ‘determined by’? Historically it is the link between cause and explanation that has commanded most attention. To explain something is to give an account of its causes. However, not all explanations are causal, and not all causes are explanatory.

Are all explanations causal?

‘Why?’ seeks an explanation and it implies ‘What was the cause?’. Cause is similar to explanation in at least three ways. Firstly, it accounts for the difference between knowing and understanding. Knowing is necessary but not sufficient for understanding. Secondly, it overcomes the why-regress, the desire to demand explanations of explanations. We can know that C caused E without kowing what caused C. Most why’s simply peter out at a point of familiarity. Thirdly, they are self-evidencing. The thing being explained is itself part of the explanation as when, seeing that the lights have been left on, I take this as itself being a reason to blame my son (in a rather circular way). But not all explanations are causal as witnessed by maths and philosophy. . Physical exlanations are also not always causal as when I describe the distribution of parts of a flower to someone without dwelling on their function.

It may be possible to expand the idea of causation to include the idea of determined outcomes but this leads towards an understanding of explanation as scientific explanation as a valid deductive argument whose premises entail scientific law that encompasses the phenomenon to be explained (deductive-nomological or hypothetico-deductive method) for which there are many objections.

Are all causes explanatory? Contrastivism

Each effect has many causes and not all of them explain it. Ultimately I can trace my headache to the fact that I exist, which itself is a consequence of the Big Bang – but we do not generally treat these factors as explanations. There is a distinction between explanations which deal with static conditions (the fire occurred in the presence of oxygen) and changing conditions (fire occurred when he struck a match) but the distinction between static and changing is intrinsic to the cause, but between explanatory and unexplanatory cause is relative to the effect requiring explanation and this devolves to the pragmatic theory of causation.

Explanations may not be of a logical form that proceeds from premises to conclusion, instead they simply answer the question ‘Why?’ – ‘Why this rather than something else?’ We intuitively think of and express causation as a two-place relation, X causes Y. But often in ambiguous cases clarity only emerges with a four-place relation e.g. C may cause E when P, or C may not cause E when Q. The contrasts or foils need to be taken into consideration. And for epistemic skepticism: if ‘knowing’ means knowing everything then we cannot know. If ‘knowing’ means having agreed shared knowledge then we all know. This approach to explanation is called contrastivism. Under contrastivism causal relations have the form ‘C rather than C* causes E rather than E*’. This accounts for most of the factors that distinguish explanatory from unexplanatory causes since the foil may be probabilistic, counterfactual and of other kinds. It appears that some causes are unexplanatory precisely because they are contrastive, where difference between fact and foil is what matters.

Why do causes explain

Some causes explain their effects but effects do not explain their causes. This asymmetry is near universal although there is the dubious claim that biological traits are often functions which are effects. Explaining the asymmetry itself can easily become circular, involving causal talk in another guise. Certainly the effects of something, P, generally tell us more about P than its causes. However, perhaps th reason why causes rather than effects explain is because effects are controlled by causes that make a difference and without which the effect would not occur – bearing in mind that not all causes make a difference. Where there are multiple causes it is the one that makes a difference that does the explaining.

Inference to best explanation

Explanations are a guide to causal inference. There may well be competing causal hypotheses based on the same data and in these situations we usually opt for the explanation that deals with the evidence most convincingly. Darwin inferred natural selection as the best explanation for his observations. This is an extension of the idea of self-evidencing explanations with their benign circularity, being supported by the very evidence they are supposed to explain. This inverts the expectation that inference is prior to explanation. However ‘best’ is problematic although it may be related to factors like mechanism, precision, scope, simplicity, and unification.

Aristotle’s causes

For ancient Greek philosopher Aristotle causes and explanations were more or less equivalent. According to Aristotle, answers to the question ‘Why is it so?‘ (in other words ‘What was the cause?‘) have four general answers, his material, efficient, formal and final causes (his four kinds of ‘becauses’). Each ’cause’ needs a particular kind of answer: the material cause the matter of which the object was made, the efficient cause concerns the mode of origin, something that makes change and motion start or stop, the thing that brings something about. This might be a mechanical interaction like one billiard ball hitting another or parents producing a child, formal causes are the factors determining particular shape, nature, or characteristics (like a definition), and the final cause which was about purposes or ends served. Regardless of the kind of cause discussed there is always something to be explained (explicandum) and the explanation itself (explanans). Some explanations depend on logic, others on empirical evidence.

Scientists of the early modern period mistrusted Aristotle’s four causes, restricting the scientific use of the word ’cause’ to Aristotle’s efficient cause which answered the question ‘What made it happen?’. Aristotle’s favourite was final cause, as a purpose, goal, or end but these early scientists regarded final cause as undesirable and this opinion has continued to the present day. Only humans have purposes, ends, and goals so final cause can only be generated by a supernatural agent, or through comparison with intentional human activity. Nature doesn’t strive for ends … And yet ‘ends’ are locked into the very notion of causation that is at the centre of science. What is an effect but an end? Nature, though not striving, still demonstrated physical constants (the laws of nature) that entail ‘ends’. The idea of a cause is incoherent without its corresponding effect (see Purpose). Be that as it may, the sense of the word ’cause’ as used in science is a technical one and we have taken this on board when engaging in serious discussion of cause. In these circumstances causal explanation is a particular kind of explanation, a scientific explanation.

Explanations, virtually without exception, identify causes. To explain is usually therefore to provide a cause – to show how an antecedent circumstance or cause, C, gave rise to a consequent situation or effect, E. It is still a moot point whether causal explanations are founded in ontology (our scientific understanding of the properties of the universe e.g. Wesley Salmon) or epistemology (the way we organize our knowledge and its logical or probabilistic relations e.g. van Fraassen). So, in general terms, we can regard causation as either a physical or metaphysical relationship between events such that either C causes E no matter how we decide to describe these events (ontology, what exists) while an explanation is a description of cause C and its effect E (epistemology, what we can know). Only particular descriptions of C and E will generate an explanation. So the description or explanation ‘The event that Peter described sank the boat’ becomes an explanation when we know that the event was ‘hitting a mine’.

Pragmatic or folk causation: narrow & broad context

I am in court faced with litigation over a car accident. If I had not been born then I would not have had the car accident. Is this relevant – isn’t it part of the causal chain leading up to the accident? A court case set up to establish the cause of my car accident would probably select one condition out of a set of conditions which are together sufficient for the accident and, for the purposes of a verdict, this would be called the cause. Background factors necessary for the event to occur at all are simply taken for granted and ignored. In such cases it is the most salient causes, given the particular context and specific interests, that are given priority. These may include the last condition that occurred before the effect took place, or the most conspicuous.

This example demonstrates how in daily life, for convenience, we ignore the complexity of causal antecedents, the multiplicity of interacting factors and counterfactual dependencies that are all part of any given situation. We can assert with confidence that what we refer to as causes are really only partial causes. In communication we habitually speak of cause in the singular. This leads us towards a cognitive bias, the assumption that in most cases when something happens there is just one cause. Our selective cognitive focus on one segment of a complex picture is discussed later.

Principle 5 – Everyday, legal, and folk causation depends on, and is related to, the particular human context and human interests, not the more general causal context.


Any event (or cause) in the universe is just one ingredient in a multiplicity of interacting objects, properties, and relations that we are aware of through the processing that goes on in our sensory apparatus and brain. This barrage of sensory information is organized into a meaningful unity by our cognition which uses some combination of automatic or innate unconscious intuition and conscious deliberation to selectively discriminate cognitive objects of special interest. The existence of foreground and background in our consciousness we can call cognitive focus and without this capacity we could not survive. Background and foreground may vary in their importance in any particular context but background is generally treated as (temporarily) irrelevant, unimportant, assumed, a black box. Background can be referred to loosely as the environment, state of affairs, boundary conditions, or conditions. For simplicity I shall refer to it as ‘context’.

If causation is not in the world but in our minds then we need a theory of what it is doing. There is no shortage of suggestions that are open to empirical investigation by experimental psychology. Hume by suggesting that we do not need the notion of cause itself, only the notion of regular conjunction. Because of constant associations we project causation onto the world as a mode of prediction (projectivism). Others have suggested that it is not our need to predict but our human agency, our human capacity to ‘make things happen’ that is doing the work (perspectivism), that we perceive the external world with the cast of our own minds as being made up of means (causes) and ends (effects). Others hold that we need causation as a mental tool for predicting, explaining, and controlling our environment. Kant thought that it was a biological necessity, like the way we see colour, it is simply the way we structure the world (innate human cognitive disposition).

Several important factors flow from our psychology. Firstly, we must be aware that we have a uniquely human species-specific interpretation of what the world is like outside our minds (see the manifest image) that is supplemented by the view of the world presented to us by science as the scientific image, a modern version of the old philosophical distinction between appearance and reality. With causation, as in so many other confusing situations concerning reality, much confusion arises over the reconciliation of our manifest and scientific images.

This is where most accounts of causation begin, with the possibility that causation is more a matter of what is going on in our minds than what exists in the world.

Cognitive categorization

If we assume the world is more or less as our best scientific explanations tell us then everything, all objects (including their properties and relations) is in a continuous state of flux. What, we must ask, in this continuous flux, is an event (cause or effect)? An event would seem to be like a snapshot of one segment of the general flux frozen in space and time. In purely physical terms our specially isolated ‘events’ involve the more obvious interactions of physical objects, their attraction, repulsion, collision, and manifestation of their properties. The universe does not recognize scale as our perceptions do, so the relation of a nebula to the rest of the universe is of the same significance (or lack thereof) as the pulsing of a caesium atom or the beating of a heart. In the sense that every moment is a consequence of the past and a cause of the future on a universal scale then the configuration of the universe or any part of it, (at any moment of any duration) is also an event, a cause, or effect. But the universe does not fragment its processes into individual events, causes and effects since it is in a continuous flux, only our minds break up this flow into segments that have special significance or impact on ourselves. We abstract from the universe the order that interests us most, trusting that this order must exist in the universe itself. Science does its best to detect the order that lies outside our minds. In recent times it has also paid particular attention to the order that might be imposed by our minds themselves.

Principle 6 – We observe order in the general flux of the universe and our daily lives but the fragmentation of the universe into cause, effect, and event is added by our minds.

The Manifest & Scientific images

We interpret the world from two key perspectives, that of the Manifest Image and that of the Scientific Image discussed in more detail in the article Reality and representation. The Manifest Image is the world as presented to us by our uniquely evolved human cognition and perception. The world of the Manifest Image is not an illusion but it is, nevertheless, our human interpretation of the world and it contains the objects of our everyday experience or common sense such as tables, trees, cars, colours, shapes, smells, and textures. The scientific image is the world as perceived through our Manifest Image supplemented by the knowledge and language that has been accrued through the use of sophisticated technology. This has extended our senses and given us a more objective, less anthropocentric, view of things. This world contains objects, many abstract and not directly accessible to the senses, like atoms, molecules, spacetime, genetic inheritance, fields, forces, and numbers. The Manifest and Scientific images do not always sit comportably together: though the objects of the manifest image are not illusory science often reveals them in a new light (they were not what we assumed them to be).

The Scientific Image

Let’s take a look at the world through the Scientific Image. We can distinguish objects that persist in time, objects that are of various composition and assorted properties. These objects are in a state of flux, moving in relation to one-another and influencing one-another in various ways. Everything is interconnected. There can be influence at a distance as with the gravitational attraction of the Earth and Moon. We can discern order in both the structure of the objects of the universe and their mode of interaction, this order being explained (in part) by certain physical constants that we call the laws of physics. In sum we can understand this as a world of objects in flux whose configurations we can predict to a greater or lesser degree depending on the quality of our science, the scale at which we are working, and the particular part of the whole that we are investigating.

Where do we see cause in this characterization of the flux of the universe? We know that there are physical forces in operation and that objects push, pull, collide, and fragment. Collisions catch our attention but they are only one part of the overall flux. Like Hume, we note that the orderliness of interaction that is a consequence of the orderliness of the universe itself. When we speak of an ‘event’ or a ’cause’ we are referring to a minute segment of the flux isolated in time, a particular configuration of existence. The cause is not something in addition to this configuration that acts independently of it like a foot kicking a ball. When we say ‘The poison caused his death‘ we have noted that one situation follows another but we also think there must be something more, something in addition to this, the causal connection.

The Manifest Image

With the Manifest Image we isolate particular configurations of the world (as representations in our imagination) so that we can explain what is happening and predict what might happen next. Though influence are many and complex it is the collisions, loud noises, extreme movements, bright colours, and strong smells, that capture our attention. It is the manifestations of the flux that have the greatest impact on our senses that we notice. This cognitive segregation, the mental fragmentation of experience into (mostly), those things that have a direct impact on us by pushing, pulling, and colliding, that we call ’causes’ and ‘events’. Historically it is these things that we have had to be aware of if we were to survive.

We experience causation as pervasive but most conspicuous in collisions and extreme events that impact our senses. It is certainly real to us, not an illusion, but it is not what we might think it is, it is just our take on the way things are. It is not something over and above more familiar elements, it is simply the orderly reconfiguration of objects, properties, and relations that goes on constantly around us.

Principle 7 – A cause is the way we refer to an isolated segment of experience, or a particular configuration of the world flux, that proceeds in a predictable way

It is important to realize that when science presents us with different objects than the manifest image this does not necessarily make the objects of the manifest image illusory, it is just that there are different interpretations of the same objects.

If we assume scientific explanation follows a deductive model than causes are in particular circumstances must be lawfully sufficient for their effects ‘C is a cause of E iff C belongs to a set of contemporaneous events that is minimally sufficient for E‘ but there are problems here.

Cause, necessity, regularity, correlation, association, probability

Causes are exlanatory and predictive (they are an accounting for order).

Philosophical studies of causation generally begin with Scottish skeptical philosopher David Hume[1] who established the framework for discussion about causation in the modern era. Hume noted the way that we regard causation as implying a necessary connection. As an empiricist he believed that all our ideas and knowledge come ultimately from experience and our reflection on that experience. He claimed that when we see one billiard ball X moving towards and hitting billiard ball Y, which then moves away, there is no third factor C present, a cause. That is, we do not have a sense experience which can count as empirical evidence for something special, a cause. All that we can derive from our sensory experience is a regularity of occurrence between events of type A and events of type B. This, he argued, is not a necessary connection. Other things being equal, we regard events as causal when they are frequently associated in time and space. For Hume a cause does not compel or produce its effect, it does not ‘make it happen’, it is the regularity itself, the patterns that arise in the flux of events. So causation is not necessary, it is contingent. In modern terms Hume was arguing that we perceive causation by a regularity that links cause to effect as correlation. Hume noticed that cause followed effect and that cause and effect were associated in space, this leading to his famous definition of cause as temporal priority with spatial contiguity but, above all, constant conjunction of events. There is only conjunction, never connection. No matter how complicated or sophisticated a process we choose to examine, causation boils down to regularities that hold it together, these regularities generally being regarded as the law of nature. For Hume what established the facts of causation was not causation itself but the laws of nature; it is these regularities that make causal statements meaningful.

Necessity & probability


Causation presents us with a sense of compulsion, causes make their effects happen, the cause-effect relationship is not accidental. Remove support from a physical object and it falls. Species of a particular kind produce similar offspring: humans do not give birth to insects. The cause is sufficient for the effect. Hume’s is a reductive theory that removes causal necessity: it reduces causation to non-causal facts and relations. His contingent interpretation of causation was swimming against the stream. Real necessity, he believed, did not lie in the world of experience, only in ‘relations of ideas’ (logic and maths, as in 3+2=5). Our common understanding of causation as necessity, Hume argued, arises by the habit of mind associated with constant conjunction or regularity. We need repetition to claim causation. If one event always follows another we believe the first causes the second: the necessity itself is a psychological projection. The necessitarian realist singularist could point to Humes difficulty with accidental regularity and his dependent on other events at other times and the need for multiple instances. Also the way abstracting from actual cases leads to conclusions like the assumption of one cause to one effect. If a set of causes C is sufficient for effect E then if C occurs, so must E, in a genuinely necessary way.

Though causal connection can seem stronger than constant conjunction, necessity seems to go too far. The world around us is predictable – but only to a degree. A smoothly running and reliable car engine can fail when a part breaks (subtractive interference) or when water gets into the petrol (additive interference). If a connection is to be ncessary then C must cause E regardless of variation in other circumstances. In the real world this is a tall order: we can almost always think of some additive circumstance that can interfere with apparent necessity. The contrast between necessity and contingency seems too extreme though. Causation often seems to be about dispositions, propensities, and probabilities. Cusation is in some sense often indeterministic.


Throughout history causation has been strongly associated with necessity. Cause C must have effect E. Hume interpreted cause as regularity of association. . We might even relax our views on regularity making it ‘quite’ regular as when we say that ‘smoking causes lung cancer’. After all, when we take medication knowing that most people, most of the time, taking this particular medication get better aren’t we entitled to claim this as a causal relation, even though we are only dealing in probabilities?


There are problems for Hume’s account. When day follows night or thunder follows lightning is this causation? Is there no actual connection between taking a pain-killer and the relief we feel? Many modern-day Humeans recognize this shortcoming and simply adopt the view that we can distinguish between causes, correlations, and coincidences, but it is a problem that the necessitarian need not worry about. Correlation (association) is a relationship between two or more, variables. Unless the variables are properly controlled for, there may be other unknown or ignored variables affecting the relationship so a wide variety of statistical tools from multiple regression to assorted modelling techniques are used to compensate. Causation occurs when changes in one variable invariably and directly elicit changes in the other. Most research deals in correlation, to claim causation, is much more difficult to prove than correlation.

Singularist interpretation

Hume’s theory is presented to us in general terms. One thing causes another only if it is part of a general regularity or covering law (a type event). This is the broad pattern of much scientific explanation. But we can also look at events like the singularists do by ignoring everything but the factors entailed in a single event (a token event). Whether C caused E does not depend on other situations (a covering law) but whether this particular C necessitated this particular E. Such individual cases seem to arise as a consequence of an intrinsic power or disposition.

The early moderns had restricted science to material-efficient cause. But the ghost of Aristotle has never been fully exorcised. A law of nature so beloved of th erScientific Revolution was Aristotle’s formal cause in another guise. Universal laws explained the order in physical structures and processes, Aristotle’s formal cause. And final cause is implicit in causal necessity, in all things that are necessary, predictable, and with inevitable outcomes. An effect is simply an inevitable and actualized cause. Without an effect, end, or final cause (telos), the notion of cause is empty, it has no meaning. For Aristotle causes are the active originators of change ‘for the sake of’ some end, they are the initiators of change that has predictable outcomes.

Until the last few decades it was thought that to explain something scientifically was to deduce a cause based on the initial conditions and a general or universal law (see Reason & science). This might have proved acceptable for physics and chemistry but has proved intractable for biosocial phenomena.

Powers, energies, and forces are not themselves evident to the senses, only their effects.

A brief history

A brief outline of the history of causality will introduce many of the general themes that will be discussed in more detail later and provide a context for today’s research. The struggle to provide a generally-accepted definition or account of causation that could usefully inform science or metaphysics can be followed chronologically through the range of definitions quoted at the end of this article.

How has the debate between regularists and necessitarians evolved over time?

The ancients

From our early ancestors came the idea of causation as an ordering of the world as a consequence of the actions of mindful or intelligent causal agents (gods, demons, spirits, humans) so studying causal relationships was a matter of assigning cause to the appropriate agency. The pre-Socratic philosophers provided the first recorded naturalistic explanations of the world with cause attributed, not to spiritual agents, but arising out of nature itself.

In the classical Greek philosophy of Plato we encounter what later became known as the Law of Causality or Causal Maxim, that ‘Nothing can come to be without a cause‘ thus introducing the idea of orderly and exceptionless regularity, necessity, and determinism. The fatalistic Stoic philosophers then added that our notions of chance and possibility are simply an admission of our ignorance of deterministic detail.

Aristotle treated ’cause’ as a synonym of ‘because’, an alternative word for ‘explanation’, explanations being of four basic kinds: material, efficient, formal, and final (also see later).

The Scientific Revolution

Aristotle’s ideas were modified somewhat during the scholasticism of the Middle Ages, but his authority was significantly challenged during the Scientific Revolution (c. 1550-1700) and Renaissance, most notably by Englishman Francis Bacon. The order in nature was perceived through the metaphor of a clock-like mechanism as supported by the famous Italian Galileo (1564-1642) and Frenchman Descartes (1596-1650). Science was now grounded less in Aristotelian logic and more in the actual substance of the world which was treated as matter in motion to be investigated by experiment and observation. Aristotle’s four causes were reduced to one, the single idea of efficient cause which for Aristotle was ‘that without which something else could not happen‘. Efficient cause was thus strongly associated with natural necessity (determinism), motion, and law-like behaviour. In a way efficient cause had now become the agency of causation.

Rationalists & empiricists of the 17th and 18th centuries

Causation was now subjected to the scrutiny of philosophers of the great traditions of rationalism (Descartes, Hobbes, Spinoza, and Liebniz) and empiricism (Locke, Newton, Hume, and Mill). Rationalists for the most part maintained that the relationship between cause and effect was a logical one while empiricists insisted that it must be grounded in experience. Emphasis was on motion since nothing happens if nothing moves. Causation was the result of contact (contiguity), there being no action at a distance, and all was a consequence of necessity. To be otherwise was inconceivable as for many the necessity was ordained by God – there was nothing without a reason, no effect without a cause. Philosopher-scientist of the early modern period Thomas Hobbes (1588-1679) promoted the idea of cause as an event by insisting that the objects of the causal relation (the relata) are not bodies or substances like billiard balls – ‘objects are not causes or effects’ – it is their movements.

The scientific genius Isaac Newton (1642-1727) used mathematics and physics to establish the universal laws of motion that we now call Newtonian mechanics. He, more than any other scientist in history, consumated the ancient Greek desire for a naturalistic explanation of order in the cosmos. He was prepared to describe motion in terms of causes imposing forces that constrain or compel, push or pull, but he refused to comment on the nature of the forces themselves: he defined causes in his Principia of 1687 as ‘forces or constraints that compel moving bodies to behave differently than they would have done without them‘. He was an instrumentalist believing that we cannot comment on objects themselves, only the way that they behave. Newton, in a sense, defied the causal maxim since with his first law of motion (that things stay the same unless acted upon by a force) he treated law-like behaviour as necessity according to a law and causal events as constraining or compelling forces.

The two greatest philosophers of the 18th century, David Hume and Immanuel Kant, challenged tradition by treating causation as subjectivist, as in some sense imposed on the world by the mind. For Hume causation was characterised by three properties: temporal priority and spatial contiguity but, above all, constant conjunction. This was a reductionist theory since he had reduced causation to something simpler and more familiar, regularity of association. Other Regularity Theories would follow including that of John Stuart Mill. For Hume cause was not something in the world but a psychological projection resulting from the regular association of certain events. Hume explained not what causation is but how we form causal concepts and he replaced necessity with regularity. His skepticism insisted that science proceeded by inductive logic which was not necessary but probabilistic since it assumed the ‘uniformity of nature’, that things in the future would follow the patterns of the past. Hume’s contention that regularity was a sufficient but not necessary reason for causation was challenged by using examples like day following night, thunder following lightning.

For Kant the universality of causation was not, as in Hume, established empirically by induction but as an a priori rule, a necessary condition of experience, an innate characteristic of human cognition according to which one state follows another necessarily. Kant believed we perceive causation because it is part of our human nature in the same way that we perceive colour. Hence, the necessary connection that every event has a cause. However, he agreed with Hume that causal laws belong to experience, not to things.

German philosopher Immanuel Kant challenged Hume’s empirical assumption by claiming that our sense of causation is an innate characteristic of our cognition. Hume’s claim was that causation was our experience of regularity (a synthetic a posteriori judgement i.e. not based on logic but in experience). Kant agreed with Hume that there is nothing in constant conjunction that implies causality but he regarded causation as a consequence of synthetic a priori judgements because causality was not learned from without by observation of constant conjunction but imposed from within by an internal Category of the mind (i.e. as a consequence of the innate characteristics of our cognition). On Hume’s view causality was contingent, on Kant’s view it was necessary. Our attribution of causation is imposed through a priori necessity, that is, it is an innate feature of our perception.

Hume had demolished the causal maxim which claimed that whatever exists must have a cause or ground for its existence. Kant agreed with Hume that substantive knowledge can only be obtained through the objective sciences and that reason could only act as a tool to organize information and draw inferences from empirical claims. But for Kant lawfulness is not perceived by our organs of sense but by our organ of cognition. It is by using our reason that we can distinguish between the constant conjunction of colliding billiard balls on the one hand, and the constant conjunction of day and night on the other, or the distinction between a moving car and the way a stationary car appears to change when we walk around it (differentiating movement in the observer and the observed) and other such cases. Furthermore for Kant a percipient can distinguish between scientific lawful causation (e.g. Newton’s law of gravitational attraction) and mere conjunction. Just as we are born with the human capacity to experience colour and smell in a particular way, so we are born with a causal perception of the world.

So, both Hume and Kant agreed that causation is a product of the mind: it is not present in the external world. Hume believed it arose as a psychological habit, a consequence of the constant conjunction of certain events, that we are aware of our own causal agency and project this onto the world, while Kant saw it as an innate property of our cognition, an a priori mental precondition of all possible experience. These views would be supported by Bertrand Russell who stated in 1913 ‘The law of causality, I believe, like much that passes muster among philosophers, is a relic of a bygone age, surviving, like the monarchy, only because it is erroneously supposed to do no harm.

In 1771 Kant decided for all time that we cannot have necessary and universal knowledge of independent reality: both reason and experience were needed for us to have knowledge but neither reason or experience or the two in combination could give us knowledge of independent reality.[7]

In the face of such a withering a priori challenge it was up to scientific ‘realists’[6] to make their case explicit and compelling. Does causation actually exist in the external world? Can the causal maxim ‘Whatever exists must have a cause or ground for its existence‘ be defended?

John Stuart Mill was a radical empiricist who followed in the tradition of Locke, Newton and Hume and he defined cause as ‘the antecedent, or concurrence of antecedents, on which a phenomenon is invariably and unconditionally consequent‘ (the ‘unconditionally’ part was stated to eliminate cases like day following night). For Mill what we usually call a cause is generally only a partial cause, often something that just happens to appear last or stand out as the most obvious. So a cause is a set of conditions sufficient to produce the effect, with each condition necessary but none of them individually sufficient. The social sciences, he believed, were simply too complex for experimentation. These subjectivist approaches received a predictable response from scientists. John Herschel (1792-1881) reacted by arguing that there are verae causae or true causes existing in nature for which we have both direct experience and inductive evidence – they are not ‘mere hypotheses or figments of the mind‘.

19th century

Henri Poincare (1854-1912) anticipated Chaos Theory by observing that minute differences in initial conditions can have a vast influence on the final phenomenon. Famous English philosopher Bertrand Russell (1872-1970) was a causal skeptic, an eliminativist who regarded causation as conceptually irrelevant. He claimed that the laws of physics were symmetrical, that physics rarely makes reference to causation, and that physical laws are mathematical not causal and do not imply agency. ‘… there is no one relationship that encompasses everything we call causation‘. However, in 1948 he tried to describe what he called process causation as causal lines ‘When two events belong to one causal line the earlier may be said to ’cause’ the latter. In this way laws of the form ‘A causes B’ may preserve a validity‘.

20th century

Through the late 1960s and early 1970s the study of causation was revitalized by several new approaches. American philosopher David Lewis consolidated the study of counterfactuals (‘If … then..’) opening up the logic of possible worlds, defining causation as ‘the events leading up to counterfactual dependence‘.

Statistician Paul Suppes launched Probability Theory by studying the predictability of events in relation to what was called ‘association’ and ‘independence’ as an alternative to regularity theories.

There was also an attempt by American Wesley Salmon to bring causation back into the scientific fold by examining various Causal Process Theories that ignored logical, probabilistic, and counterfactual accounts of causation by treating causation as a physical process involving the exchange of some quality or ‘conserved quantity’ like charge, energy, force, mass, or momentum. Sydney philosopher John Mackie extended Hume’s regularity theory by including multiple causes such that C is a cause of E when C & E are actual, C occurs before E and C is INUS (insufficient but necessary parts of an unnecessary but sufficient set of conditions) thereby drawing attention to the caiusal complexity of background conditions (1965).

By the 1980s it had become clear that the attempt to provide a universally acceptable reductive account of causation had run aground, that causation is either primitive or conceptually too opaque.

The period 1984-2004 was marked by a period of epistemology and axiomatics stimulated by statisticians and computer scientists. Statistics had now mathematicised the relative strength of causal inference by Path Analysis. In the late 1980s American statistician Judea Pearl (2000) and followers developed a theory of reasoning with uncertainty using Bayesian networks. Their approach was non-reductionist and non-metaphysical continuing Suppes work on the statistics of ‘independence’ and ‘association’. Pearl, who was concerned with causation in artificial intelligence, emphasized manipulation and interventionist theories whereby ‘Y is a cause of Z if we can change Z by manipulating Y‘.

James Woodward devised a counterfactual theory of causation (2003) like Pearl’s (although Woodward was assessing the meaning of causal claims rather than using models to determine causal relations) based on manipulation or intervention. Hume had reduced causation to regularity that was a consequence of general laws (it was a generalist theory) but philosopher David Armstrong (2004) offered a singularist theory by insisting that it is perfectly legitimate to determine the cause in each particular event, that we do not need a generalist or regularity account. For Armstrong causation was a relationship between singular events – a primitive theory defined by the folk psychology relating to the platitudes espoused by theories of regularity, agency (manipulation theory), counterfactual dependence, and probability … it is a product of the manifest image, we know it when we see it.

Nancy Cartwright in a review of causal laws (2007) concluded that since there are so many different kinds of cause we should avoid the word ’cause’ choosing verbs more appropriate to the particular circumstances.

Phillip Wolff has studied causation from the perspective of experimental psychology (2007) arguing that people infer a particular physical dynamic when they observe interactions interpreting dynamic processes as causation. This mode of perception is also, by analogy, applied to psychological and social interactions. He referred to counterfactual and probabilistic theories as ‘dependency models‘ himself espousing a process model of causation.

Cognitive psychologist Steven Pinker(2008) adopts the view that our cognitive faculties engage the world directly but our cognition is filtered through innate Kantian categories like space, time, substance, and causation that can be reinterpreted using the physical and conceptual tools of science.


Aristotle – cause and explanation are the same. The cause is the aitiai, ‘… something without which the thing could not be‘. He postulated four basic kinds of aitiai: material, efficient, formal, and final
Unknown Stoic author – ‘From everything that happens something else follows depending on it by necessity as cause
Descartes – ‘Particular causes are not the motions of the individual parts of matter, but the general principles of laws of nature‘ (1644) Causes thus cease to be particulars (tokens) and become generalities (types). Not the initiators of change but inactive instruments of God
Hobbes – cause is ‘The aggregate of accidents in the agent(s), requisite for the production of the effect’ and effect ‘ That accident, which is generated in the patient’ (1655) Causation was a relation between moving bodies occurring by contact (contiguity). There is no causation at a distance
Spinoza – ‘From a given determinate cause an effect necessarily follows (1677) …’ Free causes act from the necessity of their own nature and are therefore initiators of change’ while necessary causes are necessitated by other causes and are therefore inactive nodes in a chain
Liebniz – ‘There is nothing without a reason, or no effect without a cause‘ (1680-1684). He established both the logical ground and real cause with the principle of sufficient reason and rejected the reduction of metaphysical change to locomotion noting that in each act of causation there is an efficient and final component. The relatio between cause and effect is as strong as logical necessity. A complete knowledge of causes would yield premises from which reason could conclude the effects
Newton – Causes are ‘forces or constraints that compel moving bodies to behave differently than they would have done without them‘ (1687)
John Locke – That which produces any simple or complex idea we denote by the general name ’cause’, and that which is produced, ‘effect’ (1690)
Hume – The causal relation exhibits three factors: contiguity in space and time, priority in time of cause to effect, a necessary connection between cause and effect (1739). The latter he considered the most important because it distinguished causal from non-causal relations. However, he could not demonstrate either logical or empirical necessity making causation thus a habit of mind resulting from the regularity of constant conjunction of events
Kant – Causation is an a priori category of our understanding (1781). He agreed with Hume that cause is ‘not present in appearances‘ and must therefore be a priori or an empty concept, a ‘phantom of the brain‘. Experience provides rule-like regularities explored by induction but these have only comparative universality, not the absolute universality of necessity. However, our concept of cause and effect demands an absolutely universal rule ‘the concept of cause implies a rule according to which one state follows another necessarily‘, (1783), this conceptual rule is not empirically based in induction. Hume believed we first perceive temporal succession in events naming them cause and effect: Kant believed we need cause and effect relationships to establish an objective order in time. Humes conclusion was that every event has a cause that is the result of a pror event, the effect following from the cause necessarily and in accordance with an absolute universal rule that arises a priori, not from experience
Mill – cause is ‘The antecedent, or the occurrence of antecedents, on which [a given phenomenon] is invariably and unconditionally consequent’ (1874) and ‘The cause (philosophically speaking) is the sum total of all the conditions, positive and negative taken together, the whole of the contingencies of every description, which being realized, the consequent invariably follows‘ (1874)
Russell – ‘When two events belong to one causal line the earlier may be said to ’cause’ the latter. In this way laws of the form ‘A causes B’ may preserve a validity‘ (1948)
MackieC is a cause of E when C and E are actual, C occurs before E and C is INUS (insufficient but necessary parts of an unnecessary but sufficient set of conditions) (1965)
Lewis – Causation is ‘… the events leading up to counterfactual dependence’ (1973)
Dupre & Cartwright‘In virtue of their properties, things and events have the power to bring about other events or states’ (1988)
Humphreys – Causation is primarily a relation between properties (2004)
Cartwright – The term ’cause’ is highly unspecific. It commits us to nothing about the kind of causality involved nor about how the causes operate. Recognizing this should make us more cautious about investing in the quest for universal methods for causal inference. (1999)
Pearl – ‘Y is a cause of Z if we can change Z by manipulating Y‘ (2009)
Searle‘A cause is something that makes something else happen’ (2017)

Print Friendly, PDF & Email