1 Introduction: The Pervasiveness of Causal Talk

The words ‘cause’, ‘effect’ and their synonyms are used mainly for two purposes: to explain things and to help us decide what to do in order to achieve a certain desired state of affairs. Therefore, it is important to have a deeper understanding of their meaning. One may view the meaning of a term as the rules for its correct use. This is not to say that there is anything like a well defined meaning of the term ‘cause’ (and its converse ‘effect’) that is the same in all contexts; as with very many expressions, the meaning depends to some extent on context.

As observed in the previous chapter, relating events or states of affairs as cause and effect is a basic feature of our thinking, acting and use of language. This fact strongly suggests that it is not possible to define the term ‘cause’ (or its converse ‘effect’) using more primitive vocabulary; we learn to use the words ‘cause’ and ‘effect’ when we learn to talk, without being given any verbal explanations. We do not first learn a basic vocabulary that later is used in an explicit definition of ‘cause’. What we instead can do is to state a number of conditions for the correct use of ‘cause’, for example, that in a singular case a cause never happens later than its effect.

Causal claims are often made without any scientific backing. As an illustration, look at the following list of quotes, randomly selected, from New York Times 2020-08-17, where ‘cause’ and ‘effect’ are italicised by us, not by NYT:

‘But when the pandemic caused demand for bikes to jump, Giant needed to reverse course.’

‘Gov. Gavin Newsom of California called for an investigation into what he described as a major utility failure that was even more alarming set against the backdrop of the pandemic, when people, many largely confined inside, may be more dependent than ever on electricity: rolling blackouts over the weekend, caused by a record-shattering heat wave.’

‘Problems with coronavirus data reporting cause confusion in Iowa and beyond.’

‘Uncertain still was what effect, if any, the event would have on the spread of the virus.’

‘The Interstate Highway System was justified in part with dubious claims about national security, but it had the effect of reinforcing national unity.’

‘Dr. Kimberly Manning, an internal medicine doctor at Grady Memorial Hospital in Atlanta, recalled countless micro-aggressions in clinical settings. “People might not realise you’re offended, but it’s like death by a thousand paper cuts,” Dr. Manning said. “It can cause you to shrink.”’

These causal claims (understandably, several relating to the Covid-19 pandemic!) are not explicitly backed by any profound scientific analysis; the NYT journalists and the persons’ referred to take the readers to understand and accept the causal claims as correct against a background of common knowledge.

The scientific literature is no less filled with the words ‘cause’ and ‘effect’. In a paper discussing Bertrand Russell’s thesis that there are no causal laws (Russell, 1913), the authors started to look for the prevalence of the words ‘cause’ and ‘effect’ in a leading scientific journal:

A search for articles in which the word ‘cause’ appeared in the on-line archives of Science between October 1995 and June 2003 returned a list of results containing 8288 documents, averaging around 90 documents per month, in which the word ‘cause’ occurred. ‘Effect’ was more popular - 10456 documents for the same period, around 112 per month. (Ross and Spurrett, 2007, 60)

But we use not only the words ‘cause’ and ‘effect’ for talking about causation; expressions such as ‘bring about’, ‘lead to’, ‘make happen’, ‘produce’, ‘result in’, do’, etc., also express causal relations. And we have also the negative counterparts to these, such as ‘stop’, ‘prevent’ and ‘hinder’, expressions that suggest causal actions performed in order to bring an end to some undesirable state.

The same is true of ‘effect’, which in many contexts can be replaced by ‘result’, ‘outcome’, ‘final state’, ‘impact’, etc. So we submit that if the authors of the paper had included these other causal expressions indicated above, the figures would have been still higher. It is safe to say that causal thinking is deeply ingrained in our thinking, it belongs to our nature.

That causal thinking is a fundamental aspect of our being as humans was a point already made by Kant in his Critique of Pure Reason (Kant et al., 2003, (A80/B106)). He argued that our mind has certain structural features that determine the forms of our judgements. Thus, Kant identified 12 fundamental concepts, called categoriesFootnote 1 among which cause is one. These categories make up the conceptual basis for all judgements, according to Kant.

His arguments are convoluted and much debated, but one may arrive at a similar conclusion, at least in the case of causal judgements, without taking any stance about his transcendental arguments. One may, as was done above, observe ordinary language use, instead of speculating about our minds. Doing that, we realise how common and basic the use of causal expressions are in ordinary discourse.

2 Attempts to Define ‘Cause’

In science we are required to define central terms. Giving a verbal definition of a term is to give its meaning in terms of more common and well known expressions. This is hardly possible when it comes to such a basic term as ‘cause’.

At the online Merriam-Webster dictionarywe are given the following list of synonyms for ‘cause’ used as verb: ‘beget’, ‘breed’, ‘bring’, ‘bring about’, ‘bring on’, ‘catalyse’, ‘create’, ‘do’, ‘draw on’, ‘effect’, ‘effectuate’, ‘engender’, ‘generate’, ‘induce’, ‘invoke’, ‘make’, ‘occasion’, ‘produce’, ‘prompt’, ‘result (in)’, ‘spawn’, ‘translate (into)’, ‘work’ and ‘yield’.

If we attempt to define ‘cause’ in terms of e.g., ‘bring about’, we might further ask about the meaning of ‘bring about’. Then a natural response would be to say that it means ‘cause’. In common language, none of these terms is more basic and informative than the others.

In Nancy Cartwright’s How the Laws of Physics Lie we read:

Causes make their effects happen. We begin with a phenomenon which, relative to our other general beliefs, we think would not occur unless something peculiar brought it about. (Cartwright, 1983, p. 76)

Here Cartwright gives the meaning of ‘cause’ in two ways: first as synonymous to ‘make happen’, then using a counterfactual, a sentence of the form ‘If A had not occurred, B would not have happened.’ The first would suffice if the defined term is less well understood than the defining term, but that is hardly the case with ‘cause’ and ‘make happen’, as earlier pointed out. Why not reverse and explain the meaning of ’make happen’ in terms of causation? The term ‘cause’ is one of the most common and basic words in natural language. So maybe that is why she immediately moves to a counterfactual explanation.

Cartwright is not alone in explaining the meaning of the term ‘cause’ by a counterfactual expression, it is quite common in philosophy. But how do we know that a counterfactual statement is true? Obviously it can never be justified by an observation. This will be further discussed in Chap. 4.

The notion that we somehow could define ‘cause’ and its converse ‘effect’ using a more basic and clearer vocabulary seems indeed dubious. The meanings of ‘cause’ and ‘effect’ and their synonyms are determined by the rules we automatically and without any conscious justification apply when we use these terms in concrete communication situations. We talk about causes and effects mainly for two purposes; to explain things and to decide what to do in order to achieve our goals; successful doing is causing a desired event to happen. Causal thinking and causal talk is a basic trait of humans, which, by the way, also was Kant’s conclusion.

We have in ordinary language no explicit criteria of application for the terms ‘cause’ and ‘effect’. But when they are used in a scientific context, one must state criteria in order to know what can be inferred from causal statements. Thus Cartwright writes:

Like Machamer et al. (2000) I too have long followed Anscombe’s view that the ordinary concept of ‘cause’ is highly general. It is what, following Otto Neurath, I call a ‘Ballung’ concept. A Ballung concept is a concept with rough, shifting, porous boundaries, a congestion of different ideas and implications that can in various combinations be brought into focus for different purposes and in different contexts. Many of our ordinary concepts of everyday life are just like this. Ballung concepts also can, and often do, play a central role in science and especially in social science. But they cannot do so in their original form. To function properly in a scientific context they need to be made more precise. This will be done in different ways in different scientific sub-disciplines, serving different ends and to fit with the different concepts, methods, assumptions, and standards operating in these disciplines. The more precise scientific concepts that result will in general then be very different from each other and different yet again from the original Ballung concept. (Cartwright, 2017, 136)

We basically agree with Cartwright that in scientific contexts we need to clearly state the conditions for the legitimate use of ‘cause’, ‘effect’ and their near synonyms.

3 Are Causal Connections Observable?

Given the omnipresence of the term ‘cause’ (and ‘effect’) in everyday life and in science, what, then, are the criteria for it and its synonyms? Can one directly observe a causal connection between two events?

Some philosophers (for example G.E.M. Anscombe (1971)) argue that we can, in certain cases, observe causation. We are not convinced. One may reasonably be doubtful on the ground that ‘cause’ and ‘effect’ are relational terms; the basic syntax is ‘x is the cause of y’ and ‘y is the effect of x’, where the placeholders ‘x’ and ‘y’ represent events, states of affairs, or aggregates of such things described by categories or quantitative variables. We observe events and states of affairs, but do we observe their causal connection? Do we really see or hear the cause propagating its impact to the effect? Can one really say that we directly observe any kind of connection of any two things? We think not. Perception is primarily perception of objects and, derivatively, of their changes.

Empiricists are prone to say the same, they restrict the use of the term ‘direct observation’ to things we discern with our sense organs.

But how to draw the line between a direct observation and an inference from such an observation? Consider the following situation: you enter a new hotel room and want to turn on the light. You push the first button you see and the bathroom, not the room, is lit. Then you push another button and the room is lit. Can one say that you observed that pressing the first button caused the bathroom to be lit, pressing the second one caused the room to be lit? Would a person who never before has been in a modern building with electric lights, for example, a person belonging to a hunter-gatherer culture in the Amazon rainforest, say that they saw the causing? We think not.

When people are prone to say that they observed these two events being causally connected, they rely on previous experience and tacit inferences made from such experiences, and some knowledge about electricity. In general, inferences to causal relations between events are based on experiences from experimentation; if we manipulate one object in certain ways and observe changes in some other object so that one can control the states of the second object by doing things with the first, we apply the cause-effect relation. And this is an inference, not a direct perception.

If we, on the contrary, describe this situation as a direct observation of a cause-effect relation, we have in fact made the well known fallacy post hoc, ergo propter hoc.Footnote 2 We distinguish between mere succession and causation and that cannot be done by mere observations; ultimately, we must perform experiments by manipulating one variable and observing the other. A statement about a causal relation is the result of an explicit or implicit inference from such experiments.

4 Hume’s Criteria for the Use of ‘Cause’

David Humediscussed the observational basis for talk about causation (Hume, 1986/1739). His proposal was that there are basically two directly observable features of a pair of events that trigger (i.e., cause!) us to say that they are related as cause and effect:

  1. 1.

    The cause precedes its effect.

  2. 2.

    Cause and effect are contiguous.

But this is obviously not sufficient. There are many cases of pairs of events/states of affairs being in physical contact and one of them preceding the other, without us saying that the first one is the cause of the other. Hume therefore added the regularity condition, popularly stated as ‘same cause, same effect’. Expressed more carefully: A is the cause of B if and only if A belongs to a type of events/states of affairs that regularly is followed by another type to which B belongs, and if A and B satisfy the other two conditions.

Thus Hume’s analysis of the use of ‘cause’ and ‘effect’ may be summarised as that each of the three conditions: (1) the cause precedes its effect, (2) cause and effect are contiguous and (3) the same type of cause is regularly followed by the same type of effect, are necessary and that they are jointly sufficient for the correct use of sentences of the form ‘x causes y’.

Each of the conditions has been doubted and Hume in fact discussed caveats to all three (Hume, 1986/1739). Regarding timing, he accepted that cause and effect sometimes could be simultaneous. Regarding contiguity he realised that there could be intermediate events/states of affairs so that cause and effect may be indirectly connected via a chain of intermediate events/states of affairs. The causal relationis transitive. Finally, about regularity, he accepted the possibility of several causes for a particular effect, in which case we must say that a particular cause is not always followed by its effect, for other causes may also be needed. So a particular cause only increases the probability for the effect to occur.

Many philosophers have been critical towards Hume’s regularity theory, the main argument being that it does not really explain what a cause is. Many people ask for explanations of regularities, usually in terms of causal powers. For an empiricist this is reversing the order of explanation; if there are any such things as causal powers, these must be explained in terms of observations, i.e., observed regularities, see Sect. 2.5.

But there is another problem with the regularity view: there are many regularities that we do not count as instances of causation. It is a well established piece of knowledge that correlation is not causation. How to distinguish cases of correlation that indicates causation from those that do not?

The first step is to use Hume’s condition 2; individual instances of causes and effects must be in contact, provided we can give a clear meaning to the notion of contact. The problem is that events, states of affairs, properties, or other kinds of entities being related as cause and effect hardly can be said to be in contact, since they are not bodies.

It is easy to grasp the underlying idea that there must be a process, some kind of physical, chemical or biological link between cause and effect, when these are two individual events in space and time. But how to apply that to, e.g., a causal relation between two attributes?

One hint may be found by reflecting on how we distinguish between correlations and causal relations. An observed correlation between two variables, i.e., events/states of affairs of types A and B respectively, is a causal relation only if there is a mechanism, a chain of events/states of affairs connected by physical signals (remember: ‘physical’ here includes chemical and biological events) transferring information from any particular A-event to a particular B-event. And how do we know that there was a transfer from an A-event to a B-event and not the opposite? There are two factors determining this: timing and deliberate manipulation of the A-event.

The contact requirement is a necessary condition. This is illustrated by the fact that we do not believe in extra-sensory perception (ESP). Some people claim that they can acquire information about things from which no physical, chemical or biological signals could have reached their mind, but no evidence has been produced. Several well-conducted experiments performed with persons claiming to have extra-sensory capacities have been made and they have all failed; there is simply no empirical evidence for ESP. Perception requires physical signals triggering our sense organs, so if there is no physical signal from an individual A-event to an individual B-event, the A-event cannot be a cause of the B-event.Footnote 3

These reflections indicate, again, that experiments where one variable is manipulated and another is observed are crucial for establishing a causal relation. If we observe a variation in an observed variable following variations in a variable being manipulated, we infer that there is a causal link connecting the two variables. And we take it for granted that there is a physical, chemical or biological mechanism making up the connection between pairs of singular events.

So causal relations between variables, quantities and other abstract things are grounded on causal relations between those individual events making up these abstract entities; physical signals transferring information between individual events or states of affairs make perfect sense and has been discussed by several philosophers, e.g. Reichenbach and Reichenbach (1999), Salmon (1984), Salmon (1997), Salmon (2001), and Collier (1999).

This is not to say that physical links between cause and effect always is a salient aspect of a causal explanation. Talking about causes of historical events, wars for example, the physical connections between power centres, (letters or telegrams sent between presidents and prime ministers before a war) are rarely of any relevance for our questions about causes in history. But there must be such links.

The fundamental method to obtain information about physical links is to perform experiments, to be further discussed in Chap. 7. But it all depends on our ability to keep factors other than the hypothesised cause under control. How do we ascertain that, when we do not know which other factors there might be? That is often our problem when trying to decide which causal connections there are between parts of a complex social, ecological or social-ecological system.

Another way to decide whether a correlation between A and B is due to a causal link or not is to use earlier established scientific theory to describe causal mechanisms, to be discussed in Chap. 8. But this earlier established theory must in turn be based on experimental evidencefor causal links.

5 Summary

Our fundamental causal terms, (‘cause’, ‘effect’, ‘make happen’, ‘bring about’, etc.) belong to our basic vocabulary, learnt as part of learning ones mother tongue. This means that one cannot define ‘cause’, ‘effect’, etc., in some more basic vocabulary. The meaning of such words are learnt by learning their application to a number of concrete situations as experienced by the child.

Cause and effect are relational terms, they are predicates of the form ‘.... is the cause of....’ and ‘.... is the effect of.....’. We cannot directly observe causal relations; we observe physical objects and events, and under certain conditions two such observed events satisfy the predicate ‘...is the cause of......’

The conditions for saying that an event is a cause of another event was first formulated by Hume. They are (1) cause precedes its effect (or is practically simultaneous), (2) cause and effect are in contact (directly of indirectly) and (3) same types of causes are regularly followed by same type of effects. All three conditions have been extensively discussed by philosophers. We hold that Hume was almost correct. The only improvement needed is that the regularity condition should be restricted, namely, that the correlation (‘regularity’) is observed in experiments where the cause is manipulated and the effect varies accordingly. Hume’s three conditions might have been a correct description of common use of the term ‘cause’ in his days, but we have later learnt to be more restrictive.

Discussion Questions

  1. 1.

    Do you accept our conclusion that in ordinary language there is a rather big set of expressions each being roughly synonymous with ‘cause’ and such that none is more basic or more easily understood than the other in this set? If not, what is your idea about the most fundamental term? Why this choice?

  2. 2.

    One suggested definition of ‘cause’ goes: ‘A cause is that what makes its effect happen’. Do you think this is a satisfactory definition or not?

  3. 3.

    Do you agree with Hume (and us!) that it is impossible that the effect precedes its cause? Why?

  4. 4.

    In many cases we do not know the physical, chemical or biological mechanism connecting a cause and its effect, nevertheless we are convinced that the causal link exists. Give an example of such a situation and explain the reason why a causal connection nevertheless is believed to exist.

  5. 5.

    How do social mechanisms differ from physical, chemical or biological mechanisms?