How can we establish that causal relationships existed in the past?

Question

From Hume's problem of induction, it is intuitive to me that, for example, "taking aspirin in the past has relieved my headaches" is insufficient to say with certainty that "taking an aspirin in the future will relieve my headaches". This is because we are relying on the assumption that future events will be similar to past events. Furthermore, if we say that future events will be similar to past events because past future events were similar to past past events, then we are again relying on the assumption that the future will be like the past.

I am not making a criticism of this problem. I am making a criticism only of the first half. That is, how do I know that it was indeed the aspirin that relieved my headaches in the past? Can it be demonstrated that two past (successive) events had a causal relationship to each other?

So far, in my research, philosophers and scientists tend to look at causality in two ways:

1. Causality as opposed to correlation
2. Predictability of events due to assumed causal relationships

However, neither of these address the past, or the existence of causality itself.

From this article, it is stated that we can establish causality (X causes Y) in three steps:

1. Temporal sequencing (X must precede Y)
2. Non-spurious relationship (the relationship between X and Y cannot be by chance alone)
3. Eliminate alternate causes (there are no other variables contributing to the occurrence of Y)

Problematically, the article addresses point 2 only with hypothesis testing. Furthermore, the article tries to address present causal relationships, not past ones.

Is it merely taken as an axiom that causal relationships have existed? Are there any resources on the topic? Is there any example of two successive past events, A and B, which we can say with certainty that A caused B? If so, why?

Answer 1

We can establish a causal relationship either with data and statistics or with a model (which is itself born out by data and statistics).

Statistics: if we measure that 5000 out of 5000 people in a sample who weigh themselves, then drink a kilogram of water over 30 seconds, and then weigh themselves again immediately weigh 1kg more than they did before, while out of a control sample who did not drink any water over 30 seconds between two times weighing themselves, another 5000 out of 5000 measure no gain or loss of weight, we can make a trivial application of Bayes' law and conclude that the probability of gaining 1kg immediately upon drinking 1kg of water was close to 100%, hence drinking 1kg of water caused 1kg of weight gain in the drinking sample.

Model: we model conservation of mass, because out of countless experiments, all of them measured that when you weigh two things (in the same reference frame), mix them together without losing any of what comes out, and weigh the result, you get the sum of the two previous masses. Moreover, when we made extrapolations from the assumption that mass is conserved in order to predict other measurements, and then we measured them, they were as predicted. Thus conservation of mass is [part of] a very successful model, and we can confidently apply it to establish a causal relationship between an alien space monster who drank 1kg of alien space juice and gained 1kg of mass, despite having no prior data about alien space monsters or their beverages of choice.

Answer 2

Causality in a probabilistic universe does never occur with absolute accuracy. In complex systems like aspirin and headache we are talking about an immense network of causes and effects. All contributing causes are not known and all effects are inaccurate (=partially random) anyway.

Therefore we cannot say that taking aspirin causes the headache to go away. But due to theorizing and experimentation we have established probabilities and that is good enough for most practical purposes.

Answer 3

The existence of casualty is so much an everpresent observation in adult life that to questions or deny it in principle would be absurd. As children we learn what it means to be agents by acting and observing changes. Such as pushing a tower of wooden blocks and observing it tumble.

Agency would not be possible if effects could not be caused by action. Even in the mind, thinking about a problem means bringing about an insight or a plan as a causal effect of mind actions.

So while causality is challenging to describe formally in philosophy, the mere existence of causality, whatever it precisely is, cannot usefully be denied given our daily experience as beings in this universe.

Is there any example of two successive past events, A and B, which we can say with certainty that A caused B?

That depends on what level of "certainty" you require. Take the example of a toddler pushing over a tower of wooden blocks. Can we be certain that there is causality as opposed to invisible angels and demons in the rooms moving the wooden blocks in such a way to create the illusion of causality? Philosophically it is difficult to prove the nature of time and physics. But other than solipsism leading us nowhere there seems to be no alternative to go with a most reasonable paradigm.

Many other issues are like that. Take time itself. When you watch a recorded movie at home, you can stop the movie for a break, you can fast forward. The movie characters won't notice. Is the time in our universe like this, without us being able to notice? In the end, if we have no way of noticing or knowing, it's a waste of effort to think about it much.

Answer 4

Is it merely taken as an axiom that causal relationships have existed?

Axiomatic systems are the essential ingredients of scientific theories. Basically, axioms are the core of formal systems, and formal systems are the core of science (empirical knowledge).

But causal relations are not axiomatic.

Causality is a metaphysical fact, not a scientific (physical, empirical, axiomatic) fact. Causality is an a priori type of metaphysical knowledge. Causality cannot be tested or proven by science. In any case, science would just prove that causality is false (there are no 100% probabilities in the real world).

Now, metaphysical and empirical knowledge are essentially opposites.

When your mind (here, we are in the metaphysical domain) tells you that the future might ressemble the past, you tend to believe it (this becomes a metaphysical truth), because that's what your knowledge of physical experience shows (metaphysical a priori knowledge is influenced indirectly by knowledge of the real world). If you fall down and experience pain many times, you will avoid falling down. Why? Because you know that in general, the future ressembles the past, in spite of what Bertrand Russell says (his arguments again causality are strong).

Are there any resources on the topic?

Perhaps the initial paragraphs of Kant's Critique of Pure Reason, where the difference between a priori and a posteriori knowledge is presented. Take this into account: Kant knew that metaphysical knowledge had bad reputation, and he also knew that it was essential to experience the world. So, what he's trying to do with such book is to "raise metaphysics to the level of a science" (exactly what you are trying to do: take causality to the level of an axiom, which would be part of a scientific theory). Although that is strictly not possible, Kant did provide the basis for the formalization of metaphysical knowledge (although few are interested in following his task). I personally am working for long on an essay about formal metaphysics.

The Wikipedia article on formal (axiomatic) systems is a good introduction to the meaning and context of axioms. See also formal languages.

Is there any example of two successive past events, A and B, which we can say with certainty that A caused B? If so, why?

Excellent point, and here's the key:

"Certainty" implying TRUTH, consider there are two opposite but complementary types of truth:

• Scientific truth, that which is obtained by experience and the scientific method. Scientific truth is so only if it is objective (many can observe, i.e. measure the object, in the same way). Causality can be demonstrated to be FALSE by the scientific method (as said, because there's no 100% probability in the real world). So, depending on the formulation of a theory of causality, scientifically, either A -> B is FALSE (empirical TRUTH), or it cannot be shown to be TRUE OR FALSE due to the fallacy of ignorance (the lack of proof does not imply falsehood or truth).

• Metaphysical truth, that is created by the subject, so it is essentially a subjective truth. It is influenced indirectly by experience. This kind of truth is deeper, although subjective. Here, there are no "axioms", but a priori knowledge is essentially the same. This cannot be objective because metaphysical facts cannot be presented to other observers as an object. But in this realm, there are multiple truths that can't be doubted: smash your finger with a hammer (A) and you will feel pain (B). A -> B is TRUE (certain) in this realm: metaphysical TRUTH.