What Causes What?
Good decisions require understanding of causality and inference.
Posted August 24, 2017
Do greenhouse gas emissions cause global warming? Does the Zika virus cause babies to be born with small brains (microcephaly)? Asking causal questions is important for understanding why things happen, and also for suggesting solutions to problems. Removing a cause can eliminate an undesired effect such as disease and suffering. Introducing a cause can help produce a desired effect such as health and flourishing.
Causal thinking is natural for people, even in infants, but people do not always do it well. We are prone to the fallacy of false cause, thinking that an event that follows another was caused by it. For example, if you take vitamin C and your cold goes away, you might think that the vitamin caused the cure, even though colds go away by themselves. You may also be motivated to think that you have found a cure for colds because it makes you worry less about getting sick again.
How can people think better about causality? More fundamentally, what is causality? Definitions such as “cause” means “make something happen” are not very helpful. Philosophers dispute the nature of causality, with claims ranging from noting that causes raise the probability of their effects to emphasizing that manipulating causes changes effects.
Causality is better characterized using the method of 3-analysis, which is based on a new theory of concepts that describes how they combine exemplars (typical examples), typical features, and explanations. There are many familiar exemplars of causes, such as pushes, pulls, motions, collisions, actions, and diseases whose effects are symptoms.
The typical features of causality include:
- Causes happen before effects.
- Causes operate in sensory-motor-sensory patterns, e.g. when you see and feel a bike not moving, move the pedals, then see and feel the bike move.
- Cause and effects sometimes yield regularities, for example that hitting your finger with a hammer always hurts.
- Causes explain statistical dependencies, with causes increasing the probabilities of effects.
- Manipulations and interventions lead from causes to effects.
None of these typical features is a necessary or sufficient condition of causality, but matching a lot of them suggests that cause/effect relations have been identified. Such relations provide explanations of why things happen and how they can be changed.
Before concluding that A causes B, you need to consider alternative explanations such as that B has a different cause, or that A and B are both caused by something else. We cannot directly observe causal relations, but need to infer that they exist as part of the best explanation of systematic observations.
For example, a recent U.S. report on climate change concludes “it is extremely likely that human influence has been the dominant cause of the observed warming since the mid-20th century. For the warming over the last century, there is no convincing alternative explanation supported by the extent of the observational evidence.” This report considers many kinds of evidence for the conclusion that human activity has been causing global warming, while criticizing alternative explanations such as solar activity and natural fluctuations. The appropriate conclusion is that people need to reduce the cause (greenhouse gases) in order to prevent the disastrous results of the effect (global warming).
Similarly, scientists have come to the conclusion that the recent disturbing increase in microcephaly in newborns is caused by their mothers becoming infected by the Zika virus through mosquito bites. This hypothesis is the best explanation of the statistical association between the introduction of the virus in various countries and the occurrence of microcephaly, taking into account growing information about how the virus affects brain cells. There is also a plausible connection with global warming, which allows the anopheles mosquito that transmits the virus to operate in new places, such as Florida.
On the other hand, the conclusion that childhood vaccinations cause autism, for example, was based on unsystematic observations that children are often diagnosed with autism after getting vaccinated, and on a study published in the British Medical Journal that was fraudulent. Many other studies have found no association between vaccines and autism. The causes of autism are diverse and largely unknown, but there are no grounds to infer that vaccines cause autism.
Judgments about cause and effect are valuable for understanding events and intervening to promote human welfare. But they need to be based on inference to the best explanation that takes into account the full range of evidence and alternative hypotheses.
Links and References