Check out these propositions [inspired by The Knowledge Illusion]:
If my shirt is blue,
then my socks are guaranteed to be green.
My socks are green.
Therefore, my shirt is blue.
If I fall in a pigpen,
then I will need a shower.
I took a shower.
Therefore, I fell into a pigpen.
It’s likely that you were tricked in the first one but not in the second. This is because, instead of abstract facts, there are many causes for needing a shower. We can intuitively see that there are ways to evaluate the plausibility of the statement. Our logical reasoning includes what we know about the subject.
Cause and effect reasoning is the core of our cognition. Judea Pearl, winner of the Turing Award, describes causal reasoning as our native language. Thousands of years ago humans realized that certain things caused other things to happen. If you don’t brush your teeth, you’ll get cavities. If you tinker with a light switch, the light will turn on and off. No other species grasps this to the extent that humans do.
We have huge general knowledge of the mechanisms that change our world from one state to another; it’s the basis of our common sense. Our causal reasoning systems are applied when we want to understand change. We rely on detecting patterns that apply across many situations.
Reasoning works in two directions—we reason forward in time from cause to effect (prediction), and backward from effect to cause (diagnosis). Humans care deeply about the causal structure of the world because it enables us to generalize. Explanations and the stories that link cause and effect play a key role in allowing us to adapt flexibly to a changing world.
Humans may, in fact, be the only animals capable of diagnostic reasoning, although there is evidence that some species of crow can as well. The important thing about diagnostic reasoning is that we can reason about clues when they aren’t presented in a causal time series. For example, a doctor may use multiple symptoms to infer a disease, even though the causal arrow runs from diseases to symptoms.
Causal thinking is natural and humans have a tendency to find causes even where there are none. In fact, according to Jevin West, we have to work hard not to find causes. Our bias to think in cause-and-effect terms can make us see patterns in random noise—even when we find a plausible causal account we can remain vigilant to the alternative that there is none.
Great Human Strength: Cause and effect reasoning enables us to understand our world. From a very early age, we construct our lives around goal-driven causal interactions which enables us to flexibly achieve our objectives in an uncertain world.
Great Human Weakness: We make predictable errors when we confuse correlation with causation and when we reason forward from cause to effect because we fail to spot alternative causes. We can attribute causes to effects when there are none.
Machine Opportunity: Designs which engage our causal reasoning systems and help us spot errors.
Machine Threat: Designs which manipulate or trigger error-prone cognitive strategies or prompt us to confuse correlation and causation.
