Recently I have been writing about my biggest take-away from The Book of Why by Judea Pearl. The book is more technical than I can fully understand and grasp since it is written for a primarily academic audience with some knowledge of the fields that Pearl dives into, but I felt that I still was able to gain some insights from the book. Particularly, Pearl’s idea that humans are better causal thinkers than we typically give ourselves credit for was a big lesson for me. In thinking back on the book, I have been trying to recognize our powerful causal intuitions and to understand the ways in which our causal thinking can be trusted. Still, for me it feels that it can be dangerous to indulge our natural causal thinking tendencies.
However, Pearl offers guidance on how and when we can trust our causal instincts. he writes, “causal assumptions cannot be invented at our whim; they are subject to the scrutiny of data and can be falsified.”
Our ability to imagine different future states and to understand causality at an instinctual level has allowed our human species to move form hunter-gatherer groups to massive cities connected by electricity and Wi-Fi. However, our collective minds have also drawn causal connections between unfortunate events and imagined demons. Dictators have used implausible causal connections to justify eugenics and genocide and still to this day society is hampered by conspiracy theories that posit improbable causal links between disparate events.
The important thing to note, as Pearl demonstrates, is that causal assumptions can be falsified and must be supported with data. Supernatural demons cannot be falsified and wild conspiracy theories often lack any supporting data or evidence. We can intuit causal relations, but we must be able to test them in situations that would falsify our assumptions if we are to truly believe them. Pearl doesn’t simply argue that we are good causal thinkers and that we should blindly trust the causal assumptions that come naturally to our mind. Instead, he suggests that we lean into our causal faculties and test causal relationships and assumptions that are falsifiable and can be either supported or disproven by data. Statistics still has a role in this world, but importantly we are not looking at the data without making causal assumptions. We are making predictions and determining whether the data falsifies those predictions.
Novel Learning 