Judea Pearl’s book The Book of Why is all about causation. The reason human beings are able to produce vaccines, to send rockets into space, and maintain green gardens is because we understand causation. We have an ability to observe events in the world, to intervene, and to predict how our interventions produce specific outcomes. This allows us to develop tools to specifically achieve desired ends, and it is not a small feat.
In the book Pearl describes three levels of causation based on Alan Turing’s proposed system to classify cognitive systems in terms of the queries systems can answer. The three levels of causation are association, intervention, and counterfactuals. Pearl explains that many animals observe the world and detect patterns, but that fewer animals use tools to intervene in the world. Fewer still, Pearl explains, possess the ability to actually develop and improve new tools. As he writes, “tool users do not necessarily possess a theory of their tool that tells them why it works and what to do when it doesn’t. For that, you need to have achieved a level of understanding that permits imagining. It was primarily this third level that prepared us for further revolutions in agriculture and science and led to a sudden and drastic change in our species’ impact on the planet.”
The theory of tool use that Pearl mentions in the quote is our ability to see and understand causation. We can observe that rocks can be used to cut plant fibers, and then we can identify the qualities in some rocks that make them better at cutting fibers than others. But to get to the point where we are sharpening an edge of a rock to make it even better at cutting fibers, we have to have a causal understanding of what allows the rock to cut and we need sufficient imagination to predict what would happen if the rock had a sharper edge. We have to imagine an outcome in a future world where something was different, and that something different caused a new outcome.
This point is small, but is actually quite profound. Our minds are able to conceptualize causality and build hypothesis about the world that we can test. This can improve our tool usage, improve the ways we act and behave, and can allow us to achieve desired ends through study, prediction, imagination, and experimentation. The key, however, is that we have a theory of the tools and how they work, that we have an ability to intuit causation.
We hear all the time that correlation is not causation and in our modern technological age we are looking to statistics to help us solve massive problems. However, as Pearl’s quote shows, data, statistics, and information is useless unless we have a theory of the tools we can use based on the knowledge we gain from the data, statistics, and information. We have to embrace causation and our ability to imagine and predict causal structures if we want to do anything with the data.
This all reminds me of the saying, when the only tool you have is a hammer, everything begins to look like a nail. This represents an inability to understand causality, a lack of imagination and predictive prowess. Statistics without a theory of causality, without an ability to use our power to identify and predict causation, is like the hammer and nail saying. It is useless and throws the same toolkit and approach at every problem. Statistics alone doesn’t build knowledge – you also need a theory of causation.
Pearl’s message throughout the book is that statistics (tool use) and causation is linked, that we need a theory and understanding of causation if we are going to do anything with data, statistics, and information. For years we have relied on statistical relationships to help us understand the world, but we have failed to apply the same rigorous study to causation, and that will make it difficult for us to use our new statistical power to achieve the ends that big data and statistical processing promise.
Novel Learning 