Correlation, Causation, and the Ancient Quest for Causal Inference

This discussion delves into the fundamental distinction between correlation and causation, highlighting how human intuition often conflates the two. Correlation is defined as two things varying together over time, while causation implies an underlying reason for this co-variation. The inherent human tendency to seek causal explanations, even when only observing correlations, is identified as a deep-seated cognitive bias, suggesting that our minds struggle to grasp any other logic.

The conversation further explores how conditional probability can manipulate observed correlations. By conditioning on a third variable or selectively observing incidents, one can artificially create or destroy correlations between otherwise unrelated events, such as two uncorrelated coin flips. This illustrates a critical flaw not in the world's inherent workings, but in the human attempt to impose causal logic on purely correlational data without proper experimental design or mathematical tools. The hosts acknowledge that this challenge is particularly acute in disciplines like psychology, where variables are complex and controlled experiments are often difficult or unethical.

The podcast uses the example of studying human behavior in semi-autonomous vehicles to illustrate the complexities of observational studies. Researchers might observe correlations, such as drivers falling asleep more often when autonomous features are engaged, but struggle to infer a direct causal link due to uncontrolled variables and ethical limitations on experimental manipulation. This leads to a discussion of the long-standing nature of the problem, referencing an ancient experiment by Daniel from Babylonian times, which serves as an early example of a controlled study designed to infer causality regarding dietary effects on mental ability.

Ultimately, the segment emphasizes that while the human quest to understand causes is ancient, dating back to philosophers like Democritus, the formal mathematics required to rigorously capture and differentiate between correlation and causation, particularly the asymmetry of causal relationships (X causes Y but Y does not cause X), was only developed in the 1920s. This suggests a significant historical gap where science lacked the necessary tools to address this fundamental problem, leaving many disciplines to grapple with the challenge of inferring causation from symmetrical algebraic or correlational data.

correlation kills when two things very together over very long time is one way of measuring it or when you have a bunch of variables that it was very quickly then recalled we have a correlation here

underlying it somewhere is causation yes hidden in our intuition D is a notion of causation because we cannot grasp any other logic except causation

The flows comes if we try to impose causal logic on correlation it doesn't work too well

if you condition on a third variable and you can destroy or create correlations among two other variables they know it

your leap who is trying to get causation from correlation not you're not proving causation but you're sort of discussing it and implying sort of hypothesizing without liability

what is the research question do people fall asleep when the car is driving itself do they fall asleep or do they tend to fall asleep more frequently more fickle and the car not drive lines not driving it's it's a good question

it was a first experiment yes so today there was a very simple it's also the same research questions we want to know a vegetarian food assist or obstructing your mental ability

Democritus said if I could discover one cause of things I would rather discuss the one cause and be king of Persia

the mathematics of doing this was only developed in the 1920s so science has left us often okay science is not provided that with the mathematics to capture the idea of X causes Y and y does not cause X