Pattern Seeking Machines — I
On a rare evening walk around a lake, my friend asked if I believed in superstitions. She said that she believed that whoever accompanies her to a particular Hanuman Temple in Bangalore would go away from her life. I jokingly remarked that I wouldn’t want to accompany her to that temple! I shared a similar thought which was running in my mind — The last two vehicles my Dad owned had met with an accident. The connection was that both the vehicles were ‘near white’ in colour although an astrologer whom we had consulted had recommended white colour(We had to compromise on colour for some reasons). I was now being super cautious in choosing the colour of the car we were about to book.
I went on to explain the psychological basis of such interpretations and promised that I would write an article on the same. After a break of a few months, here I am, finally at it. (Suffice it to say that I was busy building cool stuff in the meantime.)
Oh man … and your eternal quest to make sense of everything …
We had just released an upgrade to our Android app. A couple of days later a few of our users complained that they were not receiving push notifications. A couple of them mentioned that it was after the upgrade that they had observed the problem. It started a frenzy in our management that the upgrade had broken our notification system, although
1. Only a couple of users, among tens of thousands, had reported the issue.
2. It was well known that the delivery rate of push notification systems is usually low.
3. The code pertaining to push notifications had not changed as a part of upgrade.
To the eyes of our data crunchers, even charts showed dip in notification volumes.
Later, the real cause of the problem was understood and it had nothing to do with the upgrade. But let us understand what actually went on in our minds.
We humans have a strong tendency to seek sense in the world — even when many outcomes are random and have no straightforward explanation. This quest for sense makes us see causations where it really does not exist. Users for whom notifications had stopped found a causal attribution in the upgrade, since it was the only perceptible change they knew. They were oblivious to far more frequent changes/upgrades to back-end systems which support the App.
With regard to my thoughts of vehicles’ colours and their connection to accidents, I was ignoring the fact that another ‘near-white’ car Dad earlier owned had never met with an accident. Our mind does not deal well with non-events. We give more importance to events that happened than to events that did not happen or could have happened.
What about our technical minds who also saw causality? Once an idea gets implanted in our minds, we unconsciously look for information that supports and confirms the idea — in pursuit of completion of the pattern and ignore the rest. Psychology calls it The Confirmation Bias. Once people started believing in a causal attribution to the upgrade, even a negligible downward slope in the chart seemed like a proof. All statistical lessons like need for statistical significance, correlation does not imply causation and need for non-interference from other factors to make causal decisions were ignored. (There were at least 5 other reasons that could have explained downward trend.)
We are pattern seeking machines in this noisy world — looking for patterns where they do not exist. We see the world as far more predictable and coherent than it is, giving very little importance to randomness and luck(it’s the work of System-1). It makes our world far more liveable. But it also makes us gullible and prone to wrong decisions. Suspecting that your partner is cheating on you? Any incidence will only aid in completing the pattern.
In the second part of the article, let us dig deep into what is it that makes us pattern seekers. It turns out that evolution designed Mankind to be exactly so and it also forms basis of our intelligence. Stay tuned ….
