First of all, I’d like to talk about illusions. We are all aware that our eyes can deceive us in numerous and varied ways, for example this colour illusion, this hollow face illusion and this movement illusion. Lots of research into vision tells us that we cannot trust what our eyes tell us. If it is important, we must check. Here’s another: the colour changing card trick (youtube).
It turns out that this is also true of hearing, and of taste and of touch. It seems that illusions of smell are harder to detect but probably do still exist.
Essentially, we really cannot truly trust any of our physical senses. Our brains have evolved to use shortcuts when interpreting signals from our senses and these shortcuts will often result in mistakes. If it is important that we get something right, we must distrust our own senses and test objectively.
But is this confined to our physical senses? What of our sense of logic? well it turns out that can also be fooled:
One example is caused by Regression To The Mean, which is where something that tends to vary within a range will occasionally have an extreme value but will then tend to return to its normal value. My favoured example is in childhood behaviour: most children will tend to behave with a certain level of discipline. Sometimes they are good and sometimes they are naughty. Occasionally they do something very disciplined or very antisocial but they will always tend to revert to their normal level afterwards.
Parents will punish bad behaviour and reward good behaviour, but how do parents perceive the effects of their punishments and rewards? Lets assume that the punishments and rewards actually have no effect. Regression To The Mean will still occur, so if a child is abnormally bad, they are likely to improve whether punished or not. And if they are abnormally good, they are likely to revert to normal even when rewarded. So the parent sees this:
- My child is exceptionally bad, I punish them and they return to normal. Punishment works.
- My child is unusually good, I praise them and they revert to normal. Praise does not work.
The parent comes to the conclusion that punishment works and praise does not. But in fact, neither action has any effect at all! So to find out what is really the case, we must disregard our own perceptions and do objective tests.
There are many more forms of cognitive errors we can make without realising.
All this winds up meaning that we cannot trust what we see and we cannot even trust what common sense or our own experience tells us! And ‘it works for me’ is not good enough. If we want to know whether something is true, we must cancel out all these potential errors…
In my next post, I will briefly summarise the ways these biases can be controlled for to give us a reliable picture of the world.
RationalPenguin 