Thursday 23 February 2012

Patterns from Nothing

Continuing on from my previous post on Cognitive Illusions I thought I would start with the ability we have as humans to be fooled into seeing patterns from nothing. It is common for people to find shapes or objects when starting at the clouds or to think that there is a pattern of luck associated with the game we are playing. We can look at a random set of data and without doubt we will naturally make a pattern. Ben Goldache talks about this in his book Bad Science[1] It is in our nature and we are over sensitive to making patterns when none exist. Look at the following example of tosses of a coin, H-Heads, T-Tails.

HHHHHHHHHHHTHHHHHHHHHHHH

Now what conclusion would you make from this set of results?

Have you come up with any?

If you have come up with a conclusion that is your natural instinct and intuition to create a pattern and a cognitive illusion. Given that the coin is true the chances that the sequence above would happen is the same as any other sequence. Take this one step further and I ask you to say what the result will be on the next coin toss.

What would you answer?

Why would this be your answer?

Using statistics the possibility of it being H or T is 50/50 or equal chance. This is the reason casinos make so much money they know we are all fallible and use that against us. We make the mistake that there is a pattern and that our luck must change. I am sad to inform you but there is no luck the chances are still the same and within a casino the odds will always be against you.

The same can be applied to those who follow sport and come across the phrase that someone is on a lucky streak; this again is our natural bias to create a pattern when none exist. For example a soccer player has the follow goal scoring record. (X means scored in the game – O – means did not score.

XXXXXOOXXXXXXXXXXOOOOOOOO

Our tendency to create a pattern means that we will take that data and say the player has had 2 lucky streaks of scoring and is currently having a dip in form. With such simple data it is so easy to create and formulate assumptions and make patterns where there is no pattern. This is especially easy to do if there is no context. The simple example above proves the need to have some context. If I gave some more information that the player above has for the last ten games been playing in the senior side instead of the juniors, would that make a difference to your conclusion?

So how does this apply to testing?

There is a talk within testing that we should trust our intuition (I am one of these people to talk about this) and go with our gut feelings. Malcolm Gladwell in his book Blink [2] describes this to great effect. However we need to be aware that our intuition can try and fool us and try to create patterns when we are carrying out our testing. The problems come when we start to see these patterns and this causes us to miss other information that may be important.

For an example of this watch the following video (Information provided by Gordon Pitz [3] )



Have you watched the video?

No?

Please go and watch it, it will help you understand the rest of this article.







Did you see the Gorilla? [4]

No?

This might be due to being distracted and focused on a task. Noticing patterns and forming inconclusive assumptions when there are none can cause the same effects and as such it does show the point that our minds can be easily distracted and miss important information. It is important when we are testing that we do not spend too much of our time looking and investigating patterns since out natural instinct is to see patterns we could end up missing a lot more important information.

This is vital when we are testing using the exploratory testing approach where it is very easy to go off track and away from our mission to investigate what we think is a pattern of behaviour within the system under test. It is best in these situations to make a note of it and continue on track.

Sometimes it is difficult to go against what is natural and some find it near impossible and this could be one of the reasons why the exploratory testing approach may not be suitable for them or they find it too difficult. I hope that this article will encourage those who have struggled to have another go knowing that sometimes that could be fighting against their own instincts and as such making it appear more difficult for them.

So are there are techniques that can be used to help resolve this bias?

The problem is that since this is a natural built in instinct, and because we are aware of it, it does not necessarily mean we can resolve it.

“Knowing that it exists does not remove it”
Gordon Pitz [3]

There are few techniques that could help

One previous described when using session based testing is to keep to your mission and make a note of interesting patterns that you think are emerging. Later when you do a feedback session to others explain your thoughts about the pattern and see if others see the same pattern. If they do not it could be a case that you see a pattern when there is none
.
Another way which may help to prevent this bias is to use paired testing, there is gathering evidence that social facilitation [5] can help to reduce cognitive bias and paired testing is one way to make use of social facilitation. We seem to be more attentive and aware when we are being observed. It should be used with caution since if the task is complex and difficult people will perform badly. So this can only really be used when the task is not over complex.

One more technique that I have found invaluable is the use of testing framing as mentioned by Michael Bolton [6]. I attended a course on this and I do recommend that people read the article on his website. Using this approach helps the tester to focus on the purpose of the test but it also has a cool side effect that it can help to remove this bias to see patterns when there are none. It works especially well when you have to justify your reasoning.

The next article will look at the cognitive illusion of regression to the mean and its possible impact on testing.

References:

[3] The Deceptive Nature of Intuition – Gordon Pitz- http://www.unc.edu/~gpitz/pdf/Chabris-Simons%20review.pdf
[4] The invisible Gorilla - Christopher Chabris and Daniel Simons - http://www.amazon.com/Invisible-Gorilla-How-Intuitions-Deceive/dp/0307459667
[6] Test framing – Michael Bolton - http://www.developsense.com/blog/2011/05/ive-been-framed/








Wednesday 8 February 2012

Cognitive Illusions

or how your mind plays tricks on you.

People who regularly read my blog may be aware that I have a keen interest in psychology and how it can relate to testing. If you have not read my blog before wow welcome first timer I hope you enjoy and come back for more articles in the future.

I have in the past written a few articles about bias (here, here, here and here) and how it can be dangerous when we are testing. Having just read an excellent book called Bad Science by Ben Goldache I thought I would revisit this subject since Ben has a whole chapter on this very subject called

‘Why Clever People Believe Stupid Things’

It is a very interesting chapter and it made me re-think about the need to be careful when we are testing and reporting what we believe has happened. The human mind is a tricky beast and there are various methods it uses to try and trick us into believing things which are not true.

For example take a look at the following picture by French artist Felice Varini (the site is in French) This is a fantastic anamorphic illusion in which our mind joins all the pieces together to make us see something that in reality is not real.




Looking at it from a different perspective shows us this.




An important lesson in testing is not to look at things from only one point of view. See how our mind tricks us in to thinking something is real when it is not.

Ben Goldache manages to breakdown some of the common tricks our mind plays into the following:

# Randomness
# Regression to the Mean
# The bias towards positive evidence
# Biased by our prior beliefs
# Availability
# Social influences

Which he concludes with the following statements

1 - We see patterns where there is only random noise.
2 - We see causal relationships where there are non
3 - We overvalue confirmatory information for any given hypothesis.
4 - We seek out confirmatory information for any given hypothesis.
5 - Our assessment of the quality of new evidence is biased by our previous beliefs.
6 - Our assessment of the quality of new evidence is biased by our social influences.

(I added the 6th one myself)

Once we become aware of these illusions that our mind plays on us we can start to put practices in place they helps to try and remove them. I should warn you it is impossible to remove them entirely since we are only human after all, but being aware that they exist is a good start.

Over the next few blog articles I will be taking each one of these topics and applying it to testing