System 1 and System 2 Redux

Having looked at the basics of System 1 and 2 last time, it's time to dig into some of the nuances. Both Kahneman himself and the Reasoning book point to some things to consider, from the point of view of constructing a more expansive view of thinking. What I mean by that last bit is that a lot of analysis focuses on short episodes of thinking, simple thoughts or decisions. (When I say simple I do not mean they may not be difficult or important, rather that they turn on a small of information and often are just binary choices.) The expansive view is about the thinking process involved in doing complicated things, with more than one step.

The first point that Kahneman is at pains to emphasise is that "System 1" and "System 2" are largely metaphorical - there are no structures in the brain, no rush of hormones or firing of neurons that is exclusive to either "System." Rather they are ways the brain mobilises itself to deal with stimuli (including internal stimuli) of various kinds. This is important because much of the time both kinds of activity are taking place. In particular System 1 activity is continuous and automatic. We can use System 2 to veto or modify the impulses that come from System 1, but we can never become Mr Spock, that character from Star Trek who operates only on logic.

The next thing to consider was highlighted in my piece about the book "Reasoning":

"Siegel reminds us that for almost any complex decision, both "Systems" are used. Life is such that even the most rigorous of us would simply be paralysed if we had to bring every aspect of a complex thought or decision to conscious, formal processing. "

Kahneman agrees that "System 1 has learned associations between ideas (the capital of France?)" and that memory is a function of System 1. However he cautions that memory can be flawed, and suggests System 1 may bring forward the first memory that seems to fit. It takes System 2 to conduct (an often inconvenient) thorough search of memory to see what applies to the current situation. Acknowledging that warning, it is still clear that the way we address complex questions is at least in part by relying on System 1 to bring to mind relevant information we have encountered previously. A theory of expansive thinking has to include a concatenation of episodes, some where the mind behaves as System 1 and some as System 2.

Kahneman writes that System 1 is always on and always working, in an automatic way. It is a "Mental Shotgun" that works on whatever it notices. Out of this comes an impulse towards action or decision. Our System 2 self can intervene but that takes concentration. It is more efficient to go along with our first reaction and so we often do. One of the challenges in making a process for expansive thinking is how and when to put the energy in to a System 2 review of the options created by System 1.

One use of System 2 is that it can follow instructions or task sets. This is something that System 1 cannot do, so in thinking about designing a process for thinking we are creating something that will involve System 2 and the extra brain effort that implies.

Something that surprised me in the book was that there is evidence that we are born with certain models of causality for the physical world. It seems that System 1 does much less learning by induction (ie generalising from observation) than we may have thought. Add to this that we store out model of our world in System 1. We are a mix of causal intuitions we were born with and those stored from our System 2 experiences. This makes those experiences - and whether they are stored or not quite important. I think this points to the role of education in quite a deep way. Kahneman adds that System 2 is not naturally good at statistical (inference) type thinking either - so the quality of our experiences (and education) in this area matters a lot. This links to work from Keith Stanovich (University of Toronto) that there are in fact two parts to System 2 - one part is labeled "algorithmic" - and the talent for this goes along with high performance on classic intelligence tests. The other part Stanovich labels "rationality" and appears to be a mix of an education in methods of inference and the readiness to put the mind to work in that way. So far there remains much to explore about this - but it may suggest something about "how smart people make stupid mistakes" [Links from Edition #006 and Edition #007]

Finally, I want to draw attention to another issue Kahneman mentions, about strain and effort. It has been found that a good mood improves our intuitive (System 1) thinking. By contrast vigilance and suspicion improve the activation of System 2. This is an interesting supporting fact to the well-known in creativity circles concept that effective divergence (using intuition and association between ideas, typical System 1 attributes) should be kept separate from convergence (using System 2 analysis.) They are not just different modes of thinking, they are also different moods.

I hope to build on the elements here in future to outline how we might assemble different modes and ways of thinking to create better overall thinking processes.