The Head and the Gut

Cognitive science researchers generally agree that we have two separate but interoperating cognitive systems. System 1 (commonly known as intuition) automatically recognizes situations when they occur and recalls relevant information from long term memory. System 2 (also known as reason or rationality) is the mechanism we associate with the conscious experience of “thinking.” System 1 runs automatically whenever we are awake, with System 2 normally idling in the background. System 2 can be consciously activated, but it is also triggered whenever System 1 detects an event that violates the model of the world it maintains, i.e. when it is surprised. System 2 then attempts to make sense of the anomaly, with further assistance from System 1. What we experience as “thinking” and “feeling” is the constantly shifting interplay of these two systems.

As we experience objects and events, System 1 generates and stores predictive models of regularities in our environment. At the same time, it’s also constantly attempting to match new stimuli to its preexisting models. This all happens automatically, effortlessly, nearly instantaneously, and beneath our conscious awareness. As a consequence of this activity, System 1 produces an unending stream of assessments, including evaluations of similarity and representativeness, determinations of causality, evaluations of the availability of associations and exemplars, among many others. The results of this processing often reaches our awareness as an emotion, or “gut feeling.” Our day to day judgment and decision making is largely driven by the emotional outputs of System 1.

The defining characteristic of System 2 is that it is a conscious, controlled process that uses working memory to operate on information retrieved by System 1. In contrast to System 1, System 2 is slow, serial, limited in capacity, and requires substantial effort to employ. It’s also easily disrupted. We really don’t like to use it, and will often go to great lengths to avoid doing so. Why, then, do we have it? Why did we evolve this capability and what good is it to us as a species?

Unlike System 1, which is very literal and context-dependent, System 2 gives us the unique ability to imagine. System 2 is able to retrieve dynamic mental models from long term memory and load them into working memory. It then performs mental simulations with them to generate alternative possibilities and explore their ramifications. Stanovich (2012) refers to this use of working memory as the Possible World Box (PWB). In this way, System 2 acts as a check on System 1’s blind, experience-based suggestions, which are frequently wrong and prone to a number of biases. System 2’s ability to engage in counterfactual thinking enables us to evaluate and correct our beliefs about the world. A more accurate understanding of the world translates to improved evolutionary fitness. Some of the key features of System 1 and System 2 are summarized in the table above.

So, that (very crudely) is what happens on the inside. How do these two systems interact with the external tools we’ve evolved (largely through trial and error) to assist our thinking? That’s a long and winding road with many side streets, but it’s possible to take a few tentative steps down it.

One of the most important functions these tools perform is to provide a kind of surrogate long term memory. I’ll have a lot more to say about external memory processes later, but one of the things sign systems (like data visualizations) make possible is the selective feeding of working memory under conscious control. In other words, external memory systems break System 2’s exclusive dependency on System 1 for historical information. You experience this benefit yourself every time you do long division with pencil and paper vs. doing it in your head.

Beyond their function as external information stores, it seems clear that visualization techniques present information in ways that stimulate and support the functioning of both systems. Exactly how is uncharted territory, but it’s possible to make some tentative speculations. One of the things System 1 excels at is detecting spatial patterns which are familiar and automatically associating meanings with them (Kahneman, 2011). It seems likely that one of the functions that visualizations perform is to rerepresent problems in ways that make them accessible to System 1 processing. In fact, this is exactly what we mean when we speak of the value of “intuitive” visualizations. System 1 also easily performs associative, metaphorical, and causal operations. It follows that developing a better understanding of how System 1 works may offer a principled approach for creating better visualization designs.

It’s important to note that because visualization tools likely serve, in part, to make information more accessible to System 1, there is a potential danger that judgments made on the basis of such representations will be prone to the same types of errors and biases as any other System 1-based judgement. Prudent designers will accommodate these tendencies, work to minimize their effect, and provide mechanisms that enable System 2-based review of insights gleaned from their tools.

[zotpress items=”54T5XQHM,WZFMEKCF,4RSHP669″ style=”association-for-computing-machinery”]