# From Behavior Graphs to Mental Models

Take a look at the graph below – why do you think someone would imagine the green line estimate?

World population estimates from 1800 to 2100, based on “high”, “medium” and “low” United Nations projections in 2010 (red, orange, green) and US Census Bureau historical estimates (black). Actual recorded population figures are in blue. According to the highest estimate, the world population may rise to 16 billion by 2100; according to the lowest estimate, it may decline to 6 billion.

This graph is called a Behavior Over Time graph and it shows exactly what its name suggests. Time is plotted on the x-axis and some metric (in this case, population) on the y-axis – hence as we read the graph from left to right, we are observing the “behavior” of something “over time”. The blue line explains actual population growth between 1950 to 2010. We might call this blue line the “observable behavior” because we can validate the truthfulness of it using actual data collected. Systems thinking is the study of the unobservable complex relationships that create the observable surface behavior. In studying systems we ask:

Step 1 – “What is happening?” – We find Behavior Over Time graphs like the one above as a starting point to understanding the state of the problem – it’s important that this data is collected over a period of time rather than a single point number of today’s population (~7.15 billion as of July 2013) so that we can identify trends – in this case, we see that population began growing at a steeper rate in the 1950s onward. The time frame you choose as your reference point matters – if we were to go back to 10,000 BC, we’d see a much more alarming exponential growth rate.

But we have to put these numbers into a holistic story. Unfortunately, people tend to define a problem as an existing condition that must be fixed – for example, people might view the causes of population growth to be high fertility, poor education, and mass agricultural production.  But systems thinkers aren’t just interested in these isolated variables. We look at longer timeframes of these behaviors (and more) to look for patterns of movement. The problem you want to understand is not what causes population to grow today, but rather, what are the system’s patterns of interaction that cause populations to grow and fall over time.  What are the system’s tendencies? Find where the system is resistant to policy fixes – these illustrate the system creating its own behavior. And that’s what you want to understand: the system’s fabric of relationships – the story of the system, not the current news bite.

You might start doing this by first disaggregating population into different societies. What are some trends that are shared in societies where populations have collapsed drastically? How about where population has grown exponentially in a short period of time? Does fertility often move in the opposite direction of education? Does the presence of religion affect family planning?

Step 2 – “Why did this happen?” – We construct Causal Diagrams, or mental models, to make sense of the story. Seeing the problem as a malfunctioning system as opposed to a laundry list of disparate problems happening at the same time allows you, the problem solver, to identify higher leverage solutions that get at deeper root causes.

Below is an example of Laundry List thinking:

Ignoring the fabric that holds all these problems together will lead to surface level policy decisions that are merely patching up one wound while the others are growing.

Systems thinkers don’t just stop at finding convenient correlations – we challenge ourselves to create causal explanations. We use historical data we can find, combine that with well-designed research of causal inferences, and sprinkle some of our own intuition about how the world works. All this creates a unique mental model, which is essentially a causal diagram that maps out all the interactions between various parts of the population system. The causal diagram below may be one person’s simplified Population mental model. Start with Population in the center, and trace the around the various loops all the way back to Population to see how they impact Population (“+”signs indicate that the variables move in the same direction). Can you think of how your mental model looks different?

Above is a simplified Causal Diagram highlighting some key Reinforcing and Balancing loops that impact population growth. Can you think of any missing feedback loops? To learn how to read causal diagrams, go here.

But because all models are limited simplifications of the real thing, no model is ever complete. So we can only call our models “hypotheses”. And you might as well call these  “hypothesis in progress” because models change frequently as you collect new information and  interact with different ideologies. You start adding new variables or new links after reading about a key incentive you hadn’t been exposed to before, and delete those variables that once seemed important but now clutter your model.

Over time, you may read more about global warming and modify your opinion about the severity of environmental degradation in your Population model; or you may live in the Silicon Valley and witness firsthand how people have the will and talent to create new technologies that significantly reduce our toll on the earth’s natural resources (I’m not referring to mobile app developers, although even they can create products that help reduce information flows that engage civic awareness of pressing issues), and so you may add “new technological advancements” as a key variable influencing the mental model above.

Step 3 “How will things unfold?” –  Finally, we imagine scenarios for the future. Since no two people have the same mental model and the wisest thinkers build in uncertainty into their own well-researched models, it’s no surprise we end up with three very different predictions from the United Nations on how population will grow in the future represented by the red, yellow, and green lines of the Population Behavior Over Time graph up top.

Having an elegant and comprehensive model is no guarantee that you can accurately predict the future, but it does mean that you have an understanding of how, why, and when the system will behave in vastly different ways.  It is what enables you to tell a cohesive story; here is my hypothesis of population growth:

Population will continue growing, and have lower growth in developed countries and higher growth in less developed countries.  But if we increase the education level and improve child mortality outcomes in developing countries, we will see a decline in fertility rates, which moderates population growth. But even if the annual population growth rate remained at 1% per year, we would still get exponential growth of the entire population base  (similar to how interest on unpaid credit card debt compounds exponentially). In other words, as long as the “Births per year” Reinforcing Loop is more dominant than the “Deaths per year” Balancing Loop, the net effect is that population will grow. And as the earth supports an increasing number of people and global standard of living increases as the developing world catches up to consumption patterns of the developed world, human activities will have a greater toll on natural resources and the environment. We are encountering problems such as deforestation, overfishing, and climate change (see the Balancing Loops, “population reaches carrying capacity” and “climate change risks”). The Tragedy of the Commons archetype will play out across masses.

If things get significantly (the balancing loops in the causal diagram activate and the public becomes aware of it), governments will create policies that aim to limit our damage on natural ecosystems and the environment. People will start paying attention as their collective consciousness will demonize activities such as pollution, wasting of food, and industrial farming believed to be unethical in a world at the verge of its limits. But things may never reach the limit because innovative technologies will emerge that find more efficient solutions, currently unimaginable, that reduce carbon emissions and make humans less dependent on natural resources. The development of artificial intelligence may alter human consumption patterns and create better information flows that significantly improve our abilities to solve complex problems. (Quiz yourself: The loops mentioned in this paragraph have not been drawn into the causal diagram above; Test yourself – where would you add these loops? Are they balancing or reinforcing?)

Whether the population can continue growing, stabilize to a carrying capacity, or collapse depends on which of the causal loops above win the race. In the simple causal diagram, there is only one Reinforcing Loop and yet it has been the most dominant loop since 1800 (evidenced by the increasing blue line in the Population Behavior Over Time graph). For at least the last 200 years, humankind has enjoyed living in a system that hasn’t activated the force of the Balancing Loops, but if those Balancing Loops become more dominant than the Reinforcing Loops, deaths/year will be greater than births/year. And that is the world imagined by the U.N. green line estimate.

The green line might look like this.
Source: Greenpacks

The red/orange line might look like this – a benign, green society where every one thrives alongside nature.
Image from Minority Report.

#### Prediction vs. Understanding

There are many “Ifs” in my Population hypotheses above, and rightfully so because no one really knows with certainty what will happen. How does one predict whether a new unforeseeable technology will make the whole problem moot? What we do achieve in constructing our mental models is we identify a series of relationships and scenarios that are possible. We see how and why things might happen, and what would follow if they did. In the more technical field of System Dynamics, modelers plug numbers into their causal models and this allows them to simulate the the likelihood that various outcome scenarios occur.

Contrast this with what we typically get – estimates that world population will be 16 billion by 2100, which gives us a single point result without any visibility into the cause-effect dynamics that led to such a conclusion. In order to trust experts’ predictions, we must know their assumptions and biases – their mental models – in order to understand why and how they got their golden numbers.

Unfortunately, the news media is fascinated by predictions without explanations, and so we watch closely as Doomsdayers proclaim an end to civilization by a certain date. When that date passes and humankind still exists, these people lose all credibility – they are perceived to be wrong.  But a wrong prediction about a significant event is different from a wrong understanding about the system’s underlying dynamics. The Doomsdayers’ mental models may contain important truths. So rather than base our beliefs of the future on a number that the expert predicted, we should aim to develop our own mental models. In what areas are most people’s mental models similar, and where do they diverge? What researchers are looking for explanations to the shaky areas of your mental model? Given that there are multiple possible outcomes for humanity’s future on Earth, how do we then design public policy to accommodate a richly complex world?

Essentially, a lot happens at the point where the blue line breaks into the 3 other colors. The blue line itself highlights a problem – a trend – but you won’t understand this problem until you organize multiple patterns, identify the tendencies of how behaviors move together, and create a tightly woven fabric of the underlying dynamics that created the historical behavior. Your ability to notice the dominant relationships and ignore the noise determines the accuracy of your understanding of the world.  Only after going through this thoughtful process will you be equipped to understand the system and embrace the many uncertain stories for the future.

++