Leveraging Complexity Science to understand Human Behavior

Final Mile
6 min readNov 25, 2021


Part 3 | Illusion of Causality: Problem with a simplistic focus on Why

By: Alok Gangaramany and Anushka Ashok

Recently, a colleague suggested watching the famous video “How Wolves Change Rivers”. A fantastic tale about the reintroduction of wolves in Yellowstone National park that transformed the ecosystem of the park along with altering the flow of rivers. Cascade and non-linear effects are common phenomenon in nature and this story fits the narrative really well. However, as expressed in this article, it turns out that nature is more complex than we think. While there is general agreement that the absence of wolves had a devastating impact on the park, its reintroduction doesn’t necessarily reverse it.

In this post, we are less interested in the accuracy of the video but more about its vast popularity. The 4+ min video was able to establish a strong causal relationship between the intervention and its effect. It was also able to explain in a fairly linear fashion about the cascading effects of the change. And gave us great confidence in our ability to reverse the damage in a relatively quick and efficient way (video suggests that we only needed a small population of wolves to trigger this effect).

Source: https://earthjustice.org/features/infographic-wolves-keep-yellowstone-in-the-balance

This illusion of causality is a common thread that we see in our work in understanding human behavior. In the book Fooled by Randomness, Nassim Taleb shares number of examples where people frequently misattribute the outcome (profit from market swings) to a specific cause (e.g. smart decision-making of the trader / fund manager ) instead of realizing that it may be truly random. We are certainly biased towards such a causal approach of understanding the world. It gives us confidence in our decisions and provides an illusion of certainty and control. As a result, we look for tools and methods to support this understanding and build elaborate frameworks and models.

Behavioral research methodologies, especially ones using qualitative techniques, focus on developing a causal understanding of a particular phenomena. Most of us must have come across methods such as Root Cause Analysis , 5 Why’s etc that attempt to map the cause and effect relationships of the behavior in focus. These methods have been borrowed from engineering disciplines where they have been found useful to breakdown a problem, address its fundamental issues and put the system back online. The intuitive nature of these methods have resulted in their adoption in other areas including complex behavioral and social problems where they may be ill suited.

What might be the problem? Approach and tools like these are meant for isolated / closed systems, a system that has little exchange of matter or energy with its surroundings [1]. Consider an insulated cooler that tries to maintain the internal temperature of the water inside it. Any change in water temperature (say by adding ice to the cooler) can be confidently linked to our intervention.

Source: Wikipedia https://en.wikipedia.org/wiki/Isolated_system

Now, as a thought experiment, let’s say that our next project is operating in a relatively isolated or closed system i.e. imagine a completely disconnected community with no interaction with the rest of the world. And as part of our project, we are tasked to implement a cash transfer intervention and assess its impact, say change in transactional sexual behaviors. The assumption is that transactional sexual behaviors between men and women is caused by economic dependence and cash transfers will reduce this dependence. After a period of time, if we notice a reduction in transactional sexual behaviors then we may be able to confidently ascertain that the change was caused by our intervention. Methods such as RCTs (Randomized Control trials) could then be used to quantitatively verify the impact of the intervention compared to a control group.

But we don’t live in isolated systems, do we? So, let’s consider a more realistic well connected community. Here again we are tasked to implement a cash transfer intervention and evaluate its impact towards reducing transactional sexual behavior. While we are executing this intervention, let’s say a random event like lottery occurs which results in someone getting rich enough to start a new business that suddenly brings new set of employment opportunities. Now, this event sounds fairly extreme and unlikely (it probably is) but volatile events (like the recent pandemic) that suddenly change the economic situation of a community for good or bad happen more often than we think. To further add to the complexity, let’s consider that a well intended non-profit organization also decides to impart vocational skills training to both men and women.

How might this combination change the community ecosystem? We should expect significant change and hopefully positive for the most part (negative events like tensions between men and newly skilled / employed women can always lead to second order effects but let’s keep that aside for the moment). As we see, all these events (cash transfer, jobs and training) happened independently to each other, that is they were not coordinated. However, the impact to the community was definitely not independent. Moreover, the impact is emergent, unpredictable and unlikely to resemble the past (refer our earlier article on this topic). So in this community, are we in a position to attribute the change to cash transfer intervention or vocational training intervention or a random event like lottery? In addition, are we in a position to easily trace the effective pathway of change? What if the lottery happened prior to cash transfer? What if the vocational skills training happened first? What else might have happened in the community that we did not know even know of?

In the real world where we are unable to suitably control the presence and interaction of multiple variables, it will always be difficult to establish tight causal relationships. In addition, we are more likely than not to miss the impact of random events, particularly their interactions with other environment variables. Complexity science make us rethink the level of explainability we can bring to a complex phenomena which can be humbling.

This takes us back to the topic of this article “Problem with a simplistic focus on Why”. Causality is often seen from the lens of Why which has driven research efforts to isolate system components and evaluate their impact on the outcome. This as we understand works well for isolated system. But in an open system, like most real world social ecosystems, a much broader approach to understand the “How” might help in creating a much more holistic understanding of systems and their elements. So instead of trying to establish causal relationships, we may want to focus on directional effect of interventions, potential list of future states and critical thresholds to effect change. For example, instead of asking the question if Cash Transfers can reduce transactional sexual behaviors , we can focus on whether the presence of Cash Transfers is helping generate a new set of behaviors desired within and by the community members. If we see positive feedback, we can then ask what else might be helpful or what may still be missing? The approach may not provide the purity of the laboratory experiment but it will recognize and be more suited to the dynamic change environment that we are working with.

Going back to where we started. The introduction of predators like wolves in Yellowstone national park may have positively contributed towards the ecosystem. But to what extent? And what was the role of other environmental variables (e.g. growth in the population of beavers, clear thresholds for ecosystem recovery such as willow taller than 6 feet )? And what was the combination that generated this effect and in what sequence? We may not know the answers to these questions and if we work hard enough that we may end up creating misleading causal narratives that are precisely wrong. Instead, if we change our questions then we might be able to develop ideas that are useful, satisficing (i.e. less wrong) and avoid a false sense of certainty.


  1. Open , Closed & Isolated Systems: https://www.khanacademy.org/science/ap-biology/cellular-energetics/cellular-energy/a/the-laws-of-thermodynamics
  2. https://www.nsf.gov/discoveries/disc_summ.jsp?cntn_id=126853

Final Mile brings unique and proven capabilities in addressing complex behavioral challenges. As one of the first Behavioral Science & Design consultancies, Final Mile has had the opportunity to bring these to practice in a wide variety of sectors and contexts. We have executed highly complex behavior change projects across a wide variety of areas covering Global Health (HIV, TB, Maternal Health, WASH), Financial Inclusion, Safety across Africa, Asia, Europe, and the US.

Final Mile is also building a pandemic playbook that can be used as a potential toolkit by policymakers and implementors in mitigating Covid19 and future such pandemics.

Reach out to us at contact@thefinalmile.com.



Final Mile

Final Mile is a research and consulting firm solving tough and relevant behavioral problems across the globe