Why Do We Make Decisions? Introduction To Neuroeconomics
Thanks for your patience with the two-part post. The benefit of breaking this up is allowing me to go in-depth on the problems with current models and how basic neuroscience improves our experimental or study design.
People are rarely normative decision-makers. We are faced with ICU: Instability, Complexity, and Uncertainty.
Instability is the rate or frequency of change.
Complexity is the number of features or quantity of information we need to understand the systems we interact with fully.
Uncertainty is how well we understand the link between a decision or decision chain and the short and long-term outcomes created by them.
We create heuristics to overcome ICU. When we build churn models or predict price sensitivity, we are trying to reverse engineer those heuristics. As I outlined above, we have ways to handle that partially, but none are ideal.
What Is Neuroeconomics?
Neuroeconomics is a line of research into the process of decisions and the sub-processes by which decisions are made. Studies are designed to map mathematical constructs to our brains, creating a structural model of decision-making with direct biological measurements creating evidentiary support. It is a combination of neuroscience, psychology, and economics.
Our field is more connected to the economics side of research. It is interesting to explain where events happen in the brain, but I won’t be going down any of those rabbit holes. Our focus is on the structural models of decision-making.
I split neuroeconomics into two time periods, pre-2016 and post. Early work is largely unreliable. The field was learning how to do research and many publications from the pre-2016 period are unsupportable.
Economists and psychologists have legitimate criticisms for neuroeconomics because of the early research period. Neuroeconomics is now emerging as a more rigorous version of economics and psychology. It is still imperfect, but the approaches are more stringent than behavioral economics or offshoots. For that reason, I think the field’s structural models are more reliable.
We only have a small number of supported features in the structural models. However, we don’t need many to build an effective (for business use cases) understanding of decision-making. Models are simpler and built with fewer assumptions.
The Structural Model Of Decision Making And Nike’s Behavioral Experiments
Enough theory; let’s move back to examples. Structural models are complex, but I want to demonstrate how understanding decision model features leads to more generalizable experiments. First, let’s review Nike’s experimental validity using structural models of decision making.
Nike’s experiments are moving slowly because they are too granular. Exclusive access tests higher-level decision components, loss aversion, and effort discounting. Loss aversion happens when an individual’s structural model for decision-making puts a higher weight on avoiding loss than ensuring gains.
The objective is the same for a sneaker collector with either side of loss aversion. They want to optimize their chances of getting a shoe. The sneaker collector believes exclusive access achieves this. A loss averse sneaker collector will make decisions to minimize their chances of missing out, while someone who is not loss averse will make decisions to maximize their chances of getting.
Effort discounting is the concept that our perception of a reward’s value decreases as our perception of the work required to get the reward increases. A loss averse sneaker collector will perceive the value of avoiding losing exclusive access to decrease as their work to achieve that goal increases. A non-loss averse sneaker collector will perceive the value of gaining exclusive access to decrease as their work to get it increases.
Effort discounting also has a temporal component. The further away from the reward, the more significant the impact of effort discounting on our perception of value for the reward.
Let’s explore this from a logic standpoint instead of using math.
Are sneaker collectors driven by loss or gain? Even though it sounds like a decision question, this is a segmentation or stratification question. The immediate impact of decision-making models guides our stratification process and usually moves us from unrelated features to causal features.
Stratification starts with demographic data and moves to behavioral data. That’s our typical maturity progression and data gathering progression. Experimentation begins once we have behavioral data. Nike’s behavioral hypothesis for their first experiment is, “If we provide exclusive access for a limited time, users will enable notifications at a higher rate than they do now.”
The experiment was carried out, and they accepted the hypothesis. Nike’s second experiment focused on the same reward and more complex behavior. The hypothesis is, “If we tell users that exclusive access is given to people who engage with content, users will engage with content at higher rates than they currently do.”
The experiment was carried out, and they accepted the hypothesis. The structure of Nike’s experiments tells me they are focused on the reward and the behaviors they can cause with that reward.