Thomas Z. Ramsøy

The Limits of Behavioral Economics

We are biased! We are neither rational nor utility maximizing! We are not fully informed or fully conscious when we make our choices! That’s basically the message when we use psychology to understand economic behavior — also known as behavioral economics. Indeed, as the 2002 Nobel prize in economics was awarded to Daniel Kahneman for his work, as a psychologist, on demonstrating that economics premises were flawed, it also proved the power of psychology. As the Executive Director for Science at the American Psychological Association, Kurt Salzinger, stated upon the prize announcement in 2002:

“We are all delighted that one of our own won a Nobel Prize–not only because he well deserves it but also because it reminds the world that psychology is science.”

One of the core foundations of behavioral economics is the demonstration of decision biases. That is, that we as humans deviate from rational, utility-maximizing choice. That we have inherent flaws that make us be loss averse, be affected by contextual information (framing), and we behave very differently when just believing that we are observed by others. Indeed, the list of so-called cognitive biases is long and comprehensive. In this way, we can see how psychology has empowered economics to better map out the phenomena that drive our choices. Infusing psychology into economics has been a brilliant idea all the way through.

But psychology has moved on. Today, psychological models of the mind rarely if ever exist without some mentioning of the brain. Neuroscience has become a cornerstone, or even the central chimney, of psychology and our understanding of the human mind. The discipline of neuropsychology has now existed for decades with clear, added value to patients. Psychology is incomplete without neuroscience.

Little has happened after Kahneman and others introduced psychology into economics. Our understanding of consumer behavior and other agent choices still consider deviations, nudges, and other mechanisms that can be mapped. If the adoption of psychology into economics was hard and long, the adoption of neuroscience in psychology is even slower and meets even more resistance.

Behavioral economics currently has a great success in business and academia alike. Corporations live and thrive due to the added value that this perspective brings to corporations in understanding and affecting customer behaviors. But a few notable limitations deserve mentioning — and it is exactly these cases in which we see the limits of behavioral economics and the added value of a combined neuroscience and psychology perspective into economics. Here, I will highlight a few aspects:

  • Poor explanatory power
  • Poor temporal resolution
  • Lack of measurement devices
  • Limited effects of nudging

Mapping without power

The list of cognitive biases is extensive, and a most valuable tool for those who seek to understand how we make decisions. For example, the fact that people are loss averse, and on average need twice the amount of possible gains relative to losses before they will accept a bet, is important. It allows us to understand why certain kinds of choices never occur, simply because there is a perceived unfavorable ratio in a choice. Or how a so-called anchoring effect describes how the end price of a product is affected, just by altering the first price that you suggest (e.g. €10 or €25 for a product as a bid-opener).

Mapping is simply not enough. It is, with all due respect, a poor understanding of the human mind. It does not explain why something happens in the first place. Sure, we have loss aversion, framing effects, and anchoring effects, but why do they happen in the first place? What drives us to do so? Here, we cannot escape a deeper truth about the human mind: it is a biological, evolved system. Behavioral economics does not provide any causal mechanism, no explanation as to why. This is where we need to turn to biology, to neuroscience, to neuropsychology.

One example can be taken from my 2015 co-authored paper, where we found that human loss aversion was in part driven by a deep structure of the brain called the amygdala. This structure, which is known as an “emotional” part of the brain, has previously been shown to be involved in loss aversion, and that in patients who have lesions to this part of the brain may, in fact, lose the loss aversion response seen in healthy individuals. In our fMRI study, we discovered that the amygdala was most involved when the choices were the easiest (see figure below). That is, when it was easy to accept or easy to reject the choice, the amygdala was more engaged. Furthermore, this pattern was even more pronounced for more loss averse individuals. This links not only to the causal mechanisms of loss aversion, but even can explain or predict individual differences in loss aversion.

During risky choices, the amygdala becomes increasingly engaged (A, circle), and we see that it is more engaged when choices to reject or accept are easier to make, such as at when the chance to lose is very high or low (C).

How biases unfold

A second limiting factor to behavioral economic methods and insights is a complete lack of temporal resolution. When does a bias actually occur? How does it unfold over time? Does it happen in milliseconds, seconds or even minutes? Behavioral economics methods and models do not provide a clue about this, which in turn leaves us guessing not only in how to intervene but also when to intervene.

For example, in a study published in 2016, we found that it takes only seconds for people to be affected by what they think others will do. In the study, we had people playing the Prisoner’s Dilemma game. Between each turn, a person would be instructed what the game was called — the “cooperation game” or the “competition game.” That is, the rules of the game did not change, but the social context of the game changed. What we found was that the cooperation rate changed from about 60% in the cooperation frame to about 30% in the competition frame. The game was the same, but the frame changed the participants’ behavior dramatically.

When people were affected by the name of the game, we found that a particular network of brain regions was engaged. First, the memory structure called the hippocampus was more engaged, and parts of the brain’s “social” network region (also called the mentalizing network) was more engaged. This study allowed us to go beyond mere behavior and focus on the mechanisms that drive such effects, and a determination of when they happen. Of course, neuroscience tools such as fMRI have a limited temporal resolution to focus only on seconds, while other tools such as EEG (or electroencephalography) have a temporal resolution of just a millisecond or so.

The brain network involved in swaying people in their social responses involve memory structures such as the hippocampus, and mentalizing structures such as the precuneus, dorsomedial prefrontal cortex (dmPFC) and the lateral temporal cortex.

Further examples come from business cases in which neuroscience allows us to go beyond stated behaviors. In a study for Vodafone and Ericsson, we tested the effects that delays on mobile phones could have on emotional and cognitive responses. Asking people what their tolerance to delays are, people in Northern Europe typically say something like 4-6 seconds. But in our study, also reported in Ericsson’s Global Mobility Report, we found dramatic emotional and cognitive responses after only 2 seconds! Here, even brief delays led to stress responses comparable to watching a horror movie.

Moreover, these negative responses had three devastating effects:

  1. Enjoyment of the content dropped from positive to negative — movie emotions went from positive to negative
  2. Brand emotions dropped by 20% or more for the brand believed to be the service provider during the experiment
  3. Brand emotions to competing brands increased, even though they had not been tested

Together, these studies demonstrate the importance of properly understanding how biases unfold over time. This allows not only a better understanding of the phenomena themselves, but also allow us to better diagnose and intervene towards such adverse effects.

Finding the nonverbal toolbox

As noted, behavioral economics is great at mapping biases. But it rests on a limited toolkit, mainly observing actions and, to some extent, asking people. But when we try to understand effects happening within seconds or less, behavioral mapping is rarely good, and self-reports are mainly unreliable. For self-reports this is doubly damning: either you are asking people to retrospect, in which case they are trying to remember and recreate what they did and why they did it; or you are asking people to introspect and report what they are doing, in which case you are interfering with their everyday decision-making process and making them think about every step in their choice behaviors. Who buys groceries that way?

Here, neuroscience infused with psychology is powerful, as it allows methods to track and measure how people respond without interference or meta-cognition. It’s even better than being a fly on the wall — you can basically be a homunculus inside their skull. Eye-tracking goggles allow you to measure where people are looking (and what they are missing) often with an accuracy of 20-millisecond intervals (see example below).
an example from a live demo by Neurons, showing eye-tracking live feed while a person is browsing a smartphone social feed. Click link to follow image source

Furthermore, it is possible to link this eye-tracking behavior to direct measures of brain responses, such as an EEG. This allows us to combine both methods, and thereby understand how people respond emotionally and cognitively to both general events (such as using Facebook vs using Snapchat) as well as detailed information when people look at something in particular (such as looking at the price tag, the product, and the brand in a store environment), which again allows a better measure and understanding of the mechanisms underlying choice (see below).
display of live EEG metrics and raw EEG signals during a live experiment by Neurons. This exemplifies the resolution level (1 second data update) which allows a better understanding of emotional engagement, cognitive load levels and other issues such as distraction and drowsiness. In normal study setups, data re not displayed but recorded, to ensure optimal data quality. Click image to get image source

Behavioral economics has been a landslide in terms of progress in on our understanding of decision-making. But it’s only a small step to a complete understanding of how we choose as we do. Psychology infused economics with a new stream of consciousness. Now, the proper application of neuroscience and psychology to economics is the next mature (and even overdue) step — not only in academia but particularly in business and society, in furthering our understanding of decision-making behaviors in consumers, employees and politics alike.