Reporting back: the Cognitive Economics session at the AEA conference

[Tara posting today]

We had a great response to our Cognitive Economics session at the American Economic Association’s conference a few weeks ago.

For those who weren’t there, here’s a quick summary of the four papers presented, and the discussion at the end.

Dan Benjamin, Kristen Cooper, Ori Heffetz, and Miles Kimball presented What Do People Want?

The key question of this paper is: how can we measure happiness? And how do we account for the biases and differences across different groups and demographics?

The authors have built a model that examines the multiple dimensions of wellbeing by asking individuals a huge number of different questions. Any single question about your true wellbeing will be affected by a lot of “noise” - i.e. biases. They reduced this noise by asking individuals multiple questions including tradeoffs and interpersonal comparisons.

By statistical analysis, they found a small number of underlying factors that best predict the different dimensions of wellbeing.

The authors’ broader claim is that use of survey questions to measure subjective wellbeing is an effective method and currently underutilized by economists. In the future, they believe more data of this kind will provide more accurate measures of wellbeing, and offer policymakers better levers to improve citizen happiness.

David Hagmann and George Loewenstein presented a paper on Persuasion with Motivated Beliefs.

David and George’s paper focused on how one can, and whether one should, lower other people’s psychological defences, and explained their ongoing experiments to test these questions. In general individuals have “asymmetric updating”: we listen to information and update our beliefs only if this information supports those beliefs. We find it easy to ignore information conflicting with our existing assumptions. David models this as a two-stage understanding of persuasion. Stage one is asssessing whether new information will threaten your beliefs and stage two is collecting the new information and updating existing beliefs (in a biased way, based on the degree of threat). As such, persuaders can be regarded as a potential threat: are potential persuaders going to challenge your views? Do these persuaders have the necessary expertise to challenge or comment on your beliefs? And also, how invested in this specific belief are you to be/to not be persuaded from it?

All of these questions fed into the testing model that David and George have been working on. They have built an experimental model that aims to see if messages coming from a more likeable sender, or someone who appears less certain about their opinions, reduces the defences of those receiving the message - so that the message will be more persuasive. . “Receivers” were asked to indicate agreement with certain political views (e.g. “Welfare benefits should be increased/decreased”) , and also to estimate the answer to associated questions (e.g. “What proportion of food stamp recipients have a job?”) . “Persuaders” were then incentivised to change the opinions of the receivers. Persuaders were encouraged to get receivers to acknowledge alternative views, show doubt and/or to build a rapport with the “Receiver” to see if that helped them change their mind.

The findings of the research include:

• A person who expresses uncertainty in their beliefs is more persuasive - but only if the receiver cares about the topic, and therefore has stronger defences against new information
• Likeable people are not (significantly) more persuasive
• Highly persuasive people also ended up persuading themselves of their own arguments

Emily Ho, David Hagmann and George Loewenstein presented a paper on Measuring Information Preferences.

Emily presented a fascinating paper on how individuals prefer to receive and/or avoid information and how one can measure such information preferences. The paper starts from the (rational) premise that everyone should be looking to gain more and more information as it will help us to make better decisions. In healthcare, this may mean getting testing for specific diseases, or in your social life, this may be knowing how well you performed in a certain social situation. However, as you may instinctively feel, this was not the case when surveying groups and individuals. For instance, when hypothetically asking individuals if they wanted to know the results of a test about whether they had a genetically inherited disease, 44% of people claimed they “probably didn’t want to know” or “definitely didn’t want to know” the results.

The authors went on to ask a variety of audiences from caregivers to listeners of scientific podcasts about their information preferences, and, from these surveys, they have created the Information Preferences Scale (IPS). Emily confirmed that the IPS is psychometrically and behaviourally validated for measuring individual level information preferences. The paper concludes that the IPS can predict information acquisition in hypothetical and real life tasks and scenarios. Naturally, it is easy to see how these kinds of information preferences can have some clear implications. How can one disclose information successfully if certain people or groups are not going to be receptive to such information? From individuals making better healthcare decisions to governments improving public awareness campaigns, this research has some promising real world applications.

Leigh presented a paper on Cognitive Goods, Normal Goods and the Market for Information.

Leigh’s paper starts from the premise that there are things we care about that don’t seem to fit the rules of economics. Whether it’s the fate of Lassie in the 60s TV show, or the fact that the person in the next cubicle earns less than I do, or the results of my favourite sports team or political party, these beliefs or mental states have value to me, but that value can’t be measured or priced using economic analysis.

If we are to understand the growing role these kinds of objects play in the modern economy, we need to define the economic rules they play by. Leigh set out half a dozen proposed axioms for “cognitive goods”, the mental objects that play an important role in our experience of the world but that can’t be acquired and traded like traditional material goods.

He proposed a mechanism that may operate in the mind to put value on cognitive goods, how they are used in decision making and when we mentally simulate future or hypothetical events, and empathise with other people (see the recent System 3 blog post for more details).

The goal of this theoretical work is to provide a unified view for some of the mental phenomena the other presenters have discussed; and to illuminate how beliefs, symbolic value and processes like advertising create much of the value in today’s “dematerialised” economy.


After the presentations, we had a short panel discussion between Leigh, Emily, Miles Kimball and George Loewenstein plus audience questions. A tricky question was whether “cognitive economics” is the best term for this field. Although it is clear there is a growing dissatisfaction with traditional economic models (especially in the information age), the term cognitive economics may be too cerebral. Some worry that the term negates the mental states, emotions and beliefs that form the core focus of research in this field. Nevertheless, the panel also drew attention to how this work can have some promising real-world applications. For instance, as accurate happiness measures are created, this scales could hold governments to account on their attention to citizens’ wellbeing priorities.

The session finished with two calls to action. Firstly, for academics and researchers to pursue small and well defined studies that can then be used as building blocks to create the bigger picture in this field. The second was to join Leigh and any other interested academics at the upcoming Cognitive Economics workshops. There will be two in 2019 - one in June in London and one later in the year in North America (perhaps a great chance to develop or present some of these small building blocks?!). We are planning on announcing the dates and details on how you can get involved very soon!

If any readers are interested in participating in a similar session at next year’s AEA (in San Diego - a highly recommended location for a January conference) please get in touch. We will be submitting the proposal in mid-April.