Beyond Prospect Theory: Behavioral Science for Public Health

Source: Tashatuvango/Adobe Stock

Prospect theory^{[1, 2, 3]} has perhaps been one of the most influential theories within psychology and behavioral economics. Daniel Kahneman won the Nobel prize for this work.

Prospect theory and public health messaging

The theory suggests that people are irrational in their decision-making. For example, it demonstrates that when certain messages are framed in positive or negative ways, it can bias decisions. The theory also suggests that losses are perceived as more painful than gains are satisfying.

This has been applied to the area of public health^[4], and in particular public health messaging. For example, prospect theory has been used in a way that manipulates how individuals would respond to public health messages (i.e., through framing the message in a positive way in order to alter people’s decisions about transmissible diseases). It has also been applied more recently to explaining stockpiling behavior during the COVID-19 pandemic^[5], as well as to explain the lack of social distance compliance^[6].

More recently, however, more novel approaches are now being considered, which involve behavioral science^[7]. This models how reinforcement may be utilized in public health messaging in order to encourage more positive and healthy public health behaviors.

In behavioral science, three main constructs (called functional classes) are referred to for rule-governed behavior. These are:

1. Pliance

This refers to rule-governed behavior which is controlled by the speaker, such as socially mediated consequences. For example, if a teacher said to you that you should work hard on your assignment, and you did it because you wanted the teacher’s praise, then this would be socially mediated behavior. Here, the reward in the form of praise would encourage you to follow the verbal rule. Counter-pliance is the opposite of this, where the individual behaves in the opposite way to the socially mediated consequence.

2. Tracking

This refers to rule-governed behavior that is controlled by consequences specified by the rule within its given environmental context. For example, if a mother said, “Eat your greens as it will give you energy,” if the child correctly tracks “full of energy” as being positive and rewarding, then they may be more likely to follow the rule.

3. Augmenting

This refers to rule-governed behavior, which alters the extent to which a rule’s consequences for behavior have reinforcing or punishing properties. This can take two forms: (1) motivative augmentals, where the established consequences or reinforcement of punishment are temporarily increased or decreased; and (2) formative augmentals, where a previously neutral stimulus (no reinforcing properties) now establishes reinforcing or punishing consequences. For example, if you saw a coin on the floor and thought it was worthless, it may act as a neutral stimulus, and you would be unlikely to pick up the coin. But if a friend suddenly said that it is a rare and valuable coin, this may create a reinforcing function encouraging you to pick up the coin.

These early behavioral constructs about rule-governed behavior have been further developed and established in a framework that allows for more complex context to be developed. In my own lab, work has centered around developing complex reinforcement models which utilize artificial intelligence (or machine learning) in a way where this work can be applied to predicting how effective public health messages may be.

A new model

Though Kahneman’s prospect theory and fast and slow thinking model have many useful applications to public health messaging, they have received criticism for relying too heavily on probability theories^[8], which have many problems when it comes to expressing causal assumption^[9]. More recent advances in contextual behavioral science extend the three main functional classes of rule-governed behavior to include context through mutual entailment, combinatorial entailment, and transformation of stimulus function, which explain how information can account for learning in complex relational networks of the human brain and ultimately be used to predict whether a public health message will likely be followed by the public or not, based on its context and how it affects individuals’ relationally framed networks. These networks are explained through a theory called relational frame theory (RFT), and some of its most basic principles for how relationally framed networks (knowledge contained in an individual’s mind) develop include:

1. Mutual entailment

This suggests that information can be shared between two symbolic constructs (e.g., a picture of the word fox is the same as the word “fox” spoken).

2. Combinatorial entailment

This refers to shared relations properties—which explains why we can infer information, e.g., if we are told Bill is faster than Ted, and Ted is faster than Steve, then we can combinatorially entail that Bill must be faster than Steve without being told directly.

Health Essential Reads

When (and Why) We Hide Infectious Illnesses From Others

Vitamin D: Vital to Your Health

3. Transfer of function

This suggests that a functional property from one construct can relationally pass onto another construct. For example, if you were afraid of snakes but enjoyed walking through the woods, but were told that dangerous snakes were there, then the fear of the snakes may pass onto the construct for woods—i.e., a previously neutral or positive construct (woods) now has the function of fear associated with it, and this fear may alter the person’s behavior, so they become avoidant of the woods.

In the context of public health, new models can be developed from this model, which may explain why in one context, people follow public health messages, such as keeping two meters apart during COVID times (which is what my lab is doing) and not in another situation. An example of this was in the UK, where a very simple public health message was introduced—“Stay home, protect the NHS, save lives.” This message was very successful at bringing people together for a common cause, where many people were seen clapping each week for the NHS and largely compliant with the lockdown rules. Some of the reasons for the message’s success were: (1) It was so simple, so accessible to those even with low health literacy, and promoted coherence with the public's existing knowledge (or relationally framed networks); (2) the rule was framed in a way which made contact with what most people in the public care about, i.e., saving lives and protecting the NHS (therefore exploiting motivative augmentals); (3) the message also targeted broader values, such as a sense of community and common social values, which exploited tracking the correspondence between the environment and the rule; (4) the message reinforced through contextual tracking of societal community values at both the local and global levels, where self-sacrifice to save a greater cause may be promoted as a community and national value.

Within this processing of information (or responding behaviorally to the public health message reinforcements), the mutual entailment, combinatorial entailment, and transfer of function tie the information together in complex networks within people's brains, and we are now beginning to predictively model using artificial intelligence approaches how the public may respond. Messages which do not work, as they do not exploit behavioral principles may lead to the public not following them, ultimately harming or even contributing to the death of individuals (e.g., anti-vax behavior). Messages which do follow these basic principles may save lives.