Choice Architecture: Whose Interests Matter?

Source: Image by Gerd Altmann from Pixabay

Consider the last time you signed up for a web service. It might have been a site to buy products (e.g., Amazon), monitor and pay your credit card (e.g., Chase), read subscription content (e.g., The New York Times), or receive another service for which registration is required.

At some point during that process, you may have been given a choice about whether you consent to certain types of emails (e.g., from sponsors or partners). On some sites, the default was preset with your approval, and you had to unselect that default if you didn’t consent (opt-out). On other sites, the default was set to reject such communication unless you deliberately selected the box consenting to it (opt-in).

Regardless of the default, you had the freedom to make a conscious decision about whether you consented to receive those emails, but the website designer pre-determined the default response.¹ If you failed to override the website’s default–whether intentionally or accidentally–you implicitly approved that default. This is a simple example of what is referred to as choice architecture.²

At a basic level, choice architecture involves the “context in which people make decisions” (Thaler & Sunstein, 2021). That is, choice architecture consists of how information is presented to a decision maker, which can subsequently influence the decisions made.

Want to cause people to buy more fresh fruits and vegetables? Ensure the produce department is closer to where they enter the store. Want to increase organ donation? Make organ donation the default choice and require people to opt-out if they don’t want to be a donor.

Both examples above involve the choice architect making intentional decisions about how choices are presented to the decision maker. Although with modern technology, it is possible to randomize some choices (e.g., as in a multiple-choice exam), choice architects are generally most interested in finding ways to intentionally influence people to select options desired by the choice architect (Mills & Sætra, 2021).

Choice architecture subsequently becomes an essential aspect of nudging (which I wrote about previously in the context of the pandemic)–intentionally designing choice structure, so people’s biases will make them more likely to respond the way the choice architect desires (Thaler & Sunstein, 2021).

The overall effectiveness of nudging has received mixed support (see here, here, and here), and most nudges, even if they work, show small overall effects (hence the term nudge instead of shove). Still, even small effects across a large population can yield a sizeable impact. If a given nudge improves sales by 1 percent, that may yield little obvious benefit to a small, local store but could result in sizeable revenue increases for a company like Amazon (especially if the cost of the nudge is low).

Thaler and Sunstein (2021) grounded their conceptualization of nudging in a philosophy called libertarian paternalism. The general idea is that the choice architect knows which choice is in the decision maker’s best interest (paternalism) and subsequently designs the choice architecture to increase the likelihood the decision maker will make that choice without reducing the decision maker’s freedom to choose (libertarianism.)³

But who gets to decide what is in the decision maker’s best interests? Is it the government? Your employer? Amazon? Might it be that our personal values may conflict with what the choice architect believes is in our best interest? The answer, of course, is yes, if for no other reason than the choice architect’s values or ideological biases may conflict with those of the decision maker.

Sometimes choice architects (especially in government) may premise the architecture using more of a greater good (or utilitarian) focus,⁴ which may or may not be the same criterion the decision maker would apply. For example, a choice architect might argue that renewable energy, even if more costly to the consumer, should be the default, which occurred in several cities in California back in 2019 (Roth, 2019), potentially raising the monthly utility bill by 9 percent for residents of those cities.

Although the choice architect may have decided that it was for the greater good or in the decision maker’s best interests, it is unlikely that all (or perhaps even most) decision-makers would have agreed with that conclusion (especially if they were struggling to make ends meet).

But this gets to the heart of the issue regarding choice architecture. It is difficult to know what is in the decision maker’s best interests if the choice architect is ignorant of the values and priorities of the decision maker. And this can present several ethical issues that should be addressed. Furthermore, not all attempts to influence decision-makers in this way would even be loosely labeled as a form of paternalism as the motives of various choice architects could be quite disparate from those of the decision maker (e.g., vendors, snake oil salespeople).

Intelligence Essential Reads

What DNA Can—and Can't—Tell Us About Intelligence

How Do You Know What You Know?

This disparateness has potentially even more marked implications when considering using more algorithmic and machine learning approaches to choice architecture, the subject of my next post.