Making Values Concrete

Making Values Concrete


Many thanks to B. Gabbai, A. Morris, A. Ovadya, J. Stray, & F. Noriega

(This is a more accessible version of an academic article you can cite.)

I'm in the supermarket, considering what to eat for breakfast tomorrow. Looking through the aisles, I narrow it down.

What are my criteria?

  • Cost (It's a tight month!)
  • How filling each item would be (I find it easier to work when breakfast was filling.)
  • How I'd feel after eating (Would this feel healthy in my body?)
  • Taste experience (Would tasting this item pull me out of my morning routine, into a reverie of appreciation?)

A careful observer could try to verify that these are my criteria, by watching where my eyes go, what I reach for, put in my cart, etc. She'd see me prefer a more filling item to a less filling one; set aside something less tasty and reach for something tastier. This would serve as evidence of what my values are, regarding breakfast foods.

I've just used the words 'criteria' and 'values'. But I'd like to replace those terms with something more specific. I'd like to call them 'attentional policies'.

Why? Well because, my attention wandered between these considerations, as I made my breakfast choice. And it didn't wander there by chance. Rather, I have a policy of considering these four criteria, when I'm shopping for breakfast.

So instead of criteria or values, I'll say I moved between four "attentional policies" in this choice.


Attentional Policies.

Routine places where our attention goes, as a matter of policy, when we make a given kind of choice.

Why use attentional policy and not "criteria"?

  • Attentional policies are at least theoretically observable. By watching me as I consider different options across multiple choice situations, you can check: am I really attending to factors like price or taste?
  • Furthermore, using 'attentional policy' highlights something important. Earlier—before choosing a breakfast cereal—I must have chosen these particular attentional policies. When did I decide to consider cost? Perhaps it was when I entered the supermarket and remembered the state of my bank account. When did I start to take how-I-feel-after-eating into account? That was years ago.

In this essay, I'll stress the importance of these earlier choices. It's in these moments—when we change our policies about what to attend to—that we update our values.

If we understand a value as a kind of attentional policy, this locates a person's values in their behavior and lifeline. Changes in values can be observed, verified, and disambiguated. One person can be said to have inspired a value in another, and the moment in which the second person's values changed can be narrowed-down.

That's a big advantage of thinking in terms of attentional policies. Here's another: attentional policies shed light on what's meaningful to a person, what they hold to be wise and good ways to live, and which modes of evaluation they identify with.

In the list above, some breakfast-choice attentional policies say more about my life situation (like cost and filling-ness above), while others say more about who I am, and what's important to me (such as feeling healthy in my body, and being carried away by the taste of things).

I'll show that a person's attentional policies can be sorted into these two piles, based on the kind of decision they made to adopt the policy. Policies that were adopted from one kind of decision correspond to what's meaningful to the person, and to what they feel is wise and good.

Such a list of someone's sources of meaning is extremely useful.

When we say "everyone deserves meaningful work" or "people need to participate meaningfully in a democracy" or "people have meaningful interactions on facebook" we're being vague. By collecting someone's attention policies and highlighting the meaningful ones, we can break down precisely what's meant by meaningful work, meaningful interactions, and meaningful relationships. We can assess if they're really happening, and see how they could happen more.

We can also evaluate systems by whether they support specific kinds of meaning. Many current crises (in ecosystems, global economics, media, etc) are caused by systems designed around metrics (dollars, clicks, views, votes) that are divorced from life meaning. This leads to so many problems: depression, isolation of the elderly, soul crushing workplaces. Reevaluating these systems in light of a specific account of meaning could solve these problems.

So let's try to break it down. In the next two sections, I'll take a more careful look at the word "values" and at attentional policies specifically. Then, I'll give a rule to find the "meaningful" type of attention policies, and see how they connect with phrases like "the values of science" and "democratic values". I'll conclude with some advantages of all this: in design, recommender systems, political systems, etc.


People use "values" to mean various things. I've gestured already at two: (1) whatever criteria we use in choice, and (2) the things that feel right and meaningful when we do them—such as being vulnerable, taking stage, being creative, etc.

I believe these "things that feel right and meaningful" are attentional policies that are used in choices. That is, what's meaningful is choosing to take stage, choosing vulnerable words, etc. Meaningful doing means meaningful choosing, which means meaningful attending.

We also use values (in expressions like "the values of science" or "democratic values") in a third way: a scientist's values might include intellectual humility, passionate pursuit of the truth, etc. They're things that are meaningful for scientists and also are needed to keep the institution of science on the rails. I think these, too, are attentional policies of the meaningful kind, and we can collect them in the same way.

I'll cover those three types of values in this essay.

There's are other uses of the word "values": to mean visions of what's right for everyone, or for a group (e.g., what a family should be like, how a father should behave, what a nation should be like), to mean political causes (like inclusiveness, freedom, etc), and to mean standards to which we push others to conform, or try to conform ourselves (like masculine or feminine dress-codes).

While these are related to meaningful attention policies, they aren't quite the same. I won't cover them here.

Attentional Policies

So, by policy I mean an action (like "taking out the trash on Tuesdays", "calling mom on Sundays", or "running new contracts past the lawyer") done regularly, or when in a certain context, without doing a cost-benefit analysis each time.

Attentional policies (APs), then, are policies about how to think about a thing, what to pay attention to in a context, or what to look for in selecting an action. "Taking out the trash on Tuesdays" is a normal policy, but "looking for kind words when giving feedback" is an attentional policy.

APs can be about how to treat people (honestly, openly, generously, mercilessly); how to approach things (with reverence, with levity, with skepticism); how to keep things (simple, sensual, rocking, full of surprise); or how to act more generally (boldly, thoughtfully, carefully).

APs might account for much of what’s called my “personality”: when making friends, am I cautious or bold? When considering a purchase, is my focus on price, quality, or durability? When speaking, do I try to be witty, precise, or down to earth? These may not just be “character traits” I was born with, but policies I adopted, which work together for my way of life.


Some of a person's APs—but not all—are their sources of meaning. In the next section, I'll give a rule to separate attentional policies into two piles, the meaningful kind, and the others. But first, here are some consequences of the view that values are a type of attention policy:

First, for something to be an attention policy, it must be specific. An AP about ‘honesty’ or ‘vulnerability’ is always short for a more specific policy, like “attend to what I feel about each thing we discuss, and let my feeling show”, or “attend to false impressions a listener may get from my statements, and head them off with a disclaimer”. That is: to have honesty as a policy requires a substantive interpretation of honesty, and these differ from person to person.

Second, attentional policies (and thus values) are rival. I wish I could craft my words to simultaneously be kind, honest, tactful, humble, and inspiring. I wish I could, at the same moment, be precise in my speech, aware how each word lands, aware of my own feelings; transparent, calm, and centered; passionate; physically graceful, like a dancer. But APs compete for my attention. So, my choice to look for witty things to say crowds out many other options.

Third, because of this competition, we're forced to choose the APs that are most important to us in our daily contexts. This says a lot about how we want to live, or how we believe we must live.

The Meaningful Ones

In this section, I want to distinguish between two types of APs.

Some APs, I claim, are part of our conception of what it is to live well. If is one of those, then a life in which I never attend to will seem somehow less of a life. Diminished.

Not every AP is like this. Let’s return to the four APs I used when shopping:

  1. Cost (It's a tight month!)
  2. How filling each item would be (I find it easier to work when breakfast was filling.)
  3. How I'd feel after eating (Would this feel healthy in my body?)
  4. Taste experience (Would tasting this item pull me out of my morning routine, into a reverie of appreciation?)

Would I feel my life were diminished, if I never attended to cost or filling-ness? No, I don’t think so. But my life would seem lesser, if I never attended to how I feel after eating, or to taste.

Let us give a name to APs of the latter type: APs without which my life would seem somehow lesser. Let’s call them Constitutive Attentional Policies, or CAPs.

We can go quite deep in distinguishing CAPs from other APs. At one level, this question (whether life would be diminished) marks out a set of APs which are part of a densely connected core.

We can picture our ideas about the good life as an interlocking web of values and practices, tightly connected. In this densely connected core, any particular value (such as attending to taste) acts to support many other ways of attending and living, which are also part of this core. This means that CAPs will tend to be adopted for broad benefits.

Non-CAPs (or Instrumental Attentional Policies) will, on other hand, tend to be outside this core, but pointing towards it. In the directed graph of benefits, you can think of them as thin lines that lead towards that densely connected center of the good life, but aren’t themselves densely-connected or within the core. When you ask why someone’s adopted an IAP, you’ll get narrower, specific reasons, that depend on a context.

This may sound like a difference of degree, but I believe there is also a difference in kind. Specifically, a different kind of reasoning is used to justify attending to an IAP vs a CAP.

  • IAPs are adopted due to instrumental reasoning.
  • The benefits make a case that doing a in c is functional, necessary, prudent, or expedient. It's based on expected costs and benefits. I have concrete fears of certain outcomes, or goals I want to accomplish, and adopting APs 1 and 2 will help to avoid those fears, and achieve those goals.

  • CAPs are adopted due to constitutive reasoning.
  • You could imagine it's the same with APs 3 and 4—that I'm scared to feel bad after eating, or that it's my goal to have a taste reverie once per day. But it's not so. My reasoning for adopting these latter APs is of a different sort: of a nature I'll call constitutive. Eating is a big part of life. And I see being aware of how I feel when I eat as a necessary part of doing eating how I intend to do it. Similarly, I see being pulled into reveries of appreciation as a necessary part my best life.

    Whereas an instrumental judgement relies on claims about cause and effect, this is a part-whole claim, or a claim to synergy. To make such a constitutive judgement, we ask ourselves questions like: are Xs impossible without Ys? Are Xs without Ys in some way lesser? Are there kinds of wholeness that emerge only when an X has a Y?

When we've made a constitutive judgement (e.g., that feeling good after we eat is a necessary part of doing breakfast well) it says a lot more about our identity, our sense of meaning, our wisdom—than a merely instrumental judgement about lowering costs this month.

This line—between APs adopted due to constitutive judgements vs. instrumental ones—is where I'll draw the distinction between the values that are meaningful to us vs our other attentional policies.

Indeed, I believe the experience of meaning is exactly the sense of being in touch with what we've decided is constitutive of the good life. It's the feeling that says "This is it. This is what I'm here on earth for. This is the part of breakfast that's necessary in my conception of the good life."

This is what it is to have a meaningful moment: it's a moment when one's attentional policies are of the constitutive sort. When we are attending to criteria we earlier decided on, based on constitutive reasoning.

And contrariwise, things feel meaningless when our attention is structured purely by instrumental concerns.


If the story ended there, it'd be big news for designing of meaningful experience. But I haven't gotten to the best part.

Constitutive judgments are a kind of knowledge. Take my belief that "the good life for me includes moments of sensual reverie". That's a conclusion I came to, something I could tell someone. On hearing it, they might decide it's true for their good life, too. It's knowledge.

Instrumental judgements are also a kind of knowledge. My belief that "By spending less than €1.20/day on breakfast, I'll make it through the month" is a conclusion I came to, that I could tell someone.

Knowledge that's neither constitutive nor instrumental, we call data or information. Instrumental judgements we call know-how. And constitutive judgments, we call wisdom.

Wisdom, thus defined, includes all the part/whole assessments—what makes up a good religion, a good relationship, a good life, a good day. But also, more specific constitutive judgments: what makes a good scientist? a good guitarist? What's necessary when composing a song? What's a necessary part of making a difficult policy choice?


Institutional Values

Some of this wisdom is needed for institutions to function. That brings us to another definition of values. One I mentioned at the outset (e.g., "the values of science" or "democratic values").

By watching someone's attentional policies, and highlighting the constitutive ones, you learn what's necessary to keep their institutions, relationships, and activities on the rails.

And when you design in a way that supports these constitutively-adopted attentional policies, that doesn't just foster personal experiences of meaning. It also protects institutional meanings: what's needed for good lives, good relationships, good communities, good institutions, and ultimately for a good society.

Here's an example:


Earlier in my career, I chose work colleagues for their brilliance, efficacy, and a shared sense of play. One day, with one colleague in particular, I recognized a different thing we shared: we were excited to participate in the same long-term trends. This felt meaningful.

Later, when choosing teammates, I began to look for this alignment first. My teams now seem more likely to stick together, and we have new types of conversations.

Here I adopted an attentional policy that I judge constitutive of good, long-term teams. It feels right and meaningful when I do it. But it also captures an institutional meaning needed in that kind of team.

Designing to support these chunks of institutional wisdom is especially powerful because they filter out instrumental attentional policies—what the scientist does merely to keep his job, to fit in with colleagues, or to achieve specific goals. So, a scientist's constitutively-justified APs won't include what they do merely to get tenure, to amass citations, etc. They include what's considered constitutive, but exclude the perverse incentives which clutter up institutions.


This model is powerful.

At the large scale, there are consequences for politics and machine learning. Using this approach, choices a person makes (such as votes or clicks) can be assessed as more or less meaningful. Consider my breakfast cereal choice from the introduction. Was it a meaningful choice? If my shopping was mostly about taste and what feels good in my body—constitutively-justified policies—then the choice was a meaningful one. If those aspects get crowded out by instrumental concerns like cost, it becomes meaningless.

Such a litmus test allows us to modify many systems:

  • Votes driven by instrumental concerns should perhaps count differently than votes based on meaningful choices.
  • A "like" or "share" that's downstream of a constitutively-justified attentional policy could count more.

There are also smaller-scale, interpersonal consequences of the model. When two people are articulate about values, one can easily see if her attentional policies line up with the others'. Processes of inspiration and admiration can be more explicit, new understandings of trust and conflict can emerge.

In this textbook, I won't discuss these large or small-scale consequences. Instead, I'll cover the implications for user research and design:

  • In user research, we learn to interview someone, listing their attentional policies, ask questions which sort out which policies arise from constitutive justifications (without needing the subject to be philosophical or introspective).
  • We'll look how designs structure our attention, and when our ability to pay the meaningful kind of attention gets crowded out .
  • Finally, we study which affordances or information are required in the environment to make it navigable by someone's values. This is the focus of the theory of hard steps in Quest 2.

Examples of Values

More examples


Why do we make constitutive judgements, instead of doing a cost-benefit analysis for everything?

Older Essays on the Same Subject

What the Hell are Values?, Part 1
What the Hell are Values?, Part 2

Academic ← even more precise - very optional

How to Operationalize Metaethics