Category Archives: epistemic networks

morality as a network (revisited)

4 Replies

(Syracuse, NY for a conference on Self among Selves: Emotion and the Common Life) Each of us holds many moral propositions. Some are abstract and general, like “Every person has equal moral worth.” Some look like rules or commands: “Do not kill human beings.” Some are more like positive or negative emotions about particular things or people, but we could translate them into propositions.

Some are intuitive. When they are very obvious, it can even seem obtuse to ask why they are valid. (“Why should you not randomly kill people for fun?” is a bad question.) But some of most important moral thoughts are rather particular, idiosyncratic, and in need of justification. For example, I sit here writing with a view of Tufts University, which is my employer. I feel some commitment to and fondness for Tufts, and I believe that my writing supports my work for Tufts (because it will ultimately find its way into an article). Those feelings could be translated into propositions. But each one requires justification and evidently connects to other propositions. To show that I am doing the right thing, I would owe a description of Tufts University and some premises about moral commitment (i.e., Why should one be committed to anything?). In the ensuing conversation, factual information might be relevant: Tufts actually does certain things. Morally evaluative terms would also arise and would need to be justified. The conversation about Tufts might well shift into related topics, such as the role of scholarship in a democracy, the value of democracy, the place of philosophy as scholarship, or the quality of my scholarly work.

We could view all the morally relevant propositions that I hold as nodes in a network. My relationship to each node may be a straightforward endorsement, but it may be more complicated than that. (See the concept of propositional attitude.) For instance, Anne Swidler finds (Talk of Love, 2001) that many middle class Americans have in their heads what they call a “Movie” ideal of love. They do not believe it. They view it with some irony. But they appeal to it in certain situations and they feel its force over them. If we translated that Hollywood ideal into propositions (e.g., Every individual has a perfect romantic match;  finding that person guarantees happiness), most Americans would both endorse and strongly deny those statements.

Many of my moral beliefs and ideas are connected to other nodes, but not necessarily in one way. Depending on what two nodes say, it can be the case that:

  • A logically entails B
  • A causes B
  • A makes B more likely
  • A is a step on a long path toward B
  • A and B are in fruitful tension: incompatible, yet both worthy of support
  • A resembles B
  • A prevents B
  • B echoes A
  • I am impressed by people who believe A, and they also tend to believe B
  • People with a given kind of experience tend to believe both A and B
  • A and B are examples of C

… and so on.

When we try to assess whether people are good, or whether what they does is right, we often ask about the separate propositions that these individuals hold. For instance, anyone who thinks that Jews are evil is worse for that reason, regardless of what else he may think or do. That is a node in his network, and we have a strong moral intuition about it.

But in many cases, we do not have confident intuitions about the separate nodes, nor should we, because they have no moral valence out of context. So a different question to ask about a network is: How is it structured?

I think that question has been addressed too narrowly (in the philosophical writing that I know). The question always seems to be whether the network is coherent and whether each component is entailed by broader, more abstract, more “foundational” premises. Kant, Mill, Rawls, and many others analyze morality that way. They presume that a moral network map should be organized as a tree, with abstract generalities at the root, and particular applications at the branches. But that is only one type of diagram. I see no reason to assume it’s the best; it is certainly vulnerable to skepticism about the premises that lie at the root. Nor is coherence evidently desirable: I admire more a person who is aware of moral tensions and inconsistencies than one who has simplified his principles to remove all conflict.

A very common goal of moral reflection (not only among professional philosophers) is to weed out the weaker aspects of a person’s network. “Critical thinking” is supposed to be a matter of getting rid of the mere prejudices and unsupported assumptions, conflicts, and fallacious connections. Professional philosophers often impose what Amartya Sen calls “informational restraints” to weed out nodes and links. They certainly disagree about which considerations to ignore: for example, Kantians reject consequences, Rawls erects a veil of ignorance to hide our place in society, and utilitarians screen out motives as primary evidence about what should be done. Most, however, will agree that anecdotes about specific individuals are subject to bias; that simple arguments from authority are fallacious; that strong emotional responses must be translatable into valid propositions; that evidence about consequences only matters to the extent that we can assess outcomes by an independent standard; that the use of precedents and comparisons requires justification; and that rules or principles that can be generalized are more reliable than those that are narrow and ad hoc.

As Bernard Williams writes in a slightly different context, theorists tend to criticize–and seek to delete–intuitive or conventional moral concepts, but “our major problem now is that we have not too many but too few, and we need to cherish as many as we can.”

Instead of seeking to delete nodes or connections that are unreliable, and instead of trying to make the whole network look like a flowchart with the summum bonum at the base, I would ask:

  • How extensive is the network?
  • How many connections does the person draw? (How dense is the network?)
  • Are the nodes that have the most connections intuitively correct?
  • Are the nodes that have the most connections intuitively weighty?
  • Have the conflicts been recognized and led to appropriate conclusions, or have they been ignored?
  • Are there free-standing nodes, and if so, are they justified in any way?

(I have  explored related ideas in posts on How to Save the Enlightenment Ideal and Moral Thinking as a Network.

how to save the Enlightenment Ideal

1 Reply

If there is such a thing as the “Enlightenment Ideal,” it says that individuals should hold general, publicly articulable, and correct moral principles that, in turn, guide all their opinions, statements, and actions. That is a view that–with some variations–Kant, Madison, J.S. Mill, and many others of their era explicitly defended. None of those writers was naive about the impact of “prejudice [and] voluntary ignorance” (Mill), “accident and force” (Madison), or “laziness and cowardice” (Kant) on actual people’s thought and behavior, but they presumed that ideals could have causal power, shaping actions. Reasons were supposed to be motives.

That assumption has seemed to recede into implausibility as evidence has accumulated about the scant impact of reasons or values on actions. It seems that people cannot articulate consistent moral reasons for their opinions. We choose our moral principles mainly to rationalize our decisions after we have made them.*

Scholars who reflect on this evidence seem either to dismiss the relevance of morality entirely or to defend a different model of the moral self. This alternative model presumes that our intuitive, non-articulable, not-fully-conscious, private reactions to situations can be valid, can affect our behavior, and can be improved by appropriate upbringings and institutions. The new model retains some Enlightenment optimism about the importance of morality and education, but at the cost of treating moral judgment as intuitive and non-discursive.

I would propose that we misinterpret the empirical findings and miss their normative implications if we rely on a dichotomy of conscious, logical, articulable reasons versus unconscious, emotional, private intuitions. There is more than one kind of valid, publicly articulable reason.

The Enlightenment thinkers cited above and their skeptical critics seem to share the view that a good moral reason must be highly general and abstract. They have in mind a kind of flow chart in which each of one’s concrete choices, preferences, and actions should be implied by a more general principle, which should (in turn) flow from an even more general one, until we reach some kind of foundation. This is not only how Kant thinks about the Categorical Imperative and its implications, but also how J.S. Mill envisions the “fundamental principle of morality” (utilitarianism) and the “subordinate principles” that we need to “apply it.” Consistency and completeness are hallmarks of a good overall moral structure.

But many people actually think in highly articulate, public, reflective ways about matters other than general principles and their implications. They think, argue, and publicly defend views about particular people, communities, situations, and places. They do not merely have intuitions about concrete things; they form reasonable moral opinions of them. But their opinions are not arranged in a hierarchical structure with general principles implying concrete results. Sometimes one concrete opinion implies another. Or a concrete opinion implies a general rule. That may not be post hoc rationalization but an example of learning from experience.

Moral thinking must be a network of implications that link various principles, judgments, commitments, and interests. We are responsible for forming moral networks out of good elements and for developing coherent (rather than scattered and miscellaneous) networks. But there is no reason to assume that the network should look like an organizational flowchart, with every concrete judgment able to report via a chain of command to more general principles.

I plan to support this argument by comparing two clear and reasonable moral thinkers, John Rawls and Robert Lowell. Both lapsed protestants who were educated in New England prep schools, drafted during World War II, and taught at Harvard, they shared many political views. In his writing, Rawls both endorsed and employed highly abstract moral principles, but Lowell was equally precise and rigorous. His moral thinking was a tight network of associations among concrete characters, events, and situations.

*One summary of the evidence, with an emphasis on sociology, is Stephen Valsey, “Motivation and Justification: A Dual-Process Model of Culture in Action,” American Journal of Sociology, vol. 114, no. 6 (May 2009), pp. 1675-1715.

ethical reasoning as a scale-free network

All of us have many ethical thoughts–about this person, that activity, and also about general concepts like virtues and principles. Some of our ethical thoughts are linked to other ones. One entails another, or trumps it, or incorporates it. So you could make a diagram of my moral or ethical worldview that would consist of my thoughts and links among them.

What kind of network would it be? And what kind of network should it be? These are, respectively, an empirical/psychological question (the answer to which might differ for individuals) and a moral/philosophical question (which probably has one correct answer). By the way, instead of asking these questions about individuals, one could pose them for cultures or institutions.

Ethics might turn out to involve one of three kinds of networks:

1. An ordered hierarchy. This kind of network map would resemble the organizational flowchart of the US Army. At HQ would be some very general, core principles, mutually consistent: like Kant’s Categorical Imperative or the utilitarian principle of the greatest good for the greatest number. Division commanders would be big principles like “no lying” or “spend government money to reduce suffering.” The footsoldiers would be particular judgments. The chain of command would ideally be clear. Real people might have confused structures, but then we should try to rationalize them. The purpose, for example, of trolley problems is to identify the core principles of people’s ethics so that inconsistencies can be reduced.

2. A random-looking network. In a truly random network, any node has an equal chance of being linked to any other. As in a bell curve, the node with the most links would not be that different from the mean node. Our ethical map would not be truly random, because there are reasons that one moral thought entails another. But the links among concepts and opinions might be distributed so that they were mathematically similar to those in a randomly-generated network.

I doubt that this is good description of morality. David McNaughton and Piers Rawling are correct to say that some ethical concepts are “central.” They are not just more weighty than other concepts (as rape is more weighty than jaywalking). They are also more central in the sense that they turn up more often and we rely on them more for judgments (“Unprincipled Ethics,” in Hooker and Little, eds., Moral Particularism, p. 268.)

3. A scale-free network: This is a mathematical phrase for a network in which just a few nodes have enormous numbers of links and basically hold the whole thing together. Scale-free networks have no “scale” because there’s no typical number of links that can be used to create a scale of popularity on the y-axis. Instead, popularity rises asymptotically according to a “power law.” From wikipedia:

“An example power law graph, being used to demonstrate ranking of popularity. To the right is the long tail, to the left are the few that dominate (also known as the 80-20 rule).”

 

In the case of ethics, we might find that equality, freedom, self-improvement, and compassion were power hubs with enormous numbers of links. Gratitude, fidelity, etc might appear in an important second tier. (I am drawing here on W.D. Ross’s list of prima facie duties.) Not cutting ahead in line would be out on the “long tail” of the distribution, along with reading Tolstoy and smiling at bus drivers.

Empirically, I think we could find out whether people (some or all of them) had scale-free moral network maps in their heads. One method would be to obtain a lot of text in which they reasoned about ethical issues–say, interview transcripts. One would identify and code concepts and connections among them, justifying each addition to the map with a quote. Whether the network is scale-free then becomes a mathematical question.

Philosophically, I like the idea of morality as a scale-free network. It means that some concepts are much more important than others, but everything needn’t rest on a consistent and coherent foundation. The network can be strong even though it accommodates tensions. Further, since there is no foundation, doubting any one premise doesn’t undermine morality as a whole. It just knocks out one hub and the traffic can be redirected. Finally, this metaphor helps us to think about differences in ethical thinking among individuals and among cultures. It’s not that we have incommensurable perspectives, but that our network maps have (somewhat) different hubs. That suggests that dialog is possible even though disagreement should be expected (which sounds to me like the truth).