[Content note: this is another attempt to convey one of those fundamental ideas which I feel strongly about deep down but is still a little hard to communicate, so I once again erred on the side of long and dry. Part 1, hopefully to be continued. Some political examples, especially Trump-related; how can I resist?]
Finally I’ve gotten around to writing the remaining lengthy, cerebral post I’ve been wanting to get out of my system right from the get-go (really, it’s been in my system for a lot of my life). I want to talk about object levels versus meta levels and Theory of Mind and everything that comes with it. I’m worried that this post may become overly long and sprawling because it’s such a far-reaching topic in my view, but at least there’s one thing that makes life a lot easier here: a number of people whose blogs I follow have touched on this directly or indirectly in their writings many times. By pointing attention to such things, they have done a lot of my work for me. Also, I’m going to postpone a few of the ideas I have in mind to be put in a second post.
Here is a list (nowhere near exhaustive) of what I consider to be some of the more crucial posts of Alexander’s which address the general issue of Theory of Mind / Object-Meta Distinction in one way or another:
- How Not To Lose An Argument — a particular use of Theory of Mind in debates, although even more closely related to the value of empathy
- Why I Defend Scoundrels — related to the importance of extending certain courtesies even to those whose object-level beliefs one disagrees with
- Principle of charity and n-step theories of mind — dissecting a debate into several levels and relating these distinctions to Theory of Mind
- Fetal Attraction: Abortion and the Principle of Charity — a direct application of the Theory of Mind to a concrete controversial issue
- A Comment I Posted on “What Would JT Do?” — a very current (in 2013) application of the notion that one should stick to certain standards of “niceness” regardless of object-level beliefs involved
- In Favor of Niceness, Community, and Civilization — on how to win by adhering to intellectual honesty independently of object-level beliefs involved
- The Slate Star Codex Political Spectrum Quiz — a test for being an “object-level thinker” versus a “meta-level thinker” (more on this below)
- What Developmental Milestones Are You Missing? — direct exposition on the Theory of Mind and what goes wrong if it is defective
There are many, many more essays written by Alexander and others which apply these principles without quite so directly acknowledging them. In particular, I’ve seen this from other prominent rationalist community members like Ozy (who runs the blog Thing of Things) as well as from Rob Bensinger, although off the top of my head I can’t produce any links since they both write prolifically in a lot of different places and I don’t have such a good memory for their individual articles and/or comments. This post is my attempt to unify all of these points expressed by them and others into one concept.
But first, here is a series of example scenarios of a variety of flavors in order to motivate the idea.
I. A collection of very short stories
In recent years there have been a number of controversies surrounding high-profile individuals who hold views that are unsavory in some way or other and who were punished for saying those views, by losing their job for example, or just by not being allowed a microphone. “A Comment I Posted on ‘What Would JT Do?'” addresses one of these cases, where Duck Dynasty star Phil Robertson was fired for voicing highly offensive views. In it, Alexander expresses frustration with the network for suspending Robertson, arguing that regardless of what side we’re on, we should adhere to the norm of responding to views we don’t like with counterarguments rather than silencing. Alexander later came to the defense of Brendan Eich when he was fired as CEO of Mozilla for similar reasons. Much more recently, there has been a lot of discussion in the rationalist community about the forceful protests against the very presence of certain alt-right-ish speakers at universities. Most seem to agree that regardless of how one feels about what we might call the “object-level situation” (Robertson or Eich or these speakers’ “object-level” positions that we don’t agree with), we should give priority to certain “meta-level” rules (e.g. allowing the opportunity for proponents of all beliefs to take the podium). Although it’s clearly not quite that simple. Because, waving aside the whole issue of the “free speech” defense being flawed when “freedom of speech” is understood in the most literal sense, there are some individuals, like possibly Milo Yiannopoulos, who have strayed beyond simply expressing their views into outright bullying. There seems to be a fine line between speech that is offensive to some groups and actual threats to the safety of members of those groups. So how exactly do we separate the “object level” from the “meta level” in situations like these?
There has been a particular theme in the debates I’ve (probably foolishly) gotten into with friends over a lot of things relating to the new presidential administration in America. Many are arguing that we right-thinking Americans who are anti-Trump should refuse to acknowledge Mr. Trump as our president altogether. They are more or less saying, as I understand it, that the horrid views he has Trumpeted were sufficient reason for various other authorities to have barred him from becoming president in the first place through some sort of brute force, to have refused to go to his inauguration, and get him impeached as soon as possible. It seems pretty revealing to me that in the midst of some of these “not my president” arguments, the fact that Trump has almost certainly done many highly illegal things is thrown right in with policy positions such as being anti-abortion or (allegedly) anti-gay-rights. While I agree that he’s “not my president” in the sense of not representing anything I stand for, I vehemently oppose the calls for immediate impeachment, as long as it’s motivated by pure principle rather than objective legal reasoning. My main argument has a lot to do with how the other side will view what would look like purely political strong-arming in the highly unlikely event that such efforts actually succeed. I don’t think anyone could completely deny this concern, but apparently I hold unusually strong convictions about the particular importance of considering how other people’s minds will process our behavior.
A few weeks ago I was asked an interesting question by a friend, also pertaining to the American political situation. We were talking about speculations that some Trump campaign officials engaged in illicit communications with Russian agents, thus swinging the election in his favor. My friend put forth the idea that if it is ever proven beyond reasonable doubt that Trump won the election through illegal means, then his executive orders should be considered illegal purely by virtue of the fact that he isn’t the rightful president. I replied that I disagree with this proposal. Trump’s action as president should be evaluated purely on their own merits (legal, moral, etc.), given the fact that he somehow got into the position he’s in. In other words, I want our judgments of his becoming president and each thing he does as president to be evaluated as independently as possible. That way, if we mess up our evaluation of one, this doesn’t affect how we react to the others. Besides, I believe that both the travel ban and the disastrous first attempt at executing it (these two aspects can be judged separately as well!) were despicable and deserving of harsh judgment quite independently of whether Trump’s presidency itself is legitimate, so it just doesn’t seem fitting somehow for Trump to face legal consequences for the travel ban purely on the grounds that something unlawful was done in his presidential campaign months earlier. Besides, again, one should consider what his supporters would make of us punishing him for a multitude of actions using the singular strategy of somehow convincing enough people that he never really got elected.
Now let’s move to personal drama of a sort that I’ve seen play out more times than I can count. Suppose that Alice and Bob are in some kind of close relationship, and Alice gets upset with Bob about something and, let’s say, starts berating him in a tone that somehow goes over the line or with a lot of vulgar language or just generally in a borderline-verbally-abusive way. Bob disagrees with the reasons why Alice is upset but focuses his resentment around the unacceptable way she talks to him when she’s angry. Alice’s rebuttal is to point out that Bob yells at her in an equally unpleasant way when he’s upset with her for any reason, and she gives some past examples to lend evidence to the point. Bob replies that those times were different because for X, Y, and Z reasons, he was right in those arguments and therefore justified in his nasty tone and diction, whereas today she’s wrong in her arguments and thus has no right to talk to him that way. They are — or at least Bob is — conflating two issues here which should be separate discussions: the specific things they get into arguments about, and the way they talk to each other when they get angry about such things.
I know someone who has insisted multiple times that the word “insult” refers not merely to saying nasty things about someone, but to saying nasty things about someone that are unwarranted. I have looked up the definition of the verb “to insult” in multiple dictionaries and have asked several others what they consider it to mean, and all evidence points to this person being wrong about the definition of “insult”. But setting aside explicitly agreed-upon uses of words and the confusion that results from going against them, let’s grant that we can define terms in whatever way we choose as long as we’re consistent about how we use them. To define “insult” as a valid description of a certain unpleasant behavior only as long as it is unjustified given that particular situation weakens one’s ability to separate a personal dispute into two disagreements (the particulars of why they are arguing, and the way they talk to each other when angry) as in the case of Alice and Bob above. Insisting on such a definition of “insult” betrays a certain mindset.
(Interestingly, I was corrected on my use of “flattery” several times when I was younger, because I understood it to mean, well, more or less the opposite of “insult” regardless of sincerity or validity of the claim of the flatterer, while I was told that an effusive compliment doesn’t count as flattery if it’s actually obviously true. This does seem more or less in keeping with dictionary definitions of “flattery”, although it looks slightly different from the “insult” situation situation since “to flatter” is meant to carry a connotation of insincerity.)
II. Separation of degrees
I believe that lying at the heart of all the situations described above there is a fundamental concept in common. Sometimes we might talk about it in terms of “meta levels” and “object levels” (e.g. Alice and Bob have both an object-level disagreement but also have a problem on the meta-level about how they work through disagreements). I’ve developed a habit of using this language quite a lot actually; I’m always telling myself that I’ll look back on this writing one day years from now and cringe thinking it looks sort of rhetorically immature to refer to “object” and “meta” things so often, but right now it still often seems like the best way to make my point.
At other times, we might speak of Theory of Mind as explained in some of the links I gave above (e.g. we have to operate on some consideration of the minds of Trump supporters). I claim — and I hope to argue here at least in an indirect way — that both of these ways of analyzing disagreements point to the same underlying fallacy.
Out of all the rationality-flavored topics that I care about and have been writing essays on, this one lies closest to my heart. I remember first feeling an awareness of the fact that I innately processed certain arguments in seemingly a very different way from the (equally intelligent and much more experienced) people around me at around the age of 12. These disagreements were all of the flavor of the scenarios described above, where my frustration was with those who didn’t seem to realize that there are certain general rules which we all must agree to follow regardless of who is right or wrong in a particular dispute, because all parties are equally convinced that they’re right. And that it’s no good to criticize a person you’re disagreeing with for not following some general rule on the grounds that they’re wrong about specifics when they don’t agree that they’re wrong on the specifics; in fact, it’s bound to further irritate them and push them away. By the start of my teenage years, being bothered by this was already starting to feel like a major hangup that I was almost alone in suffering from, and part of me hoped and expected to outgrow it. Yet here I am. I can’t explain precisely why I’ve always felt as intensely about this as I do, although it’s clearly related in some way to the Principle of Charity, as in Scott Alexander’s framing in some of links above (or to my modified Principle of Empathy).
When I first ran into the rationalist community, perhaps the number one reason I started identifying with the individuals therein was that they all seem to intuitively grasp what I’m getting at here. Sure, some might disagree with how I’m framing it in this essay (maybe because my framing is arguably not the most valid, but more likely due to lack of lucidity in expressing these concepts), but I never fail to feel assured that they get it. Of course, “it” is rarely directly discussed in purely abstract terms rather than in the context of a particular concrete topic. But like I said at the beginning, “it” exists as a thread running through the writing of Alexander, Ozy, and many others.
So is there a way of framing this in more definitive, purposeful language than “there’s some object- vs. meta-level thing or some Theory of Mind stuff going on here”?
Well, let’s start with Scott Alexander’s arguments on seeing issues in terms of object and meta levels in his writing which I linked to above, particularly in the “Slate Star Codex Political Spectrum Quiz”. (Warning to anyone reading this who hasn’t gone to the link yet and is interested in taking the quiz: I’m about to “spoil” it.) Here Alexander posits a series of questions, each of which describes a brief political conundrum and gives two choices as to how to proceed. The catch is that he has cleverly paired the questions into couples which depict scenarios that are very similar on some “meta” level while (very roughly) the roles of “object” level political positions are switched (e.g. a question about a visit by the Dalai Lama being protested by a local Chinese minority is paired with a question about a memorial to southern Civil War veterans being protested by a local African-American minority). The final score on the quiz is computed using a system that gives the quiz-taker one point for answering “the same way” on a pair of questions, thus displaying meta-level consistency. The final evaluation is given as follows:
Score of 0 to 3: You are an Object-Level Thinker. You decide difficult cases by trying to find the solution that makes the side you like win and the side you dislike lose in that particular situation.
Score of 4 to 6: You are a Meta-Level Thinker. You decide difficult cases by trying to find general principles that can be applied evenhandedly regardless of which side you like or dislike.
Many have undoubtedly taken this, along with Alexander’s many other articles which seem to take the “meta-level side” (applying general principles across the board including when he doesn’t like the side whose rights he’s supporting), to imply that he favors meta-level thinking over object-level thinking and that we’re all “supposed to” score a 6 on the quiz. I think I myself interpreted Alexander’s tone this way for a while. Then I realized that this isn’t necessarily the right lesson to take away from it. I can’t speak for Scott Alexander’s exact position here, but I do distinctly recall Rob Bensinger remarking in a different comment section that the Slate Star Codex Political Spectrum Quiz serves as an eloquent rebuttal to the attitude that one should always operate on the meta level. I guess it depends on how one feels about the particular questions asked in the quiz, but I do have to agree that the correct message shouldn’t be to only think on the meta level. Sometimes there are exceptional object-level circumstances which change the meta-level rules slightly. For instance, if our Alice and Bob from above are a married couple who have agreed never to try not to let their voices rise above a certain volume when fighting with each other, then one of them might be justified in bending this meta-level rule just a bit in the fight that ensues after finding out that the other, for instance, just gambled away their entire joint life savings without asking, or has been cheating with seven other partners.
Also — this is a much more superficial objection that is easy to remedy — but of course it doesn’t make sense to consider any conflict to have exactly two levels, the “object” one and the “meta” one, because real conflicts are often complicated enough to involve many degrees of “meta-ness”. For instance, two nations which are run on competing political philosophies (e.g. communism versus capitalism, in this case an object-level disagreement) may try to avoid war with each other in the absence of a particular type of threat or provocation (avoiding force is a meta-level rule), but in the case that they do declare war, they may try to follow international laws pertaining to conduct in war (as in the Geneva Conventions, meta-meta-level rules). And after all, Alexander talks about an indefinite number of “steps” in the above-linked post on an “n-step theory of mind”.
So we should view any disagreement as likely having many layers of meta-ness, like an onion. (One may consider the more “meta” layers as being closer to the center of the onion, but I sort of prefer to think of going outward as one gets more “meta”, since meta-level considerations should be a bit more all-encompassing). And there is no hard-and-fast rule as to some level which will always take precedence over all others in judging any disagreement. Instead, I think the correct message boils down to something even simpler: we should be aware that these different layers of a disagreement exist; and we should address them all separately in our arguments (even if they aren’t entirely independent). For a long time, to myself I’ve been referring to this as “separating levels” or “separating layers” or even “separating degrees of meta”.
Where does Theory of Mind come in? Well, in my experience the general way to fail at the goal I set out above involves disregard for the fact that others’ minds work independently from one’s own. After all, the most common way to conflate these layers is to insist to one’s opponents that what should be uniform meta-rules need only be applied selectively, depending entirely on the object-level situation. And it seems to me that the best way to justify this to oneself is to forget that one’s opponents hold differing convictions on the object-level situation which feel just as genuine as one’s own. That’s basically, by definition, displaying a lack of Theory of Mind.
III. What goes wrong?
When claiming something as a fallacy, I believe it’s always good form to explain why the fallacy leads one astray as well as why people persist in it despite the fact that it leads one astray. (It’s also nice to suggest a positive solution, but in this case, I don’t have any bright ideas beyond the self-evident “that mode of thinking is wrong, so don’t do that thing”.)
When thinking over why I don’t like it when people “conflate layers” of disagreements, I can’t help treating “reasons why this conflation is logically invalid” and “reasons why this conflation is bad rhetoric which will push people away rather than win arguments” as interchangeable. Here are a couple of points which may fit one or both criteria.
1) Defending one’s stance on a meta-level issue using one’s stance on object-level issues won’t actually convince anyone not already on board. If two parties disagree on the object-level issues (which I usually take to be the matter of disagreement which started the conflict in the first place), then for one party to defend their behavior of breaking some meta rule on the grounds that they are right on the object-level issue is a waste of breath. From what I’ve seen (and from what I feel when this is done to me), it only makes the other party more angry and frustrated. A valid argument uses premises that everyone involved agrees on and then uses those to convince one’s opponent of something they didn’t agree about. An attempt at an argument based on a premise that one’s opponent never agreed on is bound to completely fail at accomplishing this.
2) Upholding a principle that belongs to one “layer” of the disagreement only on grounds of being in the right at another “layer” isn’t upholding the principle at all. This can be seen in my second example with the Trump administration, where using the illegitimacy of Trump’s election to indict him for an executive order sort of implicitly excuses the illegality of the order itself. Or, going back to our friends Bob and Alice, if Bob says, “I still think you’re wrong on the issue we were fighting about, but much worse than that, the names you called me are completely unacceptable!”, and Alice points out that Bob calls her similar names from time to time (perhaps even in that same fight), and Bob replies, “But I was justified in talking to you that way because there you were wrong!”… well then Bob is essentially implying that there’s nothing innately bad about calling someone those names at all.
Or to take a slightly more universal example, when a child lies to their parent about having done something wrong, the lesson handed to them is often something along the lines of “The naughty thing you did isn’t nearly as bad as the fact that you lied about it!” But if the child soon afterwards catches their parents themselves lying to avoid getting into trouble for something they did, then justifying it on the grounds of not thinking their crime was actually bad, then there’s a risk of the child coming away very confused about the wrongness of lying. And I’m not saying that there isn’t a circumstance where the parents’ words and actions might still be completely justified — there are some things that are against the (object-level) rules but which may still be morally right and okay to lie about (i.e. these “layers” do sometimes interfere with each other). But a parent in this situation should at least be aware of the confusion that might result when laying down a blanket (meta-level) rule that lying is always wrong even when you’re trying to get out of trouble for doing something you feel was okay.
IV. Why do we go wrong?
I expect one could always cite the usual reason where people are prone to not thinking clearly, and to not having a strong Theory of Mind, especially when this allows for rhetoric which seems to work in their favor in the heat of the moment. As for something more concrete, I think “conflating layers” mainly boils down to one major temptation.
Tying together two different issues in a disagreement allows one to justify oneself based on whichever one is easier to defend. It’s easier to argue against homophobia itself than to argue purely on the meta-level that someone doesn’t deserve a public platform, so many don’t want to make the effort to separate the issue of the unsavory views of Robinson and Eich and their ilk from the issue of whether they have a right to keep their jobs despite their views. If we obtain proof that the Trump campaign actually did clinch the election illegally, it will be easier to convince everyone that Trump isn’t the rightful president than to demonstrate that his travel ban was wrong, so a lot of us would feel inclined to use the illegitimacy of Trump’s presidency to condemn his attempt at the travel ban. It may be easier during a particular argument to defend one’s object-level stance than to defend one’s use of nasty insults, so it’s tempting to define the term itself to depend on one’s rightness or wrongness on the object level.
In other words, while one can’t judge the layers of every argument completely independently, by treating them as all part of one singular issue of controversy it becomes way too easy to get away with all kinds of rhetorical shortcuts, so that one can defend one’s stance throughout the whole onion based only on the most easily justifiable layer. It enables a bait-and-switch behavior which is similar to (or perhaps just a particular flavor of) the motte-and-bailey tactic.
…and actually, I believe all of this can be generalized slightly further, but I’ll save that for another post which (I hope) will appear here soon.