Signaling: Why People Have Conversations

by liam_hinzman3 min read20th May 202021 comments

15

SignalingWorld Modeling
Frontpage

How would you convince somebody that you’re an expert in quantum mechanics?

If you’re not an expert in this field, you shouldn’t be able to effectively apply any method you can imagine. If you could, then it wouldn’t be a useful signal.

If I told you about Schrodinger’s cat to you right now, you probably wouldn’t think I’m a quantum physicist. If this wasn’t the case anybody could Google and trick you into thinking they more than they do.

Signals are actions that demonstrate evolutionarily desirable traits or possessions, such as physical fitness, material wealth, or knowledge of quantum mechanics - which is seen as a signal of intelligence, a trait that is desirable to potential allies and mates.

For a signal to be seen as honest, it has to be costly and public.

Effective signals are easy to do if they are honest and reflect traits or material you possess, and costly if not. If I actually was an expert in quantum mechanics, then showing you research papers that I’ve written would be a good signal.

Signals have to be public to benefit the one signaling. Donating $10,000 to fight breast cancer won’t convince anybody of your empathetic and pro-social nature unless people know about it.

Signaling is a primary motive behind many of our behaviours such as:

  • Conversation: talking with others lets us signal our utility as a potential ally.
  • Charity: donating shows that you have an excess of wealth, and that you’re somebody who will help their allies.
  • Consumption: sharing that you went on an expensive vacation signals to others that you have an abundance of material wealth.

Conversations are Opportunities to Signal

(This view of signaling and conversation is mostly a summary of a chapter from The Elephant in the Brain by Hanson and Simler. For a more in-depth treatment of signaling and how it effects human behaviour I'd highly recommend reading it)

Without signaling, the usual explanation for why people have conversations is that it serves as a way to trade information. I tell you about the nearby berry bush, you tell me about the dangerous lion to the West. We cooperate through conversation, and we’re better off for it.

There are a few problems with this line of reasoning.

People like talking more than listening. If we participated in conversations primarily to gain information, then you’d expect people to want to listen more than they speak. People almost always show the opposite behaviour.

Most people don’t keep track of any sort of conversational debt. If one of your friends teaches you a lot about quantum mechanics for hours, you might feel grateful, but I doubt that you’d experience an overwhelming need to repay any informational “debt”.

Relevance is important in conversations. If somebody told you that their farm was attacked by foxes, and you immediately responded by telling them about your recipe for tacos, you’d get a strange stare. Instead, maintaining relevance and conversational flow is always important. This doesn’t seem to follow from the idea that people have conversations to trade information.

But through the lens of signaling, all these pieces of etiquette and conversational quirks make sense.

People have conversations to signal their utility as a potential ally and mate. Rather than conversation being a way to cooperatively share information, this suggests that conversation is largely a selfish behaviour. People talk to signal to others their material wealth, knowledge, and fitness.

This explains why people like talking more than listening. Every time you speak you’re getting a chance to show the other people how great you are. If somebody mentions that they’re building a new house, you’ll be excited to show any knowledge you have on construction or design.

People don’t track conversational debts. Instead, people often compete to talk as much they can. You don’t care if you’re helping somebody build a house and getting nothing in return. Them knowing that you were the one who helped them, demonstrating your value as a potential ally, is a large reward.

Maintaining conversational flow is important. It would be too easy to throw random pieces of information out there otherwise. If you happen to know a lot about whatever just came up in conversation, that’s a signal to others that you probably know about a whole lot of other things as well.

So why did I write this post?

I’d like to think it was primarily because of my curiosity. Signaling has a lot of explanatory power when analyzing why people do certain things. But it also serves as an (attempted) signal of my intelligence.

Hey, look at me! I’m the type of person who spends time learning and writing about human behaviour. I hope you think of me as a good potential ally.

15

21 comments, sorted by Highlighting new comments since Today at 4:20 AM
New Comment

Useful, but I do want to warn about oversimplifying the topic. Conversations _do_ contain signals, and that may be a plurality or even a majority of the time spent. But they are not simple signals like "unidimensional fitness for cooperation". There are dozens or hundreds of signaling dimensions, and often a significant amount of actual object information, to actually communicate concepts and align beliefs, rather than just to signal fitness and capability.

How would you convince somebody that you’re an expert in quantum mechanics?

There's a LOT of unpacking to do here. what does "convince" mean? What does "expert" mean? What mutual knowledge do we have of each other? In fact, relative to many, I _AM_ an expert on quantum mechanics. Relative to time travelers, nobody born in the past is. Most importantly, why do I care whether they're convinced?

All this goes to the next step in your writeup - signaling generally has multiple levels of complexity, and they all have to land in order to be effective. Good signals are a cluster of costs on different dimensions of proxy for the underlying characteristic, and faking one is ineffective. Which leads to https://en.wikipedia.org/wiki/Countersignaling, where an entity does something that would be a bad signal out of context, but combined with other signals is actually an even more impactful costly display.

So, how did I do at the game specified at the end of your post?

I really appreciate the feedback! Agreed with all your points, there's a lot of areas I need to work on to improve my writing.

Signal-wise you come across as what I think of as the good side LW: well reasoned, thoughtful, and intelligent.

Sidenote: I think people (like you) who comment thoughtfully on other people's "content" make the internet a much better place.

I'd also be good to cite the source here, as pretty much the whole argument is copied from it: The elephant in the brain, by Hanson & Simler.

Will add, just finished reading that book. This post was my effort to solidify my main takeaway from The Elephant in the Brain for myself.

The signaling frontier moves because non-moving targets get goodharted. This is experienced as a rotation of platforms (eg social media) which start off high signal to noise and become more and more noisy over time. After all, noise you can filter for effortlessly isn't actually a problem. It would be interesting to see if such frontiers could be modeled as cycling the way game theoretic strategies do e.g. if you have a a network over saturated in one strategy it incentivizes switching to a strategy that exploits that one, until that one gets saturated etc.

Can you talk more about the movement of signaling frontiers? I'd be super appreciative of an example if possible. I assume your mention of Goodharting is the idea that as soon as something becomes legible as a reliable signal of a quality, it'll be optimized for and cease being reliable. This is the movement of the signaling frontier, I take it?

I've read through the papers you recommended in a previous comment, which I incorporated into The Dark Miracle of Optics, but I'd love to continue this conversation with you. Is there somewhere—your own post, or elsewhere—I should explore re: this moving frontier, and the constant Goodhart-led inflation of signals?

The easiest way to think about it is the creation of new platforms: radio, tv, internet, social media. Each develops its own vocabulary which is initially extremely effective due to the lack of anti-bodies in the general population, leading to rapid spread and fixation in the population (like herpes simplex). Banner ads->google ad words->instagram influencing etc.

I'm glad you linked your post, I check LW intermittently and missed it.

WRT what exists out there: AFAIK this is pretty close to the frontier and not much game theory etc work has been done yet. The airforce has funded some agent based modeling recently.

Looking at fashion might yield a lot of this pattern. You first have high class people dress a certain way. Afterwards lower class people copy the way of dressing and it stops being a high class signal. 

Yeah! That was my thought as well. Unfortunately, despite combing through fashion theory, there's not much literature on the subject; I've had to make a lot of it up as I go. I wrote a bit in the essay linked above about it:

In the Upper-Middle Paleolithic Transition, human societies and economies grow increasingly complex. Trade deals and diplomacy are performed among credible spokesmen, and social hierarchies need preservation across interactions between strangers. Fashion enters as a technology for maintaining and navigating the social graph. “By the production of symbolic artefacts that signified different social groups and kinds of relationships, Aurignacian people were able to maintain wider networks that could exist even between people who had never set eyes on each other,” giving them a competitive advantage. The practice spreads through the law of cultural evolution: “The surface of the body… becomes the symbolic stage upon which the drama of socialisation is enacted, and body adornment… becomes the language through which it was expressed.”[5] We have entered the second stage of simulacra. The territory has a map, and there are parties interested in manipulating it.
Once this association between optics and essence, between appearance and reality, between signal and quality (the biological frame) or public and private information (the economic one), is formed, it can be freeridden. It becomes, in most cases, easier to pay “lip service”—to outwardly express the associated public characteristic—than it is to to develop the private characteristic. This is not entirely the fault of the freerider; it is a difficult situation he finds himself in. Imagine he “chooses” (I’m anthropomorphizing evolution) to remain with his blue and yellow colors: even if his “product” is “good” (I’m mixing metaphors, but I mean to say, his advertising is honest), it will take some time for a trusted association between signal and quality, public and private, to form. As consumers, we may initially disbelieve an advertiser’s claims, and for good reason, since there is incentive to deceive. And thus it is with the sun-basking lizard, deciding which butterfly to eat. Far easier for a precarious insect to ride coattails, to imitate and pretend toward what he is not—and so, quite simply, it does.
The connection with fashion should come into view now. The “barberpole” metaphor of fashion, where lower classes continually imitate higher classes, who are themselves engaged in a continual quest for “distinction” from the chasing masses, is a popular one in rationalist circles for good reason. Its cyclical nature is the result of limited options and a continual evasion of freeriders who exploit an associative proxy: clothing for caste.

Recently I've been considering the frame of a discoordination game, in which it is one actor's interest to synchronize with the other actor, and in the other actor's interest to stay de-synchronized ("distinguished" or "distinct" a la Bourdieu).

I want to pick at Simler & Hanson's "relevance constraint" as meaningful evidence of a signaling thesis. There is a much simpler explanation for why conversations travel along throughlines of pertinence, which first must be dealt with, and accorded causational influence, before we get carried away with signaling: human cognition is fundamentally associative (see not just Lakoff & Hofstadter but William James). Our thought bounces from one relation to the next; this much is self-evidently clear. Why would two individuals performing cognition jointly differ significantly in this pattern? And what then is left over for the signaling explanation? Indeed, when one considers how loose the “relevance constraint” is in the first place—it really requires only one coherent point of departure to switch subjects—their argument appears all the weaker.

I agree with some of the other comments: one has to account for what type of conversation one is having, and with whom. Imagine a loving father and his daughter - their conversation wouldn't necessarily be dominated by the need to signal.

More in general, I believe that signalling could be better understood if seen in the context of polyvagal theory. The nervous systems has different "gears" depending on whether one feels oneself to belong, be safe and esteemed, or whether the environment is more indifferent or hostile towards oneself [see also the sociometer theory of self-esteem]. When the so-called social engagement system is running, we don't feel the need to one-up each other the same way we do when the fight-flight system is running. To me it seems that a lot of neo-darwinian evolutionary theory has lost sight of this nuance.

I think I know why my conversations are different from most people's now.

"People talk to Signal" is a bold claim.

Have you considered if your proposal is a testable hypothesis , or if it is speculation / opinion? If it is testable, have you considered testing? Are you trying to disprove the position?

Without signaling, the usual explanation for why people have conversations is that it serves as a way to trade information.

Is this the usual explanation? Even if it is for some theories, It sounds awfully like a straw man – an easily defeated explanation that will bolster any other.

There are a few problems with this line of reasoning.

You make three assertions about conversations here. One is "People like talking more than listening" – this doesn't hold at all from my personal experience of most people. How have you collected the data to make these assertions? Could your data have bias? Is the data robust? Are you interpreting fairly? Might other explanations fit observations?

I find that attempts to reduce everything down to signalling are usually overly simplistic, and in many cases gives you an incorrect view of peoples motivations.

What is missing here is the fact that conversations are, for most people, an enjoyable activity. You get the fuzzy chemical from being around other people, learning things from them and being able to teach them things, sharing interesting stories and perspectives, and bonding over shared experiences. Now, I'm not denying that people often want to impress the other party, but to state that this is the primary motivation doesn't really make sense. Most of my conversations are with people who I am already good friends with, for example. I listen to them because I care about them and want to help them feel better.

I could make a similar post stating that people play videogames in order to signal their ability to be skilled and focused. I'm sure that plays a part in the motivation, but the primary reason is that it's fun. The same goes for conversations.

My personal theory is that not all talk is signalling, but almost all talk about signalling is. (It signals "I am smart, sophisticated, not easily fooled, and willing to face uncomfortable realities; I see below the carefully groomed surface of things to the ugliness beneath.")

In the particularly prominent case of Literal Robin Hanson, it seems possibly significant that the uncomfortable realities he uncovers are generally much more uncomfortable for one of the two major political factions in the US than for the other, and that the "other" one is the one responsible for a lot of his funding over the years (though I think he may no longer be affiliated with the Mercatus Center now?).

(Only possibly significant, and I do actually mean that. Obviously things that are politically convenient for the person saying them can also be true.)

In this situation I like the model of player vs character. In Dungeons and Dragons you create a character sheet with abilities, stats, and motivations. This limits your options and creates preferences for certain actions but as a player you still have choice, and can still do things that are contrary to you "character sheet", it's just less likely that you will do so

I think that evolutionary psychology, specifically signaling in this case, is the reason why people enjoy conversations and it acts as our character sheet - shaping our general preferences. We as players often have different motivations for having conversations, but in aggregate the character sheet has a lot of explanatory power even if we're not consciously aware of it.

The problem is the shift from saying why a behaviour may have evolved, to saying why people do it now.

Why do people enjoy videogames? well, our brains give off the good chemicals when we overcome challenges and succeeded in difficult tasks. This was likely a result of evolution trying to incentive's hunting and gathering and so on, so we could do it more often and provide more food for our offspring. But does that mean that the primary motivation for playing videogames is to provide food for offspring? Obviously not.

So yes, it's possible that people evolved to enjoy social bonding because in pre-historic times it helped to form allies. But we live in a vastly different context from hunter-gatherer days. Now we do it for the pure enjoyment, not for the original reason that the enjoyment evolved.

This is my primary concern with evopsych in general, to be honest. Looking at evolutionary pressure from the distant past will only tell us how people thought in the distant past, it doesn't tell us how our minds have adapted to our changing context. If we want to know how people think now, we can study how they think now.

People like talking more than listening. If we participated in conversations primarily to gain information, then you’d expect people to want to listen more than they speak. People almost always show the opposite behaviour.

This seems to shift the line a bit. If the goal of conversations exchange of information I don't think we can really say who will do the most talking so should not try to draw any conclusion about conversation as info exchange based on people liking to talk more than listen. (It is perhaps interesting that is seems a lot of communication theory agrees that people who don't actively listen tend not to be good at communication or conversation.) That said, I suspect one might use the signaling view to explain why more want to talk than listen. Knowledge is power. Power is status. People like status. Therefore people want to talk more than listen.

But that behavior is fully consistent with the idea conversation is largely about info exchange.

I did enjoy reading and appreciate that your wrote up your thoughts and summaries from reading the book. So don't take my selecting one point to question as a signal otherwise ;-)

I think that this is one reason why people have conversations, but there are many others.

Things like:

-Information trading

-Coordinating

-Manipulating

-etc...

However, I like that you pointed out that relevance is important in conversations. That's something I find myself taking for granted but is actually kind of weird when you think about it. I don't think signalling fully explains why relevance is necessary. I'll put forth an alternative hypothesis:

I think conversations having a requirement for being relevant is a consequence of language being an efficient code for communicating information that also efficiently uses and changes working memory. When you talk about something, you often need a good deal of background context to get these ideas across. This can manifest itself at the lowest levels of abstraction as the need to decipher homonyms and homophones, and at the highest level when understanding why someone would want to "blow up the plane". Keeping track of this context eats up working memory. If you switch topics frequently, you'll either have to keep track of multiple smaller contexts, or waste your time wiping and repopulating working memory after every context switch. However, by tackling one topic at a time and moving smoothly between related topics, all working memory can be devoted to keeping track of the conversation, and you only need to partially recontextualize infrequently.

I think people information trading, and coordinating are good reasons for why humans evolved language, but I think that signaling gives a stronger explanation for why "casual" conversations happen so often.

Why do you think the signaling interpretation doesn't fully explain why relevance is necessary? Your hypothesis, norms for language evolving for efficiency, makes sense to me but doesn't strike me as being a more important factor than signaling.

I think people information trading, and coordinating are good reasons for why humans evolved language, but I think that signalling gives a stronger explanation for why "casual" conversations happen so often.

That sounds reasonable. I still think there's more going on in casual conversation than signalling, as evidenced by signalling in conversation getting called out as "bragging" or "humble bragging" or "flexing", indicating that people would like you to do less signalling and more of whatever else casual conversation is used for.

Why do you think the signalling interpretation doesn't fully explain why relevance is necessary?

I think it the best argument against signalling fully explaining relevance is that there are situations where signalling is pointless or impossible, this happens between people who know each other very well as any attempt to signal in those cases would either be pointless or immediately called out. However, relevance is almost a universal property of all conversation and the norm rarely if ever breaks down. (Unless you're dealing with people who are really high, but I would explain this as a consequence of these people no longer being able to keep track of context even if they wanted to.)