Hello hello readers!
Hm :). It’s one of those weeks where I’m just so happy to be here and alive and living my life. I was sitting in Computational Neuroscience class, and listening to Prof. Olshausen lecture, and thinking: no, no really, how in the world are people like me allowed to exist. Because I just get to LEARN all day. I’m not even expected to do anything except research, which involves thinking about what I’m want to think about all day anyway if given a choice. Moreover, since I’m brand-new in graduate school, I’m not even expected to be productive in research; I’m expected to learn stuff so that I may in the future be productive. Which brings me back to my point about being paid to just sit around and learn things all day. I’m in love with humanity and our society and the fact that we have a system that lets people like me exist, because by all accounts that’s a really, really improbable state of affairs and it’s incredible to be born into a time and a place that lets me inhabit this body, environment, and life.
I suspect this blog is going to be a grab-bag of thoughts as per usual. I wanted to start out with a description of research interests, though, since my mind is on a presentation I’m giving tomorrow. It’s to a public audience: I love presentations to public audiences. The first part is a description of my research interests, and then I describe the project I’m working on. It’s an unusual presentation in that way, since most of the time in talks I dive straight into the research. But since I have this special one prepared… I’ll give the transcript of the first part of my talk here :).
Hi everyone! My name is Monica; I’m a first-year graduate student in Neuroscience, in the Computational Cognitive Science lab led by Prof. Griffiths. Previously I studied color vision by analyzing electrical activity from individual neurons, and also did some neuroimaging work with fMRI. Now I’m lucky enough to study what I find most fascinating, which is social inference. I’m going to give an example on the next slide, and my first question is: how many of you have heard of “Harry Potter and the Methods of Rationality”?
“Harry Potter and the Methods of Rationality” is a fanfiction book, which means it takes characters from an existing novel, in this case Harry Potter. But in this story Harry is very, very rational. I’m going to read you one of my favorite passages.
“‘I’m joking, Professor,’ Harry said with some annoyance. Sheesh, why did she always take everything so seriously –
A slow sinking sensation began to dawn in the pit of Harry’s stomach.
Professor McGonagall looked at Harry with a calm expression. A very, very calm expression. Then a smile was put on. ‘Of course you are, Mr. Potter.’
If Harry had needed to formalise the wordless inference that had just flashed into his mind, it would have come out something like, ‘If I estimate the probability of Professor McGonagall doing what I just saw as the result of carefully controlling herself, versus the probability distribution for all the things she would do naturally if I made a bad joke, then this behavior is significant evidence for her hiding something.’
But what Harry actually thought was, Aw crap.”
I love this passage because it captures social inference so well—it beautifully describes social inference of another’s knowledge—in this case McGonagall’s—to determine Harry’s future action. Moreover, it formalizes the inference process through probability theory, which is not something you usually see in novels. Finally, it gives a description of the intuitive ease, and also the complexity of computation of making this inference. Knowing what other people are thinking is not easy. But we’re exquisitely crafted to do this sort of inference—it’s intuitive to us.
So what I’m generally interested in is anything that gets at the deeper underlying principles of what we’re trying to optimize in doing social inference. Understanding what our loss functions or utility functions are in solving social problems, what are intuitive algorithms* are and what the evolutionary basis is for their development. How we represent social problems, and if we’re optimal in our computational reasoning processes. I’m interested in how these cognitive processes operate in individuals, and also how individuals interact to create emergent group processes.
*(note: algorithms are just the rules you follow to solve a problem. We have lots of algorithms: the algorithm to making a sandwich involves taking out all of the components and assembling, usually in some order)
I think that it’s useful to integrate many fields and approaches in addressing these sorts of problems. If we can understand what our computational goals are in social reasoning, we can use algorithms and mindsets from computer science to try to figure out how to solve social reasoning. Social inference is relevant in artificial intelligence and robotics because we want our robots to be able to understand what our goals are and to be able to interact productively with us as people: we want our robots to be human-compatible. The domain of game theory is relevant; a prominent game in game theory is the Prisoner’s Dilemma, where the interaction of individual goals when interacting in a group is examined. There’s a great rationality / decision-making literature in cognitive science which discusses the biases and heuristics humans have. Social psychology is obviously relevant, and computational cognitive science integrates a lot of these fields.
I care about social inference because I find it intuitively fascinating. But why might other people, like you, care? I think that if we understand how humans reason and interact, we could increase rational-decision making, help people think smarter. There are also groups of people who aren’t as socially capable, like some people on the autism spectrum, and I think that if we were able to describe what computational goals we were fulfilling in social inference, and what strategies were using, that would probably facilitate social communication. We care about social inference with regards to our artificial intelligence, because we want them to interact with humans. And there’s something to be said about changing environments to induce better behavior—we can change the situation to make people behave in better ways.
Those are the sorts of things that I’m interested in. They’re tough questions, and previously social psychology has been done in lab-based experiments…
And from there I get into the specific platform I’m using to address these questions and a specific experiment I’m running in social memory. Man, I didn’t realize how many paragraphs of writing can be done quickly in speech! This part’s about five minutes of talking though, so I should be too surprised it’s long.
(I’m also not surprised I brought in a book quote. When I was in high school, my go-to for every argument was “Ender’s Game”, by Orson Scott Card. When you know a book as well as I knew that one, there are examples for everything buried somewhere in there… and nowadays HPMoR has so many great quotes that I identify with. Sigh. Such an excellent novel…)
Cool cool! I should probably update my website (my personal boring one) with updated research interests, but the language and framing that I use have been changing so quickly since I arrived :). Moving on now: sections are piece-meal and independent from this point forward!
My friend Carson is awesome at explaining stuff. I was recently having dinner with him and he was telling me what a pyramid scheme was, which is something I know is bad and is economic and also something which I never learned the functional definition for. (Is this normal? Maybe it’s an embarrassing omission. There’s a lot of things in my head like that—tagged with categories like “bad corruption type idea” and then never investigated.)
It’s so impressive, explanation done right. I know I bring it up frequently, but I happen to be in an environment where there are a particularly large number of good explainers around, and exciting and efficient delivery of information continues to bring me delight. I often don’t even notice effective communication until afterwards, when I look back on the situation or try to explain it myself, and subsequently draw a blank and post-hoc admiration.
In this case, my favorite part was after he’d explained the general premise in story form, and I jumped in with, “But where does the money come from?” And then he continued playing the character of this story, coming up with some unsatisfactory answer, and I asked again, “But where is the money?”
And then he said: “And you wouldn’t believe the hoops that people go through to avoid answering that question.”
It was just so beautifully presented. I didn’t know, in the beginning, that he was explaining the premise from the point of view of someone who believed in the scheme. He didn’t change his normal explaining voice at all, so I thought he was delivering the description from his perspective. But by presenting the information in the way he was—playing Devil’s Advocate to his own views, really—he got me to ask the obvious questions, made me feel like I was figuring it out, encouraged me to really engage with the problem and find the answer he was guiding me towards. It had the effect of making me as a listener feel very accomplished, and had me really internalize his point after having worked through the problem.
It definitely didn’t feel calculated—I sincerely doubt Carson went into this explanation with “I’m going to play the person involved in this pyramid scheme” in mind before he started in. But that just adds to the grace of it. When explainers have me asking questions, thinking about the material and evaluating it, I both learn information better and feel immeasurably more satisfied and self-confident in doing so. It’s a skill, constructing explanations so that listeners feel like they’re asking smart questions, and one I hope to develop in the future.
I’ve recently been employing a strategy of making my probe-questions very abstract.
Being abstract isn’t exactly unusual for me, but if I’m working through a personal problem I’ll usually start out describing the actual situation. If I do decide to frame a personal problem abstractly, I tend to cave pretty quickly when people ask “but where is this question coming from?”
Today I was experimenting with refusing to cave—or even better, giving a non-personal example that’s relevant enough that people don’t become confused and ask “where is this question coming from?” It turns out that people are willing to entertain abstract questions, as long as you can come up with real-life examples quickly enough to keep it grounded.
My experience so far is that if I can come up with a really good non-personal example, I’m then free to ask all of the follow-up probing questions I want without arousing suspicions. If I fail to come up with a good example, then I have to keep on coming up with weak semi-relevant examples, but if I do it quickly enough then people will sometimes furnish their own examples from my lead.
It produces fun results. When people don’t know where you’re coming from—that you’re asking an abstract question because it has some importance to your life—then they’re much less sensitive in giving answers. Moreover, they draw on more disparate experiences when making responses, because they don’t know what type of personal experiences you want them to target. I kind of like it, because I get more off-the-wall responses compared to what I’m expecting, and people are much more confident in their responses if they don’t think it’s a question that matters. It’s somewhat less useful because the less information you give about a situation, the less context people have to work with, so their answers won’t be as specific.
One of the weirdest parts about earlier today, when I’d constructed a great non-personal example and was performing follow-up questions, was that people totally let me do this. I had a very strong direction that my follow-up questions were targeting, but people were treating this like it was a perfectly normal interaction. In retrospect, in a sense this behavior does describe normal interactions. All of us have some strong biases and opinions that guide our questions. In fact, I suspect that we only generate good and passionate questions under specific circumstances: when we’re trying to figure something out which is personally important, or when we disagree with an explanation.
This was something new I was trying out today, though, so it’s very possible I’m overgeneralizing in my points here. I’m looking forward to collecting future data and figuring out what kinds of probes are best sent forth in abstract non-personal example form. Of course, you poor readers have to deal with me doing this sort of thing all the time—for anonymization, my blog stories are often abstracted to the point of unrecognizability. For sure, I don’t think this sort of questioning will work on everyone. The more non-abstract / grounded a question is, the more I feel qualified in answering it, and that’s sort of the feeling that can make people want to give you their advice :).
Parenting is hard. Families are fascinating. I doubt I’ll study them academically, but families are such incredibly neat social environments. You stick a bunch of people together, and they usually have to cohabit in a small space for years. You have wacked-out power dynamics, so that there are much older people who have a lot of power, and then sometimes you have people closer to your age who have lesser birth-order power. You share resources and attention. The older people have a responsibility to the younger people, though what form that responsibility should take is alarmingly underspecified. The older people often want to influence the younger people, but because the younger people aren’t great thinkers or fully developed yet, the older people can hammer lessons in through force of time rather than reasoned arguments. On the other hand, because the younger people aren’t fully developed yet, there’s far more opportunity to influence them than anyone else the older people could try to influence. People don’t choose their families, and regardless of how people come out they have to all deal with each other. People have their families for life. They exist in these long-term arrangements in which younger people are slowly developing into older people, and everyone is connected for life.
The baseline situation reads like an alien-abduction experiment scene to me. Ooh, look, we captured some humans, let’s try to figure out how they work. What will produce variable and interesting behavior? Hm, let’s see, what kind of pressures and experimental measures could we impose on them…
(…I fear my biases and interests are coming through here :)).
Regardless, families are very complicated microcosms, and if you ever want to tell me about your family dynamics I’d love to hear them. So far I’ve come up with the following opinions: I’m personally way too afraid of the amount of variability that exists to come up with an opinion. I’m amazed at all the parents who every day come up with answers. And I love the resulting environment of an astonishingly large space of familial situations.
So you know how I have social rules, right? Rules originally developed to make sure I didn’t make social mistakes, though nowadays they’re smaller optimizations to just improve upon my social interactions. I do feel like they are rules that everyone follows, but no one’s made a complete book of them that I know, so I’ve been collecting them over the years. This week I have a new one to add!
Because people usually ask me what kind of rules I’m talking about, I’ll give a few examples. Back in the good old middle school days, I had super obvious ones like:
Don’t stare at people.
Addendum: but you still are supposed to make eye contact with people.
DO NOT embarrass people in public. Good luck figuring out what actions involve embarrassing people though.
Try correctly, not just try hard.
Get people to talk about themselves.
(Don’t talk about random stuff.)
People get three checkmarks, then you have to stop being their friend. I mean it.
Don’t accuse people of stealing your stuff.
When people say things you don’t understand and then laugh, it’s definitely about sex. Somehow, even if you can’t figure it out. If you do ask, then you’re going to get incredulous looks. (That never stopped me though :)).
If you do things that are too much how you think and not enough relevant to other people or useful, people are not going to be happy with you.
Don’t use words like “manipulative” when describing others (in this case, to their face)—this is being mean, not honest.
Tact. Figure out what tact is.
Smile at people! Be friendly. People like this.
You have to answer things like “how are you”. You should say it back too.
… Middle school me was a bit behind in the social learning curve :). I look back on a lot of these and just shake my head at myself, because these are very obvious. Well, they seem obvious now. But I know they can’t have been, because I remember the situations that sparked these, and all of the analysis that was involved in developing these rules.
It’s things like this that make me so thankful that I know what I’m doing now. I’ve always wanted to make people like me, but I now know what to say and what to do to make that happen with good probability. Do you understand how special that is? To have a goal to accomplish, and to have strategies that consistently work to accomplish that goal across the myriad people I meet in my life? It’s a superpower, and we can keep improving at it. I actually marvel at this ability so much, probably because I had to work at it, but the fact remains that you can acquire these intuitions, that you can make people like you through how you act.
…(Laughs) Well, that got personal rather quickly. Not like this blog isn’t intensely personal, but it’s always interesting to reveal things from when I didn’t have a good a handle on how to present myself. Then again, I’m sure in later years I’ll look back at this blog and be vaguely horrified at how naïve the sentiments are. I’ve been pleasantly surprised looking back at my 5-years-ago blog posts, but the more time passes the sillier I often find myself :).
Anyhow, these days the rules are more specific, because I make fewer gaffes and more minor mistakes. Rules like the optimum amount of time each person should spend talking in a one-on-one conversation, or the rule about introducing people you’re with, or give-and-take rules for who’s supposed to end an ongoing text conversation. Things that won’t break a conversation, but add a bit of polish, keep people as comfortable as possible.
But I’ve got a new one which I’m happy with, which is always fun! I don’t get to add new ones very often because social situations are complicated and it’s hard to pull out consistencies that can summarized in a sentence or so.
Okay, so I like being correctly predicted, and I see this in others as well. People seem to like when you know them well, and you say something complimentary and accurate about what they’re thinking about a new situation. This seems to indicate that if we happen to have a good model of someone, we should make confident guesses about what they’re thinking, which will make them happy and make us feel closer.
But! People very much don’t like being incorrectly predicted. I can get annoyed or offended if I’m oversimplified, and I know other people can react even more strongly to this than I do. This suggests that we should be fairly tentative in our predictions, because I know I only get annoyed when people are REALLY CONFIDENT when they’re being wrong.
Stalemate. But then I thought of a win-win solution! What happens when you are really confident about predicting something and you’re wrong, but it’s a compliment? Then people feel kind of sheepish and uncomfortable and might correct you, but they’re not going to be offended. And it’s not like you’re usually going to be that wrong in your compliments.
Thus, a new rule for the box: if you have something predictive and positive to say, say it REALLY CONFIDENTLY for maximum person-happiness/closeness benefit. If you have something predictive and possibly-positive to say, say it with tentativeness to minimize offensiveness. And of course if you have something negative to say then don’t say it, but that one’s already in the rules :).
(… interestingly, I’ve always wondered how my people-pleasing tendencies interact with things like giving criticism. I actually don’t have a problem with giving negative feedback, but I’ll do so in specific situations. If I’m in a situation where I’m supposed to give feedback, I have zero problem and like doing so. The other case is if not giving feedback is making me unhappy in a long-term way, and I feel like the other person would be receptive. If neither of those environments is present, I’m unlikely to say anything.
I had a funny case recently where a moral argument was delivered that I SHOULD say something even when I don’t think the other person will be receptive (in this case, sexual harassment cases). It was an entirely foreign argument to me—all of my intuitions were telling me that of course I wouldn’t speak up in that sort of situation, but I wasn’t sure why I wouldn’t. However, under the above description of how I feel about criticism, I think that instinctive response makes sense.
I’m also pretty terrible about arguing (conflict-adverse)— my sister once threw her arms up in exasperation: “Come on, Monica, engage with me here, stop changing the subject.” (Monica about politics: “I don’t want to talk about politics” alternated with drifting out of the room.) On the other hand, I’ve recently begun doing a bit more debate with some of my friends. It’s a context thing—some people aren’t really threatened by having their opinion invalidated. But I only started questioning back when I’d already observed how they reacted when other people argued with them and they changed their minds. Of course, I can always question back when I don’t like an explanation. That’s just making people explain stuff better.
But for now, all of the non-voiced criticism I have gets dumped on the blog in anonymized form :P. Apologies, and PLEASE let me know if I’ve gotten anything wrong or interpreted badly or offended you in some way—I really really appreciate being told I’m wrong so I can do better :)).
I like doing things for the first time. I like volunteering first for things that everyone’s going to have to do, I like the satisfaction from pushing myself to do a new scary thing that I’ve never done before, I like saying I’ve completed something people have been urging me to do, I like having that curiosity quenched. I’m not going to be that big on the SECOND time, unless I’ve liked it enough that I’ve decided it’s going to be a long-term thing, but I enjoy pushing myself and trying things and having an opinion on them and telling myself that I did the difficult thing.
Novelty-seeking, sorta-not-really. It’s rarely for the thrill of it. The logic is almost always: You haven’t done anything new in a while. You need to push yourself. Plus, I’m pretty sure you’ll enjoy this activity if you put yourself out there!
Point being, if it’s inconvenient or scary but I think it’ll be vaguely fun, or even just not-that-bad, you can convince me to do a lot of stuff. I won’t do anything that has the potential for long-term impacts, which gets put squarely into the “bad” category, but this kind of logic is a great way to convince me to do many activities.
I bring this up because I’ve been wondering if I need to try something new and scary involving meeting new people or being overwhelmed with a new domain of knowledge. I’ve been really comfortable recently in the people I’ve been interacting with and the academic topics that are being covered. Then again, it’s not like it was that long ago that I was pushing myself pretty darn hard—I’m still figuring lots of things out, people and research-wise :). It’s just that I don’t have anything immediately scary on the horizon.
Ah well—I’ll keep a lookout, but for now I think I’m okay; there are plenty of directions that I want to learn about and do more in. Moreover, I’m amusing myself because my time-bar is literally on the scale of days—I didn’t do anything scary in the past few days, I don’t have anything scary planned for the next week. It’s a good life when opportunities are so available, when I can cycle between overwhelmed and comfortable, back and forth :).
ARGGG how it is bedtime already. I wanted to talk about the Dementors. Ah well, I’ll save it for next week—I unfortunately need to be responsible tonight since I’m giving my presentation tomorrow. Whee, excited :).
Anyhow, I hope you all had marvelous Thanksgivings if you’re American and happy days regardless! I wanted to give great thanks to my extended family for hosting me for a few days—they take such good care of me, and I feel very lucky to spend time with them :).
Best wishes and thank you as always for reading!