Well, welcome colleagues and friends to the Psychological Humanities and Ethics lecture series where we hope to think a little bit more humanely and practice a little bit more ethically. As we face others i'm Mookie Manalili and i'll be your host tonight with our special guests, Dr Liane thanks for joining us tonight again thanks Mookie it's nice to be here. yeah i'm a brief introduction for Liane, she is an Associate Professor of Psychology and Neuroscience at Boston College and director of the Morality Lab. She received her BA in philosophy and PhD in psychology from Harvard before doing postdoctoral research at MIT and the brain and cognitive sciences department. Her lab investigates human moral psychology using tools of social psychology and neuroscience like functional magnetic resonance imaging. A recent projects focus on how people engage in reasoning for moral thought and action how people think and learn about others across the social relational context. And how people apply principles of obligation and impartiality, the answer, it has received widespread acclaim and support so to put you on the slide including. The Stanton and William James prizes for the society of philosophy and psychology. Her lab’s work has been published in academic journals such as the Proceedings of National Academy of Sciences and Psychological Sciences. And news outlets, such as New York Times to learn more feel free to visit moralitylab.bc.edu. Additionally, on a more personal note, as a psychotherapist hoping to pursue further research in philosophical psychology Liane has been a very thoughtfully brilliant and inspiring mentor in moral psychology field. Liane, would you mind briefly sharing a little bit with the audience of what personally interested you in this field. yeah yeah Thank you Mookie for that introduction so as Mookie mentioned, I majored in philosophy as an undergrad, and the reason why was that I was really interested in. How people solve moral dilemmas, and why it is that we can have very confident intuitions about moral problems and yet disagree with people around us, including close others and so. I ended up majoring in philosophy, to be able to understand better how it is that philosophers come up with solutions to moral dilemmas and. How they're able to organize intuitions about moral cases, you know when and why is it permissible to harm one deceive many. And what are the principles that underlie our intuitions about these kinds of cases in moral scenarios and dilemmas, in particular so so my interested in psychology really came from. Those questions in philosophy and then i'm very interested in. The role of intentions in in moral intuitions and moral judgments. When intentions matter for a moral judgments how attendance make the difference between murder and manslaughter, for instance, other cases in which intention seem to matter less and so we'll talk a little bit about. How More generally, we think about other people's minds and how those processes play into our moral psychology. Thanks so much Liane yeah I know that Liane will be sharing with us, you know the wide breadth of the research and yeah in this particular lecture Dr Liane will. be discussing actually noted the field of moral psychology aspects of theory of mind and reasoning and moral judgments informed by social contexts. And another place, she noted that you know what makes morality unique is the experience of moral judgment as a flash of intuition. or a feeling of good versus bad, but underneath these kind of feelings are a complex psychological structures and now I think even my work with. Patients and other people as they kind of sort through the lived experiences, yes, you know it's a social creatures, you know we spend a lot of time thinking about mental lives of those around us, you know why did a person do this or why is my kid crying at this moment right now yeah. I know it's great. But yes so Liane will explore these broad questions and also in very focused and very insightful research on how social context, my shape these processes. From a social neuroscience perspective for those in the audience, if you have any questions at all feel free to post them in the question and answer, and we have noted that she'd be happy to take it in between. The slides and, of course, will have a more dialectical space in the end, as well, but without further ado, we. All right, thank you Mickey and yeah I don't know whether i'll be able to track the Q-and-A a once I start my screen share but. Mookie I think we'll be keeping an eye out for people's questions that i'm happy to pause at any point to answer questions as we go along alright, so let me start the screen share. There they are. Great can everyone see my screen. Yes. Great so yeah again Mookie think thanks for the introduction. I can't see everyone's faces and fortunately I saw a few familiar names filter in through the participant list, which was. Nice, so I hope that we can have a little bit of a dialogue as Mookie mentioned through the Q-and-A throughout and, at the end, but i'm really. happy to be here with all of you as Mookie mentioned the work that all present today focuses on what our lab. explores within the domain of moral psychology but i'll present as some of our lives newer work much of it is still in progress so i'll be excited to hear. folks feedback on on the work at the broadest level of the talk today will focus on how social conduct shapes social cognition how we think about other people how we think about their mental lives. depending on who they are, and our relationship to them, and we can catch this intuition that social context determines how we think about and learn about others in many instances, so this is an example from. My own experiences in academia, when we get feedback on our work, the source matters. So what I mean by that is whether the constructive criticism comes from our friendly colleague or an antagonistic reviewer. influences how we interpret that feedback and how we consider the underlying intentions and motives of the critic. So in the talk today we'll look specifically at how social context influences this process by which we think about mental states, and I will refer to this process as mentalizing or theory of mind. Liane, is it always reviewer two that gives the critical feedback. Yes, it is. That is the rule. At least reviewer two, sometimes more than review two that's right. yeah. So some of the questions that I want to explore today. Are the following so do we focus more of our attention on the minds of those we trust and respect our friends and colleagues, or those we feel we're at odds with. Do we think about these different minds in different ways, how do we tell them apart and can answering these questions, tell us about some of the key features of mental state representation, and then there are also instances in which. The people around us friends colleagues and as Mookie mentioned reviewer two. may surprise us acting in ways that are maybe inconsistent with our impressions of them and, again, who these people are in the strength and valance of our prior impressions. matter so if you hear a report of a close friend versus a stranger taking money from a tip jar you might discount or try to explain away that new bit of information, maybe your friend was making change for $1. And so, when and how do we update our representations of other minds, how does theory mind support impression updating. And then, finally, the context for other people's actions also matters for our evaluations of their actions, how do we judge a Good Samaritan who helps the stranger versus someone who helps their friend or family member. So, in the first part of the talk we'll look at how people deploy theory of mind across social and moral contexts. will look at how people think about competitive versus cooperative partners and agents with harmful or hurtful intentions in the second part of the talk will look at how people. How people update their judgments of others across contexts harmful and hurtful agents strangers and friends. will consider the underlying mechanisms and also the consequences of asymmetric and maybe biased updating across context. And then, finally, you will turn to how people judge other people who favor close versus just and others kin versus strangers, and vice versa, so first How do people think about the thoughts of others across competitive and collaborative contexts. So fortunately there's already quite a lot of relevant literature here. First, extensive research on outgroup dehumanization suggests that theory of mind may be deployed to a lesser extent during competition, then. Cooperation so people attribute less mind to out group versus in group members, for instance, even pictures of extreme out group members elicits. Reduced activation and brain regions for social cognition like the medial prefrontal cortex or or the mPFC right there, and on the flip side, the more people like and want to affiliate with an individual, the more mind, they will attribute to that individual. Different prediction is enhanced theory of mind for competition vs cooperation and here work on the evolutionary origins of. theory of mind reveals that rudimentary theory of mind capacities and non human primates are driven by the need to. outcompete conspecifics over scarce resources like food and adult behavioral work also shows that theory of mind is especially triggered by the need to understand agents' bad behavior. A third prediction is that theory of mind is robustly engaged for both cooperation and competition, including the need for strategic understanding and also the need for social. connection, so people may be motivated to predict others actions to outsmart enemies and plan attacks, but people are also motivated to. Understand others to create and maintain alliances and some work that my colleague Adam Waytz and I have done, suggesting that people might focus on. different components of mental states, depending on their motivational context and perspective so for predicting someone's action it's really important to understand their. agentive mental states, but for affiliation or social connection it's more important to understand somebody's experience. and other work that Adam and I have done with our colleague Jeremy Ginges shows certain biases and how people attribute motives across group boundaries. To in groups versus out groups and so based on this work be hypothesized that theory of mind and mental state inference would be widely deployed across both cooperation and competition. And again for some of our clinical friends Liane, the theory of mine, what would be a good succinct way to put put that and maybe what kind of regions, does that recruit particularly the RTPJ. yeah i'm about to get to this on this slide. In just a moment. yeah so i'm just hypothesis about. theory of mind across these different social context we looked at as Mookie anticipated brain regions that support this set of cognitive processes so typically we use. what's called a functional localizer test to identify these brain regions of interest, or ROI, for this cognitive capacity within individual participants and the task that we use. contrasts 10 stories describing mental states like false beliefs with 10 stories describing physical states like outdated maps or photographs in one false belief story, a character. named Sally. put a ball in the basket and left the room. Anne came into the room and removed the ball from the basket to the box and left the room. And the question for our subjects is where does Sally think her ball is. Meanwhile, in the false photo version of the story, the question is where is an object in an outdated photo and so, for this contrast we typically see a consistent network of brain regions, including as Mookie again anticipated the bilateral temporoparietal junction or the right. And left TPJ, the precuneus or PC, and the dorsal medial prefrontal cortex or DMPFC. And so, these are the brain regions that are recruited more for processing information about mental states again like false beliefs versus physical states like outdated mouse or photos. And they make up what we call this mentalizing or theory of mind brain network. And once we identify these regions, we can then examine how they are recruited, how they're activated for cooperation and competition, and also whether these regions encode information that distinguishes between these fundamental social content. So the next few studies I'll tell you about our work by my former Grad student Lily Tsoi who's a postdoc with Diana Tamir at Princeton and also interviewing for. faculty positions, right now, which is very exciting, so we scan subjects as they played a novel dyadic game modeled after rock paper scissors. As i'm sure many of you are familiar with the point of rock paper scissors is to predict what the other person will choose and then to make your own choice accordingly in a strictly competitive fashion. Our game of shapes included competitive and cooperative trials monitoring settings across all trials and the choice of shapes. So players were given a choice of a circle or a square and they were instructed to match or mismatch the other player. And we operationalize cooperation and competition in terms of players goals and rewards as i'll show you next so on. cooperative trials both players have the same goal both players have to match for both players have to mismatch if they achieve their goal, they both went $1 on that trial. On competitive trials players have opposing goals, so if one player has to match the other player has to mismatch and only one player can win on that round in a zero sum fashion, so if I were. A subject I might first see, I can't tell if you can see the highlighted green cooperate screen, but I might see a cooperate screen telling me that trial type to cooperate. and your goal, our goal is to match each other so imagine Mookie is the person that i'm playing in this game, I see. That my goal is to match Mookie Mookie’s goal is to match me. So if I predict Mookie is going to choose square, then I will also square so that then we can match each other and when $1 on that trial on competitive trials my goals different from Mookie’s goal, so my goal is to mismatch him his goal is to match me so here. It looks like I predicted that Mookie would choose circle, I wanted to mismatch him, so I chose square but he chose Square and his goal is to match me so he would win. $1 on that on that trial, so what i'm showing you here are what we're calling the active trials, the active interactions where players outcomes are determined by their active choices in this shapes game on passive trials players were still either pitted against each other or aligned. In terms of their goals and rewards, so the goal and path structures of the game are the same, but this time outcomes are determined by the computer so there's no need to predict what the other player is going to choose. As some of you might have guessed in fact. All outcomes across both kinds of trailer types are are in fact determined by the computer so it's entered into the skin session they meet a confederate who is gender match during the consenting process, they think that they're going to be playing this other person in the game. But we fix the earnings such that subjects would win on exactly half the trials and Lily also ran a purely behavioral version of this task outside of the scanner where both players were in fact actual subjects to ensure that their response patterns looked approximately. The same when the outcomes weren't fixed which they did. So first and unsurprisingly, we see that theory of mind regions were recruited more for the active interactions and the passive interactions. And what i'm showing you here in this picture is overlap in the in the whole brain between the theory of mind. localizer contrast that's regions recruited more for false beliefs over region for it for false photos and the active versus passive contrast. So what this boils down to is that subjects are more motivated to consider what the other players thinking when they need to predict what the other player is going to do, which makes sense. In the context of this task. Second theory of mind regions were recruited similarly for cooperation and competition here I’m showing you. These line graphs for the active trials, but the same was also true of the of the path of trials as well, and the takeaway here is that theory of mind is required for action on prediction across both of these contexts. Both cooperation and competition right? Yes, that's right that's right, so we see that. Both of these contexts require people to think about other people's mental states. Third, we want to ask, even though we see similar overall levels of activation is there any sense in which these regions can encode the difference between these. fundamental social contexts, can they encode on information that separates these contexts so here we used. A technique to analyze the imaging data called multi voxel pattern analysis or MVPA to examine whether the spatial patterns of activity for cooperative and competitive trials reliably distinct from each other. In theory mind regions so, as I said before, instead of averaging overall voxels within a given brain region to look at the overall pattern or the overall magnitude. Of the of the brain signals shown in the line graphs here we looked at the spatial pattern of voxelwise activity within each of these regions and ask him, these special patterns in voxels be classified about chances belonging to either one condition over the other. Okay, so for passive trials classification accuracy was that chance and all for theory of mind arise, that means that the spatial patterns look no different for cooperative and competitive trials, but for active trials, we found above chance. Discrimination across these regions within this network, so, in other words, these regions across the board, seem to encode information that separates active cooperation from active competition. So again, we see that theory mind is deployed similarly robustly, as in those line plots for action prediction in cooperative and competitive context, but in the bar graph and we see that. their including some information that separates these contexts. So, on the one hand. Actual incentivize interactions are especially salient; they might have provided us with a good opportunity to detect any difference. Between the social context, but on the other hand side because they're interacting with the same person across many trials, the same task predicting the same thing. What shape is my partner going to choose, so this could have been a fairly conservative test of the difference between cooperation and competition so Lily ran a follow up experiment which was different in a number of ways in this experiment subjects read about. Different hopeful and helpful agents from a third person perspective, and we can again ask do we see differences in the overall level of theory of mind and also do we still see that same. neural pattern discrimination, and if so. can that shed light on the the nature of the information that separates the conditions and the shapes tasks. Cooperation. So remember the task for localizing brain regions for theory of mind here in these pictures note that and is a relatively neutral moral agent in shape of in spite of the shape of her eyebrows. We could pose that and had an explicitly. Harmful intent and malicious yeah. And Sally by moving the ball or let's say she had a helpful intent and wanted to help Sally by putting the ball in the right place. So again, we can ask are brain regions for theory of mind recruited for mean-Anne or nice-Anne and also again can these brain regions tell the difference. cases. And here we also included a neutral condition as well, so simply spread 30 stories like this one here i'll show you just one more to give you a sense of the space of are stimulated beyond this classic Sally-Anne story. So here Roxanne went to bed leaving an unfinished puzzle living room so that she could solve it, the next day Max saw the puzzle and solved it that night Max wanted to show Roxanne that he. beat her to some of the puzzle Max wanted to help Roxanne solve the puzzle, and then in the neutral case Max wanted to finish the puzzle, even if it took the whole night. So, normally, as you can see, this true false question that's under are answering is exactly the same across all three conditions of the task participants are tasked with protecting the characters mental state what will the character think. So interestingly brain regions for theory of mind are recruited more robustly for harmful versus helpful agents, in contrast to the general pattern that we found in the sheets task, and when the critical mental state information is. presented for the most part, these regions again discriminate between harmful and hurtful conditions of this task. In their facial pattern of activity again using that same MVPA multi-voxel pattern analysis approach that we used in the previous shapes experiment. So, again in in the previous experiment, we saw that theory of mind reason code information separating cooperation competition. And this new vignette based experiment provides more direct evidence that the relevant difference, maybe between competitive and cooperative intent or helpful and harmful intent. we've also looked back at some of our older experiments focused on rural centers as well, and we see a similar pattern, but in a different context, comparing. Harmful intent in the case of intentional harm with neutral intent in the case of accidents so again using MVPA we find the same. pattern of neural discrimination between harmful and neutral intent in the RTPJ that right temporoparietal junction region. And importantly the extent to which this brain region discriminates in its spatial patterns between harmful in neutral intent correlates with the extent to which. People do so in their in their behavioral responses in their moral judgments, so if you're a subject who sees a very big difference between intentional or accidental harms on your RTPJ sees a big difference in it's spatial patterns of activity. So now we've seen three instances in which brain regions for theory of mind discriminate between uncooperative or helpful intent. and competitive or hurtful intent revealing a key feature dimension of mental state representation and a crucial function of theory of mind, maybe for detecting. friend and foe the way that we're able to tell a friend from foe is via theory of mind examining agents intentions and motives, whether we're interacting with them. Directly or selves as in the case of the sheets task or observing them or reading about them, as in the case of the vignette-based experiment. So I've noted here that I should I should stop and see whether anyone has any questions before I move on to the second part of the talk. Yeah we'll leave the space open for questions, it looks like there hasn't been any that has come so far, but Liane is it Okay, if I riff-off and just kind of think through some of the applications of this. yeah yeah this has me wondering too right if, one the cooperation and competition, it seems to as you noted robustly recruit. theory of mind regions, but it does seem to be some discernible difference in how we're mentalizing cooperation versus competition. This has me even think on in terms of clinical practice, how even the practice of reframing things in terms of therapy right whether it's kind of couples therapy or conflict resolution, how possibly reframing it from a competitive. scenario to a cooperative scenario. still allows for a sense of mentalizing on either way, but you know, hopefully changes, some of the affect as well. yeah that's so interesting Mookie I never thought about teaching the same two. People or the same interaction contexts either framing it as either a competitive or cooperative um but I know that there are. cases. Within social psychology that are, that are related to that very question i'm going to be interesting to see whether. The neural signatures show up in the same way as well. yeah and you know this is probably a question for another time but yeah I wonder if there are ways to implement, you know therapeutic methods, while at the same time, you know the fMRI machines are probably a little too noisy. to implement things like that, but yes, and then on the second point, in terms of friend or foe and the discernible differences there, I do wonder, too, if some of the ways that. right because intent is kind of dubious in our ability to have to perceive and then to mentalize. The person's intent, I do wonder, sometimes, especially on the societal level in terms of the things that we've been seeing in terms of prejudice right, I wonder if sometimes the intent of being placed in. kind of that color our perceptions of a given action also drive some of the ways that we interpret behavior and then kind of automatically associate friend or foe. Oh yeah absolutely I mean the tricky thing about intentions is that they're often unobservable and even when they're explicitly provided by the agent, you might have reason to. disbelieve what they're telling you, and so I think it really goes both ways, you can sort of interpret some of these intentions, based on you know preconceived notions or prior impressions. Even as you know, information about intentions will influence your judgments of our particular actions, so I think it's definitely. complicated, but I think intentions really play. an interesting role in moral judgment and communication and. Interpretation of news and current events, and all of that yeah. hundred percent. Thanks Liane, yeah looks like our questions are still a little quiet So yes. feel free to continue sharing your work. Great Okay, so, as we saw in the first experiment and the shapes task, the same individual can switch between being competitive and a cooperative partner which may be the case in many of our close relationships too as as Mookie pointed out in the case of therapy. In the in the same context, and, as we know from our own experience, people can change and maybe more importantly, we can change our minds about people. So next we'll turn to this key question of how do we update our models of agents, especially when they surprise us. and critically will look at whether the updating process is different across different social contexts, starting with harmful and helpful agents. Again we'll focus on these brain regions for theory of mind or mental state inference that we saw and recruited for action prediction, in the context of social interaction and moral judgment do these regions also support the processing of. Surprise or prediction error for moral updating so when an initially helpful or trustworthy individual leader behaves in an unhelpful or untrustworthy. way how these regions respond to that error and their predictions that people make them and then each and. And how my mental state inference or or theory of mind closes that gap between what we expect and what we end up serving. So this next experiment is work by my current fourth year grad student Minjae Kim who closely with Mookie as well, here we presented participants with fictional characters. described as performing some set of positive behaviors followed by negative behaviors or vice versa, in a moral updating paradigm. The initial behaviors were intended to build up a prior impression or prior belief either positive or negative about the agent and the switching Valence was intended to produce what we're calling. prediction error repeats that gap between expectation and observation, so we presented subjects with tickets and their photographs paired with sequences sequences of behavior so, for instance. Amy here worked in a campaign to release prisoners of war, Amy help or someone's car out of a snowbank, Amy stayed late to help a co-worker with an important project, Amy looked after her brother all day as he recovered from a bad cold, Amy stole money from a tip jar at a coffee shop, and Amy abandoned her partner during hard times. So with each behavior subjects made a judgment about how trustworthy they found the target given all the information available up to that point so in this example. We expected subjects would form a relatively strong prior impression of the target specifically. Again this positive impression based on for positive behaviors before having to update that impression based into opposite Valence negative behaviors presented after that switch point. So this is what we're calling the the strong prior positive to negative condition in the week prior positive to negative condition we presented only to. Positive behavior behaviors prior to the negative update, followed by to neutral behaviors in the week prior negative to positive condition subjects first form to week. negative impression before the positive update and then in a fourth conditions subject first formed a strong. negative impression before the the positive update in these four conditions that i've described so far meetup 80% of the trials. In the remaining 20% we presented six same valence behaviors negative on this site here positive on the side here and in these control conditions, the final information was expected rather than unexpected. So, in a given trial against subjects for saw that photograph of the target, followed by the first behavior paired with a question about the targets trustworthiness, followed by behavior-two all the way up through behavior-six, and this is the timing of the trial components in the scanner. So here are the six conditions again so first negative updating following a week positive prior negative updating following a strong positive prior. Positive updating following a weak negative prior, positive updating following a strong negative prior, positive control and negative control. And we also controlled for a number of features across valence-strength and switch-point so moral relevance emotional valence arousal perceived frequency of that action, trustworthiness, and intelligence. So our primary analyses rely on. A measure of updating intended to capture this change in both behavioral responses and also neural responses in the theory of mind network. i'm happening around that switch point so to do this, we took the average of people's responses to the two behaviors post which and subtracted the average of the responses to the two behaviors pre switch. For all four of our conditions. And i'll just walk through a quick example of calculating updating metrics and since it's key to understanding on both the behavioral in general results so on it on this particular trial, you might first read. That Thomas had all of his wedding gifts be donations to charity. You might read him a six on that seven point scale, then you might read that he spent a morning volunteering at a nursing home and read him a seven. If, then, you read at the switch point that he lied to his wife about his location when visiting an ex. You might rate him a two and then, finally, if you read that he ordered his housekeeper around in a harsh tone, you might bump him down to a one and we'll just skip those two neutral behaviors at the end here. Thomas is not looking good. yeah exactly that's the idea, so you take the average of those two post-switch ratings and subtract the average of the two pre-switch, which would be negative five. Okay, so onto the behavioral data first so you might note that all those. numbers are positive, because we multiplied the negative updates, as we saw on the previous slide by negative one. We did this instead of taking the absolute values to avoid overestimating the update magnitude in case participants updated. in the wrong direction on some trials and what we see here is an effect of update directions, such that subjects engage in more. Negative updating circled in red than positive updating and unsurprisingly subjects also update more in all four of those expectation violations or surprising valence-switch conditions, compared to the two controlled conditions at the end of the graph. We don't see an effective prior strengthened in these behavioral data, there are other hints of that in these data, but what I think is more convincing as a world power conceptual replication of. The behavioral task which we ran online on 400 mTurk participants, where we presented the same sequences. Of behaviors that scene in the scanner and found the same effect of update direction but also this time and effect of prior strength as well, so greater negative updating versus positive updating and this time, greater updating of weak beliefs, or weak priors versus strong priors. Alright, so now we'll go on to the neural data So what do we see in the neural network for theory of mind, so first, you can see that all four. theory of mind ROI show greater activity in response to post which versus pre-switch behaviors. That is these regions all respond to the switch in the valence or prediction-error that gap again between expectation and observation and this overall. response to prediction-error and the theory of my network aligns well with other work in social neuroscience i'll just briefly. highlight a paper by my former grad student Jordan Theriault who's now a postdoc at Northeastern with Lisa Feldman Barrett for the unique item analysis approach that he uses so in Jordan's work less predictable and less fact-like. statements elicit more activity in theory of mind regions and more predictable items elicit less activity, so the more unexpected moral statements moral propositions where the more activity, they elicited in this network. And in a different project we've manipulated, whether information about an agent's behavior or mental states is unexpected. or expected based on the agents past behavior past mental states and in a nutshell, we found that more unexpected behaviors and mental states elicited more activity in the same network as well. Now in Minjae’s experiment here the overall prediction error effect actually looks different across different conditions of this task and also different brain regions. So i'll turn first to the DMPFC and LTPJ so mirroring the effect of direction and the behavioral data we see an effect of update direction and both of these regions as well, that is a greater change and activity in the positive negative direction. circled in red than the negative to positive direction, so these regions are recruited more for processing surprising negative information than surprising positive information. Unlike the LTPJ, DMPFC also shows an effective prior strength so greater change and activity in response to. Violations of strong priors versus weak priors marked by those yellow stars and archie bj also shows the same effect of prior strength. And again, these patterns are broadly consistent with other work, showing that, within this network, these regions are robustly recruited for encoding social prediction air when we get people wrong. Responding more to socially unpredicted versus predicted events and these results in our work, suggests that these regions are additionally sensitive to the degree of unpredictive-ness. And then, finally, we see no effects of either prior strength or update direction in precuneus. In the precuneus region yeah. And it seems like some questions are coming in, so did you want to kind of finish this section first and or did you want to take some of these questions. And i'm happy to i'm happy to take some questions now yeah sure. Sure, I can voice it so yeah it seems Sheldon’s asking it seems as though it takes quote unquote more brain regions. To be recruited to engage in morally objectionable behavior than to do so or otherwise, is that correct? And maybe one way to interpret this is i'm wondering if that means the perception of morally objectionable behavior, but maybe over here we're finding that it is surprising behavior correct. yeah no that's that's a really great question and, and one that we're still exploring actively, so I think that. What we're finding here is that these regions which are recruited for thinking about mental states are often recruited for morally objectionable behavior because morally objectionable behavior elicit. Questions about what people are thinking and what people are intending and what they're trying to accomplish. But we're also finding that in some set of these regions we're seeing more activity for just information that is more surprising versus versus less surprising. Independent of sort of the moral status of that information and so while it's the case that often morally bad behavior is surprising because there are social norms and moral norms against bad behavior. People can also violate your expectations of them specifically and we're also seeing that these regions are recruited for answering those kinds of questions why - you know why did my friend do this? And you could also imagine it going the other way around too so someone could. behave surprisingly well and we would expect to see the same pattern there, to which we do. And I guess maybe it's more a musing. yeah Claire notes that moral updating makes me kind of think about attachment styles and she says, it would be interesting to either take an account of the. or history of the subject’s kind of attachment and yeah do you think perhaps that this would have links to their ability to kind of update robustly in terms of. These surprising event; slash yeah have you heard of work kind of in the developmental field in terms of children's updating of these kind of surprising behaviors. yeah that's a really interesting question, so my background is not developmental psychology but my grad student Minjae is currently collecting developmental data as Mookie knows on on a version of this paradigm a kid friendly version of this paradigm, to look at kids moral updating patterns I don't know that we're collecting information on. attachment on those children or like what developmental window, that would be most appropriate to do so, we haven't approached this question from a developmental angle, if I have time to get to the next part of the talk will be looking at. How updating. operates in the context of close relationships and so that might bear some semblance to attachment, although in a very different context yeah perfect. Thanks Liane. Alright, so to summarize what i've just shown you DMPFC and RTPJ are sensitive to violations of strong. versus weak prior beliefs, independent of the valence of the impression, while DMPFC and LTPJ track the direction of the impression change from good to bad versus bad to good. Now, could it be the case that surprising bad behaviors are also just more surprising than surprising positive behaviors I think this relates well to that first question from Sheldon so after the scan session. We had persons read posts which information on it's surprising this so as expected people thought that the surprising information was more surprising that is information violating strong priors versus weak priors. But they rated, the violations of the positive and negative priors as similarly surprising so together these. patterns of results against distinct roles for distinct regions in this network in tracking separate qualities of surprising social information both the change in the valence and also the degree of surprising-ness. And, as I mentioned, exploring these effects, and our interpretation of them is very much a work in progress, but here are some initial speculation, so we see that surprising negative behaviors. drive both greater theory mine and greater greater behavioral updating convert to suppressing positive behaviors One possibility is that surprising negative behaviors afford easier reinterpretation of earlier positive behaviors Am I. need. to generate reputation based explanations for someone's past positive behavior they must have just done that, to look good. vs prosocial explanations of past immoral acts another possibility is that bad behaviors might be. perceived as affording more robust inferences about intentions and good behaviors consistent with some other past behavioral work so both of these possibilities would be consistent with the patterns that we're finding here, both in behavior and in the brain. Meanwhile, violations of strong priors elicit greater theory of mind and strong priors are also resistant to updating we don't want to change our minds when we have our mind fixed about something. And so, one possibility is that strong priors lead participants to generate. Alternative explanations for surprising or prior inconsistent behavior so again take that example of your friend taking money from the tip jar if you know your friend very, very well. And you've strong evidence for your friends good moral character, then you also have reason to come up with a different explanation for your friends out of character. behavior we have some pilot data suggesting that subjects do generate more situation based versus character disposition based attributions for behaviors that are inconsistent with their strong. Prior beliefs and those attributions also correlate with their degree of trust updates so in concrete terms the more people attribute an agent's unexpected behaviour to the situation less they update their impression of the agent's moral character. And so, these alternative or what we're calling these auxiliary explanations for surprising. Information can account for both more theory of mind related activity, more mental state inference and also less updating when those strong beliefs are violated. So it's useful to remember that in many ways paradigm here participants are reading about. hypothetical fictional target, with whom they don't have any relationship they're not interacting with friends or family. But in real life, the strength of our prior knowledge often coworkers with social motivation, not only do we know more about the people that we are close with. And I've stronger evidence for their good moral character, but we also like some more were more motivated. To maintain our positive impressions of them, and so that makes it really hard to tell whether any instance of belief, maintenance is either motivated or procedurally rational. By contrast, it's easier to make that influence in the in the current work that strong priors elicit rational belief maintenance via generation of alternative explanations and social motivation is absent, you don't need to protect these fictional hypothetical characters. and other work also suggest that one close others behave badly people simply ignore or discount that new information for the sake of maintaining their impressions. disengaging from these processes, all together, and so surprising information, like our friend behaving badly that friend taking money from a tip jar could actually lead to even less. activity in this in this brain network and less behavioral updating as well, and still other work suggests that mental state inference may be especially important for overcoming. we're updating strong positive priors about close-others is that, as people successfully update their impressions of people they know and like only by recruiting these. cognitive capacities for theory of mind and that's where we'll we'll turn next, this is a figure from Minjae’s new theory paper. That walks out some of this logic, so their proposal is out there presence of mentalizing or theory mine related activity can help. diagnose different paths from prior inconsistent evidence to either rational belief updating rational belief maintenance or non rational belief maintenance. So in blue is this space that we've been occupying again procedurally rational belief updating rational belief maintenance in the case of these zero acquaintance fictional hypothetical targets. And the space in red is where we're headed next. So we've been looking at prediction error and updating, in the case of these hypothetical agents, but an outstanding question is what happens when. People interact with real people, including people they know and like so I started to talk by referencing ingroup bias. In motivating this broad question of how we deploy theory by an across group boundaries social context more generally. I'll circle back to this question now and asking what happens when our friend behaves in a surprisingly selfish way versus when a stranger behaves in the same way, what do we see for prediction error, theory of mind, and moral updating across these different social contexts. So this next study has worked by my former postdoc BoKyung Park who's now an assistant professor at UT-Dallas So here we recruited participants to bring a close gender-matched friend with them to the scanner at the scanner they also met a stranger another confederate. participants were told that they'd be playing a game in the scanner with their friend and the stranger they were told that on any given trial, they would be playing either their friend or the stranger and. In a two player modified dictator-game for folks who are familiar with economic game terminology, but i'll tell you the instructions that we give to the subject so. At the beginning of the trial both players received $20 one player, the friend or the stranger depending on the trial can freely. give money to or take money away from this subject in $5 increments from 0 to 20 participants were also told that one of those trials would be chosen at random, and they would get the money earned on that trial as a bonus payment. subjects viewed pre programmed responses on all trials and the net amount given or taken in each run with zero and on every trial participants rid of the extent to which the giver the taker was trustworthy and also how close they felt to them. So first unsurprisingly, we found that subjects rated their friends in green more positively than strangers in red and giving on the right more. positively than taking on the left and critically in the graph on the bottom, we also found that subjects updated less overall trial by trial for friends, again in green than strangers in red, across conditions. Second, we found that this evil pattern is again mirrored by activity in the RTPJ across condition so less beautiful updating and less RTPJ activity for friend in green than stranger in red, especially for taking. People. On the left. When we zoom in on their friend taking condition we see less or two days. Activities associate with more positive ratings of the friend and more activity activity as associate with more negative ratings of the friend again suggesting that you really need to recruit your theory by capacities. As supported by our to retain or overcome your bias in favor of your friend to rate them badly. I think, in the interest of time i'm going to skip over. This modeling bit. Since I think, Mookie we only have eight minutes left, right in the session. But what I will say again is that when participants are able to. overcome that bias, overcome their resistance to updating their impressions of their friend they're drawing on those same neural resources again located in RTPJ and some of these other regions in the theory of my network to be able to engage. And an update in a negative direction their impressions of their friends taking money from them in the context of this task. Alright, so a question that remains that was sort of left open and the last it also is whether reduced updating for friends versus strangers is. Procedurally rational, given the contribution of people's stronger priors their greater knowledge about their friends so in one sense if you know more about someone. It might make rational sense to not throw away that impression in the light of single interaction in the context of an experimental game. On the other hand, we could have evidence for bias, people may be motivated to maintain their positive impressions of their friends and therefore. discount or ignore evidence that suggests otherwise, as reflected by the fact that subject or learning less about their friends over the course of the. game and and again like changing their behavior less over the course of the game, so, as I said before, this is a question that we are actively exploring but preliminarily we propose that. In this contest, to the extent that impression maintenance is associated with a reduction overall reduction in RTPJ activity, which I think it should be at the very beginning. We might have evidence for motivated condition, that is reduced brain activity suggests disengagement from this mentalizing process when friends or be eating badly eating less updating overall for friends, by contrast, in Minjae’s experiment featuring those hypothetical fictional targets, an increase in mental activity in response to surprising information might signal an effort to generate those alternative auxiliary explanations to make sense of. That incoherent. Information to come up with a coherent story for how it all fits together. So returning to that tip jar example one last time, if you see or hear about your friend taking money from a tip jar you could conclude that they're a thief. Or you could explain away the evidence, by concluding that they're making change for dollar again robustly recruiting your theory of mind, your mental state inference capacities. Or you could turn a blind eye to that evidence right in well turning a blind eye to that evidence, rather than grappling with ut directly and engaging those theory of mind capacities. isn't the best path to predictive accuracy, it might have other social relationship benefits. And so BoKyung as evidence of those kinds of benefits, namely participants who resist updating their positive impressions of their friends report having more friends so show you what this looks like here. So in in another study BoKyung found that first participants who reported who updated their friend closeness ratings less negative we compared to syringe recklessness readings reported having more friends. Second participants who showed reduced neural activity when their friend took money from them engaged in less negative updating. For friend closeness ratings and stranger closeness readings and thirds are closing the circle. participants who showed reduced neural activity when their friend took money from them also reported having more friends and so these results suggest that. In this case, people's biased maintenance of their positive impressions of their friends is made possible by by ignoring rather than explaining their friends' bad behaviors and may also come with the benefit of thinking that you have more friends. mm hmm. So returning to this logic, maintaining beliefs by discounting evidence and disengaging from mentalizing can lead to maintaining. relationships of some nature, whereas updating. beliefs and maintaining beliefs via auxiliary generation generating those alternative explanations in blue may lead to having more accurate beliefs, with good good predictive power. So normally I would stop again here and and to see if there are any questions before I move on to the third part, but I might not get to the third part so maybe I should just stop here and ask for questions. yeah if you want you could you can take it to the third part if there's an insight that you wanted to share with folks. I think I can probably summarize the third part and just a couple of sentences or two, so that we have maybe at least a minute for for any final thoughts. From you Mookie or questions from the audience. That sounds good. yeah in the final part instead of instead of looking at how people. Judge close close versus distant others or update their judgments of close versus others. We look at how people judge other people who favor close versus distant others, for instance of sort of this second order judgment and here we're particularly interested in. This question of how people perceive favoritism; are there instances in which people think it is good to favor close others and are there other instances in which people think that we ought to be impartial and the answer is yes, so this really cool. phenomenon that my grad student Ryan McManus has found and again i'll just skip through to show you some of these. data, and you can just ignore the graphs, for now, but just look at those bullet points, but essentially what he finds is that, and you might have this intuition. yourself that if you hear about somebody who helps a stranger and you hear about a different person who helps their cousin you might think that the person has a stranger. deserves more moral credit because they're not obligated to help strangers, whereas people are more obligated to help their cousins. On the other hand, Ryan finds that people who choose to help the stranger. Instead of the cousin are worse than people who choose the cousin over the stranger so what he finds that this this interesting pattern of. judgments is explained by people's intuitions about obligations familial obligations in these cases. But then he also finds that. That last set of intuitions switches in cases where agents are thought to have some obligation to be impartial, so if you're a professor in a classroom you cannot choose to help a family member over. A stranger the way that you could, if you're helping a cousin move into their apartment over a stranger who is moving into an apartment, and so what these data suggest overall is that our moral judgments in these contexts are incredibly nuanced and sensitive to. principles of obligation in these cases yeah and then I guess i'll just wrap up with this. Final slide with conclusions here so again, we saw that these regions for theory of mind recruited for both competition. and cooperation, but they also distinguish between these these social contexts as well, based on their moral intent of the agent either toward us or toward others. We can think of this key feature of mental state representation as this friend versus foe dimension are regions that are for thinking about others thoughts also care about how those thoughts affect us and others. Importantly, people are also able to update their impressions of agents along this friend or foe dimension or response to prediction error surprising information indexed by activity in the same brain regions. We saw that people may be less flexible when it comes to updating impressions of close friends, but to the extent that they are able to do so, they rely on the very same brain regions. That support processing social prediction error of surprising social information and then finally. I would have shown you, but instead I just summarized for you, that people's judgments about others' treatment of close versus distant others may be explained by their intuitions about obligation across relational and especially familial context, so this capacity that we have for theory of mind or mental state inference might be this multi purpose tool shared and conserved across multiple contexts we have the same neural mechanism or set of neural mechanisms for thinking about others thoughts. distinguishing from friend versus foe in the first place, as well as for shifting agents positions around that space during moral updating and response to social prediction error so with that, thanks to the wonderful people in my lab including Mookie and everyone else on this webinar. Thank you so much Liane and before yeah I give kind of the final things and then give the mic back to Liane to just kind of give us a final kind of nugget or point of grappling yeah. Thank you there's definitely so many questions and insights that I'd love to kind of unpack one of the things and the dictums, especially in the Psychological Humanities and Ethics is that. invitation to serve the widow, the orphan, and the stranger and, as you kind of noted today here , right our ability to mentalize. Things like intent and even you know turn maybe a blind eye to our friends stealing from the tip jar rather than grappling with some of the. difficulties there in yeah it's it's kind of part and parcel of the human condition, but yes, it looks like we're definitely out of time for tonight's lecture. Just to give my gratitude this event would not have been possible without the work of an entire village and community so i'd like to thank the Psychological Humanities and Ethics. folks as well as my co-director Associate Dean David Goodman additionally i'd like to thank all on our team. In the Professional and Continuing Education Department at the School of Education and Human Development, including Lilian, Marissa, Kaitlyn, many others, and our fearless director Ashana Hurd. And, of course, our colleagues and friends in the Morality Lab. Last, but not the least, I'd love to thank Dr. Liane Young for such a rich and engaging. Dialogue on moral psychology and inviting us to think a little deeper on how we process what is right and wrong, we add, is there any kind of final thoughts or phrase that you'd want people to grapple with as they think more on this question of moral psychology. that's a tall order but I mean I think a theme running through the studies that I showed you today is how important it is to consider the broader context for people's beliefs. and behaviors, including moral updating and moral psychology psychological processes so whether that context is relationship context, or choice context, normative context. it's it's really it's really critical to sort of keep in mind that that broader framework when interpreting other people's behavior. Thank you so much Liane for that ethical invitation and thank you all for joining us in this Psychological Humanities and Ethics lecture. We hope to dialogue with you again in due time. Thank you Liane. take care. bye everybody.