 So welcome everybody to our third week of the RSP seminar series for semester two and thank you for joining us today. We have a fabulous presentation today from our cognitive and perception theme. So in a moment we'll hand over to Erin Newman who will be introducing our speakers for this week. But first I'll hand over to Michaela who'll just remind you of housekeeping for the seminar series and we hope you enjoy today. So hi everyone. Thank you for joining the seminar series for today. So just a few housekeeping rules just to remind you that this seminar is being recorded and will be available on the psychology events page and YouTube and it will also be sent via email. Upon entry of this webinar, you have all been muted and we ask that you please stay muted for the duration of this seminar. If you have any questions you can write them in the question box below at any time and we'll have a question time at the end. Thank you and I'll hand off to our theme conveners Erin Newman and Amy Dahl. Okay so hi everybody thank you for joining us today. Today's presentation is really quite wonderfully in line with the goals and the values of our seminar series. And it's really been a bit of a theme this year with having these sort of more sort of panel presentations. So today we're going to be hearing a collaborative talk from both Dr Amy Dahl who's a cognitive psychologist and clinician in the research school of psychology. And Liz Miller who is a PhD student working with Dr Dahl on why faking it isn't making it and facial recognition of recognition research. So this body of research, this emerging body of research is really challenging some of the methodological norms in this area of research. Which is rather scandalous actually in the scientific community and I'm sure Amy will get there in her presentation. And in particular this work is really forging new theoretical advances and facial expression research. And there are particularly sort of significant implications here when we think about how we understand other people's emotions. So Amy and Liz we're really lucky to have you here today to contribute to the seminar series and we look forward to hearing your presentations. So Amy I'll hand it over to you. You're still muted Amy. Thanks for that Erin. So today I'm going to be talking about a program of work that we've been building in our lab over the last three or so years. And as we sort of talk about what's been happening a bit more recently, Liz is going to chime in and talk about the work she's been doing on as a PhD in the space of that program. So very broadly in our lab what we're interested in is understanding how people perceive and respond to facial expressions of emotion. And the particular component of that that I'm focused on is really highlighting that facial expressions of emotion come in two different kinds of forms when we see them in real life. We get expressions where they're symbolic or posed so when we smile to be polite when we're doing it on purpose or perhaps in the case of a child who puts on a sad face in order to get sympathy from their parent. And we also have expressions where they're really reflecting a genuine experience of emotion. So somebody who's smiling because they're feeling happy and communicating that feeling of happiness with those around them. Before I get into talking about this distinction and a little bit more detail on how we've been addressing in the lab, I'd like to take a bit of a step back and really make a clear distinction amongst three different types of questions that the facial expression literature asks and where we sit within those. So the first set of questions ask about the production of facial expressions. These ones are really interested in how the experience of the person aligns with what physically appears on their face. And within this production system, there's a really important distinction that's not addressed most of the time. That's that it's innovative by two different systems. So what appears physically on our face is typically a combination of the emotional experience of the person displaying the expression, activating an involuntary kind of component to the facial expression or the physical appearance of it. And then the kind of voluntary personal social motivations and norms that sit around it, whether it's putting on an expression where you're not feeling something or using that voluntary system to hide what you're actually feeling. Now, the main reason that we know that these two systems exist dates back to a classical, classic article from 1984 by Rin, where he really points to evidence of double dissociations in this system. And that's with patients who either can smile involuntarily. So when they find something funny, but can't pose a smile on command or vice versa, people who can pose smiles. But when they actually experience an emotion, their face remains entirely flat. So for most of us, what's actually physically appearing on the faces is the interaction of these two systems. In our lab, though, what we're most interested in is understanding perception. Now, we're interested in understanding how that physical appearance of the face leads to a conscious interpretation of that expression as either genuine or posed and kind of the emotional content there. And then that how it eventually influences downstream responses. There is a third set of questions out there. And I often find that people think these are the main questions that we're addressing in our lab. So this is where I want to be really clear about it. And this is the sense to which our perception actually aligns with our production. So this is the kind of question where can we tell people a line? Can we really tell if an expression is genuine or not? These are very difficult questions to answer because they require having a really good understanding of what's happening for the person producing the expression and also the perceptual end and how those two marry up. So the argument I'm going to put to you today is that we really need to move beyond the way that we've traditionally studied facial expressions because we've been over-aligned on posed expressions, which our lab has shown is not perceived as showing genuine emotion. We've now got really good evidence that we can obtain stimuli that address these limitations, at least in the perceptual space. And that when we use these expressions that show genuine emotion, it can lead to quite different theoretical outcomes. And in other cases, give us a more nuanced understanding of what's going on for people when they're perceiving expressions and then responding to them. The final point I'm going to talk about and in this space to hand over to Liz quite a bit is what physical information in faces makes expressions of genuine. So what is it that our own stereotypes of genuine expressions contain in order to drive our behaviour in that way? So first of all, starting with our over-aligns on posed expressions. Many of you will have seen these expressions before they've been around since the mid-1970s from Paul Ekman's lab. And they're conceived of as prototypes of fears that are discussed to anger and also happiness and surprise not shown here. Now, these particular databases rely on Ekman's theory about what facial expressions are, that they're readouts of certain basic emotions that exist universally across cultures and across time. When we go back to our model and look at the production end of things, what Ekman's arguing for here is that facial expressions are a readout of that underlying involuntary system and that this is what he's proposing to study using these stimuli. The ironic thing about it, though, is it's the voluntary system that's actually been used to reduce them. So what Ekman's done is decided what actually belongs in the face in order to produce these expressions and then gotten people to pose them. So to kind of explain that in a little bit more detail, Ekman himself describes these expressions as the most deliberate performance of requested facial movements. To produce them, he brought people into the lab and using his facial action coding system shown here on the right, he broke down the expressions into the smallest possible units that are used to make them up. So for Anga here, for example, we have AU4. So we bring somebody in the lab and ask them to pull their eyebrows down and together, then raise their upper eyelids, AU5, and park their lips and then pull them back, AU22. And this is how all of the anger expressions in this database are produced without any mention of the word anger whatsoever. Now, this is true across five of the six expressions in the database. So anger, disgust, fear, sadness, and surprise. But where it differs is that the happy expressions were caught off guard during spontaneous moments in interactions with the photographer. So these are more naturally occurring expression rather than deliberately posed in the same way as the other emotions. The main criterion that's been used to validate and argue for these expressions and their use in perceptual work is that people have high agreement about what emotion the faces are showing, at least within Western cultures. But let's have a look at how this pans out. So we've got about 100% of people agreeing that an expression like this shows sadness. We also find it for more modern sets that were created using the same kinds of methods. But then when we turn to naturalistic expressions that look really quite different, we get that same level of consensus showing that we can get 100% agreement using things that look really quite different. As we extend into using virtual faces, we're seeing the same thing. And if anything, agreement about what emotion is being shown is higher for virtual faces. And of course, emoticons, we all agree about whether they're sad or happy or so on. So you can really take physically different things and produce the same consensus. One of the reasons that we first started thinking about whether these are really good stimuli that are tapping into what we see in everyday life and responses that come from those is that we have participants who are really spontaneously commenting on the artificial nature of them. So asking things like, are you for real? And I'm finding it hard to take these faces seriously. Now, these are comments from two 12-year-old male participants. Even people who perhaps aren't being directed to pay attention to these, like psychology undergraduates might be, are just spontaneously picking up on the artificial nature of them. There were, of course, around in the 1970s. So a more recent question we've had is about how much they're still being used in research today compared to some of the other popular databases that have emerged. So over the last year or so, we've been working with special topic students to evaluate this systematically from the literature. So screening six journals that publish a lot of work using facial expression stimuli and looking at the percentage of these that still use the ECHMEN stimuli compared to other popular post databases. And so what we see here across the last two decades is a massive increase in the number of articles being published using expression stimuli from the four most popular databases. So the ECHMEN, the KDF, the NINSTEM and the red band. And I'll give you examples of those stimuli in a second as well. And we see the ECHMEN ones increasing steadily across there, but these other newer databases kicking in as they emerged. Part of what this reflects though is that there's also been an incredible increase in the number of articles overall using facial stimuli. And so when we control for that, what we find is that across the last 20 years, it's held steady at around 60% of articles using these four main popular databases. So a huge amount of our understanding drawing on this small number of sets. These are examples of stimuli from across these four databases showing angry expressions. Now, they weren't all created in the same way as the ECHMEN pictures of facial effect were. So the KDF, for example, are created by asking people to rehearse the expressions and then evoke the emotion in the way that felt most natural to them. And the NIMSTEM1 use professional actors where they're instructed to pose a particular facial expression. So make a happy face, make a sad face, and so on. The red band database, which has emerged more recently though, was made in the same way as the ECHMEN faces were. But in this case, including the happy ones, so the happy ones were put on by directing those facial actions too. So the fundamental question that we started this body of work with is really asking about, from an empirical point of view, following on from that earlier anecdotal evidence, to what extent do these expressions and these stimuli that are being so often used in our literature perceived as showing genuine emotion. So we had a really simple task for asking this. We showed each of the expressions and asked people to rate how genuine they thought each emotional expression looked to them. Using a double-ended scale, so zero could be used to indicate whether somebody was unsure if it was genuine or not. Positive ratings indicated that it was perceived as genuine and negative ratings that it was perceived as fake. So let's have a look at what we see across these databases. And you can see here on the white exercise, zero don't point with our positive, genuine ratings and our negative, fake ratings. When we look at the happy expressions, and this is amalgamated across the four databases here, we see that they are perceived as being clearly genuine. But this isn't the case across the rest of the expressions. And in fact, for the sad, disgust and anger expressions, they're perceived as being significantly in the fake direction. Now, overall, this pattern was largely replicated across the four databases. So we've got Ekman, the KDF, Nims, Dim and Radbaum down here, all of the happy expressions being perceived as reasonably genuine, including the Radbaum ones, which we know were posed using the facial action task there. But then when we look across the other emotions, we only get a couple of exceptions, like the Radbaum surprise faces and the Ekman sad faces being perceived as genuine, with all the rest as quite ambiguous or in the strongly faked direction down here. So hopefully by now I've convinced you that a lot of the facial expressions genuinely that have been used very frequently, including over the last 20 years, are not perceived as showing genuine emotion out there. Now, one of the reasons why there's been such alliance on the Ekman faces in particular is that there was very little else out there that was standardized across laboratories and could be widely accessed. But we're now living in a social and technological age where it's much easier, of course, to access different kinds of stimuli. And so in our lab we've been demonstrating how we go about doing that. I'm going to talk about two stimuli sets which have different strengths to them. The first set is a match genuine and pose set that's derived from existing databases where we took about 500 stimuli, got people to rate the genuineness and then worked out a set that was perceived as clearly genuine and was created in ways where the person could be expecting to be experiencing a genuine emotion and pose sets from sets like the KDF and the Radbound faces. So these are examples of those stimuli from the sad sets that we have our genuine sad expressions here, combining people who are method acting and watching sad films and our pose sad expressions coming here from the KDF and the Radbound faces. And so when we look at the ratings for the individual face stimuli here, you can see the genuine ones being rated as clearly genuine, fake ones as clearly fake. And we find that this is replicated across samples. So a second sample in there of 94. Some of the work I'm going to present a little bit later in this talk looks at individual differences and so one of the things that's important to understand about this rating task is that we find we get good internal consistency and come back at self as well above 0.8 across different samples. The second set is one that we've been developing more recently where we've been really looking for naturalistic stimuli that capture the dynamic aspects of expressions, not just these still images. We think calling this set the ANU real facial expression database. Now in this case, it's not possible for us to know how people were actually feeling at the time the videos were taken. So again, it highlights that focus on the perceptual aspects in our lab. So we were drawing from various YouTube sources including reality TV, news events, blogs and you know, different advertisements, etc. So this is the advantage that we can get a very wide range of expressions showing by people who aren't familiar as well. For the static stimuli, what we were looking here was to extract the peak expression from a clip as identified by three human coders and then collecting emotion labelling and genious ratings from a number of human observers to validate this set. These are examples of what our procedure's genuine and sad expressions look like in this set and our procedure's posed ones here. Now importantly, across sort of our 2000 plus stimuli that we've now developed in this set, what you can see is that across the different emotions it's really is clearly possible to obtain expressions that are perceived as genuine. Now, some expressions are much easier to obtain naturalistically. The happy surprise is quite easy thanks to a lot of the reality TV out there and sadness. Other expressions like anger and particularly fear are very difficult to get in this space. And of course, if we talk about trying to generate these as genuinely filled expressions in the lab, this is, of course, where we also get into some ethical challenges as well. We can't genuinely simulate bringing a tiger into the lab, for example. So now that we have our stimuli, the next question and the crucial question for all of this is really whether perceived genuineness matters. Now, the logical reason why I believe it should matter is because genuine expressions are really sending out a very different signal to opposed expressions. So, for example, if you see somebody who's showing a sad expression and you interpret that as a genuine sad expression, the response is, well, this person's in distress and so we might want to approach and offer help. In the case of opposed sad expression, though, we might conceive of that as somebody who's attempting to manipulate you or kind of get the best of a situation and you might deal with this really quite differently, either with a buoyance or caution or a small union in a very different kind of way. To try and understand some of these meanings in a little bit more depth, I'm just going to touch very briefly on how we've been looking at this using qualitative methods in the lab although I won't go into the work in detail just to highlight how these methods can really complement some of the other rating work that we've been doing. So, this work originally started with SET-1, although I've moved on to working with SET-2 in this case in the last year or so, looking at the matched genuine and opposed sets and asking people to give an open-ended response about what the facial expression is telling you about what the person is thinking, feeling and or intending to do. What we see here is that we get quite different themes when we do automatic analysis here, emerging for our genuine and opposed expressions and are aiming this space to then take these themes to predict the kinds of behavioral responses that might differ across genuine and opposed expressions within an emotion category. But turning back to downstream responses and how these all fit in together, one of my key interests is in clinical traits or involved in empathy and psychopathy and so to test a major theory of psychopathy we used our genuine and opposed facial expressions that show distress, so fear and sadness to try and understand whether the effective component of psychopathy so the lack of empathy, guilt and remorse the shallow effect theme with these traits that results in reduced pro-social responses to other people who are showing distress is mediated by reduced arousal to others distress. So the basic idea here is that typically when we see somebody else feeling bad we might also feel bad and experience our own internal arousal to that and it's that internal arousal that then motivates us to respond in a pro-social or helpful kind of way. So these are traits that vary continuously across the general population taking a sample of 140 and asking people to rate how aroused they feel in response to people's facial expressions of distress so genuine and opposed sad and fearful expressions and how much they would want to be able to then help them. So what we see when we look at correlations between effective psychopathy and people's intention to help we get the kind of expected relationship particularly for genuine distress so here at the lowest effective psychopathy end we find people are highly willing to help but that decreases as we head up to the highest psychopathy end. That's quite a strong relationship for genuine distress expressions. It's still there for opposed distress expressions albeit weaker and that is the significant difference between them. So next when we look at correlations between effective psychopathy with arousal we find a quite different pattern where we get the predicted relationship between effective psychopathy and arousal for genuine distress so a decrease with higher levels of effective psychopathy but not for opposed expressions and again that's a significant difference between them. So ultimately when we put this together in our mediation model what we find is that because there's no relationship involving arousal for expressions of distress we get a failure to support Blair's theory here with no mediating role of arousal ratings. So when we look at the genuine distress expressions we do get a nice predictive mediation there. So here we've got one example where using genuine expressions leads to support for a major theory of effective deficits in psychopathy but not when we test it with opposed expressions. The second example I want to give you is about social anxiety and it's really kind of a bit more about how you can get a more nuanced insight into what's happening in some of these clinical disorders that we're studying within RISP. So here we're just looking at smiling faces and comparing people's responses to our genuine smileers and our polite smileers. We're asking people how willing they would be to approach a person to ask for directions the idea being that if you're higher on social anxiety there's a reduced tendency to approach and we're also asking them to complete our emotion genuineness rating tasks. So again a sample of 141 and here we're comparing people who are high versus typical on our social anxiety measure of the SPAI 18. Now what we have here on our Y axis this time is mean willingness to approach so low down this end high up this end and what you can see here is that for genuine smiles we get the predicted relationship where people who are high on social anxiety are less willing to approach genuine smileers than people with typical levels of social anxiety. The really interesting thing here though is that we didn't get this same effect of polite smiles we're puzzled by this knowing that there's an interaction across the two. So one of the things that we wondered about is whether maybe there's a failure to appreciate how genuine the genuine smileers were and this was maybe driving this reduced willingness to approach here. But when we look at our genuineness rating so here I don't know positive perceived as genuine negative perceived as fake we don't see any difference in the perceived genuineness across our high social anxiety and typical anxiety group so this couldn't explain what we found in those reduced approach ratings. So this really brings us back to the question of what is it in the meanings of genuine and polite smiles that are potentially driving this effect. One of the things that theorizing on social anxiety is focused on in recent years is that it's a fear of social evaluation and this is a fear of positive evaluation as well as negative. So if we're going to approach genuine smileers what is it that they're signaling around this social evaluation is that they may be signaling a greater potential for social interaction and thus exposing the person to that social evaluation that they so feel. But to test this idea we took our genuine and polite smiles and asked for directions. And indeed people do predict that genuine smileers are going to want to talk for more than average compared to polite smileers. Offering a potential explanation of why people who are high on social anxiety might be more avoidant around genuine smileers. Let's pick the polite smileer so we can get in, get the directions and get out of there as quick as we can. So finally we're just going to touch on a little bit of technical information in faces makes expressions of genuine and look this is an important question for understanding the ways in which this is being driven and potentially differences across culture. What it is about a face that makes us interpret it as expressing genuine emotion. The first idea that we're going to touch on here are particular facial muscle actions that have potentially been linked with genuine emotion. Now I'm going to hand over to Liz at this time for the PhD research in this space. Okay. I'm just going to share my screen. Okay. So the work that I did for my honours year with Amy was looking at the Duchenne marker which are these cross-fate wrinkles that we get around our eyes which typically signal genuine happiness. So that's the typical theory. There's lots of evidence to suggest that when we have a Duchenne smile, we perceive them as being more genuine compared to the non-Duchenne smile and that we also attribute more positive characteristics to them such as being more attractive, more intelligent, more social etc. But there is one challenge to this idea and that is that the Duchenne marker also appears in sad expressions. So Charles Darwin was the first to observe this in his work but more recently we've had empirical evidence to suggest that it appears in tennis players when they lose their matches and in infant cry faces. So when I started this work no studies had actually systematically manipulated the Duchenne marker in sad expressions. So that's what my honours work set out to do. So we started off with these happy expressions. So we have the Duchenne happy expression with the present and then I used Photoshop to take the neutral face from the same person and Photoshop then neutralise into the happy expression. So that we had a situation where we had a Duchenne and non-Duchenne version of the same face where the only thing that was different was the presence or absence of the Duchenne marker. So what we would predict to happen is that the Duchenne version would be rated as more genuine compared to the non-Duchenne version. And that's exactly what we find. So this is the same scale that Amy has already introduced. And you can see that the Duchenne version was being rated as more genuine compared to the non-Duchenne version. So then we also want to see how this replicates for sad expressions. So I took the from the happy face and Photoshopped them into the sad expression and then I did the same thing with the neutral eyes and put them into the sad expression. So I had Duchenne and non-Duchenne sad expressions. And what we find here is that the effect of the Duchenne marker completely disappears. So the Duchenne marker doesn't make sad expressions appear any more genuine to the distance. So at this point in my honours here I just completed two experiments to confirm this lack of effect of the Duchenne marker in sad expressions when another study came out looking at exactly the same thing but coming to the very different conclusions for sad expressions. So these are the results of Malik and Colleen's study. So their Y axis is mean sincerity ratings and their scale is 0 to 5. The scale to what we used but you can see that the Duchenne sad expression was being rated as more sincere compared to the non-Duchenne sad expression. But there was one very key and important difference between the stimuli that I used and that of Malik and Colleen's and that was that their stimuli were computer generated. So these are their stimuli. The Duchenne one is on the left and the non-Duchenne version is on the right. And this was kind of the first indication to me that just like human expressions can create different conclusions compared to posed ones computer generated faces can produce very different conclusions in research. So this is kind of what led me to led me to my PhD research which is looking at how we process computer generated faces compared to real ones. So just by way of a quick example of this this is Mikayla. She is an Instagram influencer, a virtual Instagram influencer who is posing as a computer generated being online. Some people have reactions like I would really love to meet you or why do you look so cute but other people have reactions like please reply is this human or this is literally terrifying. So computer generated faces are being used in research for things like emotional genuineness but we actually don't yet know a lot about how we process them in comparison to real faces. So that's what my research is really tapping into. Let's take that back to Amy. Thanks Liz and Amy. So Amy did you have anything else to add? Okay, I'll let you take over. Sorry, I was just a little bit slow getting it back onto the screens. Awesome, so thanks for that example Liz and taking this into a different space where we're seeing it being tested in other ways and potentially repeating some of those issues in using posed expressions. The last example I want to give you also taps onto the CG issue and how we potentially use parameters that don't occur in real life as a result of that. So in terms of what physical information and faces looks expressions genuine we're really interested in starting to understand some of the things that haven't been traditionally studied like facial actions and some of the dynamic movements and moving into things like blushing and tears but for now we're really testing some of these traditional ideas that Ekman's raised that still don't have much empirical backing behind them. Now one of his predictions is that genuine expressions have longer sets than posed expressions and this is a prediction that has been accepted in the literature just like the Duchenne smile has been for happiness and part of the evidence that's really backed this up is work by Ava Cromhuber where she's tested, perceived genuineness via ratings but again using computer generated stimuli and in this case ones that are quite old so the very early versions of the computer generated faces that were being used in research. So these are the lovely in that they give wonderful control over things so in terms of looking at the onset of an expression you could choose to evolve from a neutral expression to a full smile over different periods of milliseconds so increments of 132 up to 528 milliseconds and the prediction is that these longer durations these longer onset should be perceived as more genuine than these shorter ones and indeed with the CG stimuli this is what is found so genuineness ratings heading up to 5 up here and down to 0 at the other end for faker and we see with these shorter durations perceived more in the fake direction and longer durations rated as more genuine as Ekman's theory would predict we did think it was important to test this using our more naturalistic stimuli so this is when we get into set to our new YouTube database and for this work we went through and we extracted the onset phase from over 200 smile stimuli so the point of neutral through to the first apex with that expression and people would then ask to rate the genuineness of these short video clips here we have our genuineness ratings again over on the Y axis over here and we've just marked them as orange to make it clear which ones had ratings above one and rare ones below minus one black just sitting in the middle and as you can see when we look at onset duration here across the X axis there's really no relationship between these ratings at all and you can see our R squared is negligible in this space so you know we're really wondering about what's driving this effect we've had models that had the Shen marker in them and anything else that we kind of predict might contribute to this and found really no effect so what you'll see though is when you go back and you look at the data from the computer generated places it's really this 132 millisecond duration that's driving most of this effect that we're seeing here when we map that onto our naturalistic stimuli even across these 200 or so stimuli you see that this area is very sparsely populated in fact we only had one smile with an onset duration less than 200 milliseconds so this really sort of suggested to us that perhaps what's happening here is that these are being perceived as fake because they're just so sure that they don't exist in real life and this is one of the dangers that we see with using these artificial stimuli so overall I hope by now we've convinced you that the findings from popular posed facial expressions to me are not capturing the full story that we really need to move beyond them they do have some clear advantages and there's a lot of you know really robust knowledge that's been sitting out there by quality across multiple labs and they produce replicable findings that seem sensible but I think we need to ask what it is that they're really tapping into for me I started thinking about these posed expressions as almost human emojis there's somehow some kind of visual stereotype or prototype that we have about what these expressions look like and in some way they're tapping into kind of the conceptual knowledge that we have associated with these emotion concepts in which these visual representations are part I think we really need to be looking to expressions that are perceived as more genuine to ask questions about how we respond to other people's affect rather than just kind of a cognitive concept of emotion but naturally posed expressions are something that have been missing for them this literature very much as well so we've been studying these very artificial posed expressions but we do actually pose expressions like polite smiles for a genuine communicative purpose in everyday life we also see people showing masking smiles and of course understanding these kinds of nuances that sit around posing and are being important part of moving this work forward and having a more nuanced understanding about what it is that we're using in faces to drive our behaviour that final bit of work presented by Liz and our naturalistic dynamic stimuli really is also arguing that the testing of parameters that really occurring nature is particularly problematic now that naturalistic stimuli are more accessible than ever even though they don't have the same level of control and sometimes experimental methods are particularly important we really need to be looking to convert evidence to develop a full understanding of what's going on in our social worlds as we communicate through faces so thank you everybody for your time and Liz and I are looking forward to taking some questions and having some discussion with the group around where we've been working with this program over the last couple of years so thank you very much Amy and Liz this work is fascinating it was really neat to see the facial expressions and the different types of stimuli that have been used and Liz I thought you did an incredible job dealing with the cat that was like a big cognitive load while you were talking and clearly you have some brilliant Photoshop skills there as well so let's open up the floor for questions so just pop your digital hand up there usually it's blue I think and we can look out for those questions or you can pop something in the chat box and brilliant Michael and so what will happen is that Makayla will unmute you so you can speak so Michael would love to hear a question from you Thanks Amy that was really fascinating I was inviting you a separate email to tell you all my ideas so fantastic there's two questions one is just a generic question but one's a little more focused on your work the generic one is why these emotions and not others are these the only emotions that humans have are they the most prominent do they come from Ekman's theory which is potentially discredited in a certain sense if it is why these emotions and but more focused on your work that you presented to us can you talk a little bit about more about the no emotion because I find that really fascinating for my own work and the difference between a genuine no emotion and a non-genuine no emotion so thanks awesome thanks Michael the first point that you're getting out in there why are we looking at the six or seven basic emotions that Ekman puts forward I don't have a really good answer to that and it's really because it's pragmatic so there are lots of other theories out there that extend this range of emotions well beyond the basic six and I think there's real value in exploring those and moving into that space and getting more nuanced around it but at the same time we felt we needed a starting point for our research program and one of the key things we wanted to test is when previous findings hold when they don't hold and because they're being tested with those six emotions in the past having stimuli that fit into those categories and really contrasting that genuineness dimension in them was important for pragmatic reasons so it's not a theoretical decision for us it's a pragmatic practical one about linking with the literature and you know this is something that keeps coming up as we give this talk in fact I think this is the third time there's no I've given it now and every time we've had this question which is really flagging for us that it's an important point we need to be making as we write papers around this work as well as a limitation of it and you know highlighting for us that we don't think that there's just six emotions there and part of being out of study this physical variation physical nuance is about being able to consider the wider range of facial behavior out there. When you refer to the no emotion component sitting in there I would also argue that faces communicate a lot more than just emotion that they communicate things around mental states and behavioral intentions and we certainly see that in a lot of qualitative work that we're doing there so if you dig into kind of the more nuanced responses when people are selecting that no emotion category there you'll see things coming through like interested, bored wondering what's happening, confused labels that people with the English language wouldn't typically consider as an emotion kind of label. Great. Thanks. Fantastic. Great. Thank you. Thanks, Michael. Steph, we can see your hands raised there. Go for it. Thanks, Erin. Yeah, just one of those really great presentation Amy and Liz. I thought it was a really elegant articulation of a big picture program of research particularly like how you're being responsive to participants experience of the stimuli. I think that's something that is easy to overlook actually be a rich source of information. My question is something I think you've sort of flagged as perhaps outside of the program of research but one I find quite interesting and I wonder what the degree of correspondence is between how genuine somebody feels the experience to be and how raters consider how genuine that is. So can there be an expression that I suppose is actually genuine that people don't perceive as genuine and vice versa? Thanks, Steph. I think that's a really good question and it points to why that third set of questions is so difficult to address. Overall, the literature that sits out there in this space is mostly focused on smiling expressions so Duchenne and Nonduchenne expressions and what we see there is that people can tell above chance but they're not particularly good. One of the things that's interesting though is that when you dig in and you look at just emotion labelling data so labelling and expression and as happy, sad, angry, etc. without even thinking about genuineness the most recent work that's really rigorously evaluating the set of responses out there is finding that people aren't really that much above chance in evaluating that alone as well. So when we're making these inferences in real life I suspect it's a lot more complex that we're drawing on individual person knowledge and contextual information to understand what that person is trying to communicate to us and that at the end of the day if we want to hide our emotion we're often quite good at doing that and sometimes we're not as good at expressing them as we think we are too. Great, thanks. Fascinating. I can see a hand up I can't figure out what your name is but go for it Makayla will unmute you and perhaps you can introduce yourself, sorry. Oh sorry it's Mike Smithson. Stop by name. Yeah, sorry about that. Sorry, Mike. Amy, really great research program and I really like some of the issues you're raising about the nature of the things we want to include in an experiment the things we want to leave out so my question has kind of been pursued as you would know in a lot of the sciences experimental control is supposed to kind of strip away the complexities of the real world oftentimes in order to demonstrate that under idealized conditions we've got control of the things if you waggle this variable then that variable over there waggles as well so I'm wondering if there are any analogs like that in the difference between the genuine and the posed expressions is there a sense in which the posed expressions are stripped away just a little bit more of the complexities in the real world and therefore under some circumstances might actually be more desirable to use than the genuine expressions you mentioned one thing and that was that they're standardized but are there any other things any other aspects? That's a really interesting question Mike and it makes me think a lot more about why we pose expressions socially as well so I think one of the things about posing a smile in an extreme kind of way that's easily recognizable is it communicates cooperation and positive intent one of the things that we find is really quite interesting is when we try and find smiles that are perceived as fake it's almost impossible to do so unless they're really masking a negative emotion in there so there's something that people perceive about posed smiles, polite smiles as being genuine in that behaviour as well I'm not sure I quite get where you're going with the question did you want to elaborate a little bit more? Really just only only from the standpoint that in psychology often times I find that experiment are seem a little bit confused about whether the extent to which their experiment is trying to reproduce something in the real world with all of its complexities and contextual information all that sort of stuff versus when they're trying to demonstrate that that something can happen under ideal controlled conditions for example I ran an experiment years ago with a sociopsychological colleague where we demonstrated that under ideal conditions you could introduce a gender bias and then remove it so this was we had to make one up we had to introduce it under very carefully controlled conditions and then somehow demonstrate that it was possible to erase it by providing counter-stereotypical information that clearly had nothing to do with the real world the way that stereotypes and biases operate in the real world it was just to address the question is it possible to introduce a bias and remove it so I'm talking about the distinction between experiments that have a lot of ecological validity and those that are very idealized because it's got different purposes I think that's a really good point and so I guess I've been focusing a lot on trying to understand what's happening in the real world but can also see that there's value in the kinds of experiments you're talking about so one that's out of a completely different lab where they looked at the developmental context that sits around how we learn to recognize spatial expressions and they made up a completely bizarre expression that's never seen in the real world and looked at how children learn to label this and understand it because it's actually demonstrating how that works in development I think that's a really important area and I think that's part of what's happening in expressions in terms of our own learning and so being able to unpack it using scrolled post stimuli or even the kinds of stimuli that Liz used where they are artificially made to be able to understand those processes is more but thanks for drawing my attention to that thought because I'll give some more reflection to it too Mike thank you Okay while we're waiting for the many other hands that will be raised in the next 10 minutes I'll sneak in a question I mean Amy and Liz I'm thinking about this from maybe like a social cognitive perspective and social colleagues can disagree with me or we can hear what they think so I'm thinking about when you're getting people to view these faces so for example in some of the distrust literature showing people faces particularly sort of negative faces is a way of inducing a particular mindset or type of cognitive processing and so I'm wondering whether you think there is a way that plays a role in terms of how people evaluate your stimuli in parallel right with mood induction research you actually do see a change in the kind of cognitive processing that people engage in so is there a role there for that and the work that you're doing and differences you see between for example positive and negative faces Yeah and thanks for that question I think that it really taps onto a larger body of research where faces have been used as induces and I guess my expectation in that space is where a genuine expression is somehow inducing a stronger experience of affect in the perceiver and affect has a role in those cognitive mindsets there we can perhaps better understand some of the nuances that sit around that but I do suspect that the actual valence of the expressions of the category expression is going to be much larger than the effects that we see between genuine and poised we have to really push the envelope to make those effects observable in that space I think. Yeah it's tricky I have too much power on the webinar here so I want to keep talking but Michael fascinating yeah it's an interesting area and maybe it's something you could manipulate and play around with but maybe we should have another chat about this further so Shuan I see you have your hand up go for it you have the floor. Thank you Erin well it's something that I figure when I look at those faces during one of my labs I had a question that's you know when you're seeing an expression that is very very to the extreme of happiness or to the extreme of anger how likely that you think that the participant would like to rate that as genuine given that the extreme anger or extreme fear is not that often see and then when it's a very mild it's it's hard to well when it's very mild I guess it's very hard to perceive whether the expression was what which direction what exactly the emotion was conveying but it feels to me that it's more like when the expression is not too extreme makes me easier to feel it's likely to be genuine is there any effect like that or am I making myself clear if I answer this one Amy I know this one so two things one of them is that when we get participants to rate emotional genuineness we tell them to ignore the strength of the expressions so we have an example that says you know someone might be have a very subtle expression but it could still be genuine so we do flag that with participants when they complete the task and then the other one is that in my experiments I did get people to rate the physical intensity but also the emotional intensity of the expression so how happy they find the expressions for example and when we do that we find that the intensity can't account for the genuineness the increase in genuineness between the Duchenne and non Duchenne expressions so there is something in particular about the Duchenne marker that's signaling genuineness rather than intensity in happy expressions and then for sad expressions it seems to indicate intense sadness rather than genuineness if that makes sense thank you yes do we have any other questions I have a few sitting here that I can use but I just want to give others an opportunity let me ask one just in the meantime so I'm curious this is sort of like a broad audience question Amy and Liz you're doing work in an area that is really sort of pushing at very traditional ways of thinking about facial expression research and I'm curious and I'm sure others will be about how this has been received and the literature, what's your experience been like and perhaps even you have people who are earlier in their careers are doing research that really brings up very strong questions about the existing research and those people who are really the experts in the field may be the people reviewing the work so do you have some comments on that? thanks Erin yeah I absolutely do and it's been an interesting experience over the years as this research program developed and started off with a very strong critique of the stimuli that sit out there that really didn't allow a space for well what is it that they have contributed where are they valuable and learning to understand that and respect that and acknowledge that as we write our papers has been a really important part of getting this work accepted and I think it opened up some other important and interesting questions in that same space too so I think the advice for students who find themselves in a similar position would be to use your empathic skills put yourself in but you also choose who has perhaps been pursuing this for a number of decades and their understanding and acknowledge what it is that is of value in that space because it's never black and white and as much as we can critique and say that maybe we should be using these expressions my understanding I guess has become nuanced over time that we actually need to be doing all of these things and working it in together and understanding their strengths and weaknesses and as Mike Smithson was pointing out before being very clear in our minds about why we're using them and what we're trying to answer here and then what that actually tells us about psychology Miss did you want to comment on that one at all given your experiences going through the review process with these couple and any advice out there um I guess in terms of the reviews I guess it does kind of highlight that you have to be careful about who your audience is um because we did submit a paper and kind of the feedback that we got was a little bit um I mean we were suggesting that the whole research using pose expressions isn't sufficient so yeah it is hard that you do get a lot of yeah pushback about things like that that you know pose expressions we have these databases that are beautiful and set up and they're ready to go um so yeah and creating genuine databases are so much harder and yeah so it makes sense that you would have that pushback um yeah sorry that's more of just a stream of thoughts rather than an answer to your question but um yeah it's an interesting thing to think about that I haven't kind of consciously thought about before so I see we have another question if my memory is right I think this is Mike Smith said yeah sorry it's me again um whenever important new research is done or when a new you know technique is developed sometimes something called dual use dilemma arises in other words the realization that this this research or this technique could be used for evil as well as good so I want to raise that here um to one extent might the research you're doing on the difference between posed and genuine expressions contribute to things like deep fakes um more convincing more trustworthy avatars that sort of thing we really have to think about this it's a horrible thought um you know I was having a chat last week to a lawyer who's working on um some of the stuff that's happening in the commercial space with this kind of technology that's you know purportedly reading people's emotions based on Edmunds theory and um you know really arguing against it because of the fundamental errors that are made in there but it is a scary thought that something like what we're doing could improve that technology to the point where it's doing it more accurately and using it for commercial enterprises in that space um yeah it's it's not a nice thought um yeah I guess the um one one of the things to sort of think about in these sorts of things is what kinds of applications um could be could you know could crop up from them or develop out of them and and which of those obviously which of those you want to endorse and which of those you don't um being able to one of the things I used to do with um students and of course I used to teach on technological change was to get them to read 1984 the book and then to go out and find out all the um technological aspects of big brother surveillance that had been initiated and how they came to be and almost always um all those techniques for surveillance had been initiated through very humane um even noble um uh objectives and goals rather than you know actually expressly for the purpose of surveillance and control. Do you have any suggestions about how when you're producing knowledge like this uh you can shape where it lands and how it then kind of um proliferate proliferates into the rest of the world at all? I guess um well one one thing would be to do the kind of thing you're already doing namely to uh to uh work with people who are likely to be applying these things and to uh try to negotiate with them conditions under which uh you know these these things are going to be uh applied and for what purposes other possibilities are doing things like copywriting where possible or patenting uh to restrict uh legal usage of it anyway um but uh but mostly just negotiating with people about about you know what this should be used for and and how it should be used um one of the problems we have in in or one of the limitations we have in Australia is that there is no as far as I'm aware um governing panel or body that uh attempts to debate um legal restrictions or regulation of research applications instead it's just kind of left up to people to work it out somehow. It's a really interesting note to finish on I'm aware that we're on one o'clock we could probably go on for much longer it looks like but I mentioned that we'll develop into other discussions um so Kristin shall I hand it over to you to revert things up? Oh I have the very difficult job of saying thank you everybody that was a wonderful presentation and it's lovely to see academic staff and students um working so well together and sharing their research with all of us so thank you so much and thank you for all of the really thoughtful questions um and we will see you next week. Thank you very much. Have a great day.