Johanna Latt was at IEEE VR talking about the impact of avatar anthropomorphism on the Virtual Body Ownership Illusion effect. She found that there was some uncanny effects when the avatar looked more like a human than a more stylized robot or block humanoid.
She talks about some of the bottom-up and top-down variables for the Virtual Body Ownership Illusion effect, and some of the techniques that are used in order to help induce the effect. The research literature indicates that the biggest factor for inducing the virtual body ownership illusion is to one-to-one limb tracking. One way to test whether or not the subject is identifying with the virtual avatar is to introduce a threat within the virtual environment. If the subject reacts to a threat, then it’s a good indicator that they’ve identified the virtual avatar as their own body.
She talks about some of her conclusions from her research, and some of the future research questions about how virtual embodiment can have long-term effects on identity and behavior.
Become a Patron! Support The Voices of VR Podcast Patreon
Theme music: “Fatality” by Tigoolio
Subscribe to the Voices of VR podcast.
Rough Transcript
[00:00:05.452] Kent Bye: The Voices of VR Podcast.
[00:00:12.015] Johanna Latt: My name is Johanna, I'm from Würzburg and what we did was we wanted to find out how important is it, how realistic your virtual avatar looks like to induce the illusion of body ownership. So the illusion of body ownership real quick means that you experience a virtual body as if it was your own body. You make it up out of top-down and bottom-up factors that eventually create the illusion. And we wanted to know, is such a realistic human body a critical top-down factor so you get this illusion? So what we did was we set up an experiment, an immersive game. People were wearing a head-mounted display and they saw a virtual body from a first-person perspective. And we were completely tracking the body so they could freely move around, walk around, use their arms, use their feet, just everything. And what they do was touching spheres that were randomly appearing. Just go there, touch it, another one appeared. So they were just using the bodies a bit and had to look at the feet, look at their hands. We had a between design, so each participant saw a different avatar. Some saw just a basic human avatar, depending on their gender, a male or a female one. And then we had a block man, which was very iconic, a very basic representation people saw. And the last avatar we used was a robot. So a lot more detail, but a very mechanic version. So those two were our non-human avatars, as we called them. Moreover, we had a within variable that was a threat. Just quickly, a threat is often used, the illusion of body ownership, to test how strong it is. Because the idea is, if you feel that the virtual body is yours, you would also be scared of something that is threatening your virtual body. So we had this fire there, some constant fire during the task round and then at the end also a sudden explosion again that was surrounding people. And to the result, so we had 10 people per avatar that we were testing. And what we actually found out that the illusion was equivalently strong for all our three avatars. So there was actually no difference between them. And moreover it was quite strong with all of them. What we also did then was an equivalence testing to get a bit deeper into the topic and what we found there, what you can see on the graph, is that the block man and the robot were equivalently strong, while the human actually did a bit worse. It was not as strong in the illusion as the two non-human avatars. So that was in the first place quite surprising, since you think maybe they're equivalent or the human is better or has a stronger illusion, but it's not worse than the other two. So a possible explanation we had was maybe it's an uncanny valley effect, maybe the human looked too real to people and was a bit creepy, so they thought, no, that's not me. Or what we also thought, we had a questionnaire with also open questions and a lot of people who had the human avatars said like, oh, I can use my fingers and my shoes look different and my skin color is another one, while with the two non-human avatars, it was nothing like that. They also had fingers, but no one of them stated, oh, I couldn't use them. It was just, yeah, okay, I look different. It's not a human, but I accept it. So maybe it was just that they saw it's not human and it's done, it's my body and that's it, while if you have a human avatar you really start to compare the details, which may even diminish the illusion for you a little bit. So in conclusion we found out it's obviously possible to induce the illusion also of non-human avatars in our setup at least. So it doesn't seem like this realistic human body is any critical top-down factor. So that's quite interesting in the practical and also in the research area. So you could think of, let's say, a physiological study where you give people a very basic mannequin and then see, does it influence your self-esteem? Now, which physical looks are important to get you better self-esteem, for example? On the other hand, if you think of practical things like in the gaming area, you of course think People maybe want to play a robot in a very scientific or very futuristic surrounding environment. And that's possible, as we saw now. It's possible to do that. So a possible next step can also be induced over an animal, for example. So where you don't have this basic human skeleton we always had. So this would be maybe a next step we could do. See if that's also possible. So, yeah.
[00:03:57.104] Kent Bye: So what are some of the factors of the top-down and bottom-up of the virtual body ownership illusion?
[00:04:02.393] Johanna Latt: Yeah, we have different ones there. The bottom-up factors are, for example, that you have the synchronous vision-motor synchrony there, so it means if you move in real-time or in reality, that your virtual body moves exactly the same way you do. So this is what we had here with the full-body tracking. Another thing you could think of is the visual-tactile input. That means that your real body is touched somewhere, wherever, and your virtual body is also touched at the same spot. So you have this synchrony. This is another bottom-up factor that can also work to induce delusion. And a rather simple thing you can think of is of course the perspective you see. So you have the first-person perspective that has been found to be very important. So you should see your avatar from first-person perspective to think it's yours. The top-down factors, that's the point why we also wanted to look at this. It's not really known yet which ones are actually important. For example, top-down factors are things like your pre-existent body representation, so the picture you have of yourself in your mind. I think like expectations or what you think or interpretations of the world around you. So in this case it means how do you interpret your virtual body, what do you expect to see and what do you actually see. And that's what we wanted to find out. Is this a top-down factor that's necessary for the illusion that you look real or is it not?
[00:05:09.728] Kent Bye: And so were there any special things that you did when people are going into these experiments to help induce this virtual body ownership illusion by, for example, touching them in the real world, but also having a virtual hand touching them just to kind of get them connected there? So maybe talk about some of those things that you did in order to have them buy in to that coherence of that virtual body.
[00:05:30.035] Johanna Latt: So we decided to go for the VisionMoto Synchrony. So when they moved in real time, their avatar also moved. So we did a full body tracking. They had markers all over their arms or their legs and their shoulders and so on. So when they really moved, their avatar just moved along with them. We did not have any visual tactile input. We decided to do that because of the previous work. It was found out that the VisionMoto Synchrony is enough to give you a strong illusion. And we, of course, also had the first person perspective. We used an HMD. So when they look down, they just saw themselves from the same position. What we did was we started with an introduction phase where they first of all were in a basic room where they first of all got used to the body and we also asked them if you like put your hands up in real now do you feel like your virtual hands are in the same spot or when you look down do you feel your feet are standing on the ground so we made sure to really get them to say yeah okay this feels like it is the same what I move the avatar does too. Yeah that was pretty much what we did.
[00:06:26.563] Kent Bye: And so it sounds like that a lot of the research of Mel Slater was an inspiration and maybe you could talk a bit about, you know, some of the studies and stuff that you were referring to in terms of the specific research papers or things that he was working on that you were trying to build on.
[00:06:40.342] Johanna Latt: One that was going a bit in this direction was very interesting, it was a quite recent one, where they wanted to find out if it's possible to give people a virtual tail that they could control. That was also a bit in this direction of a non-human avatar, so people in this paper saw also a virtual body, but in addition they had such a long tail that they could control by their movements. And that was also quite interesting, and they also found out it works, the people that have the illusion that they could control it, and that's part of them. So this was a bit of this direction, which was very interesting, of course, and we liked that a lot. In general, there is just a lot we based on, especially when you go to the top-down and bottom-up factors. For example, a lot of studies have been done by him and his chair on which top-down factors are necessary bottom-up factors. There is one experiment where you have this child sitting there and you are the child and you can look in the mirror at yourself and you have your mother standing next to you, your virtual mother, who's like stroking you and eventually suddenly she slaps you in the face and that's this kind of threat they have in this version. which we also based on you have this sudden change something suddenly happens and we had this explosion in our case for example and where they found out that it's necessary to have this first-person perspective in this paper for example that's what we based on and just yeah a lot of work for them also our questionnaire was based I think all the papers were also from us later actually four different papers we used there we put together questions from these ones so yeah that's it's just basically he's in all the references
[00:08:03.304] Kent Bye: Yeah, the threat is interesting to me in terms of, you know, what types of things and what talk a bit more about the intent behind why it's important to have a threat and then maybe the spectrum of different threats that you've seen in these types of experiments.
[00:08:14.998] Johanna Latt: So the idea is that if you feel that the virtual body is yours and you see a virtual threat being a threat to this body then you should also be scared of it or afraid of it and should get a bit nervous. So our idea was okay we add fire and people had a headphone on they could hear the fire and when they came closer so the idea is okay they should be scared of the fire if they have the illusion and then we measured the galvanic skin response which should measure how nervous people are and then we should see an increase in this. We decided to go for such a sudden threat with the explosion and such a constant threat and with the fire that was just a bit explorative actually to find out if there's a difference there because so far it was mostly quite a sudden threat. They sometimes in previous studies they used a knife to suddenly stick through your hand or it was also often fire something suddenly coming at you or they suddenly moved the virtual finger just broke it or such things, quite sudden, quite violent things quite often. So that's why I had the explosion and the fire and the constant fire was a bit to just see if that also works out. In this case, what we found out, it's a bit difficult. Also in the previous papers has been found out that such a threat It can be a good measurement, but it also depends a lot on your participants. In our case, we had a lot of PC users. We had a median of 7 out of 7 PC users and also gaming experience. And then you often have the problem that people say, well, OK, I know it's virtual. So I avoid the fire, but I'm not really scared of it because I just know it's not really there. So that's a bit tricky. So we'll have to see if you maybe find a better measurement in that case. We also didn't find any differences, actually. People were not more nervous with the fire in our case. So that's the question, how you go about this, but that was also the idea behind this to just see again, does it work with a threat or does it not?
[00:09:57.681] Kent Bye: Yeah, I'd imagine that the threat may be falling into the uncanny valley as well in terms of not really being believable and what Mel said is that, and made the joke, is that the further away that your experimental studies, your VR experiment are from the computer science department, the better your results.
[00:10:12.295] Johanna Latt: we totally noticed that. Like I said we had also a lot of people who were putting on the HMD and they were used to playing HD games for example like oh okay I can see the pixels and well I'm not that impressed while other people who were not familiar with this at all were just putting the HMD and were like wow this is crazy and now can I stay here and you really notice the difference so that's something we also asked that of course in our demographic questionnaire I think you should always ask that and be sure that it doesn't influence your measurements there if you have a between variable design especially. So yeah, that's something you definitely have to consider with the illusion, I think.
[00:10:44.055] Kent Bye: So what's next with you in this research then?
[00:10:46.516] Johanna Latt: Yeah, there is a lot of future work to do, I guess. Like I said, the next step would definitely maybe be to go beyond this, to not just change the composition, but to also change the anatomy of your avatar, to really say we don't have a human skeleton maybe, but really go for, I don't know, an animal, let's say a dog or whatever, and map your movements maybe to this animal. Does this also still work? Can you still get the illusion of body ownership over this, or can you not? Or another idea we had in a discussion now was also maybe you could look at what's the role of the environment in this because we decided in our study to go for a very neutral environment. We just had such a forest and grass and just not a room because mostly so far people were in the living room and you had furniture and you had a couch and of course you'd think okay I'm a robot I'm sitting in a living room that's weird that's not gonna work. So that's already something we thought maybe we should just make sure it's a bit neutral, but no one really has studied that so far too. Maybe it matters how the environment looks like in comparison to my avatar. Maybe this can decrease or diminish the illusion to how I design my environment. So that's also definitely a very interesting setup and I think there's just so much more to come, definitely.
[00:11:52.045] Kent Bye: Yeah, and finally, I know that Mel Slater's research is looking into a lot of really big philosophical questions about the sense of self, and I'm curious from your own perspective and being involved with virtuality, some of your thoughts on that, and also where you see virtuality going in terms of what it's ultimately able to enable in people.
[00:12:11.279] Johanna Latt: Yeah, I think it's a very interesting question also. It's what they will now look at, like the long-term things that can cause, like can you really change people's perception or attributes in the long term? Because that, I mean, would be really interesting but could of course also be quite dangerous if you think of the typical examples of like I don't know military games and then people play this and of course you know what's going to be in newspapers then maybe and I think it's a very critical question that has to be researched some more but it's also very interesting I think there are so many things you could think of what you can do with this where you can find out what you never could study before like I also said if you think you can just give people very basic representations and really find out do the looks matter for your self-esteem for your self-perception for how you attribute yourself In the first time now you could replace your body and just see what does it do, what's the role of it. So I think it's quite interesting, you can really do a lot there and I think there's a lot to answer still also. I'm curious where it's going to go, really curious, and I think it's quite a big topic and quite an interesting topic. which maybe should be a bit bigger, this Illusion of Body Ownership. It's not that known. I think a lot of people have to explain it. So I hope it's going to be, soon now, a bit bigger. Also, since we have a lot more HMDs now, and it's just a topic that has to be addressed soon, since you will have games, and then you need some body representation. And then you just have to know what's the influence of this, how important is it. So yeah, I think it's a big topic, and a very interesting one. So join in.
[00:13:35.198] Kent Bye: Great. Well, thank you so much.
[00:13:37.339] Johanna Latt: Thank you, too.
[00:13:39.432] Kent Bye: And thank you for listening! If you'd like to support the Voices of VR podcast, then please consider becoming a patron at patreon.com slash voicesofvr.
