Chicago Booth Review Podcast Does Replaying Videos Change How We Watch Them?

Kristin Donnelly: And we showed it to them either once, twice, thrice, four, five, six times with the repeated conditions back to back. And what we found is that any amount of repetition was enough to get people to think this is more like a scripted scene than an improvised scene.

Hal Weitzman: Do you ever find yourself scrolling through videos on your phone? If so, does watching the same video on repeat change the way you think about it? Welcome to the Chicago Booth Review Podcast, where we bring you groundbreaking academic research in a clear and straightforward way. I'm Hal Weitzman and today I'm talking with Chicago Booth's Kristin Donnelly about her research on the replay illusion. Why does seeing video of spontaneous events more than once make them seem rehearsed or premeditated? Kristin Donnelly, welcome to the Chicago Booth Review Podcast.

Kristin Donnelly: Thank you so much for having me.

Hal Weitzman: We're delighted to have you to talk about video. Well, we all spend all our lives watching videos, don't we, nowadays? And I have children who are barely off watching videos, but there's something that you're studying about video that's called the replay illusion. Tell us about the replay illusion.

Kristin Donnelly: In order to describe this, it's a perceptual phenomenon. I kind of need to take a big step back and talk about what do humans do normally when they encounter a piece of information and how does video distort that? So in everyday life, if I see something happen, let's say you do a cartwheel, you're actually doing a cartwheel. I know this sounds obvious, but it is necessary to sort of reiterate this to explain it. If you do that cartwheel again, you've actually done it twice, but each moment that you're doing something is actually a thing that you're doing. And when you do it again, you're actually doing it again.

Okay. When we as humans, which is relatively recent in our evolution, develop the ability to capture a unique moment in time via recording and play it again and again, that's a relatively new thing, right? That's kind of a different thing than our brains are used to. And what we propose in this paper is that our brains don't fully correct for the fact that this is not a unique event happening more than once. It's one unique event that I happen to be seeing more than once. In other words, we're confusing repetition with replay. And so even though I know intellectually this is not happening more than once, what we propose is that there's still some residue maybe neurologically on how I process that video or stimulus and that bleeds into my evaluation of the content. I can get into that as well.

Hal Weitzman: Okay. Well, first of all, I should say I'm never performing one cartwheel, let alone two. If I perform one, I'll be on my back.

Kristin Donnelly: I feel the same way.

Hal Weitzman: I'd have to be taken to the hospital. So let's not talk about that. But I guess we're all somewhat familiar with this because the earliest thing I can think of that we've all grown up with is sports. You watch someone do something terrible and it looks premeditated the more you watch. And the more you watch it, the worse it looks usually. Nowadays, there's all sorts of video of people online stealing Amazon boxes on Ring cameras and those kinds of things. That is terrible, but all these actions, they look premeditated, right? But as you're saying also, once the more we look at it changes our perception of what is going on. And when your research, this must have been so fun to conduct.

Kristin Donnelly: It was so fun.

Hal Weitzman: You're watching episodes of Survivor and TikTok dances. So what was the most surprising area for you to discover this what we might call replay bias in action?

Kristin Donnelly: So I should note that the replay bias is basically that whatever you see more than once appears to be more prepared, rehearsed, less spontaneous. If you were to deliver a monologue in the exact same way more than once in real life, I may think, "Oh, he's really prepared that he's really... He's not speaking off the cuff. It's not extemporaneous speech. He's really controlling every single thing about the delivery in order to deliver that more than once." And so what we're thinking here is that when people see a video more than once, they get this impression that that person is tightly controlling their actions in order to repeat them multiple times. And so that gives the impression that this thing is more deliberate, not spontaneous, more rehearsed, controlled, and so on. And so the thing I found most fascinating, the stimulus I found most fascinating was this amazing clip we got from a 2009 episode of Cops where this woman, Dalia Dippolito, a Florida woman, I shouldn't laugh. I'm sorry.

It's this video of the police telling her that her husband has been killed and she reacts with all this anguish. She's just like, "Ah, why?" But she actually is guilty. She set up, she got a hitman, hired a hitman to try to kill her husband. And so the police told her that the hit had succeeded, but they didn't tell her obviously that it was a hit and they knew that she was behind it, but they wanted to gauge her reaction. So they told her this. She's serving time in prison right now. So we know that her reaction is not genuine. She wanted her husband killed. So them telling her, "Your husband's dead," and she's freaking out is not a genuine reaction. That video is great because what we were able to show is that by seeing it multiple times, it seems less spontaneous and real and more affected, more controlled, less genuine.

Hal Weitzman: So just talk through that. So you showing it to people who know that she has called-

Kristin Donnelly: No, they do not know.

Hal Weitzman: Oh, they don't know.

Kristin Donnelly: Yeah, that's the cool part. So I know as the experimenter.

Hal Weitzman: So you know, but they don't know, but then the more they watch it?

Kristin Donnelly: The more they watch it, well, I mean, it's really a matter of once versus any number of repetitions, the more guilty she seems. Because the idea is if she were reacting for the very first time, a spontaneous emotional reaction is tethered to the present moment. I can only have one reaction. If I appear to be doing that reaction multiple times to you, that is less genuine. That's definitionally not spontaneous, right? And so seeing her do the same thing, have the exact same reaction multiple times. If we're applying this lens where we think that we don't fully correct for the fact that we're watching something multiple times, it's not happening multiple times, she seems less genuine and thus she seems more like she's implicated in the murder. And I thought that was a really fun-

Hal Weitzman: Fascinating.

Kristin Donnelly: ... stimulus for us.

Hal Weitzman: So as you say, we haven't evolved to sort of process the idea that what we're watching is just a replay. We sort of feel like it's happening again.

Kristin Donnelly: That's what we propose.

Hal Weitzman: She's doing this again. She's back to her old tricks.

Kristin Donnelly: At least a little bit. And look, we know intellectually, we know that this is not the same moment. We know this is we're watching the same thing over and over, but the brain still has some residue from the previous watching, if you will. I'm not going to use residue in a biological sense. I'm just saying there's this impression that this has happened again and we can't quite shake that even though we know this is just a replay clip.

Hal Weitzman: So you gave a sort of negative example, I guess. She looks more guilty, but there's a positive example as well, which is TikTok dancers, the more we watch them, the more we think that the performers are skilled.

Kristin Donnelly: Yes. So any sort of movement through space is if you are able to enact it multiple times in the precisely the same way, it takes a great deal of bodily control to do that. And so things like singing and dancing are a lot about being able to control your body to produce the intended outcome with singing, right? It's the ability to change one's voice to the intended note. With dancing, it's like moving your body through space in the intended way. And if you can nail that multiple times precisely, precise repetition in the real world, that would mean you were highly skilled. If I were able to go over there and do a little TikTok dance and I did it exactly the same way, that would mean I have very tight control over my emotions. And so what we think this means is that when you have something that you see multiple times, it becomes more impressive in a sense because there's more inferences about control essentially.

Hal Weitzman: Fascinating. So you would be more skilled because you're someone who can reproduce that dance over and over again, even though you may only have done it once, but I'm just watching it over and over again. This is one of my favorite examples here because I love bad American Idol auditions and you actually got to study them. I'm so jealous. And you found that... So this is counterintuitive, right? Replaying them actually made terrible singers seem slightly better. What's going on there?

Kristin Donnelly: So that goes back to the idea of bodily control. Even a poorly hit note requires me to have some control to hit it again in precisely the same way. So I think that's what we're observing. It's a little bit unclear in this context why they get slightly better, but I'm going to propose that I think it has something to do with inferences about your ability to produce and reproduce a given note and then reproduce [inaudible 00:09:51].

Hal Weitzman: Fascinating. So you're bad, but you're consistently bad.

Kristin Donnelly: Right.

Hal Weitzman: And your ability to reproduce your badness makes you sit more skilled.

Kristin Donnelly: Right. So there's two forces that are acting on this at once and I think they're reflected in the fact this is a weaker effect than some of the other studies. There's the fact that if you're a good singer, you don't sing poorly. And then there's the fact that if you are a good singer, you're able to hit the same notes reliably. And so what we have here, I think is sort of two forces working on each other where they're giving a bad performance, but at the same time they can give that exact same precise bad performance multiple times. That's what the brain thinks at least, right?

Hal Weitzman: So in one of your studies, just about this replay illusion. So what you typically did was you showed someone a video, then you showed them other videos, so called filler videos, then you went back and showed them the original video again to get the replay. But the bias was there even when you didn't play any of that filler material. So is that illusion instantaneous? Is that what it means?

Kristin Donnelly: Yeah, I would say so. So even back-to-back replay produced the same effects. And interestingly, I can't speak to this for all context, but we did the study where we said, "Okay, is this person doing an improv theater performance or is this a scripted scene?" Because we showed them a little scene and it's from an improv theater performance. So the correct answer is it's improv. And we showed it to them either once, twice, thrice, four, five, six times with the repeated conditions back to back. And what we found is that any amount of repetition was enough to get people to think this is more like a scripted scene than an improvised scene. So in that context, it makes sense because if I am delivering a line extemporaneously and I'm able to deliver it exactly the same way again, it doesn't feel as improvised as something that is not seen multiple times.

So any amount of repetition in that case gave the impression that the behavior was more scripted. So to answer your question broadly about filler videos and I think what you're getting at too is latency. So how long in between viewings do we need? I don't have a great answer for that. I can say in our studies, it was up to a few seconds, so it's not crazy, but we're not able to study this very cleanly in the lab. I think in the real world, you would still see some of this.

Hal Weitzman: Which is fascinating because most of those platforms just play the same video over and over again until you swipe up or whatsoever, right?

Kristin Donnelly: Right. That's very true and GIFs and stuff. Yeah.

Hal Weitzman: If you're enjoying this podcast, there's another University of Chicago Podcast Network show you should check out. It's called Big Brains. Big Brains brings you the stories behind the pivotal scientific breakthroughs and research that are reshaping our world. Change how you see the world and keep up with the latest academic thinking with Big Brains, part of the University of Chicago Podcast Network. Kristin Donnelly, in the first half, we talked about your research about the replay illusion and how jealous I am that you got to watch TikTok videos and American Idol auditions and call it research and publish academic papers on it.

Kristin Donnelly: I'm very lucky.

Hal Weitzman: What a wonderful job you have. Okay. And so interesting the way you describe it. And another phenomenon that you looked at was unboxing videos. People buy products or they get their new book and they're just going to take it out. And to me, they always look totally fake. But you are actually saying that even when there's a genuine first time reaction, if we watch it over and over again, it seems less real. Does it seem less real every time? Is it increasingly unreal or does it just generally feel very staged?

Kristin Donnelly: I would say for the unboxing content, any amount of repetition would make it feel staged because any surprise reaction that we generate is inherently reflective of the present moment. I can't do that multiple times. You can't be surprised in the exact same way more than once by the same stimulus. So with the unboxing videos, what we did was we said, "Okay, here are these influencers who are opening products for the first time ostensibly, voicing their reactions. Do we think these are genuine or not?" And our participants found them to be more fake and less authentic after seeing them multiple times compared to once. But we also had access to this really interesting data set that complemented the unboxing video study where it was like videotaped reactions of participants in a lab watching content. This is called the... Oh gosh, Berkeley reactive, affective video database, something like this, don't quote me.

But it's from UC Berkeley where I got my PhD and they bring people in a lab and they're like, "Here's this video." And they videotape them watching the video. So we have videos of people's actual surprise reactions in response to something. So these are verifiably real, right? They're captured in a lab, they haven't seen the content ahead of time, nothing like that. And you still see that people think that that is less genuine.

Hal Weitzman: They're watching those reaction videos and they still think that-

Kristin Donnelly: They still find them [inaudible 00:15:29].

Hal Weitzman: ... they're staged. Fascinating. Is it possible they think it's staged because you are showing it to them in an experiment?

Kristin Donnelly: So we have various ways of trying to get around that. We change the wording of the question to try to say, is this person faking it versus being authentic? We switch it around. In psychology, how you frame the dependent measure of interest like the thing you're measuring matters because the participant will infer things about what you want from them. So we did what I think are some clever approaches to try to get around the concern that we have experiment or demand effects going on. So I think we eliminated some of that.

Hal Weitzman: Fascinating. Now you're a professor of marketing. So I mean, one of the things you're looking at is commercials, appearance of authenticity or not. And you looked at ads run by Chevrolet that are supposed to have real people, not actors. And this is a whole genre-

Kristin Donnelly: It's a whole genre.

Hal Weitzman: ... of commercials, right? So does your research suggest that these advertisers are hurting their credibility? Because even though they get supposedly real people, when you see the ad replayed over and over again, they cease to become real people in your mind.

Kristin Donnelly: Yeah, exactly. So this is a specific genre within advertising and it doesn't capture every single ad, of course. It's just this one type of ad that says, "Here's a testimonial from a real customer. Here's someone interacting with our product for the first time," whatever it is. Those types of ads, I would say based on this research, my best recommendation when launching those kinds of ads is keep in mind that if you target the same person to see it multiple times, they may find that supposedly real customer behavior to be more staged and contrived and less spontaneous. And that's interesting because advertisers have this constant tension between getting as many people as possible, getting your message out to as many people as possible and targeting a specific group or specific individual. And so I would say for that type of ad, the former approach is better.

Hal Weitzman: Sorry, the approach being getting it out-

Kristin Donnelly: The former approach where you... Yeah. Where you try to get out the actual.

Hal Weitzman: But you mean you get it one and done like on the Super Bowl?

Kristin Donnelly: Yeah.

Hal Weitzman: That would be the time to have the real people. Okay. You studied audio as well as video. What were the differences there?

Kristin Donnelly: Yeah, that was fun stimuli too. We got these audio clips of laughter that was captured in a lab by someone who was studying laughter and it was captured spontaneously between friends, which is real laughter and also in a fake setting where they're like, "Here's a mic," and you fake laugh and you're like, "Ha ha ha." I can't do it. I'm not going to do it. And so basically what we find is that people, regardless of the authenticity of that original laugh, find the laughs less authentic the more they hear them. And that study was actually, we took the laugh and we embedded it in... It was like we asked other questions about the laugh in addition to authenticity to try to hide our true intent about what are we doing here. But I thought that was interesting because we have genuinely real laughs and genuinely fake laughs and both seem less real.

Hal Weitzman: Fascinating. Yeah. I was kind of thinking when I was a kid, all the comedies had canned laughter, but I wonder if that was real, if it was fake or whether it was real laughter that was just played over and over again.

Kristin Donnelly: I don't know.

Hal Weitzman: And ended up sort of losing any kind of reality that it would actually have had. Now you looked at legal context, which of course is very, as you referred to earlier, but Cops is not, I mean, that's a TV show. But there's real legal context where people watch video of crimes and increasingly, right? And that how often we watch it will affect how we think about it. Just talk a little bit about that.

Kristin Donnelly: Yeah. So for video clips in, let's say jury decision making that are shown repeatedly, if the behavior involves something where someone could say where it's critical whether the person is behaving in a controlled or spontaneous manner, I would say replay is a bit... It has some consequences as this research shows. It's tricky, right? Because in a lot of cases you want to see the thing more than once and watching a video more than once does give us more information every time we watch it. But there's a bit of a trade-off here with perceptions of spontaneous action. And so if the video is showing something that is where whether or not the person is acting spontaneous is critical for the jury's decision, then I would say replay would matter and people should be careful with that in a courtroom. But there are of course lots of videos that exist in the courtroom that have nothing to do with spontaneous action. And in those cases, it's probably fine.

Hal Weitzman: Okay. But I'm just thinking of things like police body cameras and that sort of thing. I feel like I've heard somewhat that people are aware of this bias, but is it baked in enough in our understanding, our legal understanding, our legal system to ensure that video should be limited in its use?

Kristin Donnelly: I can't speak to that. I don't know. However, I would say there's always trade-offs between having more information, more viewings give us more information and changing impressions. If the impressions being changed are not critical for judgments of guilt or innocence, fine. And this is just something to keep in mind. This work is something to keep in mind if you do have a situation where the actions could sway the verdict.

Hal Weitzman: I'm just wondering, because I know that part of your study here was asking people to report how many times they watched a video in a legal context. Did that help neutralize the bias?

Kristin Donnelly: Yeah. So that did help neutralize the bias and that could be a very easy low lift intervention. I'm glad you're bringing that up. I don't know who is in charge of showing jurors evidence. I don't know whoever that is, maybe lawyers from the other side, whatever. If there were some low lift intervention where you say, "I acknowledge that I have seen this video X times." Reminding people of that, we found at least, seemed to reduce the effect.

Hal Weitzman: The sense that it's premeditated or that it's planned or rehearsed or whatever.

Kristin Donnelly: Yeah.

Hal Weitzman: Okay. Fascinating. I want to go back to YouTube though because it end on a lighter note. So many of us are watching video, billions of videos every day being watched on YouTube more and more all the time. And so now we're presumably thinking, "Ah, that's fake. That's not real." And now a lot of it isn't, I mean, because of AI and everything else, there's even more and more. I actually even today just saw a podcast with two completely fake hosts that was completely made up.

Kristin Donnelly: Oh my gosh.

Hal Weitzman: And it's all generated by AI. So I'm out of a job, but I'm just wondering, do you think that we are living in... Is this making us more cynical, this repetition of video that's, as I said, is baked into these platforms?

Kristin Donnelly: I think there are many forces at once making us more cynical, but one of them may be repetition for videos where you have someone talking off the cuff, for example, or just doing a simple action more than once... Or sorry, being shown more than once will change people's view on it. So I do think that that changes makes us more cynical about authenticity in general, though the larger problem is really just the number of things we're exposed to, the amount of AI slop, et cetera now. But yes, of course, I think it would definitely play into that.

Hal Weitzman: Well, Kristin, this was an entirely unrehearsed conversation.

Kristin Donnelly: Yes.

Hal Weitzman: Thank you very much. So I guess we should say to people, if you enjoyed it, don't listen again.

Kristin Donnelly: Yes.

Hal Weitzman: Thank you very much for coming on the Chicago Booth Review Podcast.

Kristin Donnelly: Thank you so much for having me, Hal. Appreciate it.

Hal Weitzman: That's it for this episode of the Chicago Booth Review Podcast, part of the University of Chicago Podcast Network. For more research, analysis, and insights, visit our website, chicagobooth.edu/review. When you're there, sign up for our weekly newsletter so you never miss the latest in business-focused academic research. This episode was produced by Josh Stunkel. If you enjoyed it, please subscribe and please do leave us a five-star review. Until next time, I'm Hal Weitzman. Thanks for listening.