Jennifer Coane and I chatted about how her life as an academic has changed as a result of COVID-19, her recent paper with Kimberly Bourne, Sarah Boland, and Grace Arnold (pictured below) published in Cognitive Research: Principles and Implications, and what the results of that research may mean during these challenging times. The paper is called “Reading the News on Twitter Source and Item Memory for Social Media in Younger and Older Adults.”

The authors found that tweets were remembered better than headlines for younger and older adults alike. Read on or listen to the interview to learn more about the research.

Transcription

Intro

Caballero: You’re listening to All Things Cognition, a Psychonomic Society podcast.

Now, here is your host, Laura Mickes.

Interview with Jennifer Coane

Mickes: I’m speaking with Jen Coane about a paper that she’s recently published with, I believe, student coauthors and the journal CR:PI or Cognitive Research: Principles and Implications.

Hi Jen! Thanks so much for talking to me.

Coane: Hi Laura. Thank you.

Mickes: Things have drastically changed because of COVID-19 and so before we talk about your paper, do you mind telling me a little bit about how your work life might’ve changed?

Coane: Yeah, it, it has changed. So like I think most faculty, I am teaching remotely. It’s been a bit of an adjustment and obviously I no longer have a lab. Research assistants were all sent home.

It’s a little interesting. Um, I do have some students who are planning on doing research. The original research plan was for them to do it on MTurk and Prolific. So we’re going to try to continue with that. So, but of course, we’re trying to negotiate all these things that normally we would be doing.

We’d have a meeting and I’d walk them through how to set up the program and how to connect it and how to post it. We’re doing all this remotely, so it’s a little more challenging. And once they get data, it’s going to be much harder I think for them to do the coding and the scoring and all the transformations just because it, they’re working in a team, but they’re all isolated from each other as well.

Mickes: Are you in Maine right now?

That’s right, yeah.

Coane: Yeah.

Mickes: Is there a lockdown now in Maine?

Coane: So we are in the shelter in place, so it’s the classic, you can only leave for exercise, walk the dog and essential business and shopping. So you can go to the grocery store.

Apparently, because this is Maine, you can also go buy a gun.

Mickes:

So clearly you bought some guns.

Coane:

Um, sure. yea.

I don’t even know where a gun shop is.

[Commentary edited out.]

I live in a small town, so you know, I walk my dog every day and see about the same number of people I normally see.

Mickes: That hasn’t changed too much.

Coane: No. And as a friend of mine put it in New England, if you’re closer than six feet to somebody, normally they kind of look at you like you’re invading their personal space.

I still get paid.

Mickes: Students are still learning.

Coane: We think, we hope. [Laughs]

I get to see them and know like Zoom meetings, but it’s okay. It’s just going to be an interesting shift.

I did do a really quick speedy batch of data collection the day after our students were told they had to leave.

Some of my students actually suggested, you know what? This would be a great flashbulb memory study.

Mickes: Ohhh.

Coane: We’re a residential college, a very tight-knit small community, 2,000 students. And so being told it’s 11 o’clock on Thursday morning, you have to be off campus by Sunday.

Mickes: Oh.

Coane: There were students bursting in tears in the middle of class and my first thought obviously was, as a professor, is wait, you’re checking your email in class? Shame on you!

Mickes: [Laughs]

Coane: So I collected data from about 60 people, which I haven’t looked at yet because I need to do a second wave of followup about four weeks out to see whether people’s memories for their COVID related things have changed or not.

Mickes: I was going to ask you that. If you’re doing any specific COVID-19 related research.

Coane: Yes.

Mickes: So yea!

Coane: It was very spur of the moment and I really have to give my students credit because I had this conversation with some of my students about how hard it is in a sense to do flashbulb memory research because you basically, something really big has to happen and your first thought has to be, well, let me collect data on what people remember about this and presumably have an IRB ready. So I did get an expedited IRB.

Yeah. And my students emailed me the evening they found out they had to leave. It occurred to us that you should do memory research on this and I thought, I have the most amazing students.

Mickes: Those are great students. It’s such a good idea.

So then you’ll email them or you’ll ask them similar questions or the same questions in four weeks.

Coane: Yup.

Mickes: Will you ask about their confidence levels, please?

Coane: [Laughs] I think I did.

Mickes: Yea!

Coane: I put this survey together really quickly. I think I asked something about how vivid their memories were and how confident they were that their memories were accurate. I asked them about their memories for the announcement and another memory for that week. So I have a comparison memory.

Mickes: I’ll have to find out in four weeks from now or whenever you can get to the data.

Now onto your paper that you published in CR:PI.

I know this paper for various reasons. Could you tell the audience a bit of background about the work?

Coane: Yes. So as you mentioned, the three coauthors on the paper are all, well my now former students and they’re all in graduate programs doing wonderful stuff. And this project started from a class project, at Colby College in the US where I teach, students are assigned a selection of target articles.

So recent publications that we think have some really interesting implications for extensions and followups. And that year and actually several other years we picked one of your papers, the Major Memory for Microblogs [Paper published in Memory & Cognition).

Mickes: A great choice. [laughs]

Coane: I agree.

Their task was to design a conceptual replication with an extension. In the original Major Memory for Microblogs across three experiments, you and your colleagues compared memory for Facebook posts or other forms of social media posts like comments on news websites and consistently found that these, off the cuff extemporaneous, gossipy kind of personal output was remembered better than pretty much anything else – faces, other sentences, other types of headlines and comments that were less gossipy.

Coane: And so my students were really interested in this and the original design that we did during that semester-long project was to take the basic, let’s compare social media posts to some other format.

And they picked tweets versus headlines to try to have something that was comparable in terms of a complete message. But clearly social media and clearly not social media. And their other interest was in seeing what happens if we give people information about the source of this material. Because in your original study that especially the first experiment, the Facebook posts were just sentences taken out of context and there were some competing hypotheses.

One is that if people know that this is a Facebook post versus a headline or a tweet in our case versus a headline from a reputable source. One hypothesis is that people recognize, Oh, this is social media, this is written by some person who has no credentials and they might just discount it. Well, there’s a lot of research suggesting that how we attend to information impacts what we remember. And there’s some research suggesting that people may not devote as much attentional resources to say user-generated material or social media stuff.

So once we decided to manipulate perceived source, so where they photo-shopped all the items to either look like tweets or look like headlines from CNN’s website [examples pictured below]. And then we crossed. So we had tweets that were presented in their original format. And so we had congruent items. So CNN headlines that were presented with the little CNN logo and all the byline and date and everything, and then tweets that were similarly presented. And then they also crossed to the stimuli. So some tweets were presented, the content, the sentence, it was a tweet, but it was dressed up to look like a headlight and vice versa.

This was also in a sense, a stronger test of: is it the content? Is there just something about the sentences, the way people write tweets or what they write about that makes them memorable regardless of whether you think it’s a tweet or a headline? And that’s pretty much what the students found.

They replicated your effect, the major memory for microblogs.

Mickes: Thank God.

Coane: [Laughs] Always good to be replicated!

Mickes: [Laughs] Yeah.

Coane: They found, again, this significant advantage. The tweets were remembered better than headlines.

Mickes: Can I stop you?

Coane: Yea.

Mickes: The tweets were remembered better than headlines. And that’s whether or not it was dressed as a tweet or a headline.

Coane: Absolutely, yes.

Mickes: Okay.

Coane: So this suggests that there is really something about the content.

So knowing that it’s a tweet or hold that it’s a headline had absolutely no effect.

Mickes: Amazing.

Coane: And the second test they did was they gave people source memory tests. So after participants said yes, I remember seeing that item, we gave them old or new recognition. They were asked when you studied this item, did you see it as a tweet or was it a CNN headline?

At test, all the items were presented just typed what sentences on a white background. So there was no source information at test.

Mickes: And how did they do?

Coane:

And what we found originally just with young adults was that we found a nice congruency effect. So source for tweets that were dressed up as Twitter posts was better. Source for headlines that were presented as CNN was better.

And overall younger adults still have this advantage. They still remember the source of better than the source of headlines.

So this major memory for microblogs also extends beyond item memory. It also extended the source memory.

Mickes: I love it.

Coane: After they finish data collection, I met with the students. Three of the four students were actually working in my lab as research assistants. Wow, these data are really cool. And I do some cognitive aging research and we thought, you know, there are a lot of interesting questions to ask with this very same design in older adults.

First of all, we know older adults tend to have impoverished source memory.

Mickes: Right.

Coane: Any memory that relies more on recollective controlled processes tends to show larger impairments in aging compared to healthy younger adults. We also know that although use of social media among older adults is increasing, they still lag pretty far behind.

Mickes: Right.

Coane: They don’t use Facebook or social media and so it might be that it’s just not as interesting to them. Maybe the content of the tweets is completely irrelevant to their personal life. And so we had some interesting questions about whether a) is this major memory for microblogs, this, you know, better memory for this social media content that seems to persist across types, is it general or are there boundary conditions in that it only occurs in the users? The end-users of the social media and older adults are like, I have no idea what you’re talking about. This is meaningless to me.

We do know from lots of lines of research that as we age, our knowledge base gets richer and richer and older adults often can capitalize on this increased knowledge base, their semantic memory to sort of compensate for some of the deficits and episodic or item-specific memory.

But it really depends on them having this rich base of knowledge that they can sort of attach information to. If the tweets or Facebook posts are about unfamiliar content that they might struggle more to make those connections.

First of all, does this effect generalize to different populations and how much does it depend on experience with the media on or familiarity with the style and content?

And so the summer after my students collected the data, we replicated the study with a sample of older adults. We categorize older adults as 60 plus.

And when we replicated the study with the older adults, we basically found – much like the younger adults, they showed the same effect for item memory. So they also showed this nice advantage or social media posts for the tweets.

Mickes: It still blows my mind that this, this is a thing. [laughs] When we first did it in the lab, we couldn’t believe it, kept redoing it.

Really. People remember these posts better than anything else, including headlines? And then you replicate it. And with older adults it’s still, it’s incredible to me.

Coane: It was really exciting.

And again, and for older adults, the formatting didn’t matter, which we thought was interesting because we sort of went to unexpecting that if older adults see a bunch of things that look like tweets, they’re going to be checked out. What are these things? Who are these people? What are they saying? And yet they remembered the items better anyway.

And on the source memory test, we found some really interesting things with our older adults. So first of all, people do make mistakes on a memory test. They’re going to say yes to something that they didn’t actually see. Overall false alarm rates, these incorrect yes responses, were pretty low. Not much of an age difference between younger and older adults.

These are meaningful, rich materials. It’s not just like an isolated word. So they’re a little bit richer. And what we found is that if you say yes to something you didn’t see, you’re still going to get the source memory question and the program is set up to then say what source did you see it in? And of course, the correct answer is neither because I didn’t actually see this item.

And so we looked at their source memory decisions for these errors and what we found was that older adults were actually in a sense better at guessing what the original source was. So they said if the item that they had seen for the very first time with no context, if that was a tweet, they were more likely to say, I saw that as a Twitter post. And if the item had been a headline, they said it was CNN.

In a sense, they were right because they were sort of skipping the whole experimental manipulation and just recognizing where we got that item from and the fact that older adults were better at this suggests that they’re better at figuring out what qualitatively makes a headline, a headline and a tweet, a tweet.

They are really sensitive to either linguistics, stylistic, content things more so than younger adults. The older adults source data, we talked about their better identification of false alarms when we asked them for the items that they had correctly identified as seen, older adults had this really, really big congruency effect. So when they saw the tweet as a tweet, they were really accurate.

If we looked at raw accuracy, they were as accurate as younger adults. But when they saw the tweet dressed up as a headline, they basically said it’s a tweet; I saw it as a tweet and vice versa. And we also did some signal detection analyses and what signal detection analysis revealed is that although looking at the raw data, older adults for the congruent items seem to be as accurate., it’s because in a sense they were just entirely responding based on content.

Mickes: Right!

Coane: So if it looked like a tweet, I must’ve seen it as a tweet. If it looked like a headline, I must have seen it as a headline. Suggesting that older adults are really relying much more on the content to make these decisions. we were considered some of the more practical implications which are if you come across a news item or a news story or a headline and it looked like a headline, let’s take an example, but it was on The Onion or the Borowitz report or one of these satirical sites, but it looks like a headline, you might misattribute to being a newsworthy item because the content has sort of a headline you feel if it’s written to sound professional and headline-ish.

Older adults seem to be more susceptible to thinking, oh, this sounds like a headline. Therefore it must be a headline. Whereas this sounds like a social media post. Therefore it must have been a social media post.

Mickes: I always ask, what are you going to do next? Are you doing something to follow up on that?

Coane: I would like to follow it up. One of the things we’re following up now in my lab is sort of extending this basic idea of how does the source influence your memory for items. And we’ve added on a twist of looking at fake versus real news headlines and we’re still in the data collection phase and we were starting with younger adults.

Mickes: So how do you think these findings might help during the COVID-19 crisis?

Coane: Well, I think obviously there is a number of ways in which it’s relevant. So obviously the fact that we’re all stuck at home and basically living online and that there is so much news and so much information going around that people are in some cases I expect almost obsessively checking social media or news sources. And this issue that we identified in our research, that source memory, in particular, seems to be disrupted when you read content on sources that are non traditionally associated. So when you read the news in a social media context, you might misremember where you actually read that.

The issue, which I think is going to be really important as a next step, is to determine the relationship between where you obtain information and the perceived credibility.

Mickes: Right.

Coane: So you know, there’s a lot of expert opinions. So Dr. Fauci is trying to do his job. There’s a lot of non-expert opinions being distributed and are they being perceived as more or less credible because now they’re appearing on this site versus that site. And so, and if you have credible experts with actual public health epidemiology, medical credentials, sort of mixed in with holistic scientists who are saying, Oh no, your body will fight off the virus, just drink ginger water or something.

Mickes: [uncomfortable laugh].

Coane: Is, is there sort of this blending of what people can on to? And you could even potentially imagine some source of confusion. Do you have one story in your social media feed from the New York Times or another source and right below it is one from another. We know that there’s a lot of source confusion.

Mickes: Right.

Coane: So we know this happens. And so I think the potential is that this could sort of aggravate the spread of either misinformation.

In an ideal world, it would facilitate the spread of accurate information because maybe you’re just being exposed to it more. And if the information is accurate, then I, I would argue source doesn’t really matter.

Mickes: Right.

Coane: Like if the information is real and verifiable and accurate, but it’s that fuzzy area of either inaccurate or flat out misleadingly false information that is disseminated for nefarious purposes and that I think is potentially problematic.

Mickes: Here’s a question that I think we could end on. If you were to write a tweet about this paper, what would it be?

Coane: Oh, let’s see. I’ve never tweeted in my life, so this is probably going to be over the character limit. [Laughs]

Mickes: I’ll check. [Laughs]

Coane: I would probably tweet something along the lines of, you may remember this tweet, but you may not remember where you saw this tweet.

Mickes: Oh, that’s really good.

I had one and mine would be: 18 to 80 you’re gonna remember this tweet.

Coane: Oh, I like that.

Mickes: I think I like yours better.

Thank you so much for talking to me about your work and I really look forward to reading and hearing about the followup work that you do.

Coane: Thank you, Laura. It’s been really fun.

And I just wanted to thank you again for doing that original research, which inspired my students to come up with this design, and I do really need to give them credit. It was their idea and they did so much of the heavy lifting, and at least one of my students got really, really good at signal detection.

Mickes: That warms my heart. That makes me feel so good. Oh, thank you.

Concluding statement

Caballero: Thank you for listening to All Things Cognition, a Psychonomic Society podcast.

If you would like to share your feedback on this episode, we hope you’ll get in touch! You can find us online at www.psychonomic.org, or on Twitter, Facebook, and YouTube.

If you liked this episode, please consider subscribing to the podcast and leaving us a review. Reviews help us grow our audience and reach more people with the latest scientific research.

Featured Psychonomic Society article:

Bourne, K. A., Boland, S. C., Arnold, G. C., & Coane, J. H. (2020). Reading the news on Twitter: Source and item memory for social media in younger and older adults. Cognitive Research Principles and Implications, 5, 11 (2020). https://doi.org/10.1186/s41235-020-0209-9