Doctored photographs create false memories of spectacular childhood events. a replication of Wade et al. (2002) with a Scandinavian twist

ABSTRACT
 Can exposure to a doctored photograph of a plausible yet fictitious childhood event create false memories in adults? Twenty years ago, (Wade, K. A., Garry, M., Don Read, J., & Lindsay, D. S. (2002). A picture is worth a thousand lies: Using false photographs to create false childhood memories. Psychonomic Bulletin & Review, 9(3), 597–603) found that half of the participants reported false beliefs or memories after multiple interview sessions about a doctored photograph of themselves as children on a fictitious hot air balloon ride. In this replication, which rigorously recreated the method and procedure of Wade et al. (2002), participants were interviewed over three interview sessions using free recall and imagery techniques about three true and one fictitious childhood event photos. The balloon ride was modified to a culturally appropriate target event – a Viking ship ride – to ensure that the doctored photograph was functionally equivalent. The results showed almost identical patterns in the two studies: 40% (n = 8) of the participants reported partial or clear false beliefs or memories compared with 50% (n = 10) in the original study. The participants who reported false memories reported detailed and coherent memory narratives of the Viking ship ride not depicted in the doctored photograph. Our study successfully replicating the results of Wade et al. (2002), suggest that memories can relatively easily be implanted, regardless of cultural setting.


Introduction
The science of the fallibility of memory has demonstrated that our childhood memories do not necessarily provide a veridical representation of flawlessly recorded events as they occurred. Rather, our recollection of past events is susceptible to errors and biases and can be fragmentary, contain incorrect details, and, oftentimes, be entirely fictional (e.g., Howe & Knott, 2015). The origins of such memory illusions are numerous and multidetermined, having the potential to emerge spontaneously, be constructed within the individual, or be implanted via exposure to fictitious events (e.g., Calado et al., 2021;Loftus, 1997;Loftus & Pickrell, 1995).
Twenty years ago, Wade et al. (2002) demonstrated in a highly cited Psychonomic Bulletin & Review publication that exposure to doctored photographs of fictitious childhood events could create false childhood memories in adults. Twenty participants were shown three pictures of realworld experienced childhood events alongside a doctored photograph in which the participants' childhood pictures were merged with a photo of a fictitious hot air balloon ride. In three interview sessions, the participants were encouraged to recall both the true childhood events and the fictitious event using free recall questions and guided imagery exercises. By the third interview session, half of the participants (n = 10) reported detailed, eventspecific memories of the balloon ride, even though it was verified prior to the exposure that they had never experienced such an event. Wade et al. (2002) was one of several studies in the body of memory implantation literature. The novel idea of using self-referent doctored photographs rather than verbal reports of fictitious narratives (e.g., Loftus & Pickrell, 1995) rapidly caught on, and variations of the hot air balloon ride were subsequently published (Garry & Wade, 2005;Lindsay et al., 2004;Strange et al., 2008). With a single exception (Hessen-Kayfitz & Scoboria, 2012), all follow-up studies involved the authors of the original study.
Close and independent replications, both confirming and countering the original findings, are crucial in cumulative science and play an important role in the advancement of empirical research and scientific progress (e.g., Asendorpf et al., 2013;Brandt et al., 2014;Nosek et al., 2012;Peels, 2019;Rosenthal, 1990;Shrout & Rodgers, 2018). However, replications rarely appear in psychology journals (Makel et al., 2012). The present study provides a close and independent replication of Wade et al. (2002) and was rigorously executed by recreating the exact methods specific to the original study's design. Only minor adjustments were made to accommodate contemporary and cultural variables, most notably substituting the hot air balloon ride with an equally spectacular, but more likely in Scandinavian countries, Viking ship ride.

Participants
The sample consisted of 20 participants with the same gender distribution (N = 10 females, 10 males) and age range (18-28 years) as in the original study. The average age of the participants (M = 21.7, SD = 2.1) was similar to the reported mean age (M = 20.2, SD = 2.3) in Wade et al. (2002). The participants were recruited using the exact same inclusion criteria as reported in the original study. The participants were a) at least 18 years old, b) had not taken a psychology class, and c) had not experienced the false target event (Viking ship ride) in the doctored photo. In recruiting the participants, we identified family members who satisfied the inclusion criteria defined in the original study who also had confederates who could provide access to a necessary number of childhood photographs. The participants and confederates were contacted via phone and invited to participate in the study. All participants signed a written consent form. Four participants were excluded during the initial interview because of having experienced a Viking ship ride in their childhood. Four additional participants who satisfied the inclusion criteria were recruited as replacements. Permission to conduct the replication was granted by the Norwegian Centre for Research Data.

Materials
The confederates provided a selection of photographs in which the participants were four to eight years old. Three true photos of moderately significant childhood events (e.g., birthday parties or family holidays) per participant were selected and digitised. The false target photo was created by digitising additional true photos, cropping the face of the participant, and pasting it into the prototype photo. The prototype photo and facial cropping were light adjusted to match. A Windows 11 operating system and Photoshop version 22.5.7 were used in this process.
In Wade et al. (2002), a hot air balloon ride was chosen as the target false event "because it is an activity available to all New Zealanders […], yet it is significant enough for family members to confirm that the subject never experienced it" (p. 598). In the present study, the original target event was modified to a culturally appropriate target event, a Viking ship ride, to ensure that the manipulated photo was functionally equivalent and effectively captured a construct similar to that of the original study. Riding in a Viking ship is considered a distinct activity that is potentially available to all Norwegians, regardless of area of residence. Thus, the target event was significant enough for family members to confirm whether the participants had experienced it. Figure 1 shows an example of a manipulated target photo, in which an image of the participant's face was doctored into the prototype picture. Informed consent to include the picture as an example of a manipulated target photo was given by the interviewee.

Interview procedure
The interviews were conducted by eight interviewers who were trained in a modified version of the stepwise interview procedure (Porter et al., 1999;Yuille et al., 1993) by an experienced interviewer (first author) and practiced mock interviews prior to data collection. The interviewers followed the same interview procedure as described in Wade et al. (2002) and the interview transcripts were checked to ensure that the interviewers adhere to the interview script. The interviews were executed in Norwegian, and the excerpts provided in the results section have been translated into English. All interviews were audiotaped, and the audio-recorded interview data were transcribed verbatim.
Each participant completed three interview sessions over a 7-16-day period in accordance with the procedure described in the original study. During the study period, the participants were instructed not to discuss the events with family members, and the confederates were instructed not to discuss the photos with the participants.
The interviews were conducted individually over Zoom, and the images were presented with the screen share function, which diverges from the original study in which the interviews were conducted in person. In the interviews, the three true and one false photograph were labelled events 1-4 and presented to the participants (one event per page) in full view during the interview sessions. As in the original study, the doctored photograph was consistently presented as picture 3.
The initial interview started with introductions to inform the participants that the purpose of the study was to investigate how and why people reminisce about childhood events. For each pictured event, the participants were asked to give a free narrative of everything they could recall from the event shown in the picture by encouraging them to "tell everything you can remember without leaving anything out, no matter how trivial it may seem." Follow-up questions were used to help the participants clarify details of recollections mentioned in the free recall phase. When the participants exhausted the recollection of events shown in a picture, their confidence regarding whether the pictured event had actually occurred was measured on a 7-point scale ranging from 1 (0% confident) to 7 (100% confident). Those participants who could not recall the pictured event were reassured by the interviewer that "many people can't recall certain childhood events at first because they haven't thought about them for such a long time," which was followed by a few-minute break to concentrate on retrieving the memory for the pictured event. If the participants still could not recall any details, they were told that another purpose of the study was to assess the efficacy of different memory retrieval techniques, including context reinstatement (CR) and guided imagery (GI).
In the CR procedure, the participant was mentally taken back into the scene of the event. For the Viking ship ride, the interviewer instructed the participants to close their eyes, imagine being back on the Viking ship, and to think for a minute about details, such as what the event might have been like, who was present, what the weather was like, and what they saw when they were onboard the ship. If the participant still could not recall any details, the interviewer introduced the next photograph and repeated the procedure. At the end of interview 1, the participants were sent copies of the four photos by email and asked to take a few minutes each night to focus on recalling the pictured events.
The second interview session was conducted 3-7 days after the first interview. The participants were asked whether they had remembered more about the pictured events they were interviewed about in the first interview, followed by a repetition of the CR and GI procedures.
The third interview session was procedurally identical to interview one and took place 3-7 days after the second interview session. It was finalised with a debriefing session in which the interviewer informed the participant that one of the photos was manipulated and that the event it depicted had not occurred.

Coding of transcripts
The first and third interviews were audiotaped and transcribed word for word. Verbatim transcriptions of the interviews constituted a total of 159 pages of interview data. The interview transcripts were coded in accordance with Hyman and Billings's (1998) criteria for categorising false memories, which was the exact same coding scheme used by Wade et al. (2002). The coding scheme make distinctions between clear and partial false memory, trying to recall, and no memory responses. To be coded as having a clear false memory, the participants had to report memories of the Viking ship ride, including consistent elaboration of information not depicted in the photograph. A partial false memory was coded when the participants consistently elaborated on the false photograph, such as reporting contextual description, emotional content or persons who were present but did not indicate memories of actually taking the Viking ship ride. Trying to recall was coded when a participant tried to recall the false event and described images of it but did not claim these images as memories. The no memory category was coded when the participants failed to recall any of the event shown in the doctored photo.
The memory reports were coded by three independent raters, who were carefully trained in the coding procedure. The raters classified all participants' responses to the false event, agreeing on 86% of the interview coding. The disputed 14% were classified into the most conservative coding category. In Wade et al. (2002), the coding agreement between raters (i.e., 85%) was almost identical to ours. Figure 2 shows the percent of true events recalled in the first and third interviews compared with the findings reported by Wade et al. (2002). In the replication, the participants recalled 52 (87%) of the 60 true events in interview 1 and 57 (95%) of the true events in interview 3. In comparison, Wade et al. (2002) reported that their 20 participants recalled 56 (93%) of the 60 true events in interview 1 and 58 (96%) in interview 3. On average, the participants recalled 2.8 memories of the true events by interview 3 in the replication compared with 2.9 true events in Wade et al. (2002). Figure 3 shows the percent of participants who reported either clear or partial false memories (combined false memory rate) in the first and third interviews in the replication compared with the findings in Wade et al. (2002). In the replication, one participant reported clear false memories, three participants reported partial false memories during the first interview (20%), six participants were trying to recall, and 10 participants had no memory of the fictitious event. In Wade et al. (2002), 7 of the 20 participants (35%) remembered the false event either partially or clearly during the first interview, of whom six participants reported partial false memories and one participant reported clear false memories.

Recall rates of false events
In the third interview in the replication, eight (40%) participants recalled the false event either partially or clearly and reported coherent details of a Viking ship ride in their childhood, compared with 10 (50%) participants who recalled similar details of a hot air balloon ride in Wade et al. (2002). In the replication, the one clear false memory report from interview 1 remained clear, and two partial false memory reports developed into clear false memories in the third interview. Two participants' trying to recall developed into a clear false memory, one no memory response from the first interview became a clear false memory in the third interview, and one no memory response in the first interview became a partial false memory in the third interview. One partial false memory remained partial in the third interview. Four participants who were trying to recall at interview 1 remained in this category in interview 3, as was the case of the seven participants who had no memory responses. One no memory response in the first interview became a trying to recall in the third interview. Figure 4 shows the percent of participants who reported either clear or partial false memory, trying to recall, and no memory in the first and third interviews in Wade et al. (2002) and the replication. A chi-square test of independence showed that there was no significant association between the ratings of the four coding categories (clear false memory, partial false memories, trying to recall, no memory) in interview 1 in the original study and the replication, X 2 (3, N = 40) = .57, p = < .05. Similarly, for interview 3, a chi-square test of independence showed no significant association between the ratings of the four coding categories in the original study and the replication, X 2 (3, N = 40) = 2.73, p = < .05.

Confidence ratings
In agreement with Wade et al. (2002), the replication showed that the participants were more confident that true events had occurred than that the false events had occurred. On average, the participants were 92.2% confident that the true events had occurred in the replication, compared with 90.8% in Wade et al. (2002).
The percent of participants who rated that they were confident that the false event had occurred was almost identical in the replication (43.8%) and original study (44.5%). Moreover, 12.4% of the participants in the replication who reported having no memory of the false event were confident that the false event had occurred, compared with 10.0% in Wade et al. (2002).

The formation of false memories during the course of the interviews
During the course of the interviews, the participants who developed clear or partial false memories recalled consistent and detailed descriptions of the Viking ship ride that were not depicted in the photograph, including naming persons who were present on the ride, what the weather was like (cold, gloomy, misty), how they rowed in the boat, the time of year (late summer, autumn), distinct emotions, and naming places that they travelled to in the ship. They also described distinct activities that they did when they went on land, such as attending a Viking market, looking at ruins, eating certain food, participating in a play battle, and wearing Viking costumes and helmets.
In the original article, Wade et al. (2002) commented on the dynamics in the interviews and pointed out that the participants' false memories tended to change over the course of the interviews: "The subjects who ultimately  reported memories of the false event typically began by offering fragmented descriptions of the balloon event and subsequently embellished the suggested information over time" (p. 600). In the replication, we noted a similar tendency in those participants who reported partial or clear false memories during the interviews that gradually developed. These participants displayed an expansion in richness and level of details between interviews 1 and 3. Three examples illustrate this process.
The first example is from a participant who reported full-fledged false memories in the first interview: Interviewer (I): Let's look at picture 3.

Participant (P): Hm. Wow, what is this? A Viking ship?
I: Now, I want you to tell me everything you remember about this event, whatever it is, and add as many details as you can remember.
P: This boat gives me a weird feeling. I am a bit uncertain where this is. It must be a school trip in elementary school, second grade, I would guess when looking at the picture. The lifejacket looks familiar.
[…] It looks cold … I don't remember anything else, just the feeling of being there, and the lifejacket that wasn't mine, but I borrowed it from someone.
I: Now, I want you to think back to the trip with the Viking ship and try to remember as many details as possible. So, I want you to think for a minute on all the details of how it was to be on the trip on the Viking ship. P: It's the lifejacket that triggers something. And that it was with my class. We rowed quite far, and my arms got tired, I remember. I wasn't particularly strong. It was cold and gloomy. I think it was late August. Still summer, I think, even though it was cold, closing in on the autumn, and the air was damp. We were unlucky with the weather, and I was nervous about the trip because of that. The boat went really fast. I remember the sea spray on the side of the boat, and my pants got wet. I started to freeze because we were out on the boat for so long. The whole day, I think. We only got home when it was dark. There was a lantern in the harbor that we entered when the trip was done.
A second participant was uncertain and did not remember anything specific about the Viking ship ride in the first initial interview but gave the following account in interview 3: Interviewer: Now, I want you to tell me anything you remember from this event and include every detail.
Participant: Yes, I think it was taken on a school trip at the cathedral, by the cathedral odd at a museum. There is a lot of medieval stuff there. And it was in elementary school on a school trip, it was one of the first trips we were on, and we took the bus with the elementary school to places that weren't nearby. I remember we learned a lot about Vikings and stuff, and it was really cool to be able to sit in a Viking boat and look at ruins.
I: Are there any other details you can remember? P: I thought it was really cool to learn about Vikings, and to try -I remember that we had, let's call it a battle on a patch of gravel, and we had foam swords or something, and pretended to fight and dressed up and got to try a Viking helmet and stuff.
A third excerpt provides an example of specific personand contextrelated elements reported during the first and the third interview: Interview 1: Interviewer: Now, I want you to tell me anything you can remember about this event, anything at all. I want you to include as many details as possible.
Participant: I have never seen this picture before. I don't know. It's shocking. I can't remember having been on a ship like this. But I only see myself, not how this picture was taken. Or who took it. I don't know. I guess it was a school trip. I don't know which. I don't really have any answer to where this could be or which school trip it was. I don't know the year or what age I was, either.
[…] I think it's in Norway.
By the coast. I don't know if it's a school camp, a museum, or a boating school. I have been to all three, but I can't remember this boat.
[…] This picture is so unique to me. It's the first time I've seen it. I can't remember anything from this event.
There isn't anything I recognize. I have not seen it before. I still don't know who took the picture, either.
I: The target of the study is also to test some memory techniques that can help you remember events from your childhood. I will now introduce you to some of these. Now, I want you to think back to the trip with the Viking ship and attempt to remember the event and any details you can think of. Think for one minute about the details of how it was to be on the trip, who you were with, what the weather was like, and what you saw. Can you remember anything else?
P: From what I can tell, it has to be one of the three things I mentioned. School camp, boat museum, or boating school. But I don't think it's the boating school because there weren't any Viking boats. In the museum, we weren't allowed to ride a boat, so I think it's school camp. We rowed in pairs. And I remember I had rowed a boat before, but not the same type of boat in misty weather. In school camp at the start of seventh grade. […] If I visualize being on a sailboat, I can remember it was late summer, early autumn. Starting to get cold and misty. I am with the school camp. I guess I am with friends who are going on the boat trip. There are different sections of what the class is going to do. And we learn to row on both sides and go fast together and stay synchronized. That is what I can remember from a boat trip that looks like a picture. I can't remember anything else. Also in interview 3, the same participant gave the following account: P: I think it was a school trip in eighth grade because the blue jacket I am wearing I only wore in eighth grade. This was probably a Viking museum that we went to. And got to test a Viking ship. Beside me are my school friends. Who I sat with, I can't remember.
[…] I can remember it was cold, probably late autumn, early winter. Very cloudy. It is hard to see the surrounding area. I still don't know where this is in Norway this is. It has to be in Oslo, close to the harbor.
I: What did you do during the trip? P: I guess, from what I can remember, they were going to show the story behind the Norwegian Vikings: design, Vikings in general, swords, wars and such, and testing boats that are built in a more modern age. Like kayaks and stuff. Like, there is teamwork in rowing. But from what I can tell from the picture, it looks like a group divided into two, so there are more groups, and this is just half the class. […] We were told to row, learning to row together at a steady pace and to be synchronized so we don't row in a circle. Learning teamwork.
I: Who took the picture? P: I guess it is the teacher. My contact teacher. I still remember how I got the picture. I think he printed it, and I got a copy of it. I guess it was at the end of the school year where we had a folder with lots of pictures that we could share with others. But yes, I obviously had this picture, even though I didn't remember that I did.

Supplementary coding and analysis
Since the Wade et al. (2002) study was conducted more than twenty years ago, the field of false memory research has evolved. The generalizability of false memory estimates reported across implantation studies has been criticised due to substantial variations in operationalising of false memories, the nature of the suggested event, and the implementation of memory induction techniques (e.g., Becker-Blease & Freyd, 2017;Blizard & Shaw, 2019;. In addition, more refined coding systems for categorisation of false memory reports have been developed for determining the prevalence and strength of false memories reports. To address this concern, we conducted supplementary analysis of the memory reports in accordance with the theoretically driven coding system for operationalising and categorising of false memory reports developed by Scoboria et al. (2016). This coding system, which encompasses more strict criteria for categorising false memories, as described in Scoboria et al. (2016) (pp. 152-154), includes seven dimensions; verbal statements of "remembering", acceptance of suggested information, elaboration beyond suggested information, presence and quality of mental imagery, coherence of memory narratives, evidence of emotional experience, and no rejection of the suggested event. A final categorisation of the memory reports is classified as either robust false memory, full false memory, partly false memory, acceptance, no memory, or active rejection.
The coding of the memory reports in accordance with Scoboria et al. (2016) showed an excellent inter-rater agreement for all coded categories: ranging from 85% to 100% agreement between raters. Inconsistent ratings were classified into the most conservative coding category.
The analysis of the supplementary coding showed that 25% of the participants met the criteria for Robust memory, 10% of the participants met the criteria for Full memory, and 10% met the criteria for Partial false memory, resulting in a combined false memory rate of 45% (n = 9). Additionally, 10% of the participants accepted the suggested event to some degree but reported low levels of elaboration beyond the suggested event. Of the remainder, 45% (n = 9) of the memory reports were categorised in the No Memory/Rejected category as there was no evidence of false memory formation. Active rejection of the suggested event was made by all the participants that were classified in this category.

Discussion
The results of the present replication of Wade et al. (2002), which was rigorously executed by recreating the procedures of the original study with a modified culturally appropriate false event, showed almost identical patterns of results to what was observed in the original study. A substantial percentage of the participants (40%) reported false beliefs or memories of nontrivial childhood events suggested to them, as compared with 50% in the original study. A supplementary analysis of the memory reports using the more recent coding scheme of Scoboria et al. (2016) showed a similar combined false memory rate of 45%, which was further increased to 55% when we included reports where the participants partly accepted the suggested event. Furthermore, this finding also agrees well with a mega-analysis of eight false memory studies which found that 30% of memory reports were classified as false memories, and further extended to 53% when including reports that were classified as having accepted the fictitious event to some degree (Scoboria et al., 2016).
The replication further showed that the percent of participants self-ratings of their confidence that the false event had occurred in the replication (43.8%) was identical to the self-ratings of confidence that was reported by Wade et al. (2002) (44.5%). The participants who reported false beliefs or memories in the replication reported mental imagery, emotional experiences, and coherent elaboration of memory narratives of the Viking ship ride that were not depicted in the doctored photograph, similar to those reported in the original study, as well as in other studies that have explored the formation of false memories (e.g., Hyman & Pentland, 1996).
The vivid imagery of detailed, event-specific information from a substantial percentage of the participants in the replication indicates that we successfully managed to modify the culturally appropriate and functionally equivalent false target event, similar to that of the original study. Also, the comparable findings of the 20-year-old original study and the current replication are astonishing, given that the replication was conducted at a time when people are much more familiar with and exposed to manipulated images and editing techniques.

Limitations
The current replication rigorously recreated Wade et al.'s (2002) methods and procedure as described in the original study. The only significant deviations from the original study were that the false target event was adjusted with a culturally appropriate target event and that the interviews were conducted digitally rather than live. Regarding the generalizability of photographically implanted false biographical memories, it should be noted that the present replication and most other studies published were conducted on populations from Western, Educated, Industrialised, Rich, and Democratic (WEIRD) countries (Vredeveldt & de Bruïne, 2022), and the phenomenon should be further explored in cross-cultural contexts and with appropriate culturally suggested events.

Conclusions
In summary, the replication of Wade et al. (2002), which showed an almost identical pattern of results as the original study regardless of coding procedure, indicate that even minimal levels of suggestion can be sufficient to lead to false beliefs and memories of self-referent events that had not occurred. The replication further adds to the growing body of evidence showing that memories about false events can relatively easily be successfully implanted regardless of cultural settings, suggesting that the phenomenon of false memory formation may be universal. However, additional independent replication studies in different cultural contexts are necessary to understand the complexity, robustness, and generalizability of the formation of false autobiographical memories as a potentially universal phenomenon.