Who Is the Better Eyewitness? Sometimes Adults but at Other Times Children.

Suggestibility is regarded as a major issue when children testify in court. Many legal professionals and memory researchers view children as inferior witnesses. Although differences in suggestibility exist between children and adults, they are much more complex than is usually assumed. We show that under certain conditions, adults are more susceptible than children to suggestion and false memories. We provide evidence that age-related shifts in suggestibility and false memory appear contingent on how quickly and automatically children and adults make associations when experiencing events. Specifically, when confronted with suggestive information about a related but nonexperienced detail, adults more frequently automatically generate links between items experienced and those already in memory, making them more susceptible to suggestion than children.

Many textbooks on child development suggest that children underperform on most, if not all, tasks related to cognition (Bjorklund & Causey, 2017). Textbooks on developmental psychology regularly stress that the ability to plan and inhibit improves with age (Zelazo et al., 2003). In the current article, we focus on the development of memory. Here, the critical assumption is that children’s memory is generally inferior to that of adults. One practical issue discussed in this context is that when suggestive interviews are used (e.g., by the police), children are more likely than adults to accept these suggestions. This topic is relevant in the legal field, where children are regularly viewed as inferior witnesses and special care is devoted to interview children in a correct manner (Bruer & Pozzulo, 2014).

Such a perspective is also shared among some memory researchers. For example, Volpini, Melis, Petralia, and Rosenberg (2016) recently echoed this by stating that “younger children are almost always more susceptible to suggestibility” (p. 104). Likewise, Kassin, Tubb, Hosch, and Memon (2001) found that the majority of psychologists with legal experience agreed that young children are more vulnerable to suggestion than adults. Are these views etched in scientific stone? We doubt this. We review literature showing that age-related shifts in suggestibility are variable but do conform to certain theoretical principles. Thus, under certain well-specified conditions, adults might be even more prone to suggestive interviewing techniques than children.

A Short History

Around 1903 or 1904, the German psychologist William Stern commented on the reliability of children’s testimony (Sporer, 1982). He reported on a case in which statements of a young boy were likely influenced by suggestive pressure and advised that these statements should be discarded. Around 1910, the Belgian psychologist Johan Varendonck (Whipple, 1913) stressed that suggestive questions might have adversely affected statements of two young child witnesses. The American professor Guy Whipple (1909, 1911) wrote the first reviews on the reliability of children’s memory. His key argument was that children’s memories were inherently unreliable even when pertaining to autobiographical experiences. This focus on the unreliability of children’s memory even led to bizarre statements such as the one by the founding father of legal psychology, Hugo Münsterberg (1908), who wrote that

experiments with school children, especially, seem to show that the girls have a better memory than the boys as far as omissions are concerned; they forget less. But they have a worse memory than the boys as far as correctness is concerned: they unintentionally falsify more. (p. 54)

Thus, already at the start of psychology as an academic discipline, children were regarded as inferior witnesses.

Scientific interest in the reliability of children’s memory increased exponentially in the 1980s and 1990s because of several high-profile day-care abuse cases in different countries such as the United States. In these cases, many children reported having been abused by the same person or persons. One notorious example was the McMartin preschool, in which hundreds of children allegedly remembered having been sexually abused by three teachers. However, the children reporting these allegations were subjected to suggestive interviews, ones that likely affected their recall, leading inexorably to erroneous recollections (Garven, Wood, Malpass, & Shaw, 1998). Discussions surrounding these day-care abuse cases reinforced the idea of children’s memory as exceptionally prone to suggestive pressure and false memories. Subsequently, studies have looked into children’s vulnerability to suggestion and false memory. Prima facie, the bulk of these studies seemed to confirm the idea that children are poor witnesses.

Children Are More Suggestible Than Adults

Cases such as the McMartin preschool case contained various interviewing techniques that likely contaminated children’s testimonies (e.g., suggestive questions, inviting speculation). There is much research attempting to mimic these circumstances and their effects on memory (Garven et al., 1998). However, for the current purpose, we have specifically focused on developmental research using various sorts of suggestive manipulations (i.e., presenting misinformation on details and events). If we home in on false memory, myriad studies exist showing that suggestion is more likely to infect younger children’s than older children’s and adults’ memory. In a pioneering study by Ceci, Ross, and Toglia (1987), 3- to 12-year-old children were read a story about a girl on her first day at school. Children received information that the girl had a stomach ache. One day later, some children were presented with misinformation implying that the little girl had a headache. These latter children were interviewed and asked whether they could remember the story about the headache. Two days later, all children received a recognition test. The youngest children (3- to 4-year-olds) most often accepted the suggestion.

This pattern has been replicated many times (e.g., Kulkofsky & Klemfuss, 2008). In one of our own experiments, 4- to 5-year-olds and 8- to 11-year-olds had to remove three pieces of clothing from a puppet (Otgaar, Candel, Smeets, & Merckelbach, 2010). Half of the children were also presented with false evidence suggesting that they had removed one extra piece of clothing. Without children seeing it, a confederate removed the extra piece of clothing, after which it was suggested to the children that they removed four pieces of clothing. During three follow-up interviews at 1-week intervals, children had to report which pieces of clothing they had taken off. Four- to 5-year-olds were more likely than 8- to 11-year old children to falsely report that they took off four pieces of clothing.

Some studies have used a false-memory-implantation paradigm to examine whether children can create what has been termed “rich false memories” (e.g., Loftus, 2005, p. 363). In contrast to typical misinformation experiments, in false-memory-implantation studies, participants are not presented with stimuli and then receive misinformation about these stimuli afterwards. Thus, participants are immediately exposed to suggestive false narratives implying that they experienced a fictitious event. For current discussions concerning developmental trajectories in suggestibility, developmental studies using this paradigm are relevant because they shed light on children’s willingness to acquiesce to external suggestions. We falsely suggested to 7- to 8- and 11- to 12-year-olds that they were abducted by a UFO or that they almost choked on a candy when they were 4 years old (Otgaar, Candel, Merckelbach, & Wade, 2009). False-memory rates for both events were statistically higher in younger (7- to 8-year-old) than older (11- to 12-year-old) children. A similar pattern emerged in a study (Otgaar, Candel, Scoboria, & Merckelbach, 2010) in which children were fed false stories such as that they received a rectal enema or got their fingers stuck in a mousetrap: False events were more likely to be implanted in younger (7- to 8-year-old) than older (11- to 12-year-old) children’s memory.

The notion that younger children are especially sensitive to including suggestion in their memory reports is confirmed when we examine the available work that has been conducted in this area over the past few decades. If, for example, we consider all studies in which false suggestive information (i.e., misinformation) was presented to different age groups, it is obvious that most of them observed an age-related decrease in susceptibility to suggestion-based false memory (Fig. 1; see also the additional material posted on the Open Science Framework at https://osf.io/cd2qg/). However, this is only one part of the story.

Fig. 1.

Number of developmental studies on the misinformation effect that show age-related decreases (younger children are more suggestible than older children), age-related increases (older children are more suggestible than older children), and no age-related differences. For information about the studies that were included, see the additional material posted on the Open Science Framework (https://osf.io/cd2qg/).

Associative Activation, Mental Representations, and False Memory

According to associative-activation theory (AAT; Howe, Wimmer, Gagnon, & Plumpton, 2009), some false memories arise because of associative activation spreading through a dense network of interrelated nodes. Essentially, AAT is a network model of memory in which nodes (e.g., concepts, not simply words) are interconnected and are also linked to overarching themes (various meanings associated with the concepts). AAT successfully predicts false memories (and their developmental trajectories from childhood through adulthood) not just in studies involving word lists but also in studies involving pictures (e.g., Howe, 2008), visual scenes (e.g., Lew & Howe, 2017), stories (e.g., Howe & Wilkinson, 2011), and autobiographical false memories. It also predicts changes in false-memory rates as a function of the emotional state of the individual and the material being remembered (e.g., Bland, Howe, & Knott, 2016; Knott, Howe, Toffalini, Shah, & Humphreys, 2018). Thus, this theory is particularly well suited to inform our understanding of developmental trends in children’s eyewitness remembering.

For example, when experiencing an event (e.g., robbery), nodes related to that event but which are not part of the current experience may become activated (e.g., seeing a gun) and increase the likelihood of a false memory. Throughout the course of life, people acquire more knowledge, resulting in faster and more automatic associative activation. The net effect of this is that under some conditions (e.g., when surrounded by associatively related cues), adults are more susceptible to false memory than children precisely because they are more likely to generate faulty associations.

Thus, AAT assumes that under these conditions, false memory follows an age-related increase, which stands in contrast to the work on false memories induced by suggestion (see Fig. 1; but also see below). So a specific form of false memory called spontaneous false memory is more likely to be evoked in adults relative to children. Spontaneous false memories are purely caused by internal mechanisms such as associative activation and are not the result of external influences such as suggestive questions. The principal procedure used to elicit spontaneous false memories is the Deese-Roediger-McDermott (DRM) paradigm (Deese, 1959; Roediger & McDermott, 1995). In this paradigm, participants receive words (e.g., “baker,” “dough,” “knife,” “flour”) that are all associated with a nonpresented word called the critical lure (e.g., “bread”). Studies have demonstrated that participants falsely recollect the critical lure at rates often indistinguishable from true memory rates (Roediger & McDermott, 1995). These false memories are more easily induced in adults than in children, a phenomenon termed developmental reversal (Brainerd, Reyna, & Ceci, 2008).

Such reversal effects have been found not only with associative word lists. Other stimuli capturing associative relations have revealed effects of developmental reversal as well. Lyons, Ghetti, and Cornoldi (2010) presented 6-, 7-, 9-, and 10-year-olds and adults with photographs depicting a common script (e.g., eating in a restaurant). A script refers to a knowledge structure containing interrelated details denoting the typical actions occurring during an event. The photographs included effects (e.g., wiping up water from a table) of nonpresented causes (e.g., knocking over a glass of water). The authors found an age-related increase in participants falsely remembering having seen the cause in the originally presented photographs.

The developmental-reversal phenomenon indicates that adults more easily generate spontaneous false memories than children, whereas for false memories evoked by suggestion, the reverse is true. However, recent work from our laboratory has revealed that developmental trends for both types of false memories are not always consistent. For example, in one of our studies (Otgaar, Howe, Peters, Smeets, & Moritz, 2014), we presented children (7- to 8-year-olds and 11- to 12-year-olds) and adults with visual scenes (e.g., a beach) containing associatively related details (e.g., sand, water). Some related details were left out (e.g., a bath towel). These not-presented related items were often spontaneously and incorrectly remembered. Furthermore, and contrary to the developmental-reversal effect, children in both age groups were more likely than adults to produce spontaneous false memories when visual scenes were used. Our explanation for this is that the visual scenes helped children to distill the underlying theme. According to AAT, visual material may compensate for children’s lack of associative networks compared with adults. For children, this additional thematic assistance might make them more prone to the creation of false memories than adults. In fact, when we presented children and adults with videos to foster spontaneous false memory, we obtained the same results. Because the theme of videos is readily identified, children rather than adults were most susceptible to spontaneous false memories (Otgaar, Howe, Peters, Sauerland, & Raymaekers, 2013).

The variability of false-memory development shows that developmental trends in false memories are not fixed and can follow a trajectory that is opposite to what one would expect. This is also what has been suggested by a developmental-representational account (Ceci, Fitneva, & Williams, 2010), which has parallels with AAT. According to this theory (and AAT as well), children’s background knowledge or mental representations about events drive developmental effects in suggestibility. That is, because young children have not acquired as much knowledge as older children and adults, they are less likely to relate misleading external information that is connected to that knowledge. Ceci and colleagues reasoned that these trends can be easily altered when considering someone’s knowledge base of an event.

In one experiment, Ceci, Papierno, and Kulkofsky (2007) provided 4- and 9-year old children with 257 sets of three pictures each, and they had to decide which picture in each set did not belong with the other two. The goal of this similarity-rating task was to map children’s representations and associations of these stimuli in order to later predict children’s proneness to suggestion. So, 1 to 3 months later, these children and other children of the same age groups listened to a story that was illustrated using pictures of objects from the first part of the experiment. After 2 days, some children received misinformation about the story. For example, in one story, a boy and a girl saw an eagle in the zoo but received the false suggestion that they saw a robin. Five to seven days later, children received a memory test including pictures that had either been presented or served as the misleading stimuli. The most interesting result was that when stimuli (e.g., eagle and robin) were highly associated, they were more likely to be misremembered than when such associations did not exist. Importantly, this effect was found irrespective of age when age-relevant similarity ratings were used. That is, older children were most susceptible to misinformation when stimuli were highly related and drawn from their similarity ratings and so, too, were younger children when those related stimuli came from their similarity ratings.

Children Are Less Suggestible Than Adults

In many child sexual-abuse cases, expert witnesses may tell legal counsels to be cautious with children’s reports as they might easily be infected by suggestion. However, developmental trends in false memory can be altered. Developmental-reversal effects have been well documented in the area of spontaneous false memory. Whether they also might occur with false memories elicited by suggestion has been a key question in our recent empirical work. Our prediction was that when children and adults are presented with stimuli containing associatively related information and receive information suggesting that a related but not presented item was shown, adults, and not children, should be most susceptible to suggestion. This prediction carries considerable ecological validity because when children and adults witness events that they then provide testimony about, the elements of the events tend to be highly interrelated.

Our counterintuitive prediction has been confirmed in a number of recent studies. For example, four experiments were conducted with children 4 to 6 years old (Experiments 2 and 4), 7 to 9 years old (Experiments 1, 2, 3, and 4), and 10 to 12 years old (Experiments 1, 2, and 4), as well as with adults (Experiments 1 and 3), in which they were shown a video (e.g., of a bank robbery) containing associatively related details (e.g., robber, vault; Otgaar, Howe, Brackmann, & Smeets, 2016). Crucially, following this, in the first two experiments, they were presented with an eyewitness-misinformation account falsely stating that, for example, a gun was present during the crime. In the last two experiments, half of the children also received the same misinformation, but now an interviewer provided the false suggestions. Next, participants received a recognition test. In all experiments, we found evidence for developmental-reversal effects. Specifically, in the first experiment, we found that adults and 11-year-olds had higher misinformation scores than 6- to 7-year-olds, and in Experiment 3, we also showed that adults were more susceptible to misinformation reporting than 7- to 8-year-olds. In Experiment 2, even younger children (4- to 6-year-olds) were included, and in that experiment, 10- to 12-year-olds produced more false memories than the 7- to 9-year-olds and 4- to 6-year-olds. Similar findings were observed in the fourth experiment, in which 11- to 12-year-olds were more likely to produce false memories than 7- to 8-year-olds and 4- to 5-year-olds.

What these studies suggest is that even very young children (i.e., 4-year-olds) can evince lower false-memory rates than older children. Why this is relevant is because previous research has shown that this younger age group of 4- to 5-year-olds are especially sensitive to external suggestions because of social factors such as accepting information from authority figures (e.g., Leichtman & Ceci, 1995). Our results show that when misinformation is presented that is directly related to one’s knowledge base, even 4- to 5-year-olds can be the least prone to suggestion.

In another study (Otgaar, Howe, Brackmann, & van Helvoort, 2017), children (7- to 8-year-olds and 11- to 12-year-olds) and adults viewed pictures (e.g., a desk) containing associatively related details (e.g., books, a laptop). They viewed the stimuli in pairs, believing they received the same pictures. Each version of the picture included a critical item that was associatively related to the scene but absent in the other picture version. Following this, participants had to discuss and retrieve all details that they could remember. The idea behind this discussion is that participants would mention (or suggest) details that while present in their pictures were absent in the pictures of the other participants. After this, participants had to individually report everything they could still recollect. Again, we found that children were not more susceptible to suggestion than adults at both recall moments. In fact, they were equally susceptible to suggestion, and when correcting for response bias, we found that adults were even more prone to suggestion.

These findings counter the default view that suggestibility is a problem primarily for children. Under some conditions—namely those fostering associations—adults are just as, or even more, suggestible than children. This perspective is not commonplace in the scientific literature, and although other researchers have also found adults to be more susceptible to memory errors than younger children (4- to 7-year-olds; e.g., Kim, Kwon, & Ceci, 2017), this work has not focused on altering developmental trends in suggestibility, something that has been a major focus of our research (but see Ceci et al., 2007).

False-Memory Development in Court

What these data show is that when suggestions involve associatively related information (similar to that used to elicit spontaneous false memories), young children are less susceptible to these suggestions than older children and adults. Because most eyewitness accounts involve information that is highly interrelated, our studies raise a crucial question: How might this more balanced idea of children’s susceptibility to false memory affect decisions in the legal arena? It is a given fact that the science of memory plays a crucial role in child sexual-abuse cases, where children’s statements are often the only piece of evidence (Howe, Knott, & Conway, 2018; Otgaar & Howe, 2018). However, and more pointedly, the key issue is whether the scientific findings reviewed here are relevant to legal cases where children’s memories serve as the only evidence.

Consider a case in which the first author (H. O.) provided his expert opinion. It involved a 6-year-old child claiming to have seen her mother being stabbed to death by her father (Brackmann, Otgaar, Sauerland, & Jelicic, 2016). H. O. was asked by the prosecution to write a report concerning the accuracy of the girl’s statement, and a clinical psychologist was hired by the defense on the same matter. The clinical psychologist argued that the child spontaneously formed a false memory of the murder. His reasoning was based on the default notion that children are exceptionally susceptible to false memories. H. O. concluded that no signs of suggestion existed and that younger children can be less likely to form spontaneous false memories. The judge deemed the statement of the girl accurate enough to convict the father to 18 years of imprisonment.

One may contend that in many cases, it is not certain what details of an event are associated with each other. However, we argue that many—if not all—of the events (e.g., having dinner) that we encounter contain interrelated details (e.g., cooking, washing the dishes). For forensically relevant events, such associations might exist as well. Indeed, we agree with Holliday, Reyna, and Brainerd (2008) who argued that

false memories induced by meaning related information embody several features of forensically relevant memories. For child witnesses of domestic violence, for example, such violence is not usually a single episode but rather a series of repeated events that are substantially similar but not exactly the same. (p. 76)

What we want to stress here is that associative activation is likely to a play a role in repeated events where children have already developed a script of the event. To provide an estimate of the number of legal cases in which children are interviewed about repeated events and, hence, to have an idea of the number of cases in which reversal effects might occur, we examined verdicts of Dutch legal cases. Specifically, in The Netherlands, one can access information (e.g., verdicts) concerning diverse cases via an online database (http://www.rechtspraak.nl). To have a rough indication of cases on repeated experiences in children, we entered the following keywords as search terms: “interview” (in Dutch: “studioverhoor”) and “child” (in Dutch: “kind”). Furthermore, we filtered the data by looking only at cases in 2017. Our search identified 38 cases. Of those 38 cases, 29 (76%) were cases in which children were interviewed (age range: 5–17). Of these 29 cases, 18 (62%) concerned events that children experienced repeatedly, and 24 (83%) referred to sexual abuse. Other cases in which children were interviewed involved domestic violence, the abduction of children, and witnessing sexual behavior of the suspect.

The data here suggest the following. First, although repeated experiences of the same event can lead to the creation of false memories of related but not experienced details in children involved in these cases, it is likely that the creation of such memory errors is less likely to occur in younger than in older children. Second, this reversal effect is also likely to take place when these children are confronted with external suggestions that are linked to the repeated experience. The conclusions that we draw from these data should not be used to argue that younger children are—by definition—less suggestible than older children and adults. The central message should be that in situations involving scripts, for example, the lack of knowledge in children might guard them from making spontaneous memory errors or going along with suggestions.[1]

Of course, in actual cases, it might be difficult to decide whether children (or adults) had sufficient background knowledge of an event. One option would be to examine whether the DRM paradigm would be a reliable and valid method in legal cases as a proxy for someone’s knowledge base. If so, it could serve a similar function as the Gudjonsson Suggestibility Scale (e.g., Gudjonsson, 2003), which is often used in legal cases to indicate whether someone (e.g., a suspect) is likely to succumb to suggestive interviewing techniques. However, although this might seem promising, recent research has failed to show any meaningful relations between false memories elicited by the DRM paradigm and false memories elicited by other paradigms (e.g., misinformation paradigm, false-memory implantation; e.g., Patihis, Frenda, & Loftus, 2018). Hence, it is important to examine whether the DRM paradigm might be helpful in reversal effects in more realistic experiences.

Concluding Remarks

The long-standing, knee-jerk response concerning children’s lack of testimonial accuracy is incorrect. Although it is widely believed to be true (Knutsson & Allwood, 2014), children do not necessarily show a heightened susceptibility to suggestion. For decades, preschoolers have been shown to be more suggestible than older children and adults (Bruck & Ceci, 1999; Goodman & Reed, 1986). However, our argument here is that young children are not always the most suggestible, and in situations that rely on the activation of scripts, stereotypes, schemas, and other forms of connected meaning, their lack of knowledge can sometimes protect them from external suggestions. That is, recent evidence shows that adults are sometimes at even greater risk of accepting suggestive information than children. The time is ripe to stop portraying children as inherently problematic eyewitnesses and acknowledge that they sometimes outperform adults, even when it comes to memory performance.

Recommended Reading

Brainerd, C. J., Reyna, V. F., & Ceci, S. J. (2008). (See References). A review of developmental reversals in false memory for word lists.

Ceci, S. J., & Bruck, M. (1995). Jeopardy in the courtroom: A scientific analysis of children’s testimony. Washington, DC: American Psychological Association. An overview of research on children’s testimonies.

Otgaar, H., Howe, M. L., Brackmann, N., & Smeets, T. (2016). (See References). An empirical demonstration that suggestibility follows an age-related increase.