Memory is not a faithful reproduction of an individual’s experiences but a reconstruction process that often leads to errors. These errors can be harmless, such as when people misremember what they had for dinner, but they can also have serious consequences. In the field of law, for example, when there is a lack of direct physical evidence relating to a crime, the evidence used to accuse and convict the defendant largely relies on memory (Brainerd and Reyna, 2019). In the field of psychotherapy, therapeutic techniques such as hypnosis and dream interpretation can increase the likelihood of inducing harmful false memories, with an average of 20–50% of individuals experiencing the induction of false events (Muschalla and Schönborn, 2021). In the medical field, patients reporting their symptoms incorrectly or doctors misremembering a condition can lead to misdiagnosis and subsequent adverse effects on health. In the existing literature, false memories can be classified into two categories (Ost et al., 2013): (1) implanted false memories induced by suggestion and (2) spontaneous false memories generated without any external pressure. Among them, the former is an external distortion, while the latter is an internal distortion. These two types of false memories are only weakly correlated with one another (Calvillo and Parong, 2016; Bernstein et al., 2018), and this article focuses on the latter.

Regarding the theory of emotions, the majority of studies probing the impact of emotions on false memory are grounded in the circumplex model of emotion (Russell, 1980), which posits that emotional experiences are composites constructed from various dimensions of valence and arousal—for instance, positive, high-intensity emotional experiences signifies happiness; positive, moderate-intensity emotional experiences denotes relaxation; negative, moderate-intensity emotional experiences signify tedium; and negative, extreme-intensity emotional experiences symbolize fear. The dominant position of the circumplex model of emotion within this sphere can largely be attributed to the availability of a standardized tools, such as the International Affective Picture Series (IAPS) (Lang et al., 1998), while available standardized materials grant researchers the capacity to manipulate the valance and arousal of the material. In recent years, empirical research has provided a wealth of evidence on the impact of emotional stimuli on false memory (Kensinger and Schacter, 2008), and the debate about when and how emotions affect false memory has continued. Regarding the theory on the influence of emotions on false memories, some theories explain the impact of emotions on false memories from different perspectives. One theory, Emotional Enhancement of Memory (Heuer and Reisberg, 1990), explains the impact of emotions on memory from a broad perspective, suggesting that emotional content can enhance the accuracy of memory; for example, people tend to remember events that are accompanied by intense emotions. In contrast, the Distinctiveness Heuristic Account makes more specific predictions, suggesting that negative emotions are less likely to generate false memories; this theory posits that people remember the distinctive details of events they have experienced and make recognition judgments based on these distinctive details (Schacter and Wiseman, 2006)—for instance, “I clearly recall that the dragon fruit I consumed last week had white flesh, not red flesh, as red-fleshed dragon fruits are quite distinctive.” Emotional content is often more distinctive and can serve as a cue for a distinctiveness heuristic (Schacter et al., 2011); specifically, negative emotional content is highly relevant to survival, and individuals are more likely to accurately remember negative emotional content, making it less prone to producing false memories. These two theories offer different predictions about the specific direction of the impact of negative emotions on false memory, which is core to the ongoing debate. The concept of emotional memory trade-off effects (Kensinger et al., 2007) outlines a contrast in how emotional and neutral components of an event are remembered; while emotional aspects of an experience are remembered more vividly and accurately, the neutral details of the same experience may be less accurately recalled. This trade-off implies a focused allocation of cognitive resources toward emotionally salient information during an event, enhancing the memory of these aspects. Conversely, less attention and hence fewer cognitive resources are directed toward neutral, peripheral details, resulting in poorer recall of these elements; for instance, an eyewitness might only remember the suspect’s fierce expression, while overlooking their attire. The Fuzzy Trace Theory (FTT) explains the impact on false memories from the perspective of different components of emotional content; it suggests that, when individuals experience an event, they store two opposing memory traces (Brainerd et al., 2018). Finally, the verbatim trace stores specific details of the experience, such as remembering specific items like “apple” and “canary,” while the gist trace involves processing the meaning, such as categorizing an apple as fruit and a canary as an animal. During the recall phase, extracting the gist trace triggers false memories, while extracting the verbatim trace inhibits false memories. In terms of valence, negative emotions strengthen gist traces, leading to more false memories. Separately, regarding arousal, a moderate level of arousal enhances verbatim traces, reducing false memories, but a greater level of arousal weakens verbatim traces, increasing false memories.

Researchers have used different paradigms to investigate emotion-related false memories. The Deese–Roediger–McDermott (DRM) paradigm dominates in studies on spontaneous false memory (Deese, 1959; Roediger and McDermott, 1995). In this paradigm, participants first learn a list of associated words (e.g., “moonlight,” “bed,” “pillow,” “night”), which are semantically related to critical lure words (e.g., “sleep”) that are not presented in the list. Higher recall or recognition of the critical lure words in the subsequent test indicates a greater generation of false memories by the participant. The DRM paradigm is also suitable for studying emotion-related false memory as researchers can readily manipulate the valence of words to examine the influence of different emotions on false memory. To verify whether the results obtained from the DRM paradigm can be generalized to other materials and paradigms, researchers have also used the picture paradigm (Koutstaal and Schacter, 1997). In this paradigm, participants learn several categories of pictures derived from real life (e.g., a set of baby photos, a set of train photos) and manipulate their emotions by changing the valence and arousal levels of the pictures. A category of pictures not previously presented is used as critical lure stimuli to measure the false-memory effect. Therefore, in both the DRM paradigm and the picture paradigm, the valence and arousal levels of the memory materials can be manipulated, and differences in the manipulation of valence and arousal may lead to different results (Chang et al., 2021). In addition, the picture paradigm uses recognition tests, while the DRM paradigm combines both recognition tests and recall tests, where researchers may use a single recognition test or recall test, or they may have participants first complete a recall test and then a recognition test.

In summary, researchers have used various materials to create false memories, but the results are mixed. This prompts the question of whether there is a distinction between false memories induced by words and pictures and whether the use of different memory tests affects the outcomes. Finally, with regard to emotional variables, some previous studies did not control for either valence or arousal, while others only controlled for one or the other. Does a difference in controlling valence and arousal have an impact on research results? This review elaborates on the specific directions of the influence of emotions on false memories in studies using words and pictures as memory materials. Respectively, the aims of this investigation were to explore (1) the differences in false memories induced by picture and word materials, (2) the impact of using different tests in experiments using words as memory materials on the generation of false memories, and (3) the impact of control over valence and arousal on the results and whether these variables should be controlled in future studies.

There are two ways in which emotions are generated when individuals experience events: either the events themselves carry emotions, or said individuals have already generated certain emotions before experiencing certain events. In other words, traumatic events can trigger negative emotions, and the emotions individuals feel before recalling memories of neutral events may also differ. Researchers mainly use word and picture memory materials to induce false memories, a process which is largely supported by the availability of standardized material databases, such as the IAPS. Studies employing word memory materials can be roughly divided into two categories: (1) content studies using DRM word lists with inherent emotional valence and (2) context studies that first induce emotions through other means (e.g., music, videos) and then ask participants to memorize neutral DRM word lists. Experiments that use pictures to naturally induce emotions in participants are considered content studies, while experiments that first induce a certain emotion in subjects and then have them recall neutral pictures are considered situational studies.

The differences in the experiments mentioned in the text are displayed in Table 1. Below, we will discuss these differences in detail.

The impact of emotions on false memory has long been a concern in the fields of law and medicine. Extensive studies have shown that emotions can affect false memory, but there is no consistent explanation of how this occurs. This article reviewed the literature and summarized the results and methodological issues of previous studies.

Studies using words as memory materials have enhanced people’s understanding of the constructive nature of memory and the susceptibility to memory errors. However, the results obtained from word lists are not consistent with situations using other materials. Researchers appear to reach the same conclusions using different materials, which is not consistent with the actual situation. Future research must clarify the inherent differences in inducing false memory using different materials such as words, pictures, videos, music, and so on. In addition, Whittlesea et al. (2005) raised doubts about the DRM paradigm from a unique angle. They embraced the DRM paradigm and found that participants’ false memories were caused by surprise induced by the critical lure (note, the critical lure was more recapitulative and connected to other learned words). Further studies revealed that, when participants consciously suppressed their surprise at the critical lure, the DRM effect vanished. Interestingly, one study suggests that surprise can elicit negative emotions (Topolinski and Strack, 2015), and whether surprise is truly an emotion remains a controversial question. Some have indicated that surprise does not always have a certain level of valence, which is a characteristic of each emotion (Gerten and Topolinski, 2019; Ortony, 2022). Therefore, it is evident that this issue is quite complex, and it is crucial to explore the impact of surprise caused by the critical lure on false memories in future research. On the other hand, the false memory induced by word lists is different from the false memory of real events in judicial and psychological therapy contexts (DePrince and Freyd, 2004), as each group of words or pictures is relatively independent and lacks logical relationships. The method mentioned earlier involves using a collection of pictures that depict a story as memory materials, which, to some extent, addresses the problem of the lack of logical relationships between learning items in traditional paradigms.

Emotion-induced false memories are sensitive to many different variables, and even subtle changes in the experimental design can lead to changes in the direction of emotion-induced memory errors. Additionally, recall and recognition tests, respectively, have different effects on memory errors, but they are often confused with each other in the existing literature. A recent study (Wiechert et al., 2024) used meta-analysis and replication research methods to uncover that negative valence does not systematically affect false memory; instead, the formation of false memories depends upon how false memories are tested; in the recall test, valence had no effect on false memory, while, in the recognition test, the effect of false memory may be attributed to response bias. Additionally, Yüvrük and Kapucu (2022) found that the effect of valence was non-significant when recognition responses were controlled for response bias. These results indicate that future research should quantify the impact and specific contributions of different manipulations on memory errors, and examining the impact of emotions on false memory from the perspective of valence and arousal alone is far from sufficient. Among previous studies, only some controlled for valence and arousal, and the conflation of the two could have caused variability in the results. Should valence and arousal be controlled? While the valence and arousal of emotions might differently influence false memories, it’s not sufficient to limit research solely to the effect of emotional arousal on memory (Levine and Pizarro, 2004). Individuals may experience specific emotions like happiness, fear, despair, or anger, but they are never merely “aroused.” Previous studies have shown that emotional experiences cannot be fully captured by just two dimensions (Panksepp, 1992; Barrett, 1998). For example, the circumplex model of emotion struggles to accurately depict rare but complex emotions like shame, guilt, or jealousy, and these emotions are common in the real situations where false memories are created. Thus, confining research to just valence and arousal essentially lacks external validity. Specific emotions have adaptive functions, enabling us to respond appropriately to environmental changes (Howe, 2011); for instance, fear and anger are similar in valence and arousal, but fear tends to make individuals avoid threats, whereas anger inclines individuals toward eliminating threats (Cunningham and Brosch, 2012). In the field of false memory, there are also numerous examples that support this view. A study on the emotion congruence of discrete emotions (Bland et al., 2016) induced participants’ fear and anger through film clips (with no significant difference in valence and arousal) and asked participants to recall DRM lists with fearful and angry themes. In the subsequent recognition test, participants erroneously remembered critical lures consistent with their emotional state. In another study on discrete emotions, (Van Damme et al., 2017) conducted an experiment in which participants were induced to experience corresponding emotions by having them empathize with the content of the slides. The results from the recognition tests showed that participants under hope and fear conditions produced more false memories than did those under happy or despair conditions. Furthermore, studies on clinical populations have evidence that individuals with PTSD, depression, and a history of trauma are more prone to false memories related to their psychological disorders (e.g., trauma-related stimuli), regardless of the valence and arousal (Otgaar et al., 2017). The above results cannot be explained solely by the dimensions of arousal and valence, indicating that limiting research to investigating only valence and arousal does not facilitate a clear understanding of the relationship between emotions and false memories.

Contrary to what previous researchers have advocated for in controlling valence and arousal levels, this paper innovatively proposes that we should go beyond the dimensions of valence and arousal to explore the impact of emotions on false memories. It is worth mentioning that this paper focuses on spontaneous emotional false memories, and similar results have also been demonstrated in studies of implanted false memories. In their work, Sharma et al. (2023) reviewed 39 studies to explore the relationship between emotions and implanted false memories and found that the impact of emotions on implanted false memories depends upon the type or aspect of emotional measurement; specifically, the valence of emotions did not affect the generation of false memories, and when information was recalled with a delay, the arousal of emotions also did not have an impact on false memory. Moreover, stress and short-term distress experienced by subjects before encoding reduced the implanted false memories produced by the subjects, while prolonged distress, anger, and greater stress increased the implanted false memories produced by the subjects. This outcome might be because the two dimensions of the circumplex model of emotions cannot explain the dynamic nature of emotional changes. Emotional experiences are often rapidly changing, comprised of both continuous and momentary variations (Scherer, 2005).

Although it may be convenient to describe emotions simply in terms of valence and arousal, two emotions with the same valence and arousal (such as fear and anger) can have different effects on false memory. Moreover, limiting research to valence and arousal neglects the adaptive nature of emotions. Therefore, future research should not only control for valence and arousal dimensions but also explore the false memories induced by specific emotions.

HY: Writing – original draft. YZ: Writing – review & editing. ZL: Writing – review & editing.