Electrophysiological evidence for the effects of emotional content on false recognition memory

Cognition 179 (2018) 298–310

Contents lists available at ScienceDirect

Cognition journal homepage: www.elsevier.com/locate/cognit

Original Articles

Electrophysiological evidence for the effects of emotional content on false recognition memory Zhiwei Zhenga,b, Minjia Langa,b, Wei Wanga,b, Fengqiu Xiaoc, Juan Lia,b,d,e,

T

⁎

a

Center on Aging Psychology, CAS Key Laboratory of Mental Health, Institute of Psychology, Chinese Academy of Sciences, Beijing, China Department of Psychology, University of Chinese Academy of Sciences, Beijing, China c China National Children’s Center, Beijing, China d Magnetic Resonance Imaging Research Center, Institute of Psychology, Chinese Academy of Sciences, Beijing, China e State Key Laboratory of Brain and Cognitive Science, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China b

A R T I C LE I N FO

A B S T R A C T

Keywords: False recognition Emotion Familiarity Recollection Event-related potentials

Two competing hypotheses attempt to explain the effects of emotional content on the production of false memory. The conceptual relatedness account posits that negative emotion increases false memory by strengthening familiarity process, whereas the distinctiveness heuristic account postulates that negative emotion reduces false memory by influencing recollection process. Here, using the categorized pictures paradigm, we examined these hypotheses by investigating emotional influences on false recognition memory performance and the event-related potential (ERP) correlates of familiarity and recollection. Participants were presented with positive, neutral, or negative pictures from various categories during encoding and later completed a recognition test while electroencephalogram data were recorded. Behavioral results revealed lower corrected false recognition rates for negative and neutral pictures than for positive ones, with no significant difference between negative and neutral pictures. In addition, negative pictures were associated with a more conservative response bias in comparison with neutral and positive pictures. Importantly, ERP results revealed enhanced recollectionrelated parietal old/new effects for negative pictures relative to positive and neutral pictures, but comparable familiarity-related early frontal old/new effects across each type of emotional valence category during both true and false recognition. Our results suggest that emotionally negative content may affect production of false memory mainly by engaging a distinctiveness heuristic. Methodological implications of these findings are discussed.

1. Introduction Episodic memory is regarded as a constructive process that may lead to memory distortions (Schacter, 1999; Schacter & Addis, 2007; Schacter, Norman, & Koutstaal, 1998). Individuals may falsely recognize or recall events which had not been previously encountered. For example, in the Deese–Roediger–McDermott (DRM) paradigm (Deese, 1959; Roediger & McDermott, 1995), participants study a series of words (e.g., thread, sewing, sharp, point, and injection) that are all related to an unpresented lure word (e.g., needle). At subsequent test, they frequently falsely recognize the lure word with high confidence. In addition, subjectively compelling memory errors can also be reliably induced using categorized stimuli (Koutstaal & Schacter, 1997; Seamon, Luo, Schlegel, Greene, & Goldenberg, 2000). In this procedure, several exemplars (words or pictures) per category are presented during

encoding. The non-studied exemplars of a given category are used as lures to measure false memory during retrieval. Research on the production of false memory provides a basis for advancing our understanding of cognitive processes underlying human memory function (Arndt, 2012; Schacter & Slotnick, 2004). It is well established that emotion has a significant impact on memory for studied events (Buchanan, 2007; Talmi, 2013). Emotionally arousing stimuli are usually remembered with greater vividness and details than neutral stimuli (Kensinger & Corkin, 2003; Kensinger & Schacter, 2016; Sharot, Delgado, & Phelps, 2004). Moreover, there is increasing evidence that negative and positive emotion may differentially affect memory retrieval processes. For example, Ochsner (2000) found that recollection contributes more to the retrieval of negative items, whereas the contribution of familiarity does not differ between negative and positive emotions during test. Event-related potential

⁎ Corresponding author at: Center on Aging Psychology, CAS Key Laboratory of Mental Health, Institute of Psychology, Chinese Academy of Sciences, No. 16 Lincui Road, Chaoyang District, Beijing 100101, China. E-mail address: [email protected] (J. Li).

https://doi.org/10.1016/j.cognition.2018.06.013 Received 4 January 2018; Received in revised form 9 June 2018; Accepted 18 June 2018 0010-0277/ © 2018 Elsevier B.V. All rights reserved.

Cognition 179 (2018) 298–310

Z. Zheng et al.

airplanes are distinctive and I would remember that”) (Gallo, 2010). Here, distinctiveness refers to the uniqueness of an item (Kensinger & Corkin, 2004). Negative stimuli may be more distinctive because of their high relevance to survival (Dolan, 2002; LeDoux, 2002). Additionally, research has consistently shown that negative items are associated with increased recollective details compared with neutral and positive items (e.g., Johansson et al., 2004; Ochsner, 2000). Therefore, participants are likely to form more detailed recollective expectations for negative stimuli. Subsequently, they may be more conservative in responding “old” to emotionally negative stimuli and avoid falsely recognizing the lures that fail to elicit expected detailed recollections. There is a great deal of prior research supporting the distinctiveness heuristic account by revealing a reduction in memory errors for negative lures compared to neutral and positive lures (Kensinger & Corkin, 2004; Kensinger, O'Brien, Swanberg, Garoff-Eaton, & Schacter, 2007; Kensinger & Schacter, 2005a, 2005b; Kensinger & Schacter, 2006; Pesta, Murphy, & Sanders, 2001). Nevertheless, it should be noted that more direct evidence for a reduction of false recognition rates for emotionally negative items comes from studies that examined reality monitoring ability (i.e., the ability to distinguish between previously imagined and perceived events; Kensinger et al., 2007; Kensinger & Schacter, 2005a, 2005b, 2006), making it difficult to make parallel comparisons with studies using the DRM or categorized lists to experimentally induce false memories. Moreover, most of the previous studies did not estimate the relative contribution of familiarity and recollection to false recognition memory (Kensinger & Corkin, 2004; Kensinger et al., 2007; Kensinger & Schacter, 2005a, 2005b, 2006). In summary, theoretical explanations regarding how emotional content affects the production of false memory remain controversial. To this end, the present study was designed to further explore how emotion modulates behavioral performance and the cognitive processes associated with false recognition memory using the categorized pictures paradigm. Specifically, participants were presented with positive, neutral, or negative pictures from various categories during encoding. Participants then performed an old/new recognition test, during which studied items were intermixed with lures from studied categories and new items drawn from non-studied categories. Electroencephalogram (EEG) data were recorded during the test phase to measure the relative contribution of familiarity and recollection to false recognition. It is well known that ERP techniques provide an objective and reliable measure of the retrieval processes involved in episodic memory. Previous ERP studies have identified two distinct ERP old/new effects that are associated with familiarity and recollection. The early frontal old/new effect from 300 to 500 ms is thought to reflect familiaritybased recognition (Curran, 2000; Rugg & Curran, 2007; but see Paller, Voss, and Boehm (2007) and Voss and Federmeier (2011) for an alternative interpretation), while the parietal old/new effect from 500 to 800 ms is believed to reflect recollection-based recognition (Rugg & Curran, 2007). In addition, there are two old/new effects reported to be associated with post-retrieval processes that occur later than the parietal old/new effect. The late frontal old/new effect has been related to post-retrieval monitoring and evaluation processes (Cruse & Wilding, 2009; Hayama, Johnson, & Rugg, 2008). The late posterior negativity, characterized by more negative-going deflections for correctly classified old items than correctly rejected new items, is regarded as reflecting reconstructive mnemonic processes when memory attributes cannot easily be recovered (Mecklinger, Rosburg, & Johansson, 2016). According to the conceptual relatedness account, gist traces are strengthened for negative content due to the enhanced conceptual relatedness among study items, relative to neutral and positive content. Accordingly, negative valence is then associated with increased false recognition of lures, and false recognition of negative lures will evoke a greater familiarity related early frontal old/new effect, relative to neutral and positive lures. Conversely, according to the distinctiveness heuristic account, participants will be more likely to adopt a distinctiveness heuristic for negative pictures due to their increased

(ERP) studies have consistently shown enhanced recollection-related electrophysiological activity for negative items than for positive items (Inaba, Kamishima, & Ohira, 2007; Inaba, Nomura, & Ohira, 2005; Johansson, Mecklinger, & Treese, 2004; Schaefer, Pottage, & Rickart, 2011; Weymar, Loew, & Hamm, 2011). These findings suggest that memory vividness is boosted for negative stimuli compared to positive stimuli in true memory. With respect to the effects of emotional content on false memory, there are two competing hypotheses which make opposing predictions. The conceptual relatedness account proposes that emotional content, particularly negative emotion, can elevate false memory (Bookbinder & Brainerd, 2016). This account evolved from fuzzy-trace theory (FTT), which suggests that gist and verbatim traces of items are stored in parallel during encoding. Gist traces represent the semantic content shared among study items, while verbatim traces represent the surface details of these items. During retrieval, gist traces foment familiaritybased false memory by increasing the similarity between lures and study items, whereas verbatim traces suppress false memory by recollecting item-specific information of study items (i.e., recollection rejection, Brainerd & Reyna, 1998, 2002; Brainerd & Reyna, 2005; Brainerd, Reyna, Wright, & Mojardin, 2003). Regarding the effects of emotion on false memory, the conceptual relatedness account supposes that emotional content enhances conceptual relatedness among study items, relative to neutral content, and conceptual cohesion are stronger for emotionally negative content than for positive content. Negative emotional content can be much more easily integrated and organized in memory due to the more conceptually overlapping features (Brainerd, Stein, Silveira, Rohenkohl, & Reyna, 2008; Talmi & Moscovitch, 2004; Talmi, 2013). Therefore, negative emotional content is associated with enhanced gist traces relative to positive content. In addition, it is argued that negative emotional content has weaker verbatim traces, but positive emotional content has stronger verbatim traces (Bookbinder & Brainerd, 2016). Consequently, it is postulated that negative emotion will increase false memory by strengthening gist traces while weakening verbatim traces. Furthermore, it is postulated that false memory for positive emotional content depends on a tradeoff between gist and verbatim traces. Consistent with this account, two representative studies, namely Bookbinder and Brainerd (2017) using the categorized pictures paradigm, and Brainerd et al. (2008) using the DRM paradigm, provided converging evidence that negatively valenced stimuli elevates false memory, relative to neutral and positive stimuli (see Bookbinder and Brainerd (2016) for a comprehensive review). Importantly, these two studies consistently showed that negative items increased engagement of familiarity process during false recognition using the conjoint recognition model (Brainerd, Reyna, & Mojardin, 1999; Brainerd et al., 2003). However, some different patterns of results also emerged. For example, Choi, Kensinger, and Rajaram (2013) found that emotion did not increase or even reduced false memory when thematic relatedness was equivalent between emotional and neutral items. In addition, Dehon, Laroi, and Van der Linden (2010) found more “Remember” responses for false recognition of negative lures (e.g., sorrow, grief, tears, despair, mourning…; lure = sadness) compared with neutral and positive lures (e.g., journey, relaxation, beach, sun, serenity…; lure = holidays) when using the Remember/Know procedure (Tulving, 1985). That is, negative false memories were associated with higher rates of recollections relative to neutral and positive false memories. The distinctiveness heuristic account offers an alternative perspective. This proposes that negative emotional content is less prone to memory distortions (Schacter, Gallo & Kensinger, 2007; Schacter & Wiseman 2006). Specifically, it is suggested that the false recognition of lures rely on the distinctiveness of study items and retrieval expectations associated with them (Dodson & Schacter, 2002; Schacter, Israel, & Racine, 1999). The distinctiveness heuristic can suppress false recognition of lures during retrieval based on metacognitive expectations (e.g., “I did not take the airplane home last Spring Festival, because 299

Cognition 179 (2018) 298–310

Z. Zheng et al.

served as lures during the test phase. The assignment of picture sets to each recognition status was counterbalanced to ensure that every picture was presented equally as often as an old or lure item. All the exemplars from the remaining nine categories for each type of emotional valence were used as new items in the test phase. The formal experiment was divided into three blocks. Each block began with a study phase, and then a distracter task followed by a test phase. Each study phase comprised 24 pictures and contained four exemplars from each of the six categories per valence. Each test phase consisted of the studied 24 pictures (old items), the remaining 24 nonstudied pictures from studied categories (lure items), and 24 new pictures drawn from three non-studied categories (new items) per valence. In addition, three untested buffers were separately presented at the beginning and end of the study and test phases in each block to avoid primacy and recency effects. A total of 216 target trials were conducted during the test phase for each type of emotional valence, consisting of 72 old pictures, 72 lure pictures, and 72 new pictures. The order of individual study-test blocks was counterbalanced across participants. Before beginning the formal experiment, each participant completed two brief study-test practice sessions; each practice session included four pictures from one category at study and 12 pictures at test for each type of valence to familiarize participants with the procedure. None of the practice stimuli appeared during the subsequent study or test phases.

recollective distinctiveness. Specifically, participants should form detailed recollective expectations for negative stimuli, and only falsely recognize the negative lures when they elicit expected detailed recollection. As a result, behaviorally, negative pictures should be associated with reduced false recognition rates. For ERP results, false recognition of negative lures should evoke a greater recollection related parietal old/new effect relative to neutral and positive lures. 2. Material and methods 2.1. Participants Twenty-four right-handed healthy young adults participated in the study. Data from three participants were excluded due to an insufficient number of artifact-free trials (< 15) for averaging the ERP in at least one relevant condition, resulting in a final sample of 21 participants (10 males, mean age = 22.76 years, SD = 1.58). All participants were native Chinese speakers with normal or corrected-to-normal vision and were free from neurological and psychiatric disorders. Each participant provided informed consent and was paid for their participation. The study was approved by the Ethics Committee of the Institute of Psychology, Chinese Academy of Sciences. 2.2. Materials

2.3. Procedure

Six hundred and forty-eight realistic pictures were selected from the International Affective Picture System (IAPS; Lang, Bradley, & Cuthbert, 2008) and Internet searches to serve as stimuli consisting of positive, neutral, and negative pictures. Ten participants (6 males, mean age 22.2 years, SD = 1.92), none of whom participated in the main memory experiment, rated the valence (negative/positive) and arousal (calm/aroused) of the pictures using two 9-point self-report scales based on the self-assessment manikin scale (Bradley & Lang, 1994). The rating results for valence and arousal for each type of emotional valence are shown in Table 1. Repeated measures analysis of variance (ANOVA) on ratings for valence confirmed a main effect of Valence [F(2, 18) = 77.63, p < .001, η2p = .90]. Bonferroni pairwise comparisons revealed most pleasant ratings for positive pictures, followed by neutral pictures, and then negative pictures (ps < .001). Repeated measures ANOVA on ratings for arousal also revealed a main effect of Valence [F(2, 18) = 39.55, p < .001, η2p = .82]. Bonferroni pairwise comparisons revealed that positive and negative pictures did not differ significantly in arousal levels (p = .33), and were more arousing than neutral pictures (ps < .001). The stimuli consisted of a total 81 categories of color pictures. There were 27 positive, 27 neutral, and 27 negative categories with eight typical exemplars per category, and thus there were 216 items per valence. The positive stimuli included categories such as a birthday party, happy families, and sports scenes; the neutral stimuli included categories such as farmland, working offices, and street scenes. The negative stimuli included categories such as scenes of aircraft accidents, air pollution, and graveyards. For each type of emotional valence, of the 27 categories, 18 were randomly selected to construct the old and lure conditions in the test phase. Four exemplars randomly selected from each category were presented during the encoding phase, and subsequently used as old items; the remaining four were reserved during the encoding phase and

The experiment was designed using E-Prime (Psychology Software Tools). All the pictures (300 × 270 pixels) were displayed in the center of a black background at a viewing distance of approximately 100 cm with stimuli at a visual angle of 7° × 6.3°. All pictures were digitally matched for brightness and contrast. Fig. 1 illustrates the procedures for the study and test phases. During each study phase, each trial began with a fixation cross (+) displayed in the center of the screen for 500 ms. A picture was then presented for 1000 ms followed by a 500 ms blank screen. The pictures were pseudorandomized in presentation order and the same type of picture (i.e., valence and category) was not presented more than three times consecutively. Participants were instructed to remember the picture for a subsequent test. To encourage the participants to actively encode and consolidate each picture, a rating interface was then presented with a maximum presentation time of 5000 ms, which prompted the participants to rate the valence of the picture using a scale from 1 (most negative) to 5 (most positive). Afterwards, the rating interface was replaced with a blank screen for the inter-trial interval (500 ms). There was a distracter task between the study phase and test phase during which participants counted backward in threes for 120 s. During each test phase, every trial began with a fixation cross presented for 500 ms followed by a picture with a maximum presentation time of 2000 ms. The pictures were pseudo-randomized in presentation order and the same type of picture (i.e., valence, category and item type) was not presented more than three times consecutively, as per the study phase. Participants were instructed to indicate whether the picture was old or new relative to the preceding study phase using the keyboard. After a response was provided, a 1500 ms blank screen was presented followed by a confidence rating interface with a maximum presentation time of 5000 ms, which prompted the participants to rate their recognition confidence using a scale from 1 (least confident) to 5 (most confident). After a response was provided, a blank screen was presented for 500 ms and the next trial began. Participants were asked to respond as quickly and accurately as possible during old/new recognition, and the key-response mappings were counterbalanced. Half of the participants made their responses of “old” by pressing the key ‘F’ with their left index finger, and of “new” by pressing the key ‘J’ using their right index finger. The other half responded “old” by pressing the key ‘J’ using their right index finger, and

Table 1 Valence and arousal ratings as a function of emotional valence (mean and standard deviations).

Valence Arousal

Positive

Neutral

Negative

6.60 (0.75) 4.89 (1.52)

5.24 (0.47) 3.21 (1.25)

3.43 (0.47) 5.29 (1.47)

300

Cognition 179 (2018) 298–310

Z. Zheng et al.

Fig. 1. Schematic illustration of the stimuli and experimental design.

‘‘new” by pressing the key ‘‘F” using their left index finger. To minimize EEG artifacts, the participants were also instructed to maintain fixation, to relax, and to avoid making head motions and eye movements other than blinks.

2.5. ERP data analyses Based on visual inspection of the grand average waveforms and previous studies (Eppinger, Herbert, & Kray, 2010; Jäger, Mecklinger, & Kipp, 2006; Rhodes & Donaldson, 2007), the ERP old/new effects were quantified by calculating the mean amplitudes over three time windows of 250–350 ms, 400–600 ms, and 650–1000 ms. These time windows estimated the frontal old/new effect, the parietal old/new effect, and the late old/new effect, respectively. The mean amplitudes in each time window were obtained from nine electrodes: frontal (F3, Fz, and F4), central (C3, Cz, and C4), and parietal (P3, Pz, and P4). To investigate the retrieval processes underlying true and false recognition, we separately measured the ERP old/new effects by comparing the ERP of true recognition with that of new items, and the ERP of false recognition with that of new items. For each time window, the initial repeated measures ANOVAs with the within-subjects factors of Valence (positive vs. neutral vs. negative), Response (true recognition vs. new or false recognition vs. new), Location (frontal vs. central vs. parietal), and Hemisphere (left vs. middle vs. right) were performed on the average amplitudes. Subsidiary ANOVAs were conducted separately on positive, neutral, and negative emotionally valenced stimuli when there were significant interactions involving the factors of Response and Valence. Then, for the interactions involving the factors of Response, pairwise comparisons were used to quantify the old/new effects at the frontal and parietal locations. Repeated measures ANOVAs employing the within-subjects factor of Valence (positive vs. neutral vs. negative) were conducted on the mean amplitudes of the difference waveforms (true or false recognition minus new items) at their representative electrodes in order to compare the valence differences in magnitude of the old/new effects. Topographical maps depicting the old/new effects were formed by subtracting the ERP of new items from the ERP of true or false recognition for each type of emotional valence.

2.4. EEG recording and processing EEG data were recorded continuously (0.05–100 Hz, sampling rate 500 Hz) from 62 Ag/AgCl electrodes embedded in an elastic cap using the Neuroscan system (http://www.neuroscan.com), based on an extended version of the international 10–20 system. The vertical and horizontal electrooculograms were recorded with electrodes placed above and below the left eye and on the outer canthi of both eyes. EEG activity was referenced online to the left mastoid. All electrode impedances were kept below 5 kΩ. Offline EEG data from the test phase were processed using Neuroscan v. 4.3 software package. Raw data was first digitally re-referenced to the average of the left and right mastoids. Eye-blink artifacts were automatically corrected using the ocular artifact reduction algorithm with a regression-based procedure (Semlitsch, Anderer, Schuster, & Presslich, 1986). Then, the data were filtered with a lowpass zero phase shift FIR filter of 40 Hz (24 dB/octave) and separated into 1700 ms epochs, including 200 ms prior to the stimulus onset for a baseline correction. Epochs with an amplitude exceeding ± 100 µV were rejected to remove the artifacts induced by eye movements, muscle artifacts, electrode drifting, or other artifacts. ERP were computed within three different response categories for each type of emotional valance: old items given old responses (true recognition), lure items given old responses (false recognition), and new items given new responses (new items). ERP recordings for each response category were collapsed across the confidence levels, and a minimum of 15 artifact-free trials was required from each participant to ensure an acceptable signal-to-noise ratio. Three participants were discarded due to limited valid trials. Thus, 21 participants remained in the dataset for subsequent statistical analyses.

3. Results All behavioral and ERP analyses were conducted with the data collapsed across confidence levels. The Greenhouse-Geisser correction for non-sphericity of data was applied as necessary. The uncorrected 301

Cognition 179 (2018) 298–310

Z. Zheng et al.

Fig. 2. Behavioral results as a function of emotional valence. Panel A shows the proportions of old responses for old, lure, and new items. Panel B shows the false recognition rates. Panel C shows the true recognition scores. Panel D shows the response bias. Panel E shows the mean response times for true recognition, false recognition, and correct rejection of new items. Panel F shows the confidence ratings assigned to true and false recognition.

degrees of freedom, corrected p-values, and effect sizes (η2p ) are reported. The p-values of the pairwise comparisons were adjusted using a Bonferroni correction. For all analyses, the significance level was set to .05.

Fig. 2B). There were significantly higher false alarm rates for positive lures than for negative lures (p < .001), higher false alarm rates for neutral lures than that for negative lures (p = .006), and higher false alarm rates for positive lures than that for neutral lures (p = .043). Raw false alarm rates for positive and neutral lures did not differ significantly. We then computed corrected false recognition rates by subtracting false alarm rates to new items (i.e., baseline false alarms) from false alarm rates to lure items (see Fig. 2B). Repeated measures ANOVA with Valence as a within-subjects factor revealed a significant main effect of Valence (F(2, 40) = 9.60, p < .001, η2p = .32). Pairwise comparisons revealed higher corrected false alarm rates for positive lures than for neutral (p = .006) and negative lures (p < .001). Corrected data for neutral and negative lures did not differ significantly. These results showed that negative lure pictures were associated with reduced false recognition, especially in comparison to positive pictures.

3.1. Valence ratings during encoding For the valence ratings, repeated measures ANOVA with Valence as a within-subjects factor revealed a main effect of Valence, F (2, 40) = 111.99, p < .001, η2p = .85. As expected, pairwise comparisons revealed lowest ratings for negative pictures [mean (SD) = 2.26 (0.35)], intermediate for neutral pictures [3.36 (0.35)], and highest for positive pictures [3.91 (0.39)], ps < .001. 3.2. Behavioral results 3.2.1. Proportions of old responses The mean proportions of old responses for the three item types (old, lure and new) as a function of emotional valence are presented in Fig. 2A. Repeated measures ANOVA with the within-subjects factors of Valence and Item Type revealed a Valence × Item Type interaction (F (4, 80) = 5.95, p < .001, η2p = .23). Pairwise comparisons for each type of valence revealed higher proportion of old responses for old pictures than for lure pictures (ps < .001). In addition, false alarm rates for lure pictures were significantly higher than those for new pictures (ps < .001), confirming the false memory effect for each type of valence. Pairwise comparisons were also conducted separately for each kind of item type. These revealed higher hit rates for positive and neutral pictures than for negative pictures (ps < .001). The hit rates for positive and neutral pictures did not significantly differ from each other. The false alarm rates of new pictures were higher for neutral pictures than those for negative pictures (p = .009). There were no significant differences in the false alarm rates of new pictures between positive and neutral pictures, or between positive and negative pictures.

3.2.3. True recognition In order to compare true recognition performance across the three types of emotional valence, we computed d′, an index of memory sensitivity: d′ (old/lure) = z(hits) − z(false alarms to lures); d′ (old/ new) = z(hits) − z(false alarms to new items), according to signal detection theory (Macmillan & Creelman, 2004, see Fig. 2C). The d′ was subjected to repeated measures ANOVA employing the within-subjects factor of Valence. For d′ (old/lure), there was no significant main effect of Valence. For d′ (old/new), the main effect of Valence reached significance (F(2, 40) = 3.27, p = .048, η2p = .14). However, although memory sensitivity for positive and negative pictures was numerically greater than neutral pictures, pairwise comparisons revealed no significant differences between positive and neutral pictures (p = .10), or between negative and neutral pictures (p = .15). Overall, no robust memory benefits for emotional over neutral pictures were identified. 3.2.4. Response bias Response bias index (C1) was computed according to signal

3.2.2. False recognition To investigate the effects of emotion on false recognition, first we performed pairwise comparisons for raw false alarm rates of lures (see

1 C (old/lure) = −[z(Hits) + z(false alarms to lures)]/2; C (old/new) = −[z(Hits) + z (false alarms to new itmes)]/2.

302

Cognition 179 (2018) 298–310

Z. Zheng et al.

Fig. 3. The grand average ERP waveforms for true recognition (red), false recognition (blue), and correct rejection of new items (black) as a function of emotion valence at three midline electrodes, showing from −200 to 1500 ms. The scale bars indicate the time windows used for the statistical analyses (250–350 ms, 400–600 ms, and 650–1000 ms). The positive voltages are plotted upwards. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

suggested that negative pictures were less likely to be endorsed as “old” relative to positive and neutral pictures.

detection theory for each type of valence (see Fig. 2D). More positive values for C are indicative of a relatively conservative response bias. For C (old/lure), repeated measures ANOVA with Valence as a withinsubjects factor revealed a significant main effect of Valence (F (2, 40) = 23.27, p < .001, η2p = .54). Pairwise comparisons revealed that negative pictures were associated with a more conservative response bias than positive and neutral pictures (ps < .001). There was no significant difference between positive and neutral pictures. For C (old/new), ANOVA also revealed a main effect of Valence (F (2, 40) = 17.62, p < .001, η2p = .47). Pairwise comparisons again revealed a more conservative response bias for negative pictures than positive (p = .001) and neutral ones (p < .001), with no significant difference between positive and neutral pictures. These results

3.2.5. Response times The mean response times for the three response categories relevant to the ERP analyses (true recognition, false recognition, and correct rejection of new items, see Fig. 2E) were analyzed with two-way repeated measures ANOVA employing the within-subjects factors of Valence and Response Category. There was a significant Valence × Response Category interaction (F(4, 80) = 10.08, p < .001, η2p = .34). First, the interaction was decomposed by conducting pairwise comparisons separately for each type of valence. For positive and neutral pictures, response times were slower for false recognition than 303

Cognition 179 (2018) 298–310

Z. Zheng et al.

Fig. 4. The early frontal old/new effect (250–350 ms) evoked by positive, neutral, and negative pictures. Panel A shows the topographical maps of the old/new effects for (a) true and (b) false recognition, which were formed by subtracting the ERP of new items from the ERP of true or false recognition. The scale bar shows the amplitude range; Panel B shows the mean amplitudes of the early frontal old/new effects for true and false recognition at electrode Fz. The error bars represent the standard error of the mean.

3.3.1. 250–350 ms 3.3.1.1. True recognition vs. new. Comparison of true recognition and new items during this time window revealed significant frontal old/new effects for positive, neutral, and negative pictures (see Fig. 3). The initial ANOVA with the within-subjects factors of Valence, Response, Location, and Hemisphere revealed a Valence × Response interaction (F(2, 40) = 8.61, p = .001, η2p = .30). Subsidiary ANOVA for positive pictures revealed a Response × Location interaction (F(2, 40) = 5.87, p = .014, η2p = .23). Pairwise comparisons revealed that true recognition was significantly more positive than new items at the frontal (p = .019) and central (p = .002) scalp regions. The old/new differences were not significant at the parietal scalp region. Subsidiary ANOVA for neutral pictures revealed a Response × Location interaction (F(2, 40) = 18.97, p < .001, η2p = .49). Pairwise comparisons revealed that true recognition was significantly more positive than new items only at the frontal scalp regions (p = .013). Subsidiary ANOVA for negative pictures revealed a Response × Location interaction (F (2, 40) = 6.62, p = .01, η2p = .25). Pairwise comparisons revealed that true recognition evoked widespread positivity, with greater old/new differences at the frontal and central scalp regions (ps < .001) than that at the parietal scalp region (p = .001). The topographic maps of the ERP old/new effects (true recognition minus new) during this time window for each type of emotional valence are illustrated in Fig. 4A.

for true recognition (ps < .005). There were trends for false recognition being slower than correct rejection of new items (ps < .06). No significant differences were found between true recognition and correct rejection of new items. For negative pictures, pairwise comparisons revealed a pattern of correct rejection of new items < true recognition < false recognition (ps < .005). Pairwise comparisons were then conducted separately for each response category. For true recognition, the results revealed that participants responded more slowly to negative pictures than to positive pictures (p = .016), and more slowly to negative pictures than to neutral pictures (p = .042). There were no significant differences in response times between positive and neutral pictures during true recognition. Pairwise comparisons for false recognition revealed that participants responded more slowly to negative lures than to neutral lures (p = .012). Participants seemingly responded more slowly to negative lures than to positive lures, but the difference did not reach significance. There was no significant difference between positive and neutral lures during false recognition. Pairwise comparisons for new items revealed that participants responded more quickly to negative pictures than to positive (p < .001) and neutral pictures (p = .002). There was no significant difference between positive and neutral pictures during correct rejection of new items. 3.2.6. Confidence ratings The mean confidence ratings assigned to true and false recognition for each type of valence were analyzed with two-way repeated measures ANOVA employing the within-subjects factors of Valence and Response Category (see Fig. 2F). The main effect of Response Category was significant (F(1, 20) = 105.37, p < .001, η2p = .84), revealing that participants were more confident in true recognition than false recognition. The main effect of Valence was also significant (F (2, 40) = 8.03, p = .001, η2p = .29), revealing that participants were more confident in their judgments for negative pictures than for positive (p = .002) and neutral pictures (p = .016). The confidence ratings for positive and neutral pictures did not differ. No significant interaction between Valence and Response Category was identified, indicating higher confidence ratings for negative pictures than for positive and neutral pictures during both true and false recognition.

3.3.1.2. False recognition vs. new. Comparison of false recognition and new items during this time window revealed significant frontal old/new effects for positive, neutral, and negative pictures (see Fig. 3). The initial ANOVA with the within-subjects factors of Valence, Response, Location, and Hemisphere revealed a Valence × Response interaction (F(2, 40) = 4.47, p = .024, η2p = .18) η2p = .10). Subsidiary ANOVA for positive pictures revealed a Response × Location interaction (F (2, 40) = 6.65, p = .012, η2p = .25). Pairwise comparisons revealed that false recognition was significantly more positive than new items at the frontal (p = .008) and central (p = .003) scalp regions. The old/ new differences were not significant at the parietal scalp region. Subsidiary ANOVA for neutral pictures revealed a Response × Location interaction (F(2, 40) = 4.72, p = .024, η2p = .19). Pairwise comparisons revealed that false recognition was significantly more positive than new items only at the frontal scalp regions (p = .033). Subsidiary ANOVA for negative pictures revealed a Valence × Response × Location interaction (F(4, 80) = 2.82, p = .04, η2p = .12). Pairwise comparisons revealed significant old/new differences between false recognition and new items broadly distributed over the scalp (ps < .005), with a maximal distribution over the middle central scalp sites (electrode Cz, p < .001). The topographic maps of the ERP old/new effects (false recognition minus new) during this time window for each type of emotional valence are illustrated in Fig. 4A.

3.3. ERP results The grand average ERP evoked by true recognition, false recognition, and correct rejections to new items for positive, neutral, and negative pictures are shown in Fig. 3. The ERP data for the three time windows of 250–350 ms, 400–600 ms, and 650–1000 ms were used to quantify the old/ new effects during memory retrieval. The first two time windows were related to the early frontal and later parietal old/new effects, respectively; the third time window was used to analyze the late ERP old/new effects. 304

Cognition 179 (2018) 298–310

Z. Zheng et al.

(2, 40) = 7.30, p = .002, η2p = .27), and a Valence × Response × Location interaction (F(4, 80) = 4.76, p = .01, η2p = .19). Subsidiary ANOVA for positive and neutral pictures revealed no significant main effects or interactions involving Response, indicating that positive and neutral pictures evoked no apparent old/new differences at any scalp regions during this time window. Subsidiary ANOVA for negative pictures revealed a Response × Location interaction (F(2, 40) = 6.41, p = .009, η2p = .24). Pairwise comparisons revealed greater old/new differences at the parietal and central scalp regions (ps < .005) than that at the frontal scalp region (p = .011). The topographic maps of the ERP old/new effects (false recognition minus new) during this time window for each type of emotional valence are illustrated in Fig. 5A.

3.3.1.3. Between-valence analyses. Between-valence comparisons were conducted with repeated measures ANOVAs employing the withinsubjects factor of Valence on the old/new difference waveforms (true or false recognition minus new) at electrode Fz separately for true recognition and false recognition. The main effect of Valence was statistically not significant during either true or false recognition, revealing comparable magnitude of early frontal old/new effects for positive, neutral, and negative pictures. The mean amplitudes of early frontal old/new effects (true or false recognition minus new) for each type of emotional valence are shown in Fig. 4B. From the above we can see that all emotionally valenced pictures (positive, neutral, and negative) evoked significant familiarity-related early frontal old/new effects during both true and false recognition. Interestingly, the old/new effects extended to parietal scalp sites for negative valence. These posterior distributed old/new differences during this time window may reflect the early emerging parietal old/ new effects.

3.3.2.3. Between-valence analyses. Between-valence comparisons were conducted with repeated measures ANOVAs employing the withinsubjects factor of Valence on the old/new difference waveforms (true or false recognition minus new) at electrode Pz separately for true recognition and false recognition. For true recognition, there was a main effect of Valence (F(2, 40) = 8.91, p = .001, η2p = .31). Pairwise comparisons revealed greater old/new differences for negative pictures than for positive (p = .049) and neutral pictures (p = .002). Old/new differences were numerically greater for positive pictures than for neutral pictures, but did not statistically differ from each other. For false recognition, there was also a main effect of Valence (F (2, 40) = 9.77, p < .001, η2p = .33). Pairwise comparisons again revealed greater old/new differences for negative pictures than for positive (p = .034) and neutral pictures (p = .001), with no significant differences between positive and neutral pictures. The mean amplitudes of parietal old/new effects (true or false recognition minus new) for each type of emotional valence are shown in Fig. 5B. In summary, the negative pictures evoked greater recollection-related parietal old/new effects than positive and neutral pictures during both true and false recognition. The positive pictures evoked significant parietal old/new effects during true recognition but not false recognition. There were no significant parietal old/new effects for neutral valence during either true or false recognition. Although the old/new differences during true recognition reached significance at the frontal scalp regions for neutral pictures, we believe that this ERP effect most likely reflects the residual familiarity-related frontal old/new effects.

3.3.2. 400–600 ms 3.3.2.1. True recognition vs. new. Comparison of true recognition and new items during this time window revealed significant parietal old/ new effects for negative and positive, but not for neutral pictures (see Fig. 3). The initial ANOVA with the within-subjects factors of Valence, Response, Location, and Hemisphere revealed a Valence × Response interaction (F(2, 40) = 10.20, p < .001, η2p = .34), and a Valence × Response × Location interaction (F(4, 80) = 11.05, p < .001, η2p = .36). Subsidiary ANOVA for positive pictures revealed a Response × Location interaction (F(2, 40) = 5.87, p = .012, η2p = .23). Pairwise comparisons revealed that true recognition was significantly more positive than new items at the parietal (p = .013) and central (p = .015) scalp regions. The old/new differences were not significant at the frontal scalp region. Subsidiary ANOVA for neutral pictures revealed a Response × Location interaction (F(2, 40) = 8.20, p = .006, η2p = .29). Pairwise comparisons revealed significant old/new differences distributed at the frontal (p = .006) and central (p = .012) scalp region, but not at the parietal scalp region. Subsidiary ANOVA for negative pictures revealed a main effect of Response (F(1, 20) = 39.12, p < .001, η2p = .66). No significant interactions involving the factors of Location, or Hemisphere were identified, indicating widespread old/ new effects for true recognition during this time window. The topographic maps of the ERP old/new effects (true recognition minus new) during this time window for each type of emotional valence are illustrated in Fig. 5A.

3.3.3. 650–1000 ms 3.3.3.1. True recognition vs. new. Comparison of true recognition and new items during this time window only revealed significant old/new differences at the frontal scalp region for negative pictures (see Fig. 3). The initial ANOVA with the within-subjects factors of Valence, Response, Location, and Hemisphere revealed a Valence × Response interaction (F(2, 40) = 4.61, p = .02, η2p = .19). Subsidiary ANOVA for positive pictures revealed no significant main effects or interactions involving Response, suggesting that positive pictures evoked no

3.3.2.2. False recognition vs. new. Comparison of false recognition and new items during this time window only revealed significant parietal old/new effects for negative pictures (see Fig. 3). The initial ANOVA with the within-subjects factors of Valence, Response, Location, and Hemisphere revealed a Valence × Response interaction (F

Fig. 5. The parietal old/new effect (400–600 ms) evoked by positive, neutral, and negative pictures. Panel A shows the topographical maps of the old/new effects for (a) true and (b) false recognition, which were formed by subtracting the ERP of new items from the ERP of true or false recognition. The scale bar shows the amplitude range; Panel B shows the mean amplitudes of the parietal old/new effects for true and false recognition at electrode Pz. The error bars represent the standard error of the mean. 305

Cognition 179 (2018) 298–310

Z. Zheng et al.

Fig. 6. The Late old/new effect (650–1000 ms) evoked by positive, neutral, and negative pictures. Panel A shows the topographical maps of the late old/new effects for (a) true and (b) false recognition, which were formed by subtracting the ERP of new items from the ERP of true, or false recognition. The scale bar shows the amplitude range; Panel B shows the mean amplitudes of the late old/new effects for true recognition at electrode Fz, and for false recognition at electrode Pz. The error bars represent the standard error of the mean.

effect of Valence (F(2, 40) = 7.42, p = .002, η2p = .27). Pairwise comparisons revealed a greater negativity for positive than for negative valence (p = .025), and a greater negativity for neutral valence than for negative valence (p = .018), with no significant differences between positive and neutral valence. The mean amplitudes of late old/new effects (true or false recognition minus new) for each type of emotional valence are shown in Fig. 6B. The above results demonstrate that positive and neutral pictures evoked posterior negative old/new differences during false recognition, and negative pictures evoked late frontal old/new effects during true recognition.

apparent old/new effects at any scalp regions during this time window. Subsidiary ANOVA for neutral pictures revealed a Response × Location interaction (F(2, 40) = 6.00, p = .016, η2p = .23). Pairwise comparisons only revealed marginally significant old/new differences at the frontal scalp region (p = .086). Subsidiary ANOVA for negative pictures revealed a main effect of Response (F(1, 20) = 5.46, p = .03, η2p = .21), and a marginally significant Response × Location interaction (F(2, 40) = 3.58, p = .059, Pairwise η2p = .12). comparisons revealed significant old/new differences distributed at the frontal (p = .018) and central (p = .013) scalp region, but not at the parietal scalp region. The topographic maps of the ERP old/new effects (true recognition minus new) during this time window for each type of emotional valence are illustrated in Fig. 6A.

4. Discussion 3.3.3.2. False recognition vs. new. Comparison of false recognition and new items during this time window revealed a significant centralparietal negativity for positive and neutral pictures (see Fig. 3). The initial ANOVA with the within-subjects factors of Valence, Response, Location, and Hemisphere revealed a Valence × Response interaction (F(2, 40) = 6.46, p = .005, η2p = .24). Subsidiary ANOVA for positive pictures revealed a main effect of Response (F(1, 20) = 7.75, p = .011, η2p = .28) and a marginally significant Response × Location interaction (F(2, 40) = 3.24, p = .078, η2p = .14). Pairwise comparisons revealed false recognition evoked more significant negative-going ERP than did new items over the parietal (p = .003) and central (p = .017) scalp regions, but not over the frontal regions. Subsidiary ANOVA for neutral pictures revealed a marginally significant main effect of Response (F (1, 20) = 3.31, p = .084, η2p = .14), indicating trends for false recognition being more negative than new items. Exploratory pairwise comparisons revealed that the old/new differences reached significance at the parietal scalp region (p = .005). Subsidiary ANOVA for negative pictures revealed no significant main effects or interactions involving Response, suggesting that negative pictures evoked no apparent old/new effects at any scalp regions during this time window. The topographic maps of the ERP old/new effects (false recognition minus new) during this time window for each type of emotional valence are illustrated in Fig. 6A.

Recently, researchers have shown an increased interest in how emotional content modulates false memory. Two current hypotheses (conceptual relatedness account vs. distinctiveness heuristic account) make opposing predictions on the emotional effects on memory distortions. The present study directly tested these competing hypotheses by examining how emotion affects the production of false memory for categorized pictures. EEG was recorded during the recognition test, and ERP old/new effects were measured to quantify the contributions of familiarity and recollection to false recognition for positive, neutral, and negative pictures. To the best of our knowledge, this is the first study to investigate how emotion influences the ERP correlates of retrieval process underlying false recognition. The key findings of the present study are as follows. First, in line with previous ERP studies of emotional memory (Inaba et al., 2007; Inaba et al., 2005; Johansson et al., 2004; Weymar et al., 2011), we observed greater recollection-related parietal old/new effects for negative pictures than for positive and neutral pictures during true recognition, suggesting that negative emotion was associated with increased recollective distinctiveness during retrieval. Second, the ERP results showed that negative pictures also evoked greater parietal old/ new effects during false recognition, suggesting that negative lures eliciting expected detailed recollections were falsely recognized as old items. Third, we observed slower response times and higher confidence ratings for negative pictures compared with positive and neutral pictures during both true and false recognition. Given that recollection process may underlie slower and higher confidence recognition responses (Yonelinas, 2002), these results converge to suggest that recollection process may contribute to both true and false recognition of negative pictures. Fourth, the negative pictures were associated with a more conservative response bias in comparison with positive and neutral pictures, suggesting that participants may set a higher criterion for responding old to emotionally negative stimuli. Finally, the behavioral results revealed lower raw false recognition rates for negative pictures than for positive and neutral pictures. For the corrected false

3.3.3.3. Between-valence analyses. For true recognition, betweenvalence comparison was conducted with repeated measures ANOVA employing the within-subjects factor of Valence on the old/new difference waveforms (true recognition minus new) at electrode Fz. Although the old/new differences were greater for negative valence (1.39 μV) than for positive or neutral valence (0.76 μV and −0.15 μV, respectively), the main effect of Valence was not significant (F (2, 40) = 2.38, p = .121, η2p = .11). For false recognition, repeated measures ANOVA on the old/new difference waveforms (false recognition minus new) at electrode Pz revealed a significant main 306

Cognition 179 (2018) 298–310

Z. Zheng et al.

old/new recognition during the test phase in the present study may shift participants away from relying on gist memory toward a reliance on verbatim memory (Brainerd, Nakamura, Reyna, & Holliday, 2017). Taken together, one plausible explanation for these inconsistent findings is that negative valence increases false memory when experimental manipulations favor gist processing but decreases false memory when experimental manipulations favor verbatim processing. Intriguingly, a recent study by Farris and Toglia (in press) offers support for this explanation. Specifically, Farris and Toglia failed to observe increased false memory for negative stimuli when their experimental manipulations presumably promoted item-specific verbatim processing. As in the present study, participants were presented with high-arousing emotional and low-arousing neutral pictures in an interspersed order during encoding in Farris and Toglia (in press). Nevertheless, further research is necessary to systematically examine the variables or conditions influencing the production of false recognition. As discussed above, the experimental manipulations of the present study may be more likely to place emphasis on the role of verbatim traces during memory retrieval. Accordingly, we found that negative pictures evoked similar recollection-related parietal old/new effects during false recognition as true recognition in the present study, which is consistent with previous behavioral studies that found more “Remember” responses for the negative lures (Dehon et al., 2010; Pesta et al., 2001). Our results suggest that episodic details of negative pictures are erroneously retrieved during false recognition, and thus that illusory recollections underlie false recognition for negative emotional content. One interpretation regarding illusory recollection is that extremely strong gist traces can support high levels of illusory recollective experience for lures (Brainerd & Reyna, 2002; Brainerd, Wright, Reyna, & Mojardin, 2001). The present findings are inconsistent with this argument, as equivalent familiarity-related old/new effects were evoked for each type of emotional valence, but only the negative pictures evoked recollection-related old/new effects during false recognition in the present study. An alternative theoretical interpretation of illusory recollection is activation-monitoring theory (Roediger, Balota, & Watson, 2001). Following this theory, the representations of lures are activated within the semantic network (Collins & Loftus, 1975), and critically, can become associated with the encoding context, when study items are presented. Thus, the recollection process supports false recognition memory because the encoding context is improperly retrieved during test (Gallo & Roediger, 2002; McDermott & Watson, 2001; Roediger, McDermott, Pisoni, & Gallo, 2004; Roediger, Watson, McDermott, & Gallo, 2001; see Arndt (2012) for a review). In the present study, it is plausible that the negative lures’ representations were more strongly activated during the presentation of negative targets, and these representations were more tightly bound with the encoding context. During test, the encoding context together with the lures’ representations thus underlies the production of false recognition. For positive and neutral pictures, we found significant early frontal but not parietal old/new effects during false recognition, which suggests that false recognition for positive and neutral pictures is driven by familiarity process. These findings are compatible with the FTT proposing that familiarity process contributes to false recognition of lures based on gist traces (Brainerd & Reyna, 1998, 2002, 2005; Brainerd et al., 2003). As a result, the present findings demonstrate divergent effects of emotional valence (negative vs. positive) on retrieval processes underlying false recognition memory. While false recognition for negative pictures was to a greater extent driven by recollection process, false recognition of positive pictures was predominantly based on familiarity process. Interestingly, ERP results for false recognition of positive and neutral pictures also revealed late posterior negativity during the later time widow (i.e., 650–1000 ms), which is considered as an ERP correlate of the continued reconstruction process of study episodes when memory features may not be readily recovered (Mecklinger et al., 2016). Late posterior negativity has usually been observed in ERP studies of false recognition memory (Chen, Voss, & Guo, 2012; Geng

recognition rates, negative pictures were still lower than positive ones, despite the corrected data did not significantly differ between negative and neutral pictures. Overall, the ERP and behavioral results in the present study are largely in agreement with the distinctiveness heuristic account. This account postulates that negative emotional content is less prone to memory distortions (Schacter et al., 2007; Schacter & Wiseman 2006). As previously noted, more direct evidence supporting the argument that emotionally negative content can decrease the likelihood of memory distortions relative to positive and neutral content, has come from studies examining reality monitoring ability (Kensinger et al., 2007; Kensinger & Schacter, 2005a, 2005b, 2006). The present behavioral results extend previous findings using the categorized pictures paradigm to experimentally induce false memories. According to the distinctiveness heuristic account, when participants rely on a distinctiveness heuristic, they focus on the distinctive details associated with the test item itself, and determine whether the retrieved information passes their expected criteria (Gallo, 2010). Negative emotion is argued to be associated with increased recollective expectations, thus participants may be more conservative in making old responses to emotionally negative stimuli and only falsely recognize the negative lures when they elicit expected detailed recollection. Consistent with this speculation, we observed a more conservative response bias for negative pictures compared with positive and neutral pictures. Importantly, the current study estimated the contribution of recollection to false memory by measuring the ERP correlates of the retrieval processes underlying the production of false recognition. Our ERP results revealed greater contribution of recollection process to false recognition of negative pictures compared with positive and neutral pictures, which provides direct evidence for the distinctiveness heuristic account. Our results may appear incompatible with the conceptual relatedness account, which postulates that negative content can increase false memory due to the enhanced retrieval of gist traces (Bookbinder & Brainerd, 2016). In particular, it is noteworthy that Bookbinder and Brainerd (2017) recently found that negatively valenced pictures elevated false memory relative to positive and neutral pictures using a similar paradigm as that in our study. In addition, their mathematical model revealed enhanced gist-based familiarity process during false recognition of negative pictures. In contrast to Bookbinder and Brainerd (2017), the present study did not reveal increased false memory for negative pictures. Critically, our ERP results revealed that the familiarity-related early frontal old/new effects were equivalently evoked by positive, neutral, and negative lures while recollection-related parietal old/new effects were enhanced for negative lures during false recognition. These different patterns of results may be due to several methodological differences between the studies. First, Bookbinder & Brainerd manipulated the valence of pictures while controlling their arousal levels, whereas the emotional pictures were more arousing than neutral ones in the present study. Second, Bookbinder & Brainerd included repetitions of exemplars that had been slightly transformed (e.g., mirror images, hue changes) for each category, whereas the exemplars from each category were similar but unique in our study. Third, there was a longer study list (more than one hundred pictures) preceding the recognition test in Bookbinder and Brainerd (2017) compared with the present study (seventy-two pictures per study list). These manipulations may have enhanced gist processing in the study by Bookbinder and Brainerd (2017), but increased verbatim processing in the present study. Besides, a larger number of members (eight exemplars) per category were presented in a blocked order during encoding in Bookbinder and Brainerd (2017), whereas a relatively smaller number of members for all categories (four exemplars per category) were presented in an interspersed order in the present study. It has been suggested that increasing the number of exemplars per category (Powell, Roberts, Ceci, & Hembrooke, 1999) and presenting the study items in a blocked method (Dewhurst, Bould, Knott, & Thorley, 2009) will create stronger gist traces. Finally, confidence ratings following the 307

Cognition 179 (2018) 298–310

Z. Zheng et al.

distinctiveness heuristic. The present study has important theoretical and methodological implications for understanding emotional false memory.

et al., 2007; Nessler & Mecklinger, 2003; Nessler, Mecklinger, & Penney, 2001), and may reflect participants striving to make additional effort to recover prior episodic details when memory attributes of positive and neutral lures relevant for the old/new discrimination may be not easily retrieved during test. The present findings indicating an absence of recollection related parietal old/new effects for false recognition of positive and neutral lures provide further support for this account. We did not find robust emotional enhancement effects for true memory performance in the present study. The exact reasons for the absence of emotional enhancement effects remain unclear. Comparable memory performance for emotional and neutral stimuli has also been observed in previous studies (Dehon et al., 2010; Johansson et al., 2004; Kissler & Hauswald, 2008; Kensinger & Schacter, 2016; Langeslag & Van Strien, 2008; Weymar, Loew, Schwabe, & Hamm, 2010). Nevertheless, consistent with previous ERP studies (Inaba et al., 2007; Inaba et al., 2005; Johansson et al., 2004; Schaefer et al., 2011; Weymar et al., 2011), the present ERP results revealed greater recollection-related parietal old/new effects for emotional pictures, especially for negative pictures, than for neutral pictures during true recognition, indicating that negative stimuli are more likely to remembered with vivid and detailed memory features. In addition, accurate recognition of negative pictures evoked significant late frontal old/new effects that were related to post-retrieval monitoring and evaluation processes (Cruse & Wilding, 2009; Hayama et al., 2008). On the one hand, the enhanced recollection and late monitoring processes may be attributed to the increased distinctiveness of negative pictures. On the other hand, the present findings are consistent with the evolutionary perspective that negative items may contain more survival-relevant information, and thus negative content will be remembered with more episodic details because it is important for individuals to facilitate avoidance when they re-encounter similar events in the future (Dolan, 2002; Ochsner, 2000). Several limitations should be noted. First, the same number of typical exemplars per category was selected for each type of emotional valence in the present study. Even though pilot rating of the valence and arousal were performed on these pictures, it becomes difficult to ensure that other attributes, for example, visual similarity between items within each category, were well matched across the three types of emotional valence. As a result, it is difficult to completely rule out the effects of other confounding factors on the present behavioral and ERP results. Second, we included a small sample of participants (N = 10) who performed the pilot rating of the valence and arousal of the pictures, which may result in a less reliable rating data. Fortunately, the pilot ratings and ratings during encoding revealed consistent pattern for the valence results. Nevertheless, it is necessary to include a large sample of participants to rate the valence and arousal of the pictures in future studies. Third, due to the high baseline false alarms, the corrected false alarm rates to neutral pictures did not significantly differ from negative ones, which to some extent weaken the evidence for the distinctiveness heuristic account. However, the significant differences in corrected false alarm rates between negative and positive pictures, and the ERP results still provide strong support for this account.

Conflict of interest The authors declare no conflict of interest. Acknowledgments This work was supported by the National Natural Science Foundation of China (31271108, 31470998, 31711530157, and 31700974), Beijing Municipal Science and Technology Commission ( Z171100000117006, Z171100008217006), the Pioneer Initiative of the Chinese Academy of Sciences, Feature Institutes Program (TSS-201506), National Key Research and Development Program of China (2016YFC1305900), and the CAS Key Laboratory of Mental Health, Institute of Psychology (KLMH2014ZK02, KLMH2014ZG06). We would like to thank anonymous reviewers for their thoughtful comments regarding this manuscript. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.cognition.2018.06.013. References Arndt, J. (2012). False recollection: Empirical findings and their theoretical implications. In B. H. Ross (Eds.), Psychology of learning and motivation (pp. 81-124). doi:10. 1016/B978-0-12-394393-4.00001-7. Bookbinder, S. H., & Brainerd, C. J. (2016). Emotion and false memory: The contextcontent paradox. Psychological Bulletin, 142(12), 1315–1351. http://dx.doi.org/10. 1037/bul0000077. Bookbinder, S. H., & Brainerd, C. J. (2017). Emotionally negative pictures enhance gist memory. Emotion, 17(1), 102–119. http://dx.doi.org/10.1037/emo0000171. Bradley, M. M., & Lang, P. J. (1994). Measuring emotion: The self-assessment manikin and the semantic differential. Journal of behavior therapy and experimental psychiatry, 25(1), 49–59. http://dx.doi.org/10.1016/0005-7916(94)90063-9. Brainerd, C. J., Nakamura, K., Reyna, V. F., & Holliday, R. E. (2017). Overdistribution illusions: Categorical judgments produce them, confidence ratings reduce them. Journal of Experimental Psychology: General, 146(1), 20–40. http://dx.doi.org/10. 1037/xge0000242. Brainerd, C. J., & Reyna, V. F. (1998). When things that were never experienced are easier to “remember” than things that were. Psychological Science, 9(6), 484–489. http://dx. doi.org/10.1111/1467-9280.00089. Brainerd, C. J., & Reyna, V. F. (2002). Fuzzy-trace theory and false memory. Current Directions in Psychological Science, 11(5), 164–169. http://dx.doi.org/10.1111/14678721.00192. Brainerd, C. J., & Reyna, V. F. (2005). The science of false memory. Oxford University Press doi:10.1093/acprof:oso/9780195154054.001.0001. Brainerd, C. J., Reyna, V. F., & Mojardin, A. H. (1999). Conjoint recognition. Psychological Review, 106(1), 160–179. http://dx.doi.org/10.1037//0033-295x.106.1.160. Brainerd, C. J., Reyna, V. F., Wright, R., & Mojardin, A. H. (2003). Recollection rejection: False-memory editing in children and adults. Psychological Review, 110(4), 762–784. http://dx.doi.org/10.1037/0033-295x.110.4.762. Brainerd, C. J., Stein, L. M., Silveira, R. A., Rohenkohl, G., & Reyna, V. F. (2008). How does negative emotion cause false memories? Psychological Science, 19(9), 919–925. http://dx.doi.org/10.1111/j.1467-9280.2008.02177.x. Brainerd, C. J., Wright, R., Reyna, V. F., & Mojardin, A. H. (2001). Conjoint recognition and phantom recollection. Journal of Experimental Psychology-Learning Memory and Cognition, 27(2), 307–327. http://dx.doi.org/10.1037//0278-7393.27.2.307. Buchanan, T. W. (2007). Retrieval of emotional memories. Psychological Bulletin, 133(5), 761–779. http://dx.doi.org/10.1037/0033-2909.133.5.761. Chen, H., Voss, J. L., & Guo, C. (2012). Event-related brain potentials that distinguish false memory for events that occurred only seconds in the past. Behavioral and Brain Functions, 8. http://dx.doi.org/10.1186/1744-9081-8-36. Choi, H. Y., Kensinger, E. A., & Rajaram, S. (2013). Emotional content enhances true but not false memory for categorized stimuli. Memory & Cognition, 41(3), 403–415. http://dx.doi.org/10.3758/s13421-012-0269-2. Collins, A. M., & Loftus, E. F. (1975). Spreading activation theory of semantic processing. Psychological Review, 82(6), 407–428. http://dx.doi.org/10.1037/0033-295x.82.6. 407. Cruse, D., & Wilding, E. L. (2009). Prefrontal cortex contributions to episodic retrieval monitoring and evaluation. Neuropsychologia, 47(13), 2779–2789. http://dx.doi.org/ 10.1016/j.neuropsychologia.2009.06.003. Curran, T. (2000). Brain potentials of recollection and familiarity. Memory & Cognition,

5. Conclusions In summary, using a categorized picture paradigm with experimental manipulations that may favor verbatim processing, we observed enhanced recollection process for negative pictures relative to positive and neutral pictures, but comparable familiarity process across each type of emotional valence category during both true and false recognition. Combined with the behavioral results of lower false recognition rates to negative pictures (especially compared with positive pictures), the present findings suggest that negative emotional stimuli may be less susceptible to memory distortions by engaging a 308

Cognition 179 (2018) 298–310

Z. Zheng et al.

Mecklinger, A., Rosburg, T., & Johansson, M. (2016). Reconstructing the past: The late posterior negativity (LPN) in episodic memory studies. Neuroscience and Biobehavioral Reviews, 68, 621–638. http://dx.doi.org/10.1016/j.neubiorev.2016.06.024. Nessler, D., & Mecklinger, A. (2003). ERP correlates of true and false recognition after different retention delays: Stimulus- and response-related processes. Psychophysiology, 40(1), 146–159. http://dx.doi.org/10.1111/1469-8986.00015. Nessler, D., Mecklinger, A., & Penney, T. B. (2001). Event related brain potentials and illusory memories: The effects of differential encoding. Cognitive Brain Research, 10(3), 283–301. http://dx.doi.org/10.1016/s0926-6410(00)00049-5. Ochsner, K. N. (2000). Are affective events richly recollected or simply familiar? The experience and process of recognizing feelings past. Journal of Experimental Psychology-General, 129(2), 242–261. http://dx.doi.org/10.1037//0096-3445.129.2. 242. Paller, K. A., Voss, J. L., & Boehm, S. G. (2007). Validating neural correlates of familiarity. Trends in Cognitive Sciences, 11(6), 243–250. http://dx.doi.org/10.1016/j.tics.2007. 04.002. Pesta, B. J., Murphy, M. D., & Sanders, R. E. (2001). Are emotionally charged lures immune to false memory? Journal of Experimental Psychology-Learning Memory and Cognition, 27(2), 328–338. http://dx.doi.org/10.1037//0278-7393.27.2.328. Powell, M. B., Roberts, K. P., Ceci, S. J., & Hembrooke, H. (1999). The effects of repeated experience on children's suggestibility. Developmental Psychology, 35(6), 1462–1477. http://dx.doi.org/10.1037/0012-1649.35.6.1462. Rhodes, S. M., & Donaldson, D. I. (2007). Electrophysiological evidence for the influence of unitization on the processes engaged during episodic retrieval: Enhancing familiarity based remembering. Neuropsychologia, 45(2), 412–424. http://dx.doi.org/10. 1016/j.neuropsychologia.2006.06.022. Roediger, H. L. III, Balota, D. A., & Watson, J. M. (2001). Spreading activation and arousal of false memories. In H. L. Roediger III, J. S. Nairne, I. Neath, & A. M. Surprenant (Eds.), Science conference series. The nature of remembering: Essays in honor of Robert G. Crowder (pp. 95–115), doi:10.1037/10394-006. Roediger, H. L., III, & McDermott, K. B. (1995). Creating false memories - Remembering words not presented in lists. Journal of Experimental Psychology-Learning Memory and Cognition, 21(4), 803–814. http://dx.doi.org/10.1037/0278-7393.21.4.803. Roediger, H. L., III, McDermott, K. B., Pisoni, D. B., & Gallo, D. A. (2004). Illusory recollection of voices. Memory, 12(5), 586–602. http://dx.doi.org/10.1080/ 09658210344000125. Roediger, H. L., III, Watson, J. M., McDermott, K. B., & Gallo, D. A. (2001). Factors that determine false recall: A multiple regression analysis. Psychonomic Bulletin & Review, 8(3), 385–405. http://dx.doi.org/10.3758/bf03196177. Rugg, M. D., & Curran, T. (2007). Event-related potentials and recognition memory. Trends in Cognitive Sciences, 11(6), 251–257. http://dx.doi.org/10.1016/j.tics.2007. 04.004. Schacter, D. L. (1999). The seven sins of memory - Insights from psychology and cognitive neuroscience. American Psychologist, 54(3), 182–203. http://dx.doi.org/10.1037/ 0003-066x.54.3.182. Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering the past and imagining the future. Philosophical Transactions of the Royal Society B-Biological Sciences, 362(1481), 773–786. http://dx.doi.org/10. 1098/rstb.2007.2087. Schacter, D. L., Gallo, D. A., & Kensinger, E. A. (2007). The cognitive neuroscience of implicit and false memories: Perspectives on processing specificity. In J. S. Nairne (Ed.). The foundations of remembering: Essays in honor of Henry L. Roediger, III (pp. 353–377). New York: Psychology Press. Schacter, D. L., Israel, L., & Racine, C. (1999). Suppressing false recognition in younger and older adults: The distinctiveness heuristic. Journal of Memory and Language, 40(1), 1–24. http://dx.doi.org/10.1006/jmla.1998.2611. Schacter, D. L., Norman, K. A., & Koutstaal, W. (1998). The cognitive neuroscience of constructive memory. Annual Review of Psychology, 49, 289–318. http://dx.doi.org/ 10.1146/annurev.psych.49.1.289. Schacter, D. L., & Slotnick, S. D. (2004). The cognitive neuroscience of memory distortion. Neuron, 44(1), 149–160. http://dx.doi.org/10.1016/j.neuron.2004.08.017. Schacter, D. L., & Wiseman, A. L. (2006). Reducing memory errors: The distinctiveness heuristic. In R. R. Hunt, & J. Worthen (Eds.). Distinctiveness and memory (pp. 89–107). New York: Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/ 9780195169669.003.0005. Schaefer, A., Pottage, C. L., & Rickart, A. J. (2011). Electrophysiological correlates of remembering emotional pictures. Neuroimage, 54(1), 714–724. http://dx.doi.org/10. 1016/j.neuroimage.2010.07.030. Seamon, J. G., Luo, C. R., Schlegel, S. E., Greene, S. E., & Goldenberg, A. B. (2000). False memory for categorized pictures and words: The category associates procedure for studying memory errors in children and adults. Journal of Memory and Language, 42(1), 120–146. http://dx.doi.org/10.1006/jmla.1999.2676. Semlitsch, H. V., Anderer, P., Schuster, P., & Presslich, O. (1986). A solution for reliable and valid reduction of ocular artifacts, applied to the P300 ERP. Psychophysiology, 23(6), 695–703. http://dx.doi.org/10.1111/j.1469-8986.1986.tb00696.x. Sharot, T., Delgado, M. R., & Phelps, E. A. (2004). How emotion enhances the feeling of remembering. Nature Neuroscience, 7(12), 1376–1380. http://dx.doi.org/10.1038/ nn1353. Talmi, D. (2013). Enhanced emotional memory: Cognitive and neural mechanisms. Current Directions in Psychological Science, 23(1), http://dx.doi.org/10.1177/ 0963721413519415 83-83. Talmi, D., & Moscovitch, M. (2004). Can semantic relatedness explain the enhancement of memory for emotional words? Memory & Cognition, 32(5), 742–751. http://dx.doi. org/10.3758/BF03195864. Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26(1), 1–12. Voss, J. L., & Federmeier, K. D. (2011). FN400 potentials are functionally identical to

28(6), 923–938. http://dx.doi.org/10.3758/bf03209340. Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal of Experimental Psychology, 58(1), 17–22. http://dx.doi.org/ 10.1037/h0046671. Dehon, H., Laroi, F., & Van der Linden, M. (2010). Affective valence influences participant's susceptibility to false memories and illusory recollection. Emotion, 10(5), 627–639. http://dx.doi.org/10.1037/a0019595. Dewhurst, S. A., Bould, E., Knott, L. M., & Thorley, C. (2009). The roles of encoding and retrieval processes in associative and categorical memory illusions. Journal of Memory and Language, 60(1), 154–164. http://dx.doi.org/10.1016/j.jml.2008.09.002. Dodson, C. S., & Schacter, D. L. (2002). When false recognition meets metacognition: The distinctiveness heuristic. Journal of Memory and Language, 46(4), 782–803. http://dx. doi.org/10.1006/jmla.2001.2822. Dolan, R. J. (2002). Emotion, cognition, and behavior. Science, 298(5596), 1191–1194. http://dx.doi.org/10.1126/science.1076358. Eppinger, B., Herbert, M., & Kray, J. (2010). We remember the good things: Age differences in learning and memory. Neurobiology of Learning and Memory, 93(4), 515–521. http://dx.doi.org/10.1016/j.nlm.2010.01.009. Farris, E. A., & Toglia, M. P. (in press). Conjoint recognition procedures reveal verbatim processing enhances memory for emotionally-valenced pictorial stimuli. Emotion. Gallo, D. A. (2010). False memories and fantastic beliefs: 15 years of the DRM illusion. Memory & Cognition, 38(7), 833–848. http://dx.doi.org/10.3758/mc.38.7.833. Gallo, D. A., & Roediger, H. L. (2002). Variability among word lists in eliciting memory illusions: Evidence for associative activation and monitoring. Journal of Memory and Language, 47(3), 469–497. http://dx.doi.org/10.1016/s0749-596x(02)00013-x. Geng, H., Qi, Y., Li, Y., Fan, S., Wu, Y., & Zhu, Y. (2007). Neurophysiological correlates of memory illusion in both encoding and retrieval phases. Brain Research, 1136(1), 154–168. http://dx.doi.org/10.1016/j.brainres.2006.12.027. Hayama, H. R., Johnson, J. D., & Rugg, M. D. (2008). The relationship between the right frontal old/new ERP effect and post-retrieval monitoring: Specific or non-specific? Neuropsychologia, 46(5), 1211–1223. http://dx.doi.org/10.1016/j.neuropsychologia. 2007.11.021. Inaba, M., Kamishima, K., & Ohira, H. (2007). An electrophysiological comparison of recollection for emotional words using an exclusions recognition paradigm. Brain Research, 1133(1), 100–109. http://dx.doi.org/10.1016/j.brainres.2006.07.010. Inaba, M., Nomura, M., & Ohira, H. (2005). Neural evidence of effects of emotional valence on word recognition. International Journal of Psychophysiology, 57(3), 165–173. http://dx.doi.org/10.1016/j.ijpsycho.2005.01.002. Jäger, T., Mecklinger, A., & Kipp, K. H. (2006). Intra- and inter-item associations doubly dissociate the electrophysiological correlates of familiarity and recollection. Neuron, 52(3), 535–545. http://dx.doi.org/10.1016/j.neuron.2006.09.013. Johansson, M., Mecklinger, A., & Treese, A. C. (2004). Recognition memory for emotional and neutral faces: An event-related potential study. Journal of Cognitive Neuroscience, 16(10), 1840–1853. http://dx.doi.org/10.1162/0898929042947883. Kensinger, E. A., & Corkin, S. (2003). Memory enhancement for emotional words: Are emotional words more vividly remembered than neutral words? Memory & Cognition, 31(8), 1169–1180. http://dx.doi.org/10.3758/bf03195800. Kensinger, E. A., & Corkin, S. (2004). The effects of emotional content and aging on false memories. Cognitive Affective & Behavioral Neuroscience, 4(1), 1–9. http://dx.doi.org/ 10.3758/cabn.4.1.1. Kensinger, E. A., O'Brien, J. L., Swanberg, K., Garoff-Eaton, R. J., & Schacter, D. L. (2007). The effects of emotional content on reality-monitoring performance in young and older adults. Psychology and Aging, 22(4), 752–764. http://dx.doi.org/10.1037/08827974.22.4.752. Kensinger, E. A., & Schacter, D. L. (2005a). Emotional content and reality-monitoring ability: fMRI evidence for the influences of encoding processes. Neuropsychologia, 43(10), 1429–1443. http://dx.doi.org/10.1016/j.neuropsychologia.2005.01.004. Kensinger, E. A., & Schacter, D. L. (2005b). Retrieving accurate and distorted memories: Neuroimaging evidence for effects of emotion. Neuroimage, 27(1), 167–177. http:// dx.doi.org/10.1016/j.neuroimage.2005.03.038. Kensinger, E. A., & Schacter, D. L. (2006). Reality monitoring and memory distortion: Effects of negative, arousing content. Memory & Cognition, 34(2), 251–260. http://dx. doi.org/10.3758/bf03193403. Kensinger, E. A., & Schacter, D. L. (2016). Memory and emotion. In L. Feldman Barrett, M. Lewis, & J. A. Haviland-Jones (Eds.). Handbook of emotions (pp. 564–578). New York: The Guildford Press. Kissler, J., & Hauswald, A. (2008). Neuromagnetic activity during recognition of emotional pictures. Brain Topography, 20(4), 192–204. http://dx.doi.org/10.1007/ s10548-008-0044-7. Koutstaal, W., & Schacter, D. L. (1997). Gist-based false recognition of pictures in older and younger adults. Journal of Memory and Language, 37(4), 555–583. http://dx.doi. org/10.1006/jmla.1997.2529. Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (2008). International affective picture system (IAPS): Affective ratings of pictures and instruction manual. Technical Report A-8. University of Florida, Gainesville, FL. Langeslag, S. J. E., & Van Strien, J. W. (2008). Age differences in the emotional modulation of ERP old/new effects. International Journal of Psychophysiology, 70(2), 105–114. http://dx.doi.org/10.1016/j.ijpsycho.2008.07.022. LeDoux, J. E. (2002). Cognitive-emotional interactions: Listen to the brain. In R. D. Lane, & L. Nadel (Eds.). Cognitive neuroscience of emotion (pp. 129–155). New York, NY: Oxford University Press. Macmillan, N. A., & Creelman, C. D. (2004). Detection theory: A user’s guide (2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates Inc. McDermott, K. B., & Watson, J. M. (2001). The rise and fall of false recall: The impact of presentation duration. Journal of Memory and Language, 45(1), 160–176. http://dx. doi.org/10.1006/jmla.2000.2771.

309

Cognition 179 (2018) 298–310

Z. Zheng et al.

Weymar, M., Loew, A., Schwabe, L., & Hamm, A. O. (2010). Brain dynamics associated with recollective experiences of emotional events. Neuroreport, 21(12), 827–831. http://dx.doi.org/10.1097/WNR.0b013e32833d180a. Yonelinas, A. P. (2002). The nature of recollection and familiarity: A review of 30 years of research. Journal of Memory and Language, 46(3), 441–517. http://dx.doi.org/10. 1006/jmla.2002.2864.

N400 potentials and reflect semantic processing during recognition testing. Psychophysiology, 48(4), 532–546. http://dx.doi.org/10.1111/j.1469-8986.2010. 01085.x. Weymar, M., Loew, A., & Hamm, A. O. (2011). Emotional memories are resilient to time: Evidence from the parietal ERP old/new effect. Human Brain Mapping, 32(4), 632–640. http://dx.doi.org/10.1002/hbm.21051.

310