The fate of emotional memories
Multiple trace theory (MTT) proposes that emotional memories are revised during the reconsolidation process. Additionally, the effects of stress on memory reconsolidation are still controversial, and the effects of emotion on memory reconsolidation is still unclear. Moreover, it has been suggested that men and women differ in neural expression and behavior during enhancement of memory consolidation during emotional events. Here, we examined in healthy participants the impact of arousal during reconsolidation of neutral memories, and the possible gender differences, related to the alteration of an emotional expression during reconsolidation of a memory trough an amygdala-dependent learning task (an object-context-episode task). The results showed that participants did not accurately recognized more objects that were reconsolidated and related to a shock, than reactivated without a shock and non reactivated. However, we found an effect of reliving on recognition accuracy. No differences were found between genders. Taken together, the current findings suggest that further research should be done in order to evaluate the effect of arousal during reconsolidation of neutral memories.
INTROCUCTION
At reunions, you might meet a bunch of new people. However, which faces are you going to remember? Perhaps, you will remember the woman who made you feel embarrassed or the guy that made you laugh. But, does this means that emotions help us to remember? In humans,
emotionally arousing events are usually better remembered than neutral ones. This hypothesis has been called as “emotional enhancement of memory” (EEM) (Okruszek, et al., 2017).
The EEM hypothesis has been demonstrated in a large number of laboratory studies with different techniques, using stimuli varying from words to pictures, to narrated slide shows (Bradley et al, 1992;; Hamann, 2001;; Christianson, 1992;; Finn & Roediger, 2011), as well as autobiographical memory studies (Conway et al., 1994). The explanations of this emotional arousal are still unclear. However, some psychologist suggests that the emotional arousal enhances novelty, focus attention and can be rehearsed, being thus easier to remember. Nevertheless, no experimental evidence has been found to support this idea (Christianson, 1992). Other authors suggest that Pavlovian conditioning can be used in the study of processes mediating emotional responses, particularly, fear conditioning (Davis, 1992). In addition, Dunsmoor et al, suggested that initially weak memories can be strengthened if this information later gains relevance, as result of an adaptation of memories.
However, memories are not just enhanced due to the attachment of an emotion, moreover, memories can become more emotional over time. For example, post- traumatic stress disorder patients remembering the painful memories in an unsafe space without a therapist may see
the trauma situation more dangerous than it used to be (Foa et al., 2008).
In general terms, the most conspicuous model of memory formation of the last century assumes that information is perceived by the brain –in order to be stored temporarily¬– forming the “short- term memory” processing. This type of information will be stored between minutes and hours. Then, the memory can make a progressive transition from short- term memory (STM) to long-term memory (LTM) as a result of (synaptic) memory consolidation (Nader and Hardt, 2009;; Cowan, 2008).
During consolidation, the cascade of processes that follows initial registration of information and the memory trace is stabilized, becoming increasingly resistant to interference by newer information or other disturbances (McGaugh, 2000), turning from a labile memory into a stable memory. The model of consolidation assumes that memory and its mechanisms will gradually and irreversible decline over time (Nader and Hardt, 2009). If so, this model cannot explain such observations related to the plasticity of memories as the ones mentioned before. In order to understand how do memories can become more emotional with time, neuroscience has to recognize memory as a dynamic process.
In summary, we could conclude that something about the relation of arousing emotion and processing of emotions might contributes to therapeutic uses. However, the specifications of what are the things about emotions that actually brings changes are still unclear. That is the reason why a relevant part in the study of
memory -as a dynamic of arousing process- will be the understanding on the origin of the plasticity of neutral memories. This will help us to prevent situations that make memories more emotional. For example, during PSTD or anxiety treatment, the reactivation of some memories could lead to a retrogression in therapy, instead of an improvement due to the lack in the etiology of the traumatic memory.
Memory as a dynamic process.
Pioneer studies during the late 1960’s showed that consolidated memories could turn into labile memories again if it is exposed to a reminder cue (Misaning et at.,1986). The work by Misaning et al. opened a new track in research and encouraged further work to study the possibility of reactivation of memories after “retrieval”, a process caused by the re-exposure to salient training-related information whereby a memory is passed from an active state into an active one (Gisquet-Verrier & Riccio, 2012).
Later, at the beginning of the XXI century, Nader and co, studied the role of the amygdala in the synaptic consolidation of the Pavlovian association of a shock with a tone. Showing that the tone presented long after consolidation was able to change the susceptibility of a memory (Nader et al., 2000).
At the same time, Nadel and Moschovitch developed an alternative theory of memory consolidation, known as the multiple trace theory (MTT). This theory suggests the memory can turn again into a fragile and labile state when the memory is retrieved (Nadel et al.,2000),
These studies gave place to a new hypothesis which explains memory as a dynamic process, adding a process now known as “reconsolidation”, to the old and static model (Nadel et al., 2000;; Sara, 2000).
Broadly speaking, reconsolidation is the process by which memories –previously consolidated– are recalled and actively consolidated. This process might help to maintain, enhance and change long-term memories by the reactivation of them after retrieval (Nadel et al., 2000;; Sara, 2000). This reconsolidation process provides an additional chance of amending or even disrupt access to the memory under appropriate circumstances (Lane et al., 2015)
Several cellular and molecular processes, as well as several anatomical regions, were found to be exclusively enrolled during reconsolidation (Lee et al., 2004;; Tronel & Sara, 2002;; Kelly et al., 2003;; Bahar et al., 2003;; Gutiérrez et al., 2003). This allowed opening a new level of the understanding of the potential role of reconsolidation during memory updating.
As we mentioned before, the consolidation process cannot explain how does memories could be updated with emotionality. However, the observations on the reconsolidation hypothesis, completely change the theories about fate memories, into a more plastic and malleable idea of memories that we expected. So, in these terms, it looks like our study of interest could be mediated by reconsolidation?
How does emotional arousal can affect consolidation and reconsolidation
In consolidation
The enhancement of memories attached to an emotional event has been attributed to the activation of the amygdala during arousal by noradrenaline. In general terms, the amygdala is the structure in humans that encodes emotional information (Stegener et al., 2005), and the noradrenaline mediates the amygdala by the modulation of diverse hormones and neurotransmitters, as the adrenal stress hormones, used during consolidation (Schwarze et al., 2012)
Specifically, it has been shown that the molecular basis of enhancement due to emotional arousal on long-term memory is related to the ß-Adrenergic receptor, when this is blocked by its antagonists, the memory enhancing effects are blocked as well (Cahill et al., 1994;; Nielson & Jensen, 1994;; van Stegeren et al., 1998). At the same time, in human subjects with selective lesions of the amygdala, emotional arousal also does not enhance long-term memory of the arousing material (McGaugh, 2000), pointing out that enhancement of memory due to emotionally arousing events may then critically be based on the specific interactions between the amygdala and the hippocampus (Anderson et al., 2006).
Parallel, fMRI studies in humans have shown that the influence of emotions on memory is regulated by multiple brain systems, in which their activation depends on the stage of information processing. During the study of the relationship between declarative memories and
emotions, frontotemporal brain regions act together to promote the retention of emotionally arousing events (LaBar & Cabeza, 2006). Additionally, the enhancement of memory due to emotional arousal involves interactions between subcortical and cortical structures and the activation of central and peripheral neurohormonal systems that are modulated by the amygdala (LaBar & Cabeza, 2006).
There is still some controversy related on what does EEM hypothesis is base on. For example, Taimi et al., have found that emotional stimuli not only promote arousal but also, intensifies cognitive processes that contribute to the enhanced memory (Taimi et al., 2007). In this case, the EEM hypothesis will be based on cognitive characteristics of emotional stimuli rather than on arousal. In contrast, Schwarze et al., studied the effects of arousal on memory formation independent of cognitive processes, finding only enhancement of neutral memories by arousal after consolidation in item familiarity but not in recollection, proving that memory do not differ with respect of cognitive factors as distinctiveness or semantic relatedness (Schwarze et al., 2012). The dualities of this controversy might rely on when does the EEM occurs during the memory information process.
In reconsolidation
As we mentioned before, reconsolidation occurs after retrieval. According to the MTT hypothesis, every time that an episodic memory is retrieved, there will be a more detailed memory trace or an expanded representation of the details of the old memory that will make more
accessible to be effectively retrieved in the future. Specifically, every time that a memory is retrieved and re-encoded, the updated trace will incorporate information, including emotions (Lane et al 2015).
Furthermore, studies on the dynamic interplay between retrieval and reconsolidation of memories have shown that the reactivation of a memory can be given just using a reminder of the spatial context of the original event to control fearful responses, and when the consolidation is blocked with ansyomycin followed by the reminder, showed no conditional fear (Nadel et al., 2012). These results suggest that after every retrieval of a fear memory, a reconsolidation process will occur and that the disruption of the reconsolidation process can occur due to the elimination of the previous fear response (Nader et al., 2012).
Moreover, MTT provides a way to understand how afflictive emotional memories can be altered or enhanced through the corrective experience (Lane et al., 2015). In this case, the emotional reaction integrates the memory, as the spatial and temporal contexts do, creating an autobiographical memory. So, the more highly arousing the emotional reaction, the more likely the evoking situation will be remembered later on (McGaugh 2003).
Likewise, it has been shown that when the memory is recalled, the emotional response incorporated into this memory will be re-engaged, reactivating the sympathetic response of the amygdala. Thus, after a traumatic incident, the memory will be strengthened, and the
emotional response attached to it will intensify after reconsolidation (Lane et al., 2015).
At the same time, MTT proposes that emotional memories are revised during the reconsolidation process. With this, during psychotherapy, patients can re- evaluate the original experience and even turn it into a more positive one (Lane et ., 2015).
Taking into account the reconsolidation process, researches started to investigate a way to modify traumatic memories previously consolidated, using drugs, such as propranolol, to block the emotional response during reconsolidation. In rats, propranolol –a beta-adrenergic antagonist– has found to have an indirect effect on protein synthesis in the amygdala, blocking the reconsolidation process (Debiec & Ledoux, 2004). At the same time, some studies in humans demonstrated the effect of propranolol in reconsolidation, specifically, blocking or decreasing the emotional response attached to ferly memories (Schwabe et al., 2012;; Pitman et al., 2002;; Brunet et al., 2008). Although the use of propanol as a treatment in psychotherapy offers a promising effect, there is still a debate regarding the ethical and legal issues that this represents (Tenenbaum & Reese 2007).
In conclusion, the MTT hypothesis suggests that during reconsolidation the memories are not created again, instead, the memories are transformed in fundamental ways, including the emotional responses associated to it (Lane et al., 2015).
Taking these criteria into consideration, arousal during reconsolidation might influence the strength of neutral memories, due to the beta-adrenergic receptors activation in the amygdala that this causes?
Effects of manipulation on arousal,
emotion, or stress during
reconsolidation
To solve the question previously exposed, some authors have work in the relation between emotional learning and strengthening of memories. For example, Finn and Roediger provided novel evidence of memory improvement by manipulation of the reconsolidation process in humans. In which, the relationship between the amygdala and the hippocampus may play an essential role. Specifically, they demonstrate that retrieval is essential for emotional enhancement of memories during reconsolidation, showing that emotion influences the accuracy of memory. However, the study occurred in the same day and giving participants a short period of time to study the words. In this case, we can question, if the information was even consolidated, because for episodic details to persist in long-term memory, the memories have to stabilize by long-term potentiation process, a process that involves a prolonged period after learning. (McKenzie & Eichenbaum, 2012;; Dunsmor et al., 2015)
When relates to the relationship between stress and reconsolidation of memories, various authors have found a positive relation. For example, Cocccoz et al., found that a naturalistic mild stressor can improve reconsolidation by the
enhancement of the long-term expression of the declarative memory. Similarly, Bos et al (2014b) found that exposure to stress after reactivation of memories leads to the improvement of recalling, implying that mild stress can enhance reconsolidation, which in turn, strength declarative memory. When relates to the “stress hormone”, i.e. cortisol, studies in humans suggest that cortisol response mediates the effect of the post-reactivation stress manipulation on contextualization of emotional memories (Bos et al., 2014a). And, exposure to stress after reactivation of memories leads to the improvement of recalling, implying that mild stress can enhance reconsolidation, which in turn, strength declarative memory (Bos et al., 2014b).
In contrast, other studies showed that the effects of stress on memory in humans are negative, impairing the reconsolidation process (Schawabe & Wolf 2010), and having a negative impact in reactivated components of declarative memory (Hupbach & Dotskind, 2014).
As we saw, the effects of stress on memory reconsolidation is still controversial, and the effects of emotion on memory reconsolidation is still unclear. So, our refined knowledge until know about the conditions under which reconsolidation of memories may occur still unclear. We thus will examine the impact of arousal during reconsolidation of neutral memories. Based on the MTT theory, we expect that arousal during the reconsolidation window will render an initial neutral memory into a more emotional one.
Gender differences
In humans, during consolidation women tend to produce memories faster, or a more intense response to cues than men during reconsolidation. At the same time, women report more vivid memories and recall more emotional autobiographical events in a timed test (Herz and Cupchik, 1992;; Canli et al., 2002). Canli and co- workers mentioned some explanations of this gender differences, such as women experience life more intensively, resulting in a better memory;; or gender differences relate to encoding, rehearsing and thinking of emotional experiences, that can guide into a different response of memory.
Furthermore, it has been shown that with MRI studies, the activation pattern of neural networks during emotional experience and memory encoding differs between men and women (Canli et al., 2012).
Moreover, it has been suggested that there is an estrogen-mediated modification of emotional enhancement via the amygdala. Pruis et al., showed that higher levels of estrogen in older women resulted in higher arousal for negative stories and images, without affecting memory. They propose that this hormone could affect the response of emotion by the amygdala and prefrontal cortex, and the reason of why they didn’t find any differences in memory it could be due to the age-related changes in the hippocampus of older women.
In contrast, Anderson et al. did not find significant differences between men and women in subjective arousal reactions to
emotionally arousing events. However, they explained that this might be due to the lack of power of their test.
In addition, according to LaBar and Cabeza (2006), the hemispheric distribution of encoding-related amygdala activity differs from men to women. Specifically, the distribution in women shows a left-lateralized effect meanwhile in men there is a right-lateralized effect. This lateralization pattern is more conspicuous when relates to the activation of the amygdala by memory, and is found less often as an effect of emotion on perceptual processing. The explanations for this emotional memory gender differences remain unclear. However, they constitute an active area of current research.
In conclusion, men and women differ in neural expression and behavior during enhancement of memory consolidation during emotional events. If the differences are present during the encoding of emotional experiences in different parts of the brain, then this differences could be also present during post-encoding events and retrieval. Suggesting that the gender differences already observed during consolidation can be also present during reconsolidation.
In the present work, we will also analyze if there are gender differences related to the alteration of an emotional expression during reconsolidation of a memory. Expecting that reactivated memories added to arousal will show a stronger physiological reaction as compared to a memory that was reactivated but without arousal, and this difference will be bigger for women than for men.
In summary, to investigate the effect of arousal during reconsolidation of neutral memories and their gender differences, we used an amygdala-dependent learning task (an object-context-episode task form Pavlovian fear conditioning). The encoding session occurred in three phases, during three consecutive days. In phase 1, also called learning phase, subjects were introduced to a set of visual stimuli –each consisting on a context image followed by the appearance of a central image within it– of 120 trials. Shock electrodes were not attached during phase 1 and there was no explicit motivation or instruction to remember any of the pictures. However, participants were asked to try to remember the background image in combination with the object as vivid as possible and rate it. Elapsed 24 hours, in phase 2, electric shock electrodes were attached to the wrist of the hand opposite to the dominant one, and just from the one hundred and twenty background images were presented, eighty of them were paired with a shock (reactivated+shock, RS+), while
the fourty images left
(reactivated+noshock, RS-) were unpaired. The participants were asked – while looking at the background image– to remember the whole mental image of the previous day and to rate the extend in which the background image gives them a feeling of complete reliving. After conditioning, the following day, in phase 3, subject classified 240 central images as novel or old, based on the images presented during day 1. Skin conductance and heart rate responses were acquired during the 3 consecutive days.
METHODS
1.1 Participants.
A total of 44 healthy participants were recruited via online advertisements to participate under the present study (Age=23.18 ± 4.90 years (mean ± s.d.), 12 males). Six subjects were removed from the analysis for equipment failures with stimulus presentation software. The final sample included 38 subjects (Age=22.58 ± 3.77 years (mean ± s.d.), 12 males). Sample size was based on prior studies. No statistical method was used to predetermine sample size. Prior to the study, candidate participants receive information about the study conducted on three consecutive days and a screening form via email. Candidates were eligible for inclusion if they meet the following criteria: (i) age 18-35 years;; (ii) Dutch speakers, (iii) first time as participants in an experiment ruled by Vanessa van Ast. Participants were excluded from the trial if they met the following criteria: (i) any neurological or psychiatric illnesses;; (ii) be under any psychiatric disorder treatment;; (iii) any cardiovascular problems;; (iv) pregnancy;; (v) drug or alcohol abuse. Information about these criteria was obtained by questionnaires (explained below). The participants received partial course credit or financial compensation (€40) in return for their participation. The study was approved by the Ethics Department of the Psychology Department of the University of Amsterdam.
1.2 Stimuli
The images used in this study were taken from Internet. The images were shown in
random order to each subject with the exception of the first ten images. Images were presented on a standard computer screen located approximately 2 feet in front of a subject.
One hundred and twenty pictures of different backgrounds served as neutral background scenes indoor, while 240 different images of objects served as neutral object pictures.
1.3 Skin conductance response
Electrodermal activity (EDA) was measured by two curved Ag/ACl electrodes of 20 by 16 mm that were attached with Micropore Surgical Tape ½ inch to the medial phalanges of the first and third fingers of the hand opposite to the dominant one. The amplifier, built by the University of Amsteredam, applied a sineshaped excitation voltage (1 V peak- peak) of 50 Hz derived from the mains frequency to the electrodes in order to detect changes in the electrodermal activity. The signal from the input device was led through a signal-conditioning amplifier. The analog output was digitized at 1000 S/s by a 16-bit ADconverter (National Instruments, NI-6224). The signal was recorded with the software program VSSRP98 v6.0 (Versatile Stimulus Response Registration Program, 1998;; Technical Support Group of the Department of Psychology, University of Amsterdam).
1.4 Heart rate activity
Heart rate variability was collected using three Ag–Ag electrodes. The electrodes were fixed to the participant skin using adhesive patches (3M Red Dot Electrode
with Micropore Tape 2239). The electrodes were stuck on three specific points of their torso: one was placed below the right clavicle, the other one on the left side of the chest, just below the sixth rib, and the ground electrode was fixed under the left clavicle. The signal was recorded with the software program VSSRP98 v6.0 (Versatile Stimulus Response Registration Program, 1998;; Technical Support Group of the Department of Psychology, University of Amsterdam), acquiring at a sampling rate of 1000 samples per second.
1.5 Shock administration.
One 30 per 40 mm Ag/AgCl electrode with a conductive gel (Signa, Parker) was attached to the wrist of the hand opposite to the dominant one. Shocks were administered using a Digitimer Constant Current Stimulator DS7A (www. digitimer.com). Shock consisted of a 4mA current of 1 s. Shock intensity was adjusted to be uncomfortable but not painful per participant.
1.6 Subjective measures
Participants filled out Dutch translations of the trait portion of the state-trait anxiety inventory (STAI-T;; Spielberger et al., 1970;; Van Der Ploeg et al., 1980), perceived stress scale (PSS;; Cohen et al., 1983;; De Vries, 1998) and the survey of recent life events (SRLE;; Kohn and Macdonald, 1992;; Majella De Jong et al., 1996). Furthermore, to assess the influence of hydrocortisone on self- reported affective state, participants filled out the state-anxiety inventory (STAI-S;; Spielberger et al., 1970;; Van Der Ploeg et al., 1980) and the positive affect and
negative affect schedule (PANAS;; Watson et al., 1988). Subjective evaluation of the conditioned stimuli on arousal and valence dimensions was assessed online using self-assessment
Participants filled out Dutch translations of the trait portion of the Spielberg Trait Anxiety Inventory (STAI-T;; Spielberger and Gorsuch, 1983), and the Beck Depression Inventory (BDI;; Beck et at., 1961), the Positive And Negative Affect Scale (PANAS;; Watson et al., 1988), and the Spielberg State-Trait Anxiety Inventory (STAI-S;; Spielberger and Gorsuch, 1983) test. 1.7 Experimental task 1.7.1 Learning phase
Before the task, participants were instructed to tried to remember the background image in combination with the object as vivid as possible. They could do this through imagining how the background image interacts with the object.
Each trial began with a black background, followed by a 700 × 933 pixels background picture that was shown for 5 s. Three seconds after background picture onset, an object image appeared in the foreground within a 300 × 400 pixels frame. The object was displayed for 2 s, outlasting background picture offset for 500 ms. After that, participants were asked to rate how vivid they imagined the interaction between background image and object. They rated this on a continues scale, starting from “not vivid at all” to “very vivid”, were mouse totally left meant “not vivid at all” and click on the totally right
meant “very vivid”, in order to know their perception regarding the image combination.
This phase consisted of 120 trials, where each background image was paired with one of the object images.
1.7.2 Reactivation phase
To give enough time to consolidate the image combination, all testing was performed elapsed at least 24 hours after the learning phase. At the beginning of the experiment, electric shock electrodes were attached.
Reactivation trials began with a 3 s black- screen, followed by the presentation of 700 × 933 pixels background picture during 5 s. After that, participants were asked to rate to what extend the background image gave them the feeling of reliving. The extend of reliving was instructed to be determined by recognizing the background image and the extend in which they remembered the corresponding object of the previous day. Participants rated the extend of reliving on a continues scale, from ‘no reliving’ to ‘a lot of reliving’. The scale had the same system as the previous day and had 5 s to rate.
The reactivation phase consisted of 80 trials, whereby 40 trials were paired with a shock (reactivated+shock, RS+), while the 40 images left (reactivated+noshock, RS- ) were unpaired. The trial sequence was randomized individually. 1.7.3 Recognition phase
Recognition trials began with a black image, followed by the presentation of an object image of 300 × 400 pixels frame for 5 s. Then, a 3 s black screen appeared. After, participants were instructed to indicate whether they had seen the object during the encoding phase or not. To report their answer, participants had to rate on a 6-point scale, that reflects how certain they were about their answer. The meaning of the numbers was as followed: 1;; Very certain the object is new, 2;; pretty certain the object is new, 3;; I guess the object is new, 4;; I guess this object is old, 5;; pretty certain the object is old, and 6;; very certain the object is old.
The recognition phase consisted of 240 trials, whereby 40 trials consisted in objects presented during the learning phase, but their background image did not appear during the reactivation phase (non reactivated), 40 trials of object images presented during the learning phase, and their background image was presented and paired to a shock during the reactivation phase (reactivated+shock), 40 trials of object images presented during the learning phase, and their bacground image was presented and unpaired to a shock during the reactivation phase (reactivated+noshock), the 120 trials left, consisted of new object images. The trial sequence was randomized individually.
Figure 1 Experimental task. Adult human subjects
object in the middle. During the reactivation phase, 80 context images from the previous day will be presented, and after 50 of them, they will receive a small shock. During the recognition test, we will present randomly mix the 120 target objects with 120 new target objects.
1.8 Procedure
Participants were scheduled to three laboratory sessions separated by a one- day interval. They were informed beforehand that during the study electrical shocks were used and provided written informed consent. Experiments took place in a light room and were conducted using a personal computer. Each subject was seated in front of a 22-inch wide monitor screen located approximately 2 feet in front of a subject.
In session 1, skin conductance electrodes, heart rate electrodes, and shock bracelet were attached, followed by the sensitiveness to shock test. After, shock bracelet was removed. Then, participants completed the STAI-S, BDI, PANAS, STAI-S (all questionnaires were computer-administrated). Subsequently, participants completed the learning phase of the contextual memory task. During the learning phase, participants were instructed to move as little as possible and to try to imagine how does the background image interacts with the object for every trial. Once completed, participants were asked to fill once again a PANAS and STAI-S.
In the second laboratory session, reactivation phase took place 24 hours after the end of the learning phase. Skin conductance sensors, heart rate electrodes, and shock bracelet were attached. Afterwards, participants
completed the PANAS and STAI-S questionnaires. Next, participants started the second task, during the task, they were instructed to move as little as possible and to try, while looking at the background image, to remember the whole mental image of the previous day for each trial. Once completed the reactivation task, they filled the PANAS and STAI-S questionnaires. Followed by the detachment of all the electrodes.
On the last session, recognition phase took place 24 hours after the end of the reactivation phase. Skin conductance sensors and heart rate electrodes were attached. Afterwards, participants completed the PANAS and STAI-S questionnaires. Thereafter, participants were instructed to move as little as possible during the task and to try to remember the object they had seen on day one and rate it. Once completed the recognition task, they filled the PANAS and STAI-S questionnaires. Followed by the detachment of all the electrodes. At the end, participants were asked to fill a questionnaire regarding the experiment and discussed with them their opinion regarding the experiment.
Figure 2. Experimental protocol. Types of
reminders. The timing of targeting objects within the context images and timing for the three days are represented.
1.9 Gender differences
To have a stronger background of women participants, during the medical screening, women had to answer some specific questions related to their menstrual cycle and contraceptive control.
2.0 Statistical analysis.
Data were analyzed using the statistical package IBM SPSS statistics 22 (Armonk, New York, USA: IBM Corp).
The recognition memory analysis was addressed with a one-way repeated measures analysis of variance (ANOVA). This analysis was used in the recognition accuracy of participants and the effects of the manipulation through physiological record responses (SCL and HR). Recognition performance was defined as percentage of correct recognized objects as “old” on day 3. Greenhouse—Geisser correction was applied when the assumption of sphericity was violated.
For the reliving amount effect analysis (containing recognition accuracy, SCL and HR measurements) an explorative, two-way, 2x3 factorial repeated measures ANOVA was conducted, with two levels for reactivation conditions (RS+, RS-) and 3 levels of reliving rates (High, Some, Low). To test the effectiveness of our reactivation manipulation, participants were asked to measure the amount of “reliving” when faced the reminder cue on day 2. The results were categorized into even 33rd percentiles.
For gender differences analysis (including recognition accuracy, SCL and HR measurements) a factorial repeated
measures ANOVA was conducted with Memory Condition Scores as within- subject factor and Gender as between- subject factor. In case of significant results, partial eta squared (hp2) is reported as a measure of effect size. An alpha-level of 0.05. RESULTS Participant characteristics
Six participants (all female) were excluded prior to analysis;; two participants were absent during the “recognition phase”, two participants indicated not having received any electric shocks during the “reactivation phase”, and two participants did not report any reliving ratings. The final sample consisted of 38 participants with a mean age of 22.58 years (SD=3.77, 12 males).
Recognition memory.
To investigate the effect of arousal during reconsolidation of neutral memories, memory performance was examined using number of hits during the recognition phase. A one-way ANOVA with RS+(0.77±0.14) RS- (0.76±0.16) and non reactivated memories (0.75±0.15), as repeated measures showed no significant effect on memory conditions on recognition accuracy (F [2, 78] = 1.31, p = 0.49, ηp2 = 0.03), as shown in Figure 3A.
Moreover, skin conductance is one of the most sensitive measures of mental activity. Specifically, in conditioning studies, the skin conductance response has been interpreted as an indicative of a neutral stimulus acquiring the signal value of the unconditioned stimulus, allowing
assessment of the degree of learning or unlearning. (Vila 2004). Therefore, we analyzed the skin conductance responses (SCR) during the recognition phase with a one-way repeated measures ANOVA of the 3 memory conditions (RS+, RS- and non reactivated). All skin conductance data of 10 participants were lost due to technical issues (malfunctions in the measuring equipment). Our results showed no significant main effect of reactivation condition on skin conductance (F [2, 54] = 1.16, p = 0.32, ηp2 = 0.04), finding RS+(0.18±0.19), RS- (0.22±0.23) and non reactivated memories (0.19±0.27). A two paired sample t-test provided confirmation that the conditioning manipulation was noneffective at generation higher arousal in RS+ (0.18 ± 0.19) than RS- (0.22 ± 0.23) trials. p=0.163.
Additionally, during recognition and recall tasks, heart rate (i.e. beats per minute) changes have been related to memory load and performance (Jennings & Hall 1980;; Subotnik et al., 2012;; Sawyer et al., 2015). Thus, differences in the heart change (see Figure 3C) were analyzed with a one-way repeated measures ANOVA, as were performed for SCR. Due to malfunctions in the measuring equipment, physiological data of nine participants had to be excluded, leaving heart-rate data of 29 participants. Same as what we observed in the other physiological test (SCR), the measures of HR change did not appear to be as sensitive to memory condition (F [1.67, 58] = 1.09, p = 0.34), having values of - 6.53±0.43 for RS+, of -6.55±0.4 for RS- and of -6.91±0.44 for non reactivated memories.
A
Per cen tage of r ec og nition ac cur ac y Reactivated + Shock Reactivated + No Shock Non reactivated 0 0.05 0.1 0.15 0.2 0.25 0.3 Per cen tage of SCL v ar ia tion Reactivated + Shock Reactivated + No Shock Non reactivatedB
-8 -7 -6 -5 -4 -3 -2 -1 0 1 Reactivated + Shock Reactivated + No Shock Non reactivatedC
BP M 0.9 0.8 0.5 0.6 0.7 0.1 0.2 0.3 0.4
FIGURE 3. Performance during recognition phase. (A) Recognition performance index by percentage
recalled images at test during day 3. Errors bars represent the SD of the mean. (B) Skin conductance responses. Mean square-root-normalized skin conductive responses for the RS+, RS- and non reactivated trials. Error bars are standard error means. (C) Heart rate (HR) response measurement. Bar height indicates beats per minute (BPM). Data are presented for RS+, RS- and non reactivated trials. The error bars represent SD. No significant differences were identified.
Reliving amount effect
Opposite to what we were expecting, as we shown above, exposure to the shock manipulations did not significantly elevated subjective arousal, did not finding differences on the effect of reactivation conditions on recognition accuracy. An extensive research has shown that in order to reconsolidate a memory, it is necessary to reactivate it (Crestani et al., 2015, Liu et al., 2012). Therefore, our results could be due to a fault during the reactivation phase, suggesting that during day 2, the reactivation of memories was not efficient, making the shock manipulations unsuccessful. Consequently, in order to test the effectiveness of our reactivation manipulation, during day 2, we asked participants to what extend the background image gave them a reliving feeling. The resulting ratings were categorized into tree standardized categories of 33.33% each, as: Low Reliving, Some Reliving and High Reliving. We assumed that High Reliving scores of objects could be categorized as a successful reactivation of the memory, improving the recognition accuracy during day 3. In contrast, the Low Reliving effect objects, will not evoke any recalling effect, affecting than the recognition accuracy of the objects. To study the effect of our manipulation, we performed a two way repeated measures ANOVA finding for
High Reliving: RS+(0.28±0.09) RS- (0.25±0.06);; for Some Reliving: RS+(0.23±0.08) RS- (0.22±0.09);; and for Low Reliving: RS+(0.23±0.06) and RS- (0.25±0.08). The results showed a main effect of reliving on recognition accuracy presenting (F [1.43, 74] = 4.38, Greenhouse-Geisser, p = 0.03, ηp2 = 0.11). Moreover, additional analyses of the interaction effect between degrees of reliving and reactivation conditions on recognition accuracy with a 2x3 repeated measures factorial ANOVA revealed marginally differences based in the degree of reliving (F [1.72, 63.63] = 2.48, p = .10, ηp2 = .06). Significant differences were revealed with evoked High Reliving rates in memory accuracy between the reactivation conditions (p = .04), as shown in Figure 4A.
The influence of the reactivation condition and reliving over the SCR variations were analyzed using a two-way reaped measures ANOVA. Differing our predictions, participants did not manifested significant main effects of reliving on SCL variability between RS+ and RS- (F[2,52] = 1.56, p = 0.21, ηp2= .06), meaning that skin conductance variation on day 3 did not differ based on whether participants reported a high (M = .23, SD = 0.05), some (M = .19, SD = .04) or low (M = .18, SD = .04) level of reliving on day 2. Moreover, no significant interaction effect was found between reliving amount and condition on SCR (F
[2, 54] = 0.24, p = 0.79 ηp2= 0.01), as show in Figure 4B.
Differences in the HR change (See Figure 4C) were analyzed with one-way repeated measures ANOVA. As the results showed of the SCR variations, the measures of HR change did not appear to be sensitive to
conditions and reliving amount (F [1, 29] = 0.14, p = 0.7 ηp2= 0.02).
FIGURE 4. Recognition accuracy for three reliving conditions. (A) The objects that evoked High Reliving
rates, compared to the Some Reliving and Low Reliving objects, showed a significant difference in recognition accuracy during day 3. *p < .04;; error bars represent S.D. (B) SCR variations for the reliving and reactivation variables;; error bars represent the S.D. (C) Differences in the minimum and maximum value of BPM for each reactivated condition.
Shock Intensity
Shock intensity was calibrated before experimental task began (see methods), measuring the level of discomfortness. The intensity of the shock differed individually, values ranging from 0.3mV and 3.8mV (M = 0.86, SD = 0.64). After the experimental task, participants rated their
level of discomfort towards the electric shocks. Values ranged from 1 to 9, with 1 being the lowest and 9 the highest level of discomfort.
To examinee the habituation effects of the shock manipulation, a paired-samples t- test was conducted to compare the difference between self-reported
0 0.05 0.1 0.15 0.2 0.25 0.3 Reactivated + No Shock Reactivated + Shock Low High Medium Reliving * A Por cen tage of r ec og nition ac cur ac y 0 0.05 0.1 0.15 0.2 0.25 0.3 Reactivated + No Shock Low High Medium Reliving Per cen tage of SCL v ar ia tions B 4,00 0 2,00 - 4,00 - 8,00 - 6,00 - 2,00 Reactivated + No shock Reactivated + Shock Minimum
value Maximum value
Reactivation type Hear t r at e (BP M) C Reactivated + Shock
discomfort prior to the experimental task and after the task. As Table 1 shows, our results revealed a significant difference in participant discomfort prior testing (M =
5.87, SD = 0.23) and post-testing (M = 4.82, SD = 0.27);; t (38) = 3.24, p = 0.002).
Mean N Std. Deviation Std. Error Mean
Pre- testing shock discomfort 5.87 39 1.436 0.230 Post- testing shock discomfort 4.82 39 1.684 0.270
Table 1. Shock discomfortness. Paired Samples Statistics of the self-reported degree of discomforts
experienced by the participants prior and post the experimental task during the Recognition Phase.
Gender differences
Previous studies have shown that women and men tend to differ their emotional enhancement of episodic memory, where women typically demonstrated higher enhancement (Cahil et al., 2001;; Cahil, 2006;; Canli et al., 2002). In the present study, a mixed Between-Within subjects repeated measures ANOVA showed for women (n=26): RS+(0.75±0.14) RS- (0.75±0.15) and non reactivated memories (0.74±0.15), and for men (n=12): RS+(0.81±0.14) RS- (0.79±0.16) and non reactivated memories (0.78±0.18). Then, did not violating the assumptions of Levene’s and Box’s tests, Willi’s Lamda test reported that the interaction effect between reactivation condition (RS+, RS-) and non reactivated, and gender was not statistically significant (F [2, 35] = 0.54, p = 0.589 ηp2= 0.03). Moreover, there was no main effect in memory performance (F [2, 35] = 1.1, p = 0.344 ηp2= 0.059), and between groups the differences were no statistical significant (F [1, 36] = 0.91, p = 0.345 ηp2= 0.025). See Figure 5A.
As a manipulation check of the conditioning procedure, we analyzed the SCR responses of the participants with a mixed Between-Within subjects repeated
measures ANOVA. Women (n=21) presented for RS+ 0.17±0.23, for RS- 0.23±0.26 and for non reactivated 0.20±0.30. Men (n=8) presented for RS+ 0.18±0.11, for RS- 0.19± 0.11, and for non reactivated 0.15±0.11. The same as our previous results, the analysis did not violate the assumptions of Levene’s and Box’s tests, Willi’s Lamda test reported that the interaction effect between reactivation condition (RS+, RS- and non reactivated) and gender was not statistically significant (F [2, 26] = 0.486, p = 0.621 ηp2= 0.036). Additionally, no main effect was found in memory performance (F [2, 26] = 0.680, p = 0.516 ηp2= 0.050). An additional analysis between groups, reported no statistical significant differences (F [1, 27] = 0.086, p = 0.772 ηp2= 0.03). (See Figure 5B below).
Differences in the HR change (Figure 5C) were analyzed with mixed Between-Within subjects repeated measures ANOVA. The minimum BPM scores for women (n=20) were: RS+(-5.75±1.78) RS- (-5.87±2.56) and non reactivated memories (- 6.01±2.68), and for men (n=8): RS+(- 8.22±2.62) RS- (-7.39±1.56) and non reactivated memories (-9.3±0.17). Later, did not violating the assumptions of Levene’s and Box’s tests, Willi’s Lamdas test reported that the interaction effect
between reactivation condition (RS+, RS- and non reactivated) and gender was not statistical significant (F [2, 25] = 0.812, p = 0.074 ηp2= 0.288). However, we did find significant main effect between minimum values in memory performance (F [2, 25] = 0.736, p = 0.02 ηp2= 0.264). A post-hoc test specified that the significant main differences were between RS- and non reactivated scores in men. Likewise, the differences between groups were statistical significant (F [1, 26] = 8.31, p = 0.008 ηp2= 0.242). The maximum BPM scores for women (n=20) were: RS+(4.23±2.27) RS- (4.22±2.2) and non reactivated memories (3.94±1.92), and for men (n=8): RS+(3.71±1.6) RS- (3.35±1.13) and non reactivated
memories (2.34±1.92). Willi’s Lamdas test reported that the interaction effect between reactivation condition (RS+, RS- and non reactivated) and gender is not statistical significant (F [2, 25] = 0.875, p = 0.189 ηp2= 0.125). In contrast, we found a main effect in memory performance (F [2, 25] = 0.734, p = 0.021 ηp2= 0.266). A post-hoc test (Bonferroni) revealed that the scores of the max values changes between RS+ and non reactivated conditions in men. The statistical analysis did not report any significant differences (F [1, 26] = 1.792, p = 0.192 ηp2= 0.064).
Figure 5. Gender differences in the effect of post-encoding arousal. (A) Differences on recognition
performance index by percentage recalled images at test during day 3 between women and men, and no differences were found. Errors bars represent the standard error means. (B) Skin conductance responses. Mean
Reactivated + No Shock
Non reactivated Reactivated
+ Shock 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Women Men Per cen tage of r ec og nition ac cur ac y
A
0 0.05 0.1 0.15 0.2 0.25 0.3B
Women Men Reactivated + No ShockNon reactivated Reactivated
+ Shock Per cen tage of SCL v ar ia tion -10 -8 -6 -4 -2 0 2 4 6 8 10 BP M Minimum values Maximum values * * *** -12 12
C
Women Men Reactivated + No ShockNon reactivated Reactivated
+ Shock
square-root-normalized of differences between women and men in skin conductive responses for the RS+, RS- and non reactivated trials. Error bars are standard error means. (C) Heart rate (HR) response differences measurements between women and men. Bar height indicates average of beats per minute (BPM) of maximum and minimum values for each memory condition. Significant differences were identified between maximum score values of non reactivated and RS+ memory conditions in men (*p=0.021). On the other hand, minimum values significant differences were found between non reactivated and RS- values (*p=0.02). Statistical significant differences between groups were found (***p=0.008). Errors bars represent the standard error means.
DISCUSSION
In contrast with our expectations, the present study was not capable to demonstrate a significant effect of arousal during reconsolidation of neutral memories with an object-context-episode task, in 38 healthy participants (Figure 3A).
Our sample was typical, so we doubt that our findings could relate to this. However, there was a possibility that our conditioning was ineffective. To test this, we analyzed the physiological responses during the recognition test.
First, previous studies have shown that during stimuli and emotional arousal, the autonomic nervous system responds and has an impact at body level. This effect can be studied through measurements of skin conductance responses (Lempert & Phelps, 2014;; Sequeira et al., 2009;; D’Hondt et al., 2010). Particularly, higher SCR magnitude responses have been observed as an effect of pleasant and unpleasant stimulus, and arousal (D’Hondt et al., 2010). Based on this, to confirm the reliability of our conditioning, we would expect basal responses in the non reactivated condition, middle scores on the RS- condition, and high responses on the RS+ condition. At first sight, our results showed that the basal levels were in fact the non reactivated condition. In contrast, the highest scores were the ones
from the RS- condition and the middle scores belonged to the RS+ condition. However, the probabilistic test showed that these differences were not significant (Figure 3B). The lack of differences between the memory conditions and the SCL responses can be translated to a lack of effectiveness of our manipulation, hence no effect on memory conditions on recognition accuracy was observed.
Second, high HR responses correspond to enhancement of memory recognition in test after consolidation (Larra et al., 2014). Then, HR responses were also used as a feature to measure the effectiveness of our manipulation. The analysis of variance applied to the results of this procedure revealed no significant differences between the three conditions and their HR responses.
Third, the frequency of responding to an incentive shock declines with repeated electrical stimulation, this is called habituation (Spear & Campbell, 2014). Habituation has been studied in different mammals, as rats, cats and humans (Peeke, 2012), and is considered a stimulus-specific and response-general phenomenon (Spear & Campbell, 2014). We discovered a significant difference in participant discomfort of the electrical shock before testing and after testing, indicating that participants habituated to electrical stimulation, suggesting no effect of the stimuli to evoke higher arousal