Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 Oct 15.
Published in final edited form as: Neuroimage. 2010 Jun 11;53(1):318–324. doi: 10.1016/j.neuroimage.2010.06.015

The Effect of Arousal on the Emotional Memory Network Depends on Valence

Katherine R Mickley Steinmetz (1),(2), Donna Rose Addis (2),(3), Elizabeth A Kensinger (1),(2)
PMCID: PMC3063443  NIHMSID: NIHMS279430  PMID: 20542121

Abstract

Some suggest that arousal is the essential element in order to engage the amygdala. However, the role of arousal in the larger emotional memory network may differ depending on the valence (positive, negative) of the to-be-remembered information. The goal of the current study was to determine the influence of arousal-based changes in amygdalar connectivity for positive and negative items. Participants were shown emotional and neutral pictures while they underwent a functional magnetic resonance imaging (fMRI) scan. The emotional pictures varied by valence (positive or negative), and arousal (high or low). Approximately 45 minutes later, outside of the scanner, participants took a surprise recognition test. Effective connectivity analysis examined how arousal affected successful encoding activity. For negative information, arousal increased the strength of amygdala connections to the inferior frontal gyrus and the middle occipital gyrus, while for positive information arousal decreased the strength of these amygdala efferents. Further, while the effect of arousal on memory for positive information was restricted to amygdalar efferents, arousal had a more widespread effect for negative items, enhancing connectivity between other nodes of the emotional memory network. These findings emphasize that the effect of arousal on the connectivity within the emotional memory network depends on item valence.

The amygdala has long been cited as an important part of the neural circuitry for processing emotion. Its importance in emotion processing has been found in studies that have focused on the processing of a variety of stimuli, including emotional faces (e.g., Morris et al., 1998), pictures (e.g., Glascher & Adolphs, 2003), and words (e.g., Hamann & Mao, 2002; Isenberg et al., 1999). When processing these emotional stimuli, the amygdala activation has been linked to modulation of other brain areas, including the fusiform gyrus, the middle occipital gyrus, and the parahippocampal gyrus (e.g., Kilpatrick & Cahill, 2003; Tabert et al., 2001; Vuilleumier et al., 2001), leading to enhancements in attention directed towards the emotional stimuli (Phan, Wager, Taylor, & Liberzon, 2002; Stein, et al., 2007). The amygdala’s modulation of these networks during emotion processing also has been shown to be important in the retention of emotional information over time (Cahill et al., 1996; Kilpatrick & Cahill, 2003; Richardson, Strange, & Dolan, 2004).

It has been debated whether the amygdala’s role in information processing and retention is modulated by arousal (the stimulating or calming nature of a stimulus) or valence (the positive or negative nature of a stimulus). Initial studies suggested that the amygdala primarily responded to threatening stimuli (Kluver & Bucy, 1937; Weiskrantz, 1956; Whalen, 1998), and there has been some evidence from lesion studies that the amygdala may be more involved in the recognition of negative stimuli than of positive stimuli (Adolphs, Russell, & Tranel, 2001; Tranel, Gullickson, Koch, & Adolphs, 2006). However, recent neuroimaging and lesion evidence has suggested that arousal may be the key factor; the amygdala may respond to any arousing stimulus regardless of valence. Further, the arousal response might be essential in order for the amygdala to modulate memory (Adolphs, Russell, & Tranel, 1999; Anderson, 2005; Anderson, Wais, & Gabrieli, 2006; Kensinger, 2004; Cahill & McGaugh, 1995; Canli, Zhao, Brewer, Gabrieli, & Cahill, 2000; Cahill et al., 2003).

The focus of the current study is to examine how arousal affects connectivity within an amygdala-modulated emotional memory network, and to examine whether the influence of arousal varies depending on whether stimuli are of positive or negative valence. As reviewed above, the link between amygdala activity and subsequent memory may not differ based upon item valence (Hamann & Mao, 2002; Kensinger & Schacter, 2006), yet this does not mean that the effect of arousal on the amygdala’s role in guiding memory must be identical for positive and negative stimuli. Based on prior research, we hypothesized that increasing stimulus intensity would have a greater impact on amygdala connectivity with regions of the emotional memory network when stimuli were negative than when they were positive. This hypothesis stemmed from a few lines of prior research. First, the effect of arousal may differ based on valence. Garavan, Pendergrass, Ross, Stein, & Risinger (2001) found that when processing emotional pictures, arousal modulated amygdala increases for negative stimuli, while the amygdala response remained high across positive stimuli. In addition, Bernsten, Bechara, Damasil, Tranel & Cacioppo (2007) found that people with amygdalar lesions did not show a typical behavioral gradient in arousal ratings to negative pictures, while the gradient in ratings for positive pictures was preserved. These findings suggest that the amygdala is particularly important for processing the arousal tied to negative stimuli. Second, amygdala damage disrupts the ability to re-experience the negative affect that accompanied prior events (Buchanan, Ezel, Adolphs, & Tranel, 2006), corroborating a strong link between amygdala function and the experience of arousal during a negative event. Third, amygdala damage has a detrimental impact on the ability to recall negative arousing life experiences, whereas it does not have as large an effect on the recall of positive arousing experiences (Buchanan et al., 2006), suggesting that the amygdala may play a particularly central role in the emotional memory network when information is both arousing and also of negative valence. These studies suggest that arousal may affect amygdala activity differently depending on the valence of the item. This differential effect of arousal may have implications for the amygdala-based processes that guide emotional memory (see McGaugh, 2004).

Arousal seems to affect memory by leading to amygdala-related modulation of other areas, including other regions of the medial temporal lobe (McGaugh, 2004; Kilpatrick & Cahill, 2003), regions with the temporo-occipital cortex (Lane, Chua, & Dolan, 1999; Dolan & Vuilleumier, 2003), and regions within the prefrontal cortex (Zald et al., 1998). Yet these studies have primarily focused on the role of arousal when information is negative in valence, leaving open the possibility that the ability for arousal to modulate the strength of connections between the amygdala and other peripheral areas may depend of the valence of the to-be-remembered item. Because negative and arousing information is often recollected with a sense of strong visual detail, and has been associated with encoding activity in the middle occipital gyrus and fusiform gyrus (Kensinger & Corkin, 2003; Mickley & Kensinger, 2008; Mickley Steinmetz & Kensinger, 2009), it is possible that for negative information, arousal may be especially likely to increase connections between the amygdala and these areas associated with visual processing for negative information. By contrast, for positive information, arousal may not elicit the same magnitude of strengthened connections. Thus, differences in arousal based memory for positive and negative items might be due to changes in the way in which arousal influences the connections between the amygdala and a broader emotional memory network. To address this hypothesis, the present study employed effective connectivity analyses using structural equation modeling (SEM).

Methods

Participants

Participants included the same nineteen right-handed young adults, aged 19 – 35, reported in Mickley Steinmetz and Kensinger (2009). All participants were native English speakers and had normal or corrected to normal vision. Participants with a history of psychiatric or neurological disorders or participants taking medication that would affect the central nervous system were excluded from the study. Informed consent was obtained in a method approved by the Institutional Review Boards at Massachusetts General Hospital and Boston College.

Materials and Procedure

The materials and procedure were identical to those described in Mickley Steinmetz and Kensinger (2009). Behavioral data can also be found in that publication. To briefly reiterate those parameters, stimuli were 350 pictures: 70 negative arousing, 70 negative nonarousing, 70 positive arousing, 70 positive nonarousing and 70 neutral pictures taken from the International Affective Picture System database (Lang et al., 1999). Positive and negative pictures were matched on visual complexity, brightness [as indicated in Adobe Photoshop (Adobe Systems, San Jose, CA)], and number pictures that included animals, people, buildings or landscapes (normative data from Kensinger & Schacter, 2006b). Positive and negative pictures were equated for arousal and for absolute valence (i.e., distance form neutral valence). High and low arousal images were equated for valence ratings.

Participants viewed 175 pictures (35 from each emotion category) for 2 sec each while undergoing a functional magnetic resonance imaging (fMRI) scan. Interstimulus intervals were jittered, ranging from 2 to 14 s (Dale, 1999). About an hour and a half later and outside of the MRI scanner, participants performed a recognition memory task. During the delay interval, all participants completed a verbal attention task. On the recognition memory task, participants were shown the 175 pictures that they had viewed in the scanner intermixed with 175 new pictures. The studied pictures varied across participants, counterbalancing whether an image appeared on the recognition test as a studied item or as a novel foil. Each picture was shown for 2 sec, and participants were asked to indicate if they recognized the picture as “old” or if it was “new.” If the participant recognized the picture as “old,” they were asked to complete a modified version of the Memory Characteristics Questionnaire (Johnson, Foley, Suengas, & Raye, 1988), rating their memory for confidence, remembered thoughts, feelings, intensity of feelings, associations, visual details, and temporal information they remembered about the item’s presentation. There were no significant valence x arousal interactions for any of the ratings. Therefore, these memory characteristic data will not be discussed further.

Image Acquisition and Preprocessing

Data were acquired on a 1.5 Tesla Siemens Avanto whole body MRI system (Erlangen, Germany) with a standard birdcage head coil. Stimuli were projected from a Macintosh iBook G4 using a Sharp200 color LCD projector with a collimating lens (Buhl Optical). The image was shown on a screen that was mounted at the end of the magnet bore. The screen was viewed via mirrors placed on the head coil.

Anatomic images were acquired with a multi-planar rapidly acquired gradient echo (MP-RAGE) sequence with a repetition time (TR) of 2730 ms, an echo time (TE) of 3.31 ms, a flip angle of 40°, a field of view of 256 x 256 mm, an acquisition matrix of 256 x 256, slice thickness of 1.33 mm, no gap, and a 1x1x1.33 mm resolution. Co-planar and high-resolution T1 weighted localizer images were acquired. In addition, a T1 weighted inversion recovery echo planar image was acquired for auto alignment.

Functional images were acquired via at T2* weighted echo planar imaging sequence sensitive to blood oxygenation dependant contrast (BOLD). The TR was 3000 ms, the effective TE was 40 ms, and the flip angle was 90°. The slices were taken in an interleaved fashion, at an axial-oblique angel (parallel to the AC-PC line). Twenty-nine slices were acquired in a 3.125 mm x 3.125 mm x 3.72 mm matrix. The slices were 3.12 mm thick and had a 0.6 mm skip between slices.

Preprocessing and data analysis were completed using SPM2 (Wellcome Department of Cognitive Neurology, London, UK). Preprocessing included: slice time correction, motion correction (which used a six parameter, rigid body transformation algorism from SPM2), normalization (to the Montreal Neurological Institute template, with resampling at 3mm isotropic resolution), and spatial smoothing (at a 7.6 mm isotropic Gaussian kernel).

Region Selection

Regions were selected a priori based on those areas known to be an important part of the emotional memory network (e.g., LaBar & Cabeza, 2006; Mickley & Kensinger, 2008). Five regions were included in our model: left amygdala, left hippocampus, left inferior frontal gyrus, left middle occipital gyrus, and left fusiform gyrus (see Table 1). We defined these ROIs using coordinates selected from a contrast that examined the regions whose encoding-related activity corresponded parametrically with a more confident response during retrieval. At the first level, the parametric modulation function was used, entering the corresponding confidence ratings for each trial in each of the five emotion conditions. Contrast images of the parametric effect of confidence (collapsed across emotion condition) were entered into a second level (group) one sample t-test to identify a network of regions involved in the encoding of confident emotional memories. This analysis collapsed across emotion types, so that regions were not biased with regard to how they processed the emotional content of the items. Further, there were no statistically significant differences in confidence ratings across the five emotion categories (F (4,72) = 1.145, p > .05, partial eta squared = 0.06). Our model included left-lateralized ROIs because the amygdalar activity revealed in the parametric analysis was left-lateralized, and because evidence suggests that the strongest amygdalar connections were likely to be ipsilateral (Amaral, Behniea, & Kelly, 2003; Kilpatrick & Cahill, 2003; Vuilleumier, Richardson, Armony, Driver & Dolan, 2004). A significance threshold of p < .001 (uncorrected for multiple comparisons), and an extent threshold of five contiguously active voxels (3x3x3mm) was applied to this parametric analysis. MNI coordinates were converted to Talairach space and regions of activation were localized in reference to a standard stereotaxic atlas (Talairach & Tournoux, 1988).

Table 1.

Brain Area Talairach Coordinates Brodmann Area
Left Inferior Frontal Gyrus (OFC) −40, 24, −18 47
Left Amygdala −24, −1, −13
Left Hippocampus −34, −9, −12
Left Fusiform Gyrus −30, −45, − 18 37
Left Middle Occipital Gyrus −26, −86, −1 18
Open in a new tab

From each region in our model, we then extracted BOLD signal from a 5 mm sphere centered on the local maximum coordinate. Using the Marsbar toolbox in SPM2 (Brett, Anton, Valabregue, & Poline, 2002), a hemodynamic response function was calculated for each individual subject and for each emotion type (positive arousing, positive non-arousing, negative arousing, and negative non-arousing) relative to baseline (fixation). Note that signal was only extracted from encoding trials of items that were subsequently remembered, so that differences in connectivity between the two item types would not be due to differences in later memory accuracy given that more arousing than non-arousing items were later remembered. Thus, if signal from all trials (remembered and forgotten) was entered into the analysis, differential connectivity between arousing and non-arousing conditions could have nothing to do with arousal-based processing but rather to changes in the proportion of processes that were associated with successful vs. unsuccessful encoding. We then computed the average of the signal change within peristimulus times 4-6 seconds, and these values were used to construct the functional models for the SEM analyses, as described below.

Effective Connectivity Analysis

We investigated the effective connectivity of the regions within the emotional memory network using structural equation modeling (SEM; McIntosh & Gonzalez-Lima, 1994). This is a multivariate technique that uses the neuroanotomical model of connections and assesses its fit with the interregional covariances associated with the BOLD signal. SEM provides information about the strength of the connections in the model, and how these connections vary under different conditions. Moreover, unlike correlations, SEM analysis can also determine the directionality of the connections between different regions given that path coefficients can be asymmetric (e.g., the influence of region A upon region B can be greater than the influence of region B upon region A). The anatomical model specified the location and potential direction of the connections to ensure that the significant functional differences were anatomically viable. The connections in this model were based on known primate and rodent anatomy (Amaral, Behniea, & Kelly, 2003; Burman, Palmer, Gamberini & Rosa, 2006; Petrides & Pandya, 1999; Rosene & Van Hoesen, 1977) and were further refined to ensure stability of the model. First a model was made in which all connections were reciprocal connections. This model did not have good stability (i.e., stability estimates were above 1, indicating an unstable model) and so it was necessary to revise that model. Because the amygdala efferents were the dominant connections, and because of our theoretical motivation for studying the amygdala’s modulation of other brain areas, we revised the model to include only amygdalar efferents, leaving the bidirectional connections between the hippocampus and fusiform and the fusiform and the middle occipital gyrus. This model revealed beta weights that were greater than 1 for the hippocampus to fusiform connection. Beta weights that are greater than 1 are theoretically impossible and indicate the presence of an inappropriate reciprocal connection. Replacing the reciprocal connection between the fusiform and hippocampus with a unidirectional connection eliminated these inflated beta weights. This model also had good stability, and so it was this anatomical model that we used in the present study. Once the anatomical model was set, a functional model was then constructed for each emotion category condition (positive arousing, positive nonarousing, negative arousing, and negative nonarousing) by correlating the percent signal change for that condition across the regions in the anatomical model.

All SEM calculations were performed using Lisrel 8.30 (Joreskog & Sorbom, 1993). Two separate SEM analyses were computed. One compared negative arousing to negative nonarousing items. The other compared positive arousing to positive nonarousing items1. These two SEM analyses were therefore designed to reveal whether there were arousal-dependent changes in connectivity for (1) negative items or for (2) positive items. For each of the two SEM analyses, estimates of path coefficients were calculated based on the correlation matrices (i.e., the functional models).

To determine whether there were significant differences in connectivity between the conditions (i.e., between negative arousing and negative nonarousing, or between positive arousing and positive nonarousing), we used a stacked-model approach (McIntosh & Gonzalez-Lima, 1994). Using an omnibus test, a null model was constructed in which the path coefficients from both conditions were assumed to be equal. The fit of this model was determined using a goodness-of-fit χ2 test. The fit of the null model was compared to the fit of an alternative model in which the path coefficients were allowed to vary. The fit of the alternate model was also determined using a goodness-of-fit χ2 test, and the differences between the models were calculated by comparing these goodness-of-fit χ2 values to obtain a χ2diff. If the χ2diff was significantly lower for the alternate model relative to the null model (significance being determined by taking into account the difference in degrees of freedom between the two models), it indicated that at least some connections within the emotional memory network differed as a function of stimulus arousal (p < .05).

Thus, the next step was to determine which connections contributed significantly to the omnibus difference between the null and the alternate model. To do so, the amygdalar efferents and then the other connections were allowed to vary in a stepwise manner (following the methods of Gilboa et al., 2004). First, an alternate model was created in which only the amygdalar efferents were allowed to vary, and the significance of the decrease in the goodness-of-fit χ2 value relative to the null model was assessed (i.e., a χ2diff was computed as described above for the omnibus analysis). If the χ2diff was not significant (p > .05), then the strength of the amygdalar efferents were permanently set as equivalent for the two conditions (e.g., between negative arousing and negative nonarousing) in all subsequent alternate models. If the χ2diff was significant (p < .05), then the amygdalar efferents were permanently allowed to vary between the two conditions (e.g., between negative arousing and negative nonarousing) for all subsequent models. Then a second alternate model was created, in which the extra-amygdalar connections were now allowed to vary, and if this caused a further reduction in the goodness-of-fit χ2 value, then those connections also were set to vary between the conditions in the final model; otherwise the strength of these connections were fixed to be equivalent.

Results

Effects of Arousal on the Connectivity for Negative Items

The first SEM analysis compared the connectivity among regions in the emotional memory network for subsequently remembered negative items that were either arousing or non-arousing. The omnibus SEM analysis revealed that the goodness-of-fit was significantly better for the alternate model relative to the null model fit (χ2diff = 16.48, df = 7, p < .05), indicating there was a significant effect of condition on the effective connectivity of the emotional memory network. A stepwise assessment of connections was conducted to determine which connections differed significantly across conditions. This assessment revealed that the networks differed in both the amygdalar efferents (p < .01) and also the extra-amygdalar connections (p < .01; See Figure 1). The encoding of subsequently remembered negative arousing items was associated with stronger connections from the amygdala to the left inferior frontal gyrus and left middle occipital gyrus than was the successful encoding of negative nonarousing items. Connectivity between the fusiform and middle occipital gyrus was also stronger when encoding negative arousing versus non-arousing items. Finally, there was evidence that when successfully encoding negative low arousing items relative to negative arousing items, amygdala influences on the hippocampus became weakly negative, while positive influences of the fusiform upon the hippocampus became stronger.

Figure 1.

Figure 1

Open in a new tab

Anatomical model for the effective connectivity analysis of negative items. Arrows represent anatomical connections used in the model based on known primate and rodent neuroanatomy. Connections that are significantly different between the two arousal conditions are in blue. All connections were significantly different in this model. For Talairach coordinates of peak voxels see Table 1. L=Left, IFG = Inferior Frontal Gyrus, Amyg = Amygdala, HC = Hippocampus, Fusi = Fusiform, MOG = Middle Occipital Gyrus.

Effects of Arousal on the Connectivity for Positive Items

An omnibus SEM analysis comparing the connectivity within the emotional memory network during the successful encoding of positive arousing versus nonarousing items revealed that the goodness-of-fit was significantly better for the alternate model relative to the null model fit (χ2diff = 14.68, df = 7, p < .05), indicating there was a significant effect of condition (p < .05). A stepwise assessment of connections was revealed that the networks differed in terms of the connectivity of amygdalar efferents (p < .01) but not extra-amygdalar connections (See Figure 2). The encoding of subsequently remembered positive nonarousing items was associated with stronger connections from the amygdala to left inferior frontal gyrus, left hippocampus, and left middle occipital gyrus (See Figure 2) than was the successful encoding of positive arousing items.

Figure 2.

Figure 2

Open in a new tab

Anatomical model for the effective connectivity analysis of positive items. Arrows represent anatomical connections used in the model based on known primate and rodent neuroanatomy. Connections that are significantly different between the two arousal conditions are in green. For Talairach coordinates of peak voxels see Table 1. L=Left, IFG = Inferior Frontal Gyrus, Amyg = Amygdala, HC = Hippocampus, Fusi = Fusiform, MOG = Middle Occipital Gyrus.

DISCUSSION

The current study demonstrates that the level of arousal of successfully encoded items modulates the connectivity between the amygdala and other regions of the emotional memory network. Although past studies have suggested that the amygdala may play a particularly important role during the encoding of high-arousal stimuli, this study adds the important caveat that the effect of arousal may critically depend on the valence of the stimuli. To briefly summarize the results, for both positive and negative items, arousal modulated the strength of amygdalar efferents. However, the effects of arousal on those efferents differed for negative and for positive items. For negative items, there were stronger amygdalar efferents when encoding arousing items as compared to nonarousing items, both to the left middle occipital gyrus and the left inferior frontal gyrus. For positive items, on the other hand, it was nonarousing items that resulted in stronger amygdala connectivity to the left middle occipital gyrus, the left inferior frontal gyrus and the left hippocampus. Furthermore, only for negative items did arousal modulate the connections among regions aside from the amygdala, with arousal modulating connections between the middle occipital gyrus and the fusiform gyrus, as well as from the fusiform to the hippocampus.

Only a small number of studies have assessed the encoding of low-arousal positive and negative stimuli as well as high-arousal positive and negative stimuli, and this is the first study to examine changes in the neural connectivity of the emotional memory while using this broader range of stimuli. Therefore, this study opens a number of questions for future research. The most notable question is why amygdala connectivity would strengthen as arousal increases for negative stimuli but weaken as arousal increases for positive stimuli, a point which we will return to later in this discussion.

The finding for the negative stimuli is generally consistent with prior proposals regarding the role of the amygdala in encoding: It makes sense that amygdala connectivity would strengthen with increasing arousal if the amygdala plays a particularly important role in the encoding of negative, high-arousal stimuli (e.g., Adolphs, Russell, & Tranel, 1999; Kensinger & Cokrin, 2004; Cahill & McGaugh, 1995). It also makes sense that if this enhanced connectivity were disrupted – such as by damage to the amygdala – then this could disrupt the perception of negative arousal, or the preferential retention of negatively arousing memories (e.g., Bernsten et al., 2007; Buchanan et al., 2006).

A particularly interesting finding for the negative items was that arousal seemed to change the way in which valence modulated the neural connectivity that would be tied to sensory processing. For the encoding of negative items that are low in arousal, the amygdala exerted its influence on the hippocampus via the fusiform gyrus. Thus, even when negative items did not elicit high arousal, there was a pathway through which the amygdala could influence the amount of sensory input being received by the hippocampus. This finding may explain why negative events are sometimes remembered vividly, and with sensory detail, even when they are not high in arousal (e.g., Kensinger & Corkin, 2003) although it is important to note that in the present study, subjective perceptual memory vividness was not greater for negative pictures than it was for positive or neutral pictures. Because differences in connectivity are evident despite a lack of difference in the subjective vividness ratings, this connectivity does not seem to necessitate that a perceptually vivid memory will be maintained. However, previous studies using neutral and emotional stimuli have suggested that differences in fusiform-hippocampal connections may be associated with differences in memory for perceptual details (Dickerson et al. 2007; Kensinger & Schacter, 2007). Thus, in general these changes in connectivity may increase the likelihood that perceptual information is retained, thereby explaining why it is more common to find perceptual enhancements in memory for negative low-arousal stimuli than for other types of low-arousal information.

For negative items that were high in arousal, the connection through the fusiform gyrus was not as strong, and instead the middle occipital gyrus played a significant role, forming a loop from the amygdala to the middle occipital gyrus and the fusiform and then on to the hippocampus. Numerous studies have found amygdala activity to be coupled with activity in the visual cortex when processing negative information (Hamann, Ely, Hoffman & Kilts, 2002; Morris et al., 1998; Tabert et al., 2001; Vuilleumier, Armony, Driver, & Dolan, 2001; Vuilleumier, et al., 2004), and studies have revealed that the encoding of negative information that is high in arousal is specifically associated with activity in the middle occipital gyrus (Mickley & Kensinger, 2008; Mickley Steinmetz & Kensinger, 2009). It therefore makes sense that the connectivity between the amygdala and these occipital regions would be enhanced during the encoding of negative information that is arousing. This finding may also help to explain why there is such perceptual vividness associated with memories for items that are negative and high in arousal (Dewhurst & Parry, 2000; Doerksen & Shimamura, 2001; Kensinger & Schacter, 2006a; Kensinger, Garoff-Eaton, & Schacter, 2007). The perceptual quality associated with negative information that is high in arousal may be due to the overall strength of connections through the occipital cortex, or it may be due to the alternate pathway through the occipital gyrus, or a combination of the two. Future research will be necessary to directly examine whether these changes in connectivity modulates the specific phenomenological qualities of a memory.

Although the findings for the negative stimuli can be interpreted within existing frameworks regarding the role of the amygdala in emotional memory, the findings for the positive stimuli are not consistent with existing frameworks and suggest that the link between high-arousal and amygdala engagement may be more nuanced than previously discussed. The finding that high arousal decreases the strength of most amygdala efferents when stimuli are positive cannot be accounted for by the hypothesis that it is the high arousal level of stimuli that strengthens the amygdala’s role in memory encoding, nor can it be accounted for by any hypothesis that suggests that the amygdala only is engaged for negative stimuli and not for positive stimuli. The present data suggest that amygdala efferents can be strong for both positive and negative stimuli, and that the strength of those efferents can be modulated by arousal for both valences of stimuli, yet the direction of that modulation is opposite depending on the item valence.

As indicated in the introduction, studies have shown that arousal may have different effects depending on the valence of the item. For example, Bernsten, et al. (2007) found that people with amygdalar lesions did not show a typical behavioral gradient in arousal ratings to negative pictures, while the gradient in ratings for positive pictures was preserved. In addition, Garavan et al. (2001) found that arousal modulated amygdala activity for processing negative information, but amygdala activity remained constant for the processing of all positive stimuli, regardless of arousal. The present study extends these prior findings, revealing that arousal can have fundamentally different effects on amygdala connectivity with other regions of the emotion memory network, depending upon the valence of the information. Although arousal enhances amygdala connectivity for negative items, it does not do so for positive items. The present study cannot explain the basis for the enhanced connectivity for positive items that are low in arousal, but one interesting possibility is that the amygdala efferents for positive stimuli could be tied to more controlled or elaborative processes, which may be particularly likely to occur for low-arousal stimuli, whereas the amygdala efferents for negative stimuli may be linked to a different type of processing, perhaps one that is more automatic in nature, and thus that is more likely to occur for high-arousal stimuli. Although speculative, this proposal is consistent with the fact that amygdala efferents to the prefrontal cortex were particularly strong for positive, low-arousal items, whereas bottom-up inputs from sensory cortices to the amygdala were particularly strong for negative, high-arousal items.

Positive information has often been associated with a more global, heuristic, and elaborative processing than negative information (Fredrickson, 2004; Gasper & Clore, 2002), and the engagement of these types of processes seems to lead to the successful encoding of positive information (reviewed by Kensinger, 2009). Because elaborative processes have been proposed to be particularly important for the encoding of low-arousal information (e.g., Kensinger & Corkin, 2004), it is possible that these elaborative processes are most likely to be implicated in successful encoding when information is positive and of low arousal, whereas they may be less likely to correspond with successful encoding when arousal level is increased. Although the tie to elaborative processing is a speculative one at this point, the present results clearly demonstrate that arousal influences amygdala connectivity differently for negative and positive items.

A final point of interest is that, while the effect of arousal on memory for positive information was restricted to the amygdalar efferents, arousal had a more widespread effect for negative items. It was only for negative stimuli that arousal modulated connections among regions beyond the amygdala, between the fusiform and the hippocampus and the fusiform and the middle occipital gyrus. Thus, it seems that for the encoding of negative items, arousal has a more global effect, while the effect of arousal on positive items is more focused upon the amygdala efferents. The reason for this difference across valences is an open question for further research but it emphasizes that the power for arousal to enhance neural connectivity may be exaggerated for items that are of negative valence.

In summary, the current study is the first to reveal that the effect of arousal on the connectivity within the emotional memory network differs depending on the valence of the to-be-remembered item: while amygdala efferents ramp up in strength as arousal increases for negative items, this is not the case for positive items. This finding sheds light on previous findings which indicate that amygdalar involvement and encoding processes may differ depending on valence (Berntson et al., 2007; Dolcos, LaBar, & Cabeza, 2004; Mickley & Kensinger, 2008; Mickley Steinmetz & Kensinger, 2009), revealing that the way in which arousal influences how the amygdala is incorporated into the emotional memory network depends critically upon the valence of the information. These results underscore the importance of considering how the effects of arousal may be qualified by the valence of the item.

Acknowledgments

This research was supported by grant MH080833 from the National Institute of Mental Health (to E.A.K.) and by a National Defense Science and Engineering graduate fellowship (to K.R.M.S.). The authors thank Keely Muscatell and Elizabeth Choi for assistance with participant recruitment and data management.

Footnotes

1

Models that compared positive to negative items (collapsing across arousal types) were also created and compared. However, the null and the alternate model in this analysis did not differ significantly. Thus, there does not appear to be an overall effect of valence on the connectivity within the network examined here, but rather a difference in the way that arousal modulates that network.

References

  1. Adolphs R, Russell JA, Tranel D. A role for the human amygdala in recognizing emotional arousal from unpleasant stimuli. Psychological Science. 1999;10:167–171. [Google Scholar]
  2. Amaral DG, Behniea H, Kelly JL. Topographic organization of projections from the amygdala to the visual cortex in the macaque monkey. Neuroscience. 2003;118:1099–1120. doi: 10.1016/s0306-4522(02)01001-1. [DOI] [PubMed] [Google Scholar]
  3. Anderson AK. Affective influences on the attentional dynamics supporting awareness. Journal of Experimental Psychology: General. 2005;134:258–261. doi: 10.1037/0096-3445.134.2.258. [DOI] [PubMed] [Google Scholar]
  4. Bernsten GG, Bechara A, Damasio H, Tranel D, Cacioppo JT. Amygdala contribution to selective dimensions of emotion, Social Cognitive and Affective. Neuroscience. 2007;2:123–129. doi: 10.1093/scan/nsm008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Burman KJ, Palmer SM, Gamberini M, Rosa MGP. Cytoarchitectonic subdivisions of the dorsolateral frontal cortex of the marmoset monkey Callithrix jacchus, and their projections to the dorsal visual areas. The Journal of Comparitive Neurology. 2009;495:149–172. doi: 10.1002/cne.20837. [DOI] [PubMed] [Google Scholar]
  6. Brett M, Anton J-L, Valabregue R, Poline J-B. Region of interest analysis using an SPM toolbox [abstract] Presented at the 8th International Conference on Functional Mapping of the Human Brain; June 2–6; Sendai Japan. 2002. Available on CD-ROM in NeuroImage 16. [Google Scholar]
  7. Buchanan TW, Etzel JA, Adolphs R, Tranel D. The influence of autonomic arousal and semantic relatedness on memory for emotional words. International journal of psychophysiology. 2006;61:26–33. doi: 10.1016/j.ijpsycho.2005.10.022. [DOI] [PubMed] [Google Scholar]
  8. Cahill L, Alkire M. Epinephrine enhancement of human memory consolidation: Interaction with arousal at encoding. Neurobiology of Learning and Memory. 2003;79:194–198. doi: 10.1016/s1074-7427(02)00036-9. [DOI] [PubMed] [Google Scholar]
  9. Cahill L, Haier RJ, Fallon J, Alkire MT, Tang C, Keator D, Wu J, McGaugh JL. Amygdala activity at encoding correlated with long-term, free recall of emotional information. PNAS. 1996;93:8016–21. doi: 10.1073/pnas.93.15.8016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cahill L, McGaugh JL. A novel demonstration of enhanced memory associated with emotional arousal. Conscious Cognition. 1995;4:410–421. doi: 10.1006/ccog.1995.1048. [DOI] [PubMed] [Google Scholar]
  11. Canli T, Zhao Z, Brewer J, Gabrieli JDE, Cahill L. Event-Related Activation in the human amygdala associates with later memory for individual emotional experience. The Journal of Neuroscience. 2000;20:99. doi: 10.1523/JNEUROSCI.20-19-j0004.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dale A. Optimal experimental design for event-related fMRI. Human Brain Mapping. 1999;8:109–114. doi: 10.1002/(SICI)1097-0193(1999)8:2/3&#x0003c;109::AID-HBM7&#x0003e;3.0.CO;2-W. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Deldin PA, Keller J, Gergen JA, Miller G. Cognitive bias and emotion in neuropsychological models of depression. Cognition and Emotion. 2001;15:787–802. [Google Scholar]
  14. Dewhurst SA, Parry LA. Emotionality, distinctiveness, and recollective experience. European Journal of Cognitive Psychology. 2000;12:541–551. [Google Scholar]
  15. Dickerson BC, Miller SL, Greve DN, Dale AM, Albert MS, Schacter DL, Sperling RA. Prefrontal-Hippocampal-Fusiform Activity During Encoding Predicts Intraindividual Differences in Free Recall Ability: An Event-Related Functional-Anatomic MRI Study. Hippocampus. 2007;17:1060–70. doi: 10.1002/hipo.20338. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Dolan RJ, Vuilleumier P. Amygdala automaticity in emotional processing. Annuals of the New York Academy of Science. 2003;985:348–355. doi: 10.1111/j.1749-6632.2003.tb07093.x. [DOI] [PubMed] [Google Scholar]
  17. Dolcos F, LaBar KS, Cabeza R. Dissociable effects of arousal and valence on prefrontal activity indexing emotional evaluation and subsequent memory: an event-related fMRI study. NeuroImage. 2004;23:64–74. doi: 10.1016/j.neuroimage.2004.05.015. [DOI] [PubMed] [Google Scholar]
  18. Doerksen S, Shimamura AP. Source memory enhancement for emotional words. Emotion. 2001;1:5–11. doi: 10.1037/1528-3542.1.1.5. [DOI] [PubMed] [Google Scholar]
  19. Fredrickson BL. Neural correlates of self-reflection. Philosophical Transactions: Biological Sciences. 2004;125:1808–1814. [Google Scholar]
  20. Garavan H, Pendergrass JC, Ross TJ, Stein EA, Risinger RC. Amygdala response to both positively and negatively valenced stimuli. Neuroreport. 2001;12:2779–2783. doi: 10.1097/00001756-200108280-00036. [DOI] [PubMed] [Google Scholar]
  21. Gasper K, Clore GL. Attending to the big picture: Mood and global versus local processing of visual information. Psychological Science. 2002;13:34–40. doi: 10.1111/1467-9280.00406. [DOI] [PubMed] [Google Scholar]
  22. Gilboa A, Winocur G, Grady CL, Hevenor SJ, Moscovitch M. Remembering our past: functional neuroanatomy of recollection of recent and very remote personal events. Cerebral Cortex. 2004;14:1214–25. doi: 10.1093/cercor/bhh082. [DOI] [PubMed] [Google Scholar]
  23. Glascher J, Adolphs R. Processing of the Arousal of Subliminal and Supraliminal Emotional Stimuli by the Human Amygdala. The Journal of Neuroscience. 2003;23:10274–82. doi: 10.1523/JNEUROSCI.23-32-10274.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hamann S. Cognitive and neural mechanisms of emotional memory. Trends in Cognitive Sciences. 2004;5:394–400. doi: 10.1016/s1364-6613(00)01707-1. [DOI] [PubMed] [Google Scholar]
  25. Hamann S, Ely TD, Hoffman JM, Kilts CD. Activation of the human amygdala in positive and negative emotion. Psychological Science. 2002;13:135–41. doi: 10.1111/1467-9280.00425. [DOI] [PubMed] [Google Scholar]
  26. Hamann S, Mao H. Positive and negative emotional verbal stimuli elicit activity in the left amygdala. Neuroreport. 2002;13:15–19. doi: 10.1097/00001756-200201210-00008. [DOI] [PubMed] [Google Scholar]
  27. Hugh G, Pendergrass JC, Ross TJ, Stein EA, Risinger RC. Amygdala response to both positively and negatively valenced stimuli. Neuropharmacology and Neurotoxicology. 2001;12:2779–83. doi: 10.1097/00001756-200108280-00036. [DOI] [PubMed] [Google Scholar]
  28. Isenberg N, Silbersweig D, Engelien A, Emmerich S, Malavade K, Beattie B, Leo AC, Stern E. Linguistic threat activates the human amygdala. PNAS. 1999;96:10456–9. doi: 10.1073/pnas.96.18.10456. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Johnson MK, Foley MA, Suengas AG, Raye CL. Phenomenal characteristics of memories for perceived and imagined autobiographical events. Journal of Experimental Psychology: General. 1988;117:371–376. [PubMed] [Google Scholar]
  30. Joreskog KG, Sorbom D. LISREL 8: Structural equation modeling with the SIMPLIS command language. Hillsdale, NJ: Earlbaum Associates; 1993. [Google Scholar]
  31. Kensinger EA. Remembering emotional experiences: The contribution of valence and arousal. Reviews in the Neurosciences. 2004;15:241–251. doi: 10.1515/revneuro.2004.15.4.241. [DOI] [PubMed] [Google Scholar]
  32. Kensinger EA, Corkin S. Memory enhancement for emotional words: Are emotional words more vividly remembered than neutral words? Memory and Cognition. 2003;31:1169–1180. doi: 10.3758/bf03195800. [DOI] [PubMed] [Google Scholar]
  33. Kensinger EA, Garoff-Eaton RJ, Schacter DL. How negative emotion enhances the visual specificity of a memory. Journal of Cognitive Neuroscience. 2007;19:1872–1887. doi: 10.1162/jocn.2007.19.11.1872. [DOI] [PubMed] [Google Scholar]
  34. Kensinger EA, Leclerc CM. Age-related changes in the neural mechanisms supporting emotion processing and emotional memory. European Journal of Cognitive Psychology. 2009;21:192–215. [Google Scholar]
  35. Kensinger EA, Schacter DL. When the Red Sox shocked the Yankees: Comparing negative and positive memories. Psychonomic Bulletin and Review. 2006;13:757–763. doi: 10.3758/bf03193993. [DOI] [PubMed] [Google Scholar]
  36. Kensinger EA, Schacter DL. Amygdala activity is associated with the successful encoding of item, but not source, information for positive and negative stimuli. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience. 2006b;269:2564–2570. doi: 10.1523/JNEUROSCI.5241-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Kensinger EA, Schacter DL. Remembering the specific visual details of presented objects: Neuroimaging evidence for effects of emotion. Neuropsychologia. 2007;45:2951–2962. doi: 10.1016/j.neuropsychologia.2007.05.024. [DOI] [PubMed] [Google Scholar]
  38. Kilpatrick L, Cahill L. Amygdala modulation of parahippocampal and frontal regions during emotionally influenced memory storage. Neuroimage. 2003;20:2091–99. doi: 10.1016/j.neuroimage.2003.08.006. [DOI] [PubMed] [Google Scholar]
  39. Kluver H, Bucy PC. “Psychic blindness” and other symptoms following bilateral temporal lobectomy in rhesus monkeys. American Journal of Physiology. 1937;119:352–53. [Google Scholar]
  40. LaBar KS, Cabeza R. Cognitive neuroscience of emotional memory. Nature Reviews Neuroscience. 2006;7:54–64. doi: 10.1038/nrn1825. [DOI] [PubMed] [Google Scholar]
  41. Lane RD, Chua PML, Dolan RJ. Common effects of emotional valence, arousal and attention on neural activation during visual processing of pictures. Neuropsychologia. 1999;37:989–997. doi: 10.1016/s0028-3932(99)00017-2. [DOI] [PubMed] [Google Scholar]
  42. Lang PJ, Bradley MM, Cuthbert BN. International affective picture system IAPS: Technical manual and affective ratings. Gainesville, FL: The Center for Research in Psychophysiology; 1999. [Google Scholar]
  43. Lewis PA, Critchley HD, Rotshtein P, Dolan RJ. Neural correlates of processing valence and arousal in affective words. Cerebral Cortex. 2007;17:742–8. doi: 10.1093/cercor/bhk024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. McIntosh AR, Gonzalez-Lima F. Structural equation modeling and its application to network analysis in functional brain imaging. Human Brain Mapping. 1994;2:2–22. [Google Scholar]
  45. McGaugh JL. The amygdala modulates the consolidation of memories of emotionally arousing experiences. Annual Review of Neuroscience. 2004;27:1–28. doi: 10.1146/annurev.neuro.27.070203.144157. [DOI] [PubMed] [Google Scholar]
  46. Mickley KR, Kensinger EA. Emotional Valence Influences the Neural Correlates Associated with Remembering and Knowing. Cognitive, Affective, and Behavioral Neuroscience. 2008;8:143–152. doi: 10.3758/cabn.8.2.143. [DOI] [PubMed] [Google Scholar]
  47. Mickley Steinmetz KR, Kensinger EA. The effects of valence and arousal on the neural activity leading to subsequent memory. Psychophysiology. 2009;46:1190–1199. doi: 10.1111/j.1469-8986.2009.00868.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Morris JS, Friston KJ, Buchel C, Frith CD, Young AW, Calder AJ, Dolan RJ. A neuromodulatory role for the human amygdala in processing emotional facial expressions. Brain. 1998;121:47–57. doi: 10.1093/brain/121.1.47. [DOI] [PubMed] [Google Scholar]
  49. Muscatell KA, Addis DR, Kensinger EA. Self-Involvement Modulates the Effective Connectivity of the Autobiographical Memory Network. Social, Cognitive & Affective Neuroscience. 2010 doi: 10.1093/scan/nsp043. In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Phan KL, Wager T, Taylor SF, Liberzon I. Functional Neuroanatomy of Emotion: A Meta-Analysis of Emotion Activation Studies in PET and fMRI. NeuroImage. 2002;16:331–348. doi: 10.1006/nimg.2002.1087. [DOI] [PubMed] [Google Scholar]
  51. Richardson MP, Strange BA, Dolan RJ. Encoding of emotional memories depends on amygdala and hippocampus and their interactions. Nature Neuroscience. 2004;7:278–285. doi: 10.1038/nn1190. [DOI] [PubMed] [Google Scholar]
  52. Rosene DL, Van Hoesen GW. Hippocampal efferents reach widespread areas of cerebral cortex and amygdala in the rhesus monkey. Science. 1977;198:315–7. doi: 10.1126/science.410102. [DOI] [PubMed] [Google Scholar]
  53. Stein JL, Wiedholz LM, Bassett DS, Weinberger DR, Zink CF, Mattay VS, Meyer-Lindenberg A. A validated network of effective amygdala connectivity. NeuroImage. 2007;36:736–45. doi: 10.1016/j.neuroimage.2007.03.022. [DOI] [PubMed] [Google Scholar]
  54. Tabert MH, Borod JC, Tang CY, Lange G, Tsechung CW, et al. Differential amygdala activation during emotional decision and recognition memory tasks using unpleasant words: an fMRI study. Neuropsychologica. 2001;39:556–573. doi: 10.1016/s0028-3932(00)00157-3. [DOI] [PubMed] [Google Scholar]
  55. Tagami K. Negative bias on self-referent processing in depression: focused on mood congruent effects. Shinrigaku Kenkyu. 2002;73:412–8. doi: 10.4992/jjpsy.73.412. [DOI] [PubMed] [Google Scholar]
  56. Talairach J, Tournoux P. In: Co-planar stereotaxic atlas of the human brain: 3-dimensional proportional system: An approach to cerebral imaging. Rayport M, translator. New York: Thieme; 1988. [Google Scholar]
  57. Tranel D, Gullickson G, Koch M, Adolphs R. Altered experience of emotion following bilateral amygdala damage. Cognitive Neuropsychiatry. 2006;11:219–232. doi: 10.1080/13546800444000281. [DOI] [PubMed] [Google Scholar]
  58. Vuilleumier P, Armony JL, Clarke K, Husain M, Driver J, Dolan RJ. Neural response to emotional faces with and without awareness: event-related fMRI in a parietal patient with visual extinction and spatial neglect. Neuropsychologia. 2002;40:1–11. doi: 10.1016/s0028-3932(02)00045-3. [DOI] [PubMed] [Google Scholar]
  59. Vuilleumier P, Armony JL, Driver J, Dolan RJ. Effects of attention and emotion on face processing in the human brain: An event-related fMRI study. Neuron. 2001;30:829–841. doi: 10.1016/s0896-6273(01)00328-2. [DOI] [PubMed] [Google Scholar]
  60. Vuilleumier P, Richardson MP, Armony JL, Driver J, Dolan RJ. Distant influences of amygdala lesion on visual cortical activation during emotional face processing. Nature Neuroscience. 2004;7:1271–1287. doi: 10.1038/nn1341. [DOI] [PubMed] [Google Scholar]
  61. Weiskrantz L. Behavioral changes associated with ablation of the amygdaloid complex in monkeys. Journal of Comparative Physiological Psychology. 1956;49:381–91. doi: 10.1037/h0088009. [DOI] [PubMed] [Google Scholar]
  62. Whalen PJ. Fear, vigilance, and ambiguity: initial neuroimaging studies of the human amygdala. Current Directions in Psychological Science. 1998;7:177–88. [Google Scholar]
  63. Zald DH. The human amygdala and the emotional evaluation of sensory stimuli. Brain Research: Brain Research Reviews. 2003;41:88–123. doi: 10.1016/s0165-0173(02)00248-5. [DOI] [PubMed] [Google Scholar]

RESOURCES