Effect of relevance on amygdala activation and association with the ventral striatum

NeuroImage 62 (2012) 95–101

Contents lists available at SciVerse ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

Effect of relevance on amygdala activation and association with the ventral striatum Olga Therese Ousdal a, b,⁎, Greg E. Reckless a, b, Andres Server c, Ole A. Andreassen a, b, Jimmy Jensen a, b, d a

Division of Mental Health and Addiction, Oslo University Hospital, Oslo, Norway Institute of Clinical Medicine, University of Oslo, Oslo, Norway c Department of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo, Norway d Department of Psychiatry and Psychotherapy, Charité Universitätsmedizin, Berlin, Germany b

a r t i c l e

i n f o

Article history: Accepted 16 April 2012 Available online 23 April 2012 Keywords: Amygdala Ventral striatum Relevance Emotion fMRI

a b s t r a c t While the amygdala historically has been implicated in emotional stimuli processing, recent data suggest a general role in parceling out the relevance of stimuli, regardless of their emotional properties. Using functional magnetic resonance imaging, we tested the relevance hypothesis by investigating human amygdala responses to emotionally neutral stimuli while manipulating their relevance. The task was operationalized as highly relevant if a subsequent opportunity to respond for a reward depended on response accuracy of the task, and less relevant if the reward opportunity was independent of task performance. A region of interest analysis revealed bilateral amygdala activations in response to the high relevance condition compared to the low relevance condition. An exploratory whole-brain analysis yielded robust similar results in bilateral ventral striatum. A subsequent functional connectivity analysis demonstrated increased connectivity between amygdala and ventral striatum for the highly relevant stimuli compared to the less relevant stimuli. These findings suggest that the amygdala's processing profile goes beyond detection of emotions per se, and directly support the proposed role in relevance detection. In addition, the findings suggest a close relationship between amygdala and ventral striatal activity when processing relevant stimuli. Thus, the results may indicate that human amygdala modulates ventral striatum activity and subsequent behaviors beyond that observed for emotional cues, to encompass a broader range of relevant stimuli. © 2012 Elsevier Inc. All rights reserved.

Introduction Historically, the amygdala was assigned to the limbic system based on its connections mainly to the hypothalamus and the brainstem. However, a surge of neuroanatomical studies since this early notion has documented that the amygdala has a wide-reaching network of connections and, in fact, is one of the most densely interconnected areas in the primate forebrain (Davis and Whalen, 2001). Equivalently, the reported range of stimuli that activate the amygdala has gradually expanded. Initial studies suggested that the amygdala was specialized for detection of threat signals based on its involvement in fear conditioning (LeDoux, 2000) and the reduced fear observed in lesioned nonhuman primates (Prather et al., 2001). However, this valence-specific conceptualization was challenged by findings relating amygdala to positive affect and reward related learning as well (Baxter and Murray, 2002). Moreover, the human amygdala has been reported to be engaged by non-emotional stimuli of novel (Schwartz et al., 2003), arousing (Anderson et al., 2003) or ambiguous (Hsu et al., 2005; Whalen et ⁎ Corresponding author at: Psychosis Research Section — TOP, Building 49, Division of Mental Health and Addiction, Oslo University Hospital, Kirkeveien 166, N-0407 Oslo, Norway. Fax: + 47 23 02 73 33. E-mail address: [email protected] (OT. Ousdal). 1053-8119/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2012.04.035

al., 1998) character, suggesting a functional specialization beyond emotion per se. The amygdala response to various sensory stimuli is not static, but rather sensitive to current goals (Cunningham et al., 2008), internal states (LaBar et al., 2001), personal traits (Hariri, 2009) and context (Hindi Attar et al., 2010). Consequently, amygdala responses to a cue signaling food or a positive face vary according to state of satiety (LaBar et al., 2001) or individual extroversion score (Canli et al., 2002), respectively. Given the wide variety of stimuli that activate the amygdala, defining the precise operating characteristics of this region has not been straightforward. One intriguing hypothesis, put forward by Sander et al. (2003), suggested amygdala to be a “relevance detector” responding primarily to stimuli which are of central importance to the organism and its wellbeing (Sander et al., 2003). Notably, the relevance hypothesis entitles the amygdala with a more general evaluation function than several of the previous theories, and thus integrates many of the diverse findings from animal and human studies which have been difficult to reconcile with existing hypothesis. Importantly, relevant events include both emotional and non-emotional ones, with the amygdala activity reflecting their relevancy, regardless of their emotional valence (Santos et al., 2011). Moreover, the proposal stresses how context and current goals may shape the relevance and thus amygdala response to a stimulus within a personal situation. For example, amygdala may respond to an emotional neutral face (Wright and Liu, 2006), a simple geometrical

96

OT. Ousdal et al. / NeuroImage 62 (2012) 95–101

shape (Laurens et al., 2005) or even a neutral auditory stimulus (Kiehl et al., 2001a, 2001b) if relevant for the task. The relevance theory has recently been indirectly supported by neuroimaging studies in humans (Hindi Attar et al., 2010; Ousdal et al., 2008; Santos et al., 2011), but there is still a need for a direct operationalization of this hypothesis separating relevancy from other task qualities that activate the amygdala. According to Sander's definition, “an event is relevant if it can significantly (positively or negatively) influence the attainment of his or her goals” (Sander et al., 2003). The contention that amygdala responds to relevancy is compatible with the surge of work linking amygdala to for instance emotion (Phelps and LeDoux, 2005; Zald, 2003) and reward (Baxter and Murray, 2002; Murray, 2007). However, as much of this work did not separate relevance from the emotional or rewarding qualities of the stimuli, the stimulus quality that engages the amygdala still remains unresolved. In an attempt to directly test this hypothesis we used a functional magnetic resonance imaging (fMRI) paradigm where the opportunity to respond for a reward (the goal) was either contingent or not contingent on the response accuracy in a preceding task. Importantly, the reception of reward was separated from the task of interest, which was a neutral stimulus–response task. We hypothesized that where opportunity for reward was dependent on response accuracy of the task, the task would be more relevant as compared to when reward opportunity was unrelated to performance. Consequently, neural activity in amygdala, which has been implicated in relevance encoding, would reflect relevancy, with enhanced neural responses for high relevance compared to low relevance conditions. Materials and methods Subjects Twenty-five healthy subjects (mean age ± SD = 27 ± 5 years; 14 women) were recruited for the study in accordance with local ethics committee guidelines and provided written informed consent. Before participating in the study, subjects were screened to exclude somatic and psychiatric illness, substance abuse, MRI-incompatibility or serious head trauma. Subjects were paid a minimum of 300 NOK for their participation (150 NOK for a screening interview and 150 NOK for the fMRI session) and they got to keep any additional money won in the experiment described below. fMRI task To address the hypothesis that amygdala may respond to relevance, a 4-choice stimulus–response task was created. The experiment consisted of two 6 min and 42 s runs. The paradigm rationale is presented in Fig. 1. Trials belonged to one of two conditions, Low relevance and High relevance. The conditions were randomly presented. Each trial consisted of two tasks: a relevance task followed by a reward receipt task. Trials were separated by a jittered inter-trial interval lasting 5 ± 2 s. In both conditions, the relevance task was characterized by the appearance of four white boxes against a black background. Four sequentially presented colored circles, each lasting 800 ms, appeared in the boxes, in a randomized order. Only one circle appeared at a time. The color of the circles varied according to the two conditions, so that the circle appeared in black in Low relevance trials and in purple in High relevance trials, respectively. The participant had to press the key corresponding to the box in which the circle appeared. The relevance task was followed by reward receipt task, where a number, corresponding to the amount the subject could win in NOK (i.e. 0 NOK or 5 NOK; 1 NOK equals approx. 0.17 USD), appeared in one of the four boxes for 2.0 s. A response terminated the task. Reward was received when the participant correctly indicated in which box the number appeared. A jittered interstimulus interval lasting 3.5±1.5 s separated the relevance task and the reward receipt task.

Fig. 1. The paradigm. The subject was instructed to respond with the key corresponding to the box in which a stimulus (i.e. circle or number) appeared. In the relevance task, four sequentially presented circle stimuli appeared for 800 ms each in one of the boxes. Only one circle appeared at a time. The participants had to respond while each circle stimulus was presented on the screen in order to obtain a correct response. Then a jittered pause lasting 3.5 + 1.5 s followed. After the pause, a number appeared in one of the boxes for 2.0 s, the number representing the actual monetary reward in NOK which was either 0 NOK or 5 NOK. A correct key press had to be made in order to receive the reward.

The subjects were given verbal instructions prior to the scanning, and they did a practice version of the experiment. They were told about the two possible colors of the circles, and the consequences of a wrong response for the black and purple circles, respectively. In the High relevance condition, reward opportunity in the reward receipt task was contingent on accuracy in the relevance task. If the participant did not successfully indicate the location of all four stimuli, they would not receive an opportunity to respond for reward (i.e. 5 NOK). This was indicated by the appearance of 0 NOK in one of the boxes. In contrary, in the Low relevance condition, the opportunity to respond for reward (i.e. 5 NOK) was independent of performance on the relevance task and was instead presented in 80% of the trials. The 80% distribution was chosen based on the accuracy for both the Low and High relevance conditions in a pilot version of the task, thus rewarding trials were fairly balanced across conditions. For the remaining 20% of trials, 0 NOK appeared in one of the boxes. Consequently, the only difference between the High and Low relevance conditions was whether response accuracy was necessary for a subsequent chance to respond for reward. In each run 14 trials of the High and 14 trials of the Low relevance condition was presented, giving a total of 28 trials for each condition. Apparatus E-prime software (Psychology Software Tools, Inc.; Pittsburgh, PA, USA) controlled stimuli presentation. In the scanner, stimuli were presented using VisualSystem (NordicNeuroLab, Bergen, Norway) and responses collected using ResponseGrip (NordicNeuroLab, Bergen, Norway). Image acquisition Functional MRI scans were acquired by a 3 T scanner (General Electric Company; Milwaukee, WI, USA) supplied with a standard eightchannel head coil. A total of 192 volumes were acquired for each session, using a T2*-weighted EPI sequence sensitive to the BOLD contrast (TR = 2000 ms, TE = 25 ms, flip angle 90°, 260 mm × 260 mm field of view, 64 × 64 matrix). The first 3 volumes were discarded as dummies to ensure homologous tissue magnetization. Each volume consisted of 36 slices covering the whole brain acquired parallel to the AC–PC plane (sequential acquisition; 3.5 mm thick with a 0.5 mm gap). In order to better localize our findings, FSPGR T1-weighted anatomical

OT. Ousdal et al. / NeuroImage 62 (2012) 95–101

images (TR = 7.7 ms, TE = 3.0 ms, flip angle 12°) were acquired for coregistration prior to the functional imaging. Behavioral data analysis The behavioral data was analyzed using SPSS (Statistical Package for Social Sciences 16.0. SPSS Inc., Chicago, USA). To investigate whether there were differences between the High and Low relevance conditions of the relevance task for response time or response accuracy, paired-sample t-tests were performed.

97

individual t-contrast images of the interaction gained from the PPI were then entered into a random effects one-sample t-test. Due to the robust activations of ventral striatum in the second-level analysis, we applied small volume correction using an anatomically defined right ventral striatum mask. Since an appropriate ventral striatum ROI was not available in the PickAtlas toolbox, a mask created by Nielsen et al. (2004) using probability density estimates from the BrainMap database was used (Fox and Lancaster, 1994). Results

fMRI data analysis

Behavioral results

The images were visually inspected for signal dropout in the amygdala as this area is prone to magnetic susceptibility. None of the participants had to be excluded due to signal dropout. All the functional MRI volumes were preprocessed and analyzed using SPM8 software package (http://www.fil.ion.ucl.ac.uk/spm). All volumes were unwarped and realigned to the first volume (Friston et al., 1995) and the anatomical image co-registered to the mean functional. No participants moved more than 3 mm in any direction. The images were spatially normalized to the Montreal Neurological Institute (MNI) template (Evans et al., 1992), resampled at 3 × 3 × 3 mm and smoothed using an 8 mm full width half maximum (FWHM) isotropic kernel. Data were high-pass filtered using a cut-off value of 128 s. The model was built by convolving short boxcar functions for the onsets of the relevance task with a canonical hemodynamic response function. The duration of the task was 3200 ms (i.e. four circles, each with a fixed duration of 800 ms). The High and Low relevance conditions were modeled as separate regressors, and our contrast of interest was the High relevance condition > Low relevance condition. The onsets for the reward stimuli (i.e. 5 NOK and 0 NOK) as part of the reward task were modeled as regressors of no interest. Also, trials in which one or more wrong responses were given in either the High or Low relevance condition were modeled as a separate regressor. The individual contrast images were moved up to a second-level random effects model. To correct for multiple comparisons, whole-brain family-wise error (FWE) correction was used (FWEb 0.05, k > 10 voxels). In addition, as we had an a priori hypothesis regarding the amygdala, small volume correction based on anatomically defined bilateral amygdala region of interest (ROI) and FWE corrected p-values was used. The anatomically defined ROIs were created using the SPM Wake Forest University (WFU) PickAtlas toolbox (http://www.fmri.wfubmc.edu/cms/ software, version 2.3) (Maldjian et al., 2003).

In order to be included in the analyses, accuracy in each condition (i.e. High and Low relevance) had to be above 50%. Three subjects were discarded due to behavioral performances not fulfilling these criteria. The remaining twenty-two subjects successfully completed the task (total accuracy: 80.9 ± 15.7%) and scanning procedure. The response time and accuracy by condition are presented in Table 1. Paired-sample t-tests revealed no significant differences in response time (t = 0.9) or accuracy (t = 1.3) between the conditions of the relevance task.

Psychophysiological interaction (PPI) analysis In addition to the whole-brain random effects analysis, a supplementary psychophysiological interaction (PPI) analysis (Friston et al., 1997) was performed with amygdala as a seed region. Based on the results from the second-level analysis, it was hypothesized that amygdala and ventral striatum would be more functionally connected during processing of High relevance stimuli relative to Low relevance stimuli. For each subject¸ mean corrected activity was extracted from volumes of interest (first eigenvariate from the activated voxels within the anatomically (WFU PickAtlas (Maldjian et al., 2003)) defined right amygdala. Right amygdala was chosen based on the proposed predominance of the right hemisphere in detection of behavioral relevant stimuli (Mormann et al., 2011; Vallortigara and Rogers, 2005). The psychological variable represented the contrast between High and Low relevance conditions. The aim was to test for differences in regression slopes for two levels of relevance (i.e. High and Low relevance) as a measure of difference in regional connectivity (i.e. between seed region and other areas). To test for this, we generated a GLM in which the explanatory variable was the interaction term, and the main effects of time-course, the task regressors and the motion regressors were included as covariates of no interest. The

Imaging results A priori ROI analyses yielded bilateral significant amygdala responses for the High vs. Low relevance conditions (peak coordinates: right amygdala; x = 24, y = −7, z = −20, Z= 3.14, p(SVC) =0.02, left amygdala; x = −24, y= −7, z = −23, Z = 2.87, p(SVC) = 0.04; Fig. 2). An additional whole-brain analysis for the same contrast yielded significant responses in bilateral ventral striatum (peak coordinates: right ventral striatum; x = 6, y= 14, z =-2, Z= 5.48, p(FWE) b 0.001, cluster size= 50 voxels, left ventral striatum; peak coordinates x = −9, y = 14, z = −5, Z= 5.00, p(FWE) = 0.008, cluster size = 25 voxels) as the only regions surviving whole-brain FWE correction (Fig. 3). To investigate if the coding of relevance by amygdala was reflected in the behavioral data, the individual differences in mean amygdala response to High vs. Low relevance conditions were correlated with the equivalent behavioral measures. Specifically, we extracted the individual beta values from the group peak voxel of right and left amygdala for the High and Low relevance conditions. Then we correlated the individual differences in amygdala response with the individual difference in response time and accuracy for the same contrast. The analysis revealed a trend in left (r=0.42, p=0.05) and right (r=0.36, p=0.10) amygdala for accuracy, but no significant results for reaction times. This might suggest that subjects with greater amygdala response to High vs. Low relevance conditions also demonstrated the greatest differences in accuracy between these conditions. Psychophysiological interaction (PPI) analysis The PPI-analysis with right amygdala as seed region revealed a significantly increased connectivity with right ventral striatum (peak voxel coordinate; x = 9, y = 8, z = −11, Z = 3.03, p(SVC) = 0.03) during the High relevance as compared to the Low relevance condition. The results are displayed in Fig. 4. In order to explore how the “coupling” differed between amygdala and ventral striatum for the two conditions, we extracted the individual beta values using the group level peak voxel for right ventral striatum and the individual peak voxel within the anatomically defined Table 1 Accuracy and response time by conditions in the relevance task.

High relevance condition Low relevance condition

Response time (ms)

Accuracy (%)

1576 ± 177 1565 ± 176

82.5 ± 15.3 79.4 ± 18.3

98

OT. Ousdal et al. / NeuroImage 62 (2012) 95–101

Fig. 2. BOLD fMRI activation in the amygdala for the contrast ‘High relevance > Low relevance’. (A) Statistical parametric maps (SPM) demonstrating significant small volume corrected (FWE) activations in bilateral amygdala. The colors refer to t-values as coded in the bar at the lower right corner (B) Beta values for the peak voxel in right amygdala for the conditions Low relevance and High relevance.

right amygdala ROI (WFU PickAtlas (Maldjian et al., 2003)). The results are displayed in Fig. 4. For the High relevance condition there was a trend positive correlation between right amygdala and right ventral striatum activity (r = 0.39, p = 0.08), while in the Low relevance condition no correlation appeared (r = 0.19, p = n.s.). Discussion The main finding of this study was a differential amygdala response to emotionally neutral stimuli depending on their relevance. The dependence on performance for a subsequent opportunity to respond for reward may have increased the relevance of the experimental task compared to when reward opportunity was unrelated to task performance. Since both conditions potentially had the same rewarding outcome, the findings support how relevant stimuli with no explicit emotional content may enhance amygdala neuronal responses. This is in line with the proposed role for the amygdala in encoding an event's relevance (Sander et al., 2003) and our previous results (Ousdal et al., 2008). In addition, robust responses in bilateral ventral striatum were obtained for the same contrast, and amygdala and ventral striatum were more functionally associated for the more relevant events. Based on their anatomical connections (Haber and Knutson, 2010), the current findings may represent amygdala modulation of ventral striatum responses and subsequent behaviors to relevant events. According to the relevance hypothesis of the amygdala, an event is relevant if it can significantly influence (positively or negatively) the attainment of one's goals or one's wellbeing (Sander et al., 2003). In the current task, the goal was to receive a monetary reward, and thus the relevance task was maximally relevant if opportunity to respond for reward was contingent on the tasks' response accuracy. In contrary, the task was minimally relevant if opportunity of reward was independent of experimental task performance. This manipulation

of relevance was reflected in the amygdala BOLD responses, supporting that amygdala processes a stimulus dimension related to relevance. In line with the present findings, Zink et al. (2004) have previously demonstrated significant amygdala activations when reception of a monetary reward was contingent upon subject's behavior compared to when the reward was received unrelated to the task. However, receiving a reward was not separated from the task of interest in that study, thus there is a possibility that amygdala responses to reward per se influenced the results. Moreover, an event may also be relevant by virtue of its importance in the ongoing experimental task, without any reward related to it. This is supported by two recent studies which demonstrate that the amygdala responds to target stimuli that require a specific behavioral response, and are embedded in a stream of non-target stimuli (Hindi Attar et al., 2010; Santos et al., 2011). Supposing the subjects' aim of optimal performance of the task, the targets represent events that can significantly influence overall performance and thus are more relevant during the experiment than the non-targets. The notion that amygdala encodes the relevance of stimuli and events was based on a collection of fMRI experiments in humans and neurophysiological recordings in monkeys indirectly demonstrating that amygdala responds to relevant as opposed to non-relevant events. Importantly, relevant stimuli includes, but is not limited to, emotional or rewarding ones which is supported by a surge of work showing that amygdala responds to emotional neutral stimuli of social relevance as well (Herry et al., 2007; Hsu et al., 2005; Schwartz et al., 2003). Though the relevance hypothesis has gained some indirect support (Hindi Attar et al., 2010; Ousdal et al., 2008; Santos et al., 2011; Wright and Liu, 2006), it has been difficult to operationalize so that the conditions compared only differ according to relevance. In the present study, the High and Low relevance conditions did not differ in motor requirements, visual appearance (except from color), emotional properties or reward receipt (i.e. none of them leads to directly receiving a reward). Therefore we believe the observed response differences

Fig. 3. BOLD fMRI responses in the bilateral ventral striatum obtained for the contrast ‘High relevance’ > ‘Low relevance’. (A) Statistical parametric maps (SPM) demonstrating whole-brain significant (FWE-corrected, p b 0.05) activations in bilateral ventral striatum. The colors refer to t-values as coded in the bar at the lower right corner. (B) Beta values for the peak voxel in right ventral striatum for the conditions Low relevance and High relevance.

OT. Ousdal et al. / NeuroImage 62 (2012) 95–101

99

Fig. 4. The results of the psychophysiological interaction (PPI) analysis. (A) Statistical parametric map (SPM) showing the region in ventral striatum that showed condition-specific BOLD signal changes with right amygdala activity. The results are corrected for multiple comparisons using small volume correction (FWE). The colors refer to t-values as coded in the bar at the lower right corner. (B) Scatter plot with regression lines demonstrating the significant interaction obtained with PPI. The solid line indicates the Low relevance condition, while the dotted line represents the High relevance condition. Further, the x-axis represents activity in the right amygdala (beta values) and the y-axis represents activity in right ventral striatum (beta values). In the High relevance condition there was a trend positive correlation between right amygdala and right ventral striatum activity (r = 0.39, p = 0.08). However, in the Low relevance condition, no correlation appeared (r = 0.19, p = n.s.).

reflect degree of relevance of the two conditions. Thus, the present results directly support the notion that encoding of relevance is an important amygdala function. The contention that amygdala responds to relevant stimuli is compatible with a wealth of animal and human data linking amygdala to reward learning (Baxter and Murray, 2002; Murray, 2007) and to identification of the emotional significance of stimuli and events (Zald, 2003). Evaluation of the stimulus' relevance in relation to a subject's goals or wellbeing is an essential part of identifying the emotional significance or reward value of an event (Sander et al., 2003). Consequently, the amygdala activity elicited by an emotional or rewarding stimulus is probably more closely tied to its relevance in predicting significant outcomes than to its emotional or rewarding qualities per se. For example, the amygdala response to an angry face varies according to its prediction of an aversive electrical shock, reflecting a response to the relevance of the face, and not just to the facial expression itself (Schiller et al., 2008). Similarly, the relevance of a food item increases whenever a subject desires it to satisfy its needs, hence the amygdala response elicited by a foodrelated stimulus and its reward value increases in states of hunger (LaBar et al., 2001). In summary, the amygdala is probably not necessary for all facets of emotion and reward, but for those aspects that amygdala is essential, evaluation of relevance may be vital. An alternative interpretation of the present findings is that participants found High relevance trials to be relatively more rewarding or had an overall greater value, in line with recent findings linking amygdala to relative value encoding (Morrison and Salzman, 2010; Paton et al., 2006). However, events that are highly relevant may also have a greater current value along a positive and negative continuum, thus making these two qualities difficult to disentangle. In order minimize any differences in value for the two conditions, the conditions were carefully matched according to absolute reward value and reward occurrence rate. The amygdala is also activated by reward outcome and in anticipation of rewards (Baxter and Murray, 2002; Murray, 2007). We do not think that the presently observed responses are attributable to receipt of reward per se. Our task aimed at separating brain responses to receipt of reward from presentation of the relevance task. In addition, the reward value after correct responses to the targets in the High relevance condition and after any response in the Low relevance condition was identical (i.e. 5 NOK). We have no reason to believe that the change of color itself triggered elevated attention and amygdala activity, based on findings from a previous study in our group (Ousdal et al., 2008). We therefore argue that any elevated attention is secondary to increased relevance of the purple compared to the black circle condition. In addition, there were no significant behavioral differences between the two conditions,

which would have been indicative of attentional differences (Lim et al., 2009). Interestingly, only the bilateral ventral striatum survived wholebrain correction for multiple comparisons for the contrast High relevance > Low relevance, indicating a strong and specific activation in ventral striatum in response to the current paradigm. It is possible that this response reflects a greater salience of the High relevance condition, in line with the ventral striatum's coding of salience (Jensen et al., 2003, 2007; Zink et al., 2003, 2004). Alternatively, as previous neuroimaging studies have demonstrated that action requirements dominate over reward expectation in the ventral striatum (Elliott et al., 2004; Guitart-Masip et al., 2011), this finding may represent the difference in response accuracy requirements for the two conditions. To date there are no known projections from the striatum to the amygdala in primates. However, the amygdala projects to both the neostriatum (i.e. the caudate and putamen) (Russchen et al., 1985) and ventral striatum (Fudge et al., 2002), and has thus been implicated in modulating responses in the latter (Ambroggi et al., 2008). For example, the amygdala response to reward-predictive cues precedes that of ventral striatum (Ambroggi et al., 2008) and an intact basolateral amygdala is necessary for the ventral striatum to generate the excitatory firing pattern that encodes such cues (Nicola et al., 2004). Moreover, the amygdala–ventral striatum projections facilitate reward related behavior initiated by nucleus accumbens in both animals (Stuber et al., 2011) and humans (Passamonti et al., 2009), supporting their strong interrelationship in guiding motivated behaviors. The psychophysiological interaction (PPI) analysis revealed that activity in right amygdala covaried more strongly with right ventral striatum when facing highly relevant events. There are at least two ways to interpret this result. The finding may reflect how amygdala both directly and indirectly, via prefrontal sites, modulates responsiveness in ventral striatum (Sesack and Grace, 2010) according to the aforementioned anatomical pathways. Though the relation between amygdala and ventral striatum in motivated behaviors has been thoroughly investigated (Ambroggi et al., 2008; Stuber et al., 2011), the type of information transferred from amygdala to ventral striatum is still debated. If the present results represent amygdalamodulation of ventral striatum, it suggests that during motivated behaviors, the amygdala supplies relevance information to the ventral striatum, which subsequently may use this information to modulate actions (Elliott et al., 2004). By such, highly relevant events encoded by the amygdala have the potential to influence both cognition (Phelps, 2006) and motor actions, due to this brain area's rich interconnectivity with cortical and subcortical targets. An alternative explanation is that

100

OT. Ousdal et al. / NeuroImage 62 (2012) 95–101

another independent neural system could encode the relevance and further modulate the amygdala–ventral striatum projections accordingly. Both the amygdala and the ventral striatum are densely innervated by dopaminergic neurons from the ventral tegmental area (VTA) (Bjorklund and Dunnett, 2007; Haber and Knutson, 2010). Though previously assumed that such dopaminergic neurons carry a reward signal, more recent reports expand this view to encompass a large category of salient events of novel, intense or aversive nature (Horvitz, 2000). Hence, dopamine release from VTA in response to relevant events may activate neural circuits including the amygdala and ventral striatum, with their downstream effect on cognition and behavior. However, a final conclusion based on data from a PPI analysis is not possible, thus these interpretations remain speculative. Though the present findings support interrelated amygdala–ventral striatum neural activity, it is important to distinguish between neural processes which normally are supported by the amygdala on one hand, and processes that require amygdala involvement on the other. Accordingly, the amygdala may modulate striatal responsiveness parallel to that observed for hippocampus in explicit memory processes (Phelps, 2004; Richter-Levin and Akirav, 2000). Alternatively, one could speculate that an intact amygdala is necessary for ventral striatum responses to motivational relevant events, parallel to what has been observed for rewardrelated cues (Nicola et al., 2004). However, this warrants further investigations in future studies. In summary, the results indicate that amygdala responds to motivational relevant stimuli without explicit emotional properties. The current findings suggest that the amygdala's computational profile goes beyond detection of threats and emotion per se, and supports the proposed role in relevance detection. In addition, the study supports the notion that there is a close relationship between amygdala and ventral striatum neural activity, which may be dysfunctional in numerous neuropsychiatric disorders. Hopefully, a growing understanding of amygdala's functions may increase our understanding of such neural circuits implicated in both normal behavior and psychopathology. Acknowledgments This work was supported by grants from the University of Oslo, the Research Council of Norway (#167153/V50, #163070/V50) and South East Norway Health Authority (#39386/6051). The authors would like to thank the TOP study group and Anne Hilde Farstad for assisting with data collection. References Ambroggi, F., Ishikawa, A., Fields, H.L., Nicola, S.M., 2008. Basolateral amygdala neurons facilitate reward-seeking behavior by exciting nucleus accumbens neurons. Neuron 59, 648–661. Anderson, A.K., Christoff, K., Stappen, I., Panitz, D., Ghahremani, D.G., Glover, G., Gabrieli, J.D., Sobel, N., 2003. Dissociated neural representations of intensity and valence in human olfaction. Nat. Neurosci. 6, 196–202. Baxter, M.G., Murray, E.A., 2002. The amygdala and reward. Nat. Rev. Neurosci. 3, 563–573. Bjorklund, A., Dunnett, S.B., 2007. Dopamine neuron systems in the brain: an update. Trends Neurosci. 30, 194–202. Canli, T., Sivers, H., Whitfield, S.L., Gotlib, I.H., Gabrieli, J.D., 2002. Amygdala response to happy faces as a function of extraversion. Science 296, 2191. Cunningham, W.A., Van Bavel, J.J., Johnsen, I.R., 2008. Affective flexibility: evaluative processing goals shape amygdala activity. Psychol. Sci. 19, 152–160. Davis, M., Whalen, P.J., 2001. The amygdala: vigilance and emotion. Mol. Psychiatry 6, 13–34. Elliott, R., Newman, J.L., Longe, O.A., William Deakin, J.F., 2004. Instrumental responding for rewards is associated with enhanced neuronal response in subcortical reward systems. Neuroimage 21, 984–990. Evans, A.C., Marrett, S., Neelin, P., Collins, L., Worsley, K., Dai, W., Milot, S., Meyer, E., Bub, D., 1992. Anatomical mapping of functional activation in stereotactic coordinate space. Neuroimage 1, 43–53. Fox, P.T., Lancaster, J.L., 1994. Neuroscience on the net. Science 266, 994–996. Friston, K.J., Holmes, A.P., Poline, J.B., Grasby, P.J., Williams, S.C., Frackowiak, R.S., Turner, R., 1995. Analysis of fMRI time-series revisited. Neuroimage 2, 45–53. Friston, K.J., Buechel, C., Fink, G.R., Morris, J., Rolls, E., Dolan, R.J., 1997. Psychophysiological and modulatory interactions in neuroimaging. Neuroimage 6, 218–229.

Fudge, J.L., Kunishio, K., Walsh, P., Richard, C., Haber, S.N., 2002. Amygdaloid projections to ventromedial striatal subterritories in the primate. Neuroscience 110, 257–275. Guitart-Masip, M., Fuentemilla, L., Bach, D.R., Huys, Q.J., Dayan, P., Dolan, R.J., Duzel, E., 2011. Action dominates valence in anticipatory representations in the human striatum and dopaminergic midbrain. J. Neurosci. 31, 7867–7875. Haber, S.N., Knutson, B., 2010. The reward circuit: linking primate anatomy and human imaging. Neuropsychopharmacology 35, 4–26. Hariri, A.R., 2009. The neurobiology of individual differences in complex behavioral traits. Annu. Rev. Neurosci. 32, 225–247. Herry, C., Bach, D.R., Esposito, F., Di Salle, F., Perrig, W.J., Scheffler, K., Luthi, A., Seifritz, E., 2007. Processing of temporal unpredictability in human and animal amygdala. J. Neurosci. 27, 5958–5966. Hindi Attar, C., Muller, M.M., Andersen, S.K., Buchel, C., Rose, M., 2010. Emotional processing in a salient motion context: integration of motion and emotion in both V5/ hMT+ and the amygdala. J. Neurosci. 30, 5204–5210. Horvitz, J.C., 2000. Mesolimbocortical and nigrostriatal dopamine responses to salient non-reward events. Neuroscience 96, 651–656. Hsu, M., Bhatt, M., Adolphs, R., Tranel, D., Camerer, C.F., 2005. Neural systems responding to degrees of uncertainty in human decision-making. Science 310, 1680–1683. Jensen, J., McIntosh, A.R., Crawley, A.P., Mikulis, D.J., Remington, G., Kapur, S., 2003. Direct activation of the ventral striatum in anticipation of aversive stimuli. Neuron 40, 1251–1257. Jensen, J., Smith, A.J., Willeit, M., Crawley, A.P., Mikulis, D.J., Vitcu, I., Kapur, S., 2007. Separate brain regions code for salience vs. valence during reward prediction in humans. Hum. Brain Mapp. 28, 294–302. Kiehl, K.A., Laurens, K.R., Duty, T.L., Forster, B.B., Liddle, P.F., 2001a. An eventrelated fMRI study of visual and auditory oddball tasks. J. Psychophysiol. 15, 221–240. Kiehl, K.A., Laurens, K.R., Duty, T.L., Forster, B.B., Liddle, P.F., 2001b. Neural sources involved in auditory target detection and novelty processing: an event-related fMRI study. Psychophysiology 38, 133–142. LaBar, K.S., Gitelman, D.R., Parrish, T.B., Kim, Y.H., Nobre, A.C., Mesulam, M.M., 2001. Hunger selectively modulates corticolimbic activation to food stimuli in humans. Behav. Neurosci. 115, 493–500. Laurens, K.R., Kiehl, K.A., Liddle, P.F., 2005. A supramodal limbic–paralimbic–neocortical network supports goal-directed stimulus processing. Hum. Brain Mapp. 24, 35–49. LeDoux, J., 2000. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, 155–184. Lim, S.L., Padmala, S., Pessoa, L., 2009. Segregating the significant from the mundane on a moment-to-moment basis via direct and indirect amygdala contributions. Proc. Natl. Acad. Sci. U.S.A. 106, 16841–16846. Maldjian, J.A., Laurienti, P.J., Kraft, R.A., Burdette, J.H., 2003. An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. Neuroimage 19, 1233–1239. Mormann, F., Dubois, J., Kornblith, S., Milosavljevic, M., Cerf, M., Ison, M., Tsuchiya, N., Kraskov, A., Quiroga, R.Q., Adolphs, R., Fried, I., Koch, C., 2011. A category-specific response to animals in the right human amygdala. Nat. Neurosci. 14, 1247–1249. Morrison, S.E., Salzman, C.D., 2010. Re-valuing the amygdala. Curr. Opin. Neurobiol. 20, 221–230. Murray, E.A., 2007. The amygdala, reward and emotion. Trends Cogn. Sci. 11, 489–497. Nicola, S.M., Yun, I.A., Wakabayashi, K.T., Fields, H.L., 2004. Cue-evoked firing of nucleus accumbens neurons encodes motivational significance during a discriminative stimulus task. J. Neurophysiol. 91, 1840–1865. Nielsen, F.A., Hansen, L.K., Balslev, D., 2004. Mining for associations between text and brain activation in a functional neuroimaging database. Neuroinformatics 2, 369–380. Ousdal, O.T., Jensen, J., Server, A., Hariri, A.R., Nakstad, P.H., Andreassen, O.A., 2008. The human amygdala is involved in general behavioral relevance detection: evidence from an event-related functional magnetic resonance imaging Go–NoGo task. Neuroscience 156, 450–455. Passamonti, L., Rowe, J.B., Schwarzbauer, C., Ewbank, M.P., von dem Hagen, E., Calder, A.J., 2009. Personality predicts the brain's response to viewing appetizing foods: the neural basis of a risk factor for overeating. J. Neurosci. 29, 43–51. Paton, J.J., Belova, M.A., Morrison, S.E., Salzman, C.D., 2006. The primate amygdala represents the positive and negative value of visual stimuli during learning. Nature 439, 865–870. Phelps, E.A., 2004. Human emotion and memory: interactions of the amygdala and hippocampal complex. Curr. Opin. Neurobiol. 14, 198–202. Phelps, E.A., 2006. Emotion and cognition: insights from studies of the human amygdala. Annu. Rev. Psychol. 57, 27–53. Phelps, E.A., LeDoux, J.E., 2005. Contributions of the amygdala to emotion processing: from animal models to human behavior. Neuron 48, 175–187. Prather, M.D., Lavenex, P., Mauldin-Jourdain, M.L., Mason, W.A., Capitanio, J.P., Mendoza, S.P., Amaral, D.G., 2001. Increased social fear and decreased fear of objects in monkeys with neonatal amygdala lesions. Neuroscience 106, 653–658. Richter-Levin, G., Akirav, I., 2000. Amygdala–hippocampus dynamic interaction in relation to memory. Mol. Neurobiol. 22, 11–20. Russchen, F.T., Bakst, I., Amaral, D.G., Price, J.L., 1985. The amygdalostriatal projections in the monkey. An anterograde tracing study. Brain Res. 329, 241–257. Sander, D., Grafman, J., Zalla, T., 2003. The human amygdala: an evolved system for relevance detection. Rev. Neurosci. 14, 303–316. Santos, A., Mier, D., Kirsch, P., Meyer-Lindenberg, A., 2011. Evidence for a general face salience signal in human amygdala. Neuroimage 54, 3111–3116. Schiller, D., Levy, I., Niv, Y., LeDoux, J.E., Phelps, E.A., 2008. From fear to safety and back: reversal of fear in the human brain. J. Neurosci. 28, 11517–11525.

OT. Ousdal et al. / NeuroImage 62 (2012) 95–101 Schwartz, C.E., Wright, C.I., Shin, L.M., Kagan, J., Rauch, S.L., 2003. Inhibited and uninhibited infants “grown up”: adult amygdalar response to novelty. Science 300, 1952–1953. Sesack, S.R., Grace, A.A., 2010. Cortico-basal ganglia reward network: microcircuitry. Neuropsychopharmacology 35, 27–47. Stuber, G.D., Sparta, D.R., Stamatakis, A.M., van Leeuwen, W.A., Hardjoprajitno, J.E., Cho, S., Tye, K.M., Kempadoo, K.A., Zhang, F., Deisseroth, K., Bonci, A., 2011. Excitatory transmission from the amygdala to nucleus accumbens facilitates reward seeking. Nature 475, 377–380. Vallortigara, G., Rogers, L.J., 2005. Survival with an asymmetrical brain: advantages and disadvantages of cerebral lateralization. Behav. Brain Sci. 28, 575–589 (discussion 589–633).

101

Whalen, P.J., Rauch, S.L., Etcoff, N.L., McInerney, S.C., Lee, M.B., Jenike, M.A., 1998. Masked presentations of emotional facial expressions modulate amygdala activity without explicit knowledge. J. Neurosci. 18, 411–418. Wright, P., Liu, Y., 2006. Neutral faces activate the amygdala during identity matching. Neuroimage 29, 628–636. Zald, D.H., 2003. The human amygdala and the emotional evaluation of sensory stimuli. Brain Res. Brain Res. Rev. 41, 88–123. Zink, C.F., Pagnoni, G., Martin, M.E., Dhamala, M., Berns, G.S., 2003. Human striatal response to salient nonrewarding stimuli. J. Neurosci. 23, 8092–8097. Zink, C.F., Pagnoni, G., Martin-Skurski, M.E., Chappelow, J.C., Berns, G.S., 2004. Human striatal responses to monetary reward depend on saliency. Neuron 42, 509–517.