Contextual fear conditioning in humans using feature-identical contexts

Neurobiology of Learning and Memory 121 (2015) 1–11

Contents lists available at ScienceDirect

Neurobiology of Learning and Memory journal homepage: www.elsevier.com/locate/ynlme

Contextual fear conditioning in humans using feature-identical contexts Christian Baeuchl ⇑,1, Patric Meyer 1, Michael Hoppstädter, Carsten Diener 2, Herta Flor Department of Cognitive and Clinical Neuroscience, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, Germany Bernstein Center for Computational Neuroscience Heidelberg/Mannheim, Germany

a r t i c l e

i n f o

Article history: Received 16 December 2014 Revised 19 February 2015 Accepted 8 March 2015 Available online 16 March 2015 Keywords: Associative learning Context Functional magnetic resonance imaging Hippocampus Psychophysiological interaction

a b s t r a c t Contextual fear conditioning studies in animals and humans found an involvement of the hippocampus and amygdala during fear learning. To exclude a focus on elements of the context we employed a paradigm, which uses two feature-identical contexts that only differ in the arrangement of the features and requires configural processing. We employed functional magnetic resonance imaging to determine the role of the hippocampus and neocortical areas during the acquisition of contextual fear in humans. For contextual fear acquisition, we paired one context (CS+) with an aversive electrical stimulus, whereas the other (CS ) was never followed by aversive stimulation. Blood oxygen level dependent activation to the CS+ was present in the insula, inferior frontal gyrus, inferior parietal lobule, superior medial gyrus and caudate nucleus. Furthermore, the amygdala and hippocampus were involved in a time-dependent manner. Psychophysiological interaction analyses revealed functional connectivity of a more posterior hippocampal seed region with the anterior hippocampus, posterior cingulate cortex and superior parietal lobule. The anterior hippocampus was functionally coupled with the amygdala and postcentral gyrus. This study complements previous findings in contextual fear conditioning in humans and provides a paradigm which might be useful for studying patients with hippocampal impairment. Ó 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction In fear conditioning, an initially neutral conditioned-stimulus (CS) is paired with an aversive unconditioned stimulus (US) that evokes fear or anxiety responses. Repeated pairings of the CS with the US result in an association of both stimuli that causes the occurrence of the CS alone to elicit an emotional response. While cue conditioning requires only a single feature to be associated with the US, contextual conditioning demands the association of the US with a whole set of features. Consequently, these two variants of classical fear conditioning also differ in the way in which the CS–US association occurs on a behavioral and neural level. The dual-systems theory provides a mechanistic framework for contextual representations in the mammalian brain (Nadel & Willner, 1980; Rudy & O’Reilly, 2001). According to this account, a single stimulus is thought to be represented in the neocortex and bound into an association with a threatening event in the amygdala (Fanselow, 2010; Rudy, 2009). Several co-occurring ⇑ Corresponding author at: Department of Cognitive and Clinical Neuroscience, Central Institute of Mental Health, J5, 68159 Mannheim, Germany. Fax: +49 621 1703 6305. E-mail address: [email protected] (C. Baeuchl). 1 These authors contributed equally to this work. 2 Present address: School of Applied Psychology, SRH University of Applied Sciences Heidelberg, Heidelberg, Germany.

stimuli, in contrast, first need to be consolidated into a hierarchical, conjunctive representation which necessitates the binding capacity of the hippocampus (Rudy, 2009). This representation is then transferred to the amygdala to drive the associative process. However, studies showed that lesioning of the hippocampus shortly after the learned CS–US association severely impairs the expression of contextual fear, whereas damage to the hippocampus prior to conditioning has little effect (Maren, Aharonov, & Fanselow, 1997; Wiltgen, Sanders, Anagnostaras, Sage, & Fanselow, 2006). These findings have led to the hypothesis that if the hippocampus is damaged, single cues, which are stored in the neocortex, still can represent the context. This is referred to as ‘elemental processing’ as opposed to the hippocampus-dependent ‘configural processing’ (Iordanova, Burnett, Aggleton, Good, & Honey, 2009). Configural or relational learning theories state that the formation of the representation of context relies on the integration of multiple cues into a unified or configural representation and it is assumed that the hippocampus plays a major role in this process (Eichenbaum, 2004; Moses & Ryan, 2006; Nadel & Willner, 1980; Sutherland & Rudy, 1989). However, in rats, hippocampal damage only seems to affect performance in those configural learning paradigms that require discrimination between visual scenes containing common elements (Albasser et al., 2013; Dumont, Petrides, & Sziklas, 2007; Sanderson, Pearce, Kyd, & Aggleton, 2006). Albasser et al. (2013) suggest that stimuli with

http://dx.doi.org/10.1016/j.nlm.2015.03.001 1074-7427/Ó 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

2

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11

common elements will be individually structured by binding together common cues in unique spatial ensembles. Hippocampal lesions can spare configural discriminations when item-location binding is not integral to the problem (Bussey, Warburton, Aggleton, & Muir, 1998; Sanderson et al., 2006; Saksida, Bussey, Buckmaster, & Murray, 2007). Amnesic patients compared to matched controls show deficits in reconstructing the spatial locations of a small array of objects after a short delay (Watson, Voss, Warren, Tranel, & Cohen, 2013). They were particularly impaired when two objects swapped places during the delay phase, which demanded object identity-to-relative-location bindings. A further study showed that hippocampal damage results in poor memory for the change in location of a single item embedded in a scene, even though the memory for the scene itself was intact (Hannula, Tranel, & Cohen, 2006). Similarly, Olson, Moore, Stark, and Chatterjee (2006) reported that amnesic patients had a specific deficit in remembering object-location conjunctions, while the memory for objects and individual locations was preserved. These results are consistent with the finding that hippocampal place fields show global remapping after the presentation of familiar cues in changed places (Leutgeb et al., 2005). In humans, previous contextual fear conditioning paradigms utilized virtual reality contexts (Alvarez, Biggs, Chen, Pine, & Grillon, 2008; Grillon, Baas, Cornwell, & Johnson, 2006), spatial picture contexts (Marschner, Kalisch, Vervliet, Vansteenwegen, & Büchel, 2008) or color background contexts (Lang et al., 2009; Pohlack et al., 2012a; Pohlack, Nees, Ruttorf, Schad, & Flor, 2012b) during fMRI. These studies did not focus on the question of elemental versus configural processing and thus did not employ stimulus material that included identical elements between the context scenes. This could lead to unclear results, especially in subjects with impaired hippocampal functioning, as these contextual stimuli could be processed without reverting to a configural, hippocampus-dependent strategy. To create an experimental conditioning scenario that requires configural processing we constructed a cue-array context paradigm that is comprised of two featureidentical picture stimuli, which are only differing in the arrangement of their context components. This paradigm should ensure that focusing on single elements is not a sufficient strategy to distinguish between the two context pictures and thus to predict the CS–US association. We expected that fear-related neocortical brain regions would be constantly active during acquisition, whereas learning-related regions in the medial temporal lobe should show an initial activation that would decrease over time (Büchel, Morris, Dolan, & Friston, 1998; Marschner et al., 2008). Furthermore, the coupling patterns of the hippocampus with other brain regions were of interest to delineate the contextual fear conditioning process, assuming that functional connections with regions involved in emotional (e.g. amygdala) as well as cognitive (e.g. parietal cortex) processing should emerge.

2. Materials and methods 2.1. Participants Seventeen healthy young adults participated in the study after giving written informed consent (8 male, age range: 22–36; mean age: 28.5 ± 3.52 SD). They were all right-handed and reported no history of mental or neurological disorders. Two participants were excluded from further data analysis due to their inability to identify which of the two picture-stimuli was actually associated with an aversive stimulus, leaving 15 participants (7 male) for the fMRI analysis. Due to technical problems during recording of skin conductance responses (SCR), the data of one participant were discarded, reducing the number of participants for the SCR analysis

to 14 (6 male). All participants were German native speaking university students or graduates. The study was approved by the Ethics Committee of the Medical Faculty Mannheim and adhered to the Declaration of Helsinki. 2.2. Experimental design The two context-picture stimuli were created using the virtual reality software NeuroVR (version 2.0; www.neurovr2.org) and depict a living room in which 4 elements (TV set on a cabinet, bookshelf, wall picture and a door) had a different spatial arrangement in picture one compared to picture two (Fig. 1). Three other elements (couch, chair and a floor lamp) remained stationary in both pictures. The experimental procedure in this event-related design consisted of three conditions: one picture that was never associated with an electric stimulus (CS ) and a second picture where a painful electric stimulus was pseudorandomly applied in 50 percent of the trials (CS+paired and CS+unpaired, respectively). The assignment of the pictures to CS+ and CS was counterbalanced between participants. The condition CS+unpaired was created to investigate hemodynamic responses evoked by the CS+ without the confounding effects of the US. Pictures were presented for the duration of four seconds and appeared in a pseudo-randomized order with every picture being shown 40 times during the entire experimental run. The same stimulus (e.g. CS+) occurred maximally three times in a row and the US was never administered in two consecutive trials. Inter-stimulus intervals were randomly jittered between 8 and 12 s resulting in trials of 12, 13, 14, 15 and 16 seconds length (Fig. 2). As a US we used an electric stimulus, which was administered to the right thumb via a pair of surface electrodes and occurred within an interval of 0.5–3.5 s during the presentation of the CS+. US onset was randomized within the described interval to ensure that participants perceived the occurrence of the US as unpredictable, a prerequisite for inducing anxiety in aversive context conditioning (Grillon, Baas, Lissek, Smith, & Milstein, 2004). The US consisted of a train of 6 electric pulses that were applied in a frequency of 12.2 Hz over the duration of 480 ms. US intensity was individually adjusted to be aversive but not too painful. The magnitude of the stimulation was initially set at 80 percent of the difference between the individually assessed pain threshold and pain tolerance level. The electric stimulus of this magnitude was then administered to the subject’s right thumb and had to be rated on painfulness and unpleasantness on a 9-point scale (from 1 = not painful/not unpleasant to 9 = very painful/very unpleasant). The magnitude of the stimulation was adjusted if ratings for painfulness and unpleasantness did not reach 7 or 8 points on both scales. Before the experiment started, participants were instructed to view the pictures attentively during the session while they would occasionally receive a painful stimulus. The net scanning time for a single subject session was 19 min. The experimental procedure included neither a habituation (presentation of CSs and US without pairing prior to acquisition) nor an extinction phase (presentation of CS+ and CS without delivery of US during CS+ after the acquisition phase). 2.3. Skin conductance response (SCR) Skin conductance was recorded continuously by two Ag/AgCl electrodes from the thenar and hypothenar of the left hand with a sampling rate of 5000 Hz. Before mounting of the electrodes, the skin was prepared with an isotonic saline solution (0.9 percent saline) and electrode paste was applied to the electrodes, which contained 0.5 percent saline in a neutral base. The signal was amplified using a BRAINAMP ExG MR device in combination with a GSR MR module (BRAIN PRODUCTS, Gilching, Germany).

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11

3

Fig. 1. The two pictures of a room used as contexts in the experiment. Both rooms contain the same cue-elements of which only four – TV set, bookshelf, door and painting – are arranged in a different constellation in context-picture 1 compared to context-picture 2. Thus the mere presence of cues in the pictures does not provide sufficient information to differentiate them. This is only possible if the relation of cues to each other is taken into account.

Fig. 2. The design was comprised of 3 conditions using two contextual stimuli: during the CS condition (40 trials) one of the contexts was never associated with aversive electrical stimulation (US), while in the CS+paired condition (20 trials), the second context was paired with the US and in the CS+unpaired condition (20 trials) the second context was presented without the US being administered. Each context stimulus presentation lasted for 4 s and the inter-stimulus interval varied randomly between 8 and 12 s (var. ISI).

2.4. MRI data acquisition MRI data were collected on a 3 Tesla Siemens MAGNETOM Trio scanner (Erlangen, Germany) using a 12 channel head coil. Functional images were obtained with a T2⁄ weighted echo-planar imaging sequence (40 axial slices, co-planar with AC–PC; TR = 2700 ms; TE = 30 ms; FA = 90°; FOV = 220  220 mm; matrix size = 96  96 mm; voxel size = 2.3  2.3  2.3 mm) in an interleaved order. Each functional scan resulted in 420 volumes of which the first 5 were discarded to allow for magnetic saturation. Additionally, T1 weighted anatomical (MP-RAGE) images were acquired (TR = 2300 ms; TE = 3 ms; FA = 9°; FOV = 240  256  192 mm; voxel size = 1  1  1 mm). The stimuli were presented using Presentation (version 14.9; Neurobehavioral Systems, Inc., Albany, USA).

2.5. Rating data and SCR analysis After the experiment, the participants rated the two contextual pictures on emotional valence and arousal using a 9-point scale ranging from ‘‘1’’ (very pleasant/not arousing) to ‘‘9’’ (very unpleasant/very arousing). In addition, the participants were asked about

the perceived likelihood that the US occurred during the presentation of each picture (contingency awareness), on a 9-point scale ranging from ‘‘1’’ (very unlikely) to ‘‘9’’ (very likely). The participants were classified as aware of the CS+/US contingency if they gave a contingency awareness rating for the CS+/US that was at least 50 percent higher than their rating for CS /US (difference > 4 points), all other participants were coded as unaware. This criterion is similar to that applied by Lovibond, Liu, Weidemann, and Mitchell (2011). Two participants who were unaware of the contingency were excluded from data analysis, leaving 15 participants (7 male). All ratings were given verbally by the participants. Paired t-tests were calculated for the ratings on CS+ and CS in the dimensions valence, arousal and contingency awareness. All results were considered to be significant if they surpassed a Bonferroni-corrected (two-tailed) threshold of p = 0.0166. The skin conductance response (SCR) was assessed as a peripheral indicator of conditioning (Boucsein et al., 2012), indicating sympathetic activation. Since the classic trough-to-peak method might underestimate phasic peaks in the raw SCR signal (Benedek & Kaernbach, 2010; Boucsein, 1992), we applied a continuous decomposition analysis (CDA) which is based on deconvolution of the original data into continuous tonic and phasic activity to reduce the possible impact of superposition effects (Benedek & Kaernbach, 2010). The data were downsampled to 10 Hz and CDA was performed using the software package Ledalab (version 3.3.0; http://www.ledalab.de/). Event-related SCRs were analyzed in a response window from 1 to 4 s after stimulus presentation, which denotes the first interval response (FIR). The conditioning literature often differentiates between FIR and second interval response (SIR), with SIR assumed to more reliably reflect conditioned responses (Stewart, Stern, Winokur, & Fredman, 1961). However, studies, which employed differential conditioning, suggest that there is no significant difference between FIR and SIR in their ability to detect conditioning effects (Pineles, Orr, & Orr, 2009). Extreme cases were excluded from further analyses (cut-off 3 SDs; 0.48% of the data). The magnitude of the SCRs was quantified for each subject using the time integral of the deconvoluted phasic activity over the whole response window. The SCRs were then logarithmically transformed in order to normalize the data (Boucsein et al., 2012) and CS+unpaired and CS trials were split into three non-overlapping time bins. Since CS+unpaired contained 20 and CS contained 40 trials, bin sizes were chosen to be 7 or 14 sample points for the first and last bin and 6 or 12 sample points for the second bin, respectively. The data were averaged within each time bin and the condition CS+unpaired was compared to CS using separate paired t-tests. All results were considered to be significant if they surpassed a Bonferronicorrected (two-tailed) threshold of p = 0.0166.

4

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11

2.6. Functional MRI preprocessing and BOLD activity analysis The functional data were preprocessed and analyzed using SPM 8 (Statistical Parametric Mapping; http://www.fil.ion.ucl.ac.uk/ spm/). First, the functional images were realigned to correct for head motion. Then the anatomical image was coregistered to the mean functional image and segmented into gray matter and white matter using the New Segment algorithm. The segmented images were used to normalize the functional images to the standard space of the Montreal Neurological Institute (ICBM 152 MNI template) via SPM’s DARTEL toolbox. Functional images were resampled to 1.5 mm cubic voxels and spatially smoothed (8 mm FWHM Gaussian kernel). Data that were used for the connectivity analyses were additionally slice-time corrected prior to the other preprocessing steps using the Fourier phase shifting interpolation of SPM. BOLD responses were analyzed within the framework of the general linear model (GLM). To this end, the time series of all conditions (CS+paired, CS+unpaired and CS ) were modeled as stick function regressors and convolved with the canonical hemodynamic response function and its time derivative, thus creating 6 regressors (3 canonical and 3 time derivative). The 3 canonical regressors depict BOLD responses that were relatively constant throughout the course of the experiment (sustained activity). Previous studies identified a decay of neural responses in the amygdala (Büchel et al., 1998; Quirk, Armony, & LeDoux, 1997) and hippocampus (Büchel, Dolan, Armony, & Friston, 1999; Marschner et al., 2008) during fear conditioning. Therefore, we created additional regressors by parametrically modulating the main effect regressors of our 3 conditions with a demeaned linear decaying function to obtain BOLD effects that decreased over time (transient activity). Additionally, the GLM design matrix included 6 motion parameters from the realignment step as covariates of no interest. The data were high-pass filtered with a cut-off of 128 s and corrected for temporal autocorrelation using the AR(1) model. For single subject analyses, contrast estimates were constructed such that they captured the neural responses to the US-associated context stimulus relative to the safe context, without the confounding effects of the presence of the US. This was done for sustained as well as transient activity and hence yielded the following contrasts: CS+unpaired(sustained) > CS (sustained) and CS+ unpaired(transient) > CS (transient). Single subject contrast images were entered into random-effects one-sample t-test analyses to test for group-level significance. For sustained activity, statistical results are reported for contiguous voxels that exceeded a cluster-level threshold of p 6 0.02 (false discovery rate (FDR) corrected) at cluster-size k = 200. For transient activity, we carried out an anatomical region of interest (ROI) analysis since we were primarily interested in BOLD responses within the medial temporal lobe (MTL). The anatomical ROI comprises bilateral hippocampus and bilateral amygdala and was created using the Anatomy toolbox (Eickhoff et al., 2005). Statistical results for the ROI analysis are reported for contiguous voxels that exceed a cluster-level threshold of p < 0.05 (family-wise error (FWE) corrected) at a cluster-size k = 200. In addition, the contrast CS+paired(sustained) > CS (sustained) was calculated with a cluster-level threshold of p < 0.02 (FDR corrected) at cluster-size k = 200, to compare BOLD activation patterns to the CS+ with and without US application. 2.7. Functional connectivity analysis To estimate functional coupling between brain regions involved in contextual fear conditioning, we employed psychophysiological interaction (PPI; Friston et al., 1997; Gitelman, Penny, Ashburner, & Friston, 2003) analyses as implemented in SPM 8. PPI detects regionally specific responses in terms of an interaction of a seedregion (extracted mean time-series of a functionally or

anatomically defined region of interest (ROI)) with an experimental factor, using the difference in regression coefficients. Because interactions in the brain do not occur on a hemodynamic, but on a neural level, a deconvolution step is incorporated into the PPI algorithm (Gitelman et al., 2003). Seed regions for the analyses were functionally determined from activity in the right MTL, which was significant for the transient activity contrast at the group level.3 Although the cluster-level-corrected result yielded one contiguous activation that spanned from the anterior hippocampus/ amygdala to the posterior hippocampus, the voxels were not equally distributed and concentrated on opposite ends of the cluster. Therefore, we selected two spherical ROIs from the anterior and a more posterior part of the right hippocampus as seed regions, with the former centered around the peak activation with a 2 mm radius and the latter within the posterior cluster with a 5 mm radius. To ensure that only voxels were included, which were with high probability within the hippocampus, we utilized maximum probability maps for the MTL as implemented in the Anatomy toolbox (Eickhoff et al., 2005). This procedure resulted in a small radius for the anterior hippocampus-ROI (capturing peak activation while avoiding overlap with the adjacent amygdala) and a more posterior hippocampus-ROI that was not centered around the peak activation, but instead positioned so that all voxels within the ROI were assigned to hippocampus (HC) with a probability of P70%. The label ‘‘more posterior’’ for the latter cluster was chosen following the suggestion by Poppenk, Evensmoen, Moscovitch, and Nadel (2013) to term hippocampal foci at or posterior to y = 21 mm in MNI space as falling into the posterior hippocampus. Even though the aforementioned hippocampal cluster resides mainly in the posterior part of this division, it also stretches out beyond the anterior–posterior border to the y = 18 plane. From these seed regions, time series were extracted as the first eigenvariate of the filtered and adjusted response in all voxels. Interaction regressors were then created by computing the element-wise product of the experimental event time course and the seed region time series for the following interactions: HCanterior  CS+unpaired, HCposterior  CS+unpaired. On a singlesubject level, the effects of these interaction regressors were tested against baseline and the ensuing contrast images were entered into random-effects one-sample t-test analyses for group level inference. Statistical results are reported for contiguous voxels that exceeded a threshold of p < 0.001 (uncorrected) at a cluster-size of k = 30.

3. Results 3.1. SCR and rating data The skin conductance responses evoked by the conditions CS+unpaired and CS for the three non-overlapping time bins showed significant differences of the SCRs over the course of the experiment. While there were no significant differences between the two conditions in the first (t(13) = 0.0756; p = 0.8328) and last time bin (t(13) = 0.3365; p = 0.7418), larger SCRs for CS+unpaired versus CS were significant in the second time bin (t(13) = 2.9446; p = 0.0114). These results indicate that successful conditioning occurred only after one third of the experiment, while SCRs for CS+unpaired and CS converged toward the end of the experiment. Mean SCRs for CS+unpaired, CS and the US across all three time bins are depicted in Fig. 3. The context stimuli ratings were significantly higher for CS+ than CS on emotional valence 3 This does not lead to an inferential bias due to non-independent selective analyses (Kriegeskorte, Simmons, Bellgowan, & Baker, 2009), since the regressor for the seed region was included in the design matrix and inference was drawn for the interaction between the seed region and the experimental factor. Therefore, PPI tests for effects that cannot be explained by other regressors, including the physiological regressor derived from the seed region.

5

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11 Table 1 Statistical comparisons of the ratings of the context-picture stimuli. CS

Emotional valence Arousal Contingency awareness

CS+

Mean

SD

Mean

SD

3.33 2.27 1.33

1.91 1.79 0.72

6.33 6.27 7.8

1.99 1.44 0.77

Ratings for CS+ (pain-associated context) and CS using a paired t-test. SD, standard deviation.

Table 2 Sustained activation for the pain-associated context versus the non-painful one.

t

p

Region

3.81 5.98 27.36

0.002 3.4  10 1.5  10

R IFG (pars ORB, pars TRIA)/R MFG R IFG (pars TRIA, pars OPER)/R insula R MFG L IPLa L caudate nucleus/L thalamusa Superior medial gyrus L insula R middle temporal gyrus R IPLa R caudate nucleus

MNI coordinates X

5 13

(safe context) were compared

Y 45 51 51 63 12 3 31 62 54 9

Z 48 21 14 39 2 39 26 41 48 5

3 0 45 27 17 41 2 3 54 11

Peak t-value 8.60 8.39 7.49 7.02 6.07 5.88 5.86 5.79 5.61 5.42

Cluster-level significance: p < 0.02 (FDR (false discovery rate) corrected), cluster extent threshold: 200 voxels. a Region labeled using maximum probability maps from the Anatomy toolbox (Eickhoff et al., 2005); all other regions were labeled using the Automatic Anatomical Labeling (AAL) software (Tzourio-Mazoyer et al., 2002). MNI, Montreal Neurological Institute; IFG, inferior frontal gyrus; IPL, inferior parietal lobule; L, left; R, right; MFG, middle frontal gyrus, ORB, orbitalis, OPER, opercularis; TRIA, triangularis.

Fig. 3. Mean SCR (skin conductance response) amplitudes (ln [lS  s]) are shown for CS+unpaired and CS trials and the US across three time bins of the experimental time course. Error bars denote the standard error of the mean. ⁄ p = 0.0114.

connections of these regions with other brain areas. An interaction of the activation in the right aHPC with the condition CS+unpaired revealed significant connections of the seed region with the right amygdala (basolateral and superficial nucleus) and left somatosensory cortex (Brodmann areas (BA) 3a and 3b). Detailed results are reported in Table 4 and depicted in Fig. 6. The right, more posterior part of the hippocampus was significantly functionally connected with the right aHPC and the left superior parietal lobule (BA 7a) during CS+unpaired. Detailed results are reported in Table 5 and depicted in Fig. 6.

4. Discussion (t(14) = 3.8129; p = 0.002), arousal (t(14) = 5.9787; p = 3.4  10 7) and contingency awareness (t(14) = 27.3577; p = 1.5  10 13). Detailed results for the comparison between context stimuli ratings are reported in Table 1. 3.2. Functional MRI results BOLD responses that were evoked by the fear conditioning process and sustained over time (CS+unpaired(sustained) > CS (sustained)) were pronounced in the bilateral anterior insula, superior medial gyrus, bilateral caudate nucleus/thalamus, right inferior parietal lobule (IPL), left IPL and inferior frontal gyrus (bilateral: pars triangularis, pars opercularis; right: pars orbitalis). Detailed results for this contrast are reported in Table 2 and depicted in Fig. 4. Brain activity that was triggered by the same differential contrast but decreased linearly over time (CS+unpaired(transient) > CS (transient)) was found in a cluster of contiguous voxels that included the basolateral amygdala and right anterior hippocampus and a more posterior located right hippocampal cluster. Detailed results for this contrast are reported in Table 3 and depicted in Fig. 4. An additional comparison of BOLD effects related to the contrast CS+unpaired(sustained) > CS (sustained) versus CS+paired(sustained) > CS (sustained), is provided in Fig. 5. This figure depicts common and distinct activations related to the presence or absence of the US. 3.3. Functional connectivity results PPI analyses with seed regions in the right anterior (aHPC) and more posterior part of the hippocampus were performed to investigate the effects of context conditioning on the functional

In the current study we investigated differential contextual fear conditioning in humans, using two cue-array picture contexts, which contained the same visual cues but with a different arrangement in each picture. This ensured that mere detection of the presence of cues in the pictures did not provide sufficient information to differentiate them. CS+ contexts were rated as more arousing and more unpleasant after the experiment. The SCRs were significantly higher in CS+unpaired compared to CS trials in the second third of the experiment, but not in the last third, probably due to habituation effects. Differential brain responses in the right hippocampus and right amygdala followed a decay over the course of the experiment, contributing further evidence for their involvement in contextual fear conditioning (Alvarez et al., 2008; Lang et al., 2009; Marschner et al., 2008). Furthermore, activity within the right hippocampus was spread from the anterior to the posterior part of its longitudinal axis. Numerous proposals (see Poppenk et al., 2013 for a review) for the specialization of the human hippocampus along an anterior–posterior axis exist in the literature, the most prominent being the suggestion of a more specific role of emotional processes in aHPC and mnemonic processes in posterior hippocampus (pHPC), which received support from animal studies (Fanselow & Dong, 2010). A meta-analysis revealed that better retrieval for emotional than neutral stimuli involved the aHPC in humans (Murty, Ritchey, Adcock, & LaBar, 2010). However, trait anxiety was specifically linked to higher activity in the pHPC during a task condition, in which subjects experienced threat of a painful electric stimulus (Satpute, Mumford, Naliboff, & Poldrack, 2012), questioning the emotion–cognition differentiation related to the anterior and posterior hippocampus. Since the involvement of the hippocampus is required for the processing of both context-pictures,

6

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11

Fig. 4. Sustained brain activity (A) for the contrast CS+unpaired > CS at a cluster-level threshold of p < 0.02 (false discovery rate corrected), cluster-size k = 200. A list of all significant activations for this contrast can be found in Table 2. Transient brain activity (B) (linearly decaying over time) for the contrast CS+unpaired > CS at a cluster-level threshold of p < 0.05 (family-wise error corrected), cluster-size k = 200. Plane coordinates are in MNI (Montreal Neurological Institute) space. A list of all significant activations for this contrast can be found in Table 3.

Table 3 Transient activation for the pain-associated context versus the non-painful one (ROI analysis). Region

MNI coordinates X

R (more posterior) hippocampus (CA, DG)a R amygdala (BL)a R (anterior) hippocampusa

30 29 27

Y

Peak t-value

Z 19 6 10

18 23 17

5.49 4.80 4.72

Cluster-level significance: p < 0.05 (family-wise error corrected), cluster extent threshold: 200 voxels. a Region (maximum peak value) labeled using maximum probability maps from the Anatomy toolbox (Eickhoff et al., 2005); ROI, region of interest; MNI, Montreal Neurological Institute; BL, basolateral; R, right; CA, cornu ammonis; DG, dentate gyrus.

stronger activation of the right hippocampus during CS+unpaired relative to CS is thought to reflect the specific contribution of this brain region to contextual fear conditioning. Particularly, neural responses in a more posterior part of the hippocampus might indicate the encoding of emotionally negative memories (Shafer &

Dolcos, 2012), whereas the anterior cluster might process gist-like associations between a threat (painful stimulation) and its context (Poppenk et al., 2013). Our PPI analyses show significant connections of the more posterior part of the hippocampus with the posterior cingulate cortex/ thalamus and precuneus (BA 7a), regions that have been shown to display increased responses in a recognition memory task under conditions of high arousal (Greene, Flannery, & Soto, 2014). Moreover, the functional connections between both hippocampal clusters might involve the integration of their respective computations while the correlation of the aHPC and amygdala possibly reflects the projection of the contextual representation to the amygdala to generate a fear response (although the PPI method does not permit directional statements). Finally, it can be hypothesized that the connectivity between the aHPC and the primary somatosensory cortex (BA 3a and 3b) of the contralateral side reflects the anticipation of the nociceptive US to the right hand. Brain regions that were active in a sustained manner in our experiment are largely part of a network known to be implicated in contextual fear conditioning. These regions included the

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11

7

Fig. 5. Comparison of BOLD effects related to the contrast CS+paired > CS (in light blue) versus the contrast CS+unpaired > CS (in red) for sustained effects. Overlapping regions from both contrasts are shown in yellow. This figure depicts common and distinct activations related to the presence or absence of the aversive stimulus. The clusterlevel threshold was p < 0.02 (false discovery rate corrected), cluster-size k = 200. Plane coordinates are in MNI (Montreal Neurological Institute) space. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 4 Psychophysiological interaction of anterior hippocampus with CS+unpaired. Region

MNI coordinates X

R amygdala (BL)/R hippocampus (SUB, CA)a R postcentral gyrus (BA 4a)/R precentral gyrus (BA 6)a R amygdala (SF)a R IPC/BA3b, 3a, 4pa L postcentral gyrus (BA 3a, 3b)a

Y

Peak t-value

Z

27

9

18

8.07

23

33

69

6.12

18 41 46

0 31 15

12 39 32

4.97 4.95 4.91

Cluster-level significance: p < 0.001 (uncorrected), cluster extent threshold: 30 voxel. a Region labeled using maximum probability maps from the Anatomy toolbox (Eickhoff et al., 2005). CS+unpaired, condition where pain-associated context picture was presented without administration of the unconditioned stimulus; MNI, Montreal Neurological Institute; BA, Brodmann area; BL, basolateral; IPC, inferior parietal cortex; L, left, R, right; CA, cornu ammonis; SUB, subiculum; SF, superficial.

bilateral anterior insula, superior medial gyrus, bilateral caudate nucleus/thalamus, right and left IPL and IFG (pars opercularis, pars triangularis and a spatially separated activation cluster more frontally that also extended into the pars orbitalis). The anterior insula is implicated in anticipatory fear (Ploghaus et al., 1999; Wager et al., 2004). The right and left IPL were found to interact with the IFG (pars triangularis, pars opercularis) to execute attentional processes (Simon et al., 2004). Interestingly, a previous study has also shown that the IFG was engaged in a Pavlovian conditioning task when outcome prediction for aversive events was ambiguous (Bach, Seymour, & Dolan, 2009). In our task subjects were confronted with uncertainty about the US application in the CS+ trials, even after they learned the CS+ US contingency, since the partial reinforcement led to uncertainty about the receipt of a US in a given trial. We also obtained neural responses in a more anteriorly located cluster that includes the right IFG and middle frontal gyrus (MFG) as well as an additional cluster situated within the right

8

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11

Fig. 6. Results for the psychophysiological interaction (PPI) analysis of the seed region right anterior hippocampus in interaction with the condition CS+unpaired (A) at a cluster-level threshold of p < 0.001 (uncorrected) and cluster-size k = 30. A list of all significant activations for this contrast can be found in Table 4. Results for the PPI analysis of the seed region right posterior hippocampus in interaction with the condition CS+unpaired (B) at a cluster-level threshold of p < 0.001 (uncorrected) and cluster-size k = 30. Plane coordinates are in MNI (Montreal Neurological Institute) space. A list with coordinates and t-values for these activations can be found in Table 5.

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11 Table 5 Psychophysiological interaction of posterior hippocampus with CS+unpaired. Region

MNI coordinates X

R (anterior) hippocampus (CA, SUB)a R posterior cingulate cortex/thalamusa L IFG (pars TRIA) (BA44)a L superior parietal lobule (BA 7a)a

Y 17 8 40 13

Peak t-value

Z 16 39 14 60

23 14 29 53

7.75 6.20 5.81 4.66

Cluster-level significance: p < 0.001 (uncorrected), cluster extent threshold: 30 voxel. a Region labeled using maximum probability maps from the Anatomy toolbox (Eickhoff et al., 2005). CS+unpaired, condition where pain-associated context picture was presented without administration of the unconditioned stimulus; MNI, Montreal Neurological Institute; BA, Brodmann area; IFG, inferior frontal gyrus; L, left; R, right; CA, cornu ammonis; SUB, subiculum; TRIA, triangularis.

9

decay. Also, we did not assess context stimulus ratings before the experiment. Hence we cannot exclude the possibility that the CS+ would have been rated more negatively than the CS prior to conditioning. However, since CS+ and CS were counterbalanced, this seems unlikely. Finally, the short duration of our CS stimuli (4 s) did not allow for the separate analysis of CS+ trials and the US. Since we employed a passive task, we cannot be sure that subjects stayed engaged throughout the fMRI procedure, despite being instructed to attentively view the pictures. It might be beneficial to implement a behavioral component to keep the subjects’ attention to stimuli high (Indovina, Robbins, Núnez-Elizalde, Dunn, & Bishop, 2011).

5. Conclusions MFG. The MFG has previously been implicated in the acquisition of contingency awareness in trace and delay conditioning (Carter, O’Doherty, Seymour, Koch, & Dolan, 2006). In the current analyses we excluded subjects who were not aware of the CS/US contingency, so it is possible that the right MFG activation in our study also mirrors the acquisition of contingency awareness. However, such an interpretation has to be made with care since Carter, O’Doherty, Seymour, Koch, and Dolan (2006) did not investigate contextual conditioning and their subjects, in contrast to our study, gave trial-by-trial US expectancy ratings. The caudate nucleus is usually associated with instrumental rather than Pavlovian conditioning (O’Doherty et al., 2004), but also showed significant activation in our contextual conditioning experiment. Activity in the caudate nucleus in our subjects might be related to fear of pain (Scharmüller & Schienle, 2014). We did not observe significant BOLD activation in the anterior cingulate cortex (ACC), a structure that is usually active during fear conditioning (Sehlmeyer et al., 2009). Yet, in a recent review Etkin, Egner, and Kalisch (2011) noted that frontal activations during the acquisition of fear in classical conditioning experiments in humans are not only located in ACC but distributed throughout the frontal cortex, with clusters in the dorsal ACC, dorsomedial prefrontal cortex (dmPFC), supplementary motor area (SMA) and pre-SMA. They suggest that these activations might reflect a variety of different processes that occur simultaneously such as CS appraisal, expression of conditioned responses, consolidation and storage of fear memories. We detected a cluster in the superior medial gyrus that lies in the dmPFC (MNI coordinate of peak voxel: 3, 39, 41). This is in line with studies that support the view that the dorsal ACC and dmPFC are involved in both the generation of fear and anxiety responses as well as fear and anxiety appraisal (see Etkin et al., 2011). Taken together, brain regions which were found to be active in this study largely overlap with those from previous contextual fear conditioning studies. Since subjects with an intact hippocampus normally process contexts in a configural fashion (Fanselow, 2010), the brain activation patterns found in this study might not strongly deviate from those of previous studies using spatial contexts in healthy controls. Our current design does not permit to directly test the brain regions involved in configural processing since we did not compare it to a design that does not require a configural strategy. However, with this design it is possible to test if configural processing strategies are absent in subjects with aberrant hippocampal functioning. Our study contains several methodological limitations. The first is the rather small sample size that potentially renders our findings less safe. Furthermore, we assumed a linear decay of responses in the MTL, which probably only represents an approximation to the subjects’ individual neural time courses. A model-based fMRI approach employing a reinforcement learning model (O’Doherty et al., 2007) might provide a better fit to the data than a linear

The current study complements previous findings of the recruitment of the amygdala, hippocampus, insula and frontal and parietal structures during contextual fear conditioning (e.g. Alvarez et al., 2008; Büchel et al., 1998; Lang et al., 2009; Marschner et al., 2008; Pohlack et al., 2012a, 2012b) in a way that the critical point in our design is the requirement to process the unique spatial relationships between these cues in order to separate the contexts. This process demands a configural learning strategy, which is hypothesized to involve the hippocampus (Aggleton, Sanderson, & Pearce, 2007; Eichenbaum, 2004; Rudy & Sutherland, 1995) and which was shown to be severely disrupted in hippocampus-lesioned animals (Albasser et al., 2013; Dumont et al., 2007; Sanderson et al., 2006) and humans (Hannula et al., 2006; Watson et al., 2013). Hence, this key feature of our design – presence of the same cues with different arrangement in the contexts – make it particularly suitable for investigating subjects with hippocampal dysfunction, like patients suffering from depression, schizophrenia, post-traumatic stress disorder and pathological as well as healthy aging (Small, Schobel, Buxton, Witter, & Barnes, 2011), since the process of forming a conjunctive representation should be specifically impaired in those subjects. According to Fanselow (2010), subjects with hippocampal impairment might still be able to process these contextual stimuli by employing an extra-hippocampal circuit. This alternate circuit might, however, perform less efficiently, because the ambiguous nature of the contexts in our design makes the task of learning the pain-stimuluscontingency particularly demanding (Fanselow, 2010). Therefore, a further validation of the task would be a comparison of healthy controls and patients with circumscribed hippocampal damage and/or subjects suffering from hippocampal impairment (e.g. hippocampal atrophy in healthy aging and depression). It could also be clarified if such an extra-hippocampal circuit for configural processing exists and how brain activity differs between subjects who successfully condition from those who do not. In this regard, an interesting application of our design could be the examination of patients with post-traumatic stress disorder (PTSD) who may have an inability to adequately form conjunctive context representations (Acheson, Gresack, & Risbrough, 2012). Accordingly, PTSD patients should express fear to the CS as well because they are prone to simple elemental associations and thus respond to features present in both contexts. Future studies need to contrast this paradigm with tasks that are more geared toward the integration of cues and context, which would be the most natural learning situation to better understand the specificity of the hippocampal involvement. In addition future research on psychopathology could use this experimental design to determine neural mechanisms of impaired contextual processing in various mental disorders to unravel some of the pathophysiological processes that accompany those (Maren, Phan, & Liberzon, 2013).

10

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11

Acknowledgment This work was funded by a grant from the German Ministry of Education and Research (BMBF, 01GQ1003B). References Acheson, D. T., Gresack, J. E., & Risbrough, V. B. (2012). Hippocampal dysfunction effects on context memory: Possible etiology for posttraumatic stress disorder. Neuropharmacology, 62(2), 674–685. Aggleton, J. P., Sanderson, D. J., & Pearce, J. M. (2007). Structural learning and the hippocampus. Hippocampus, 17(9), 723–734. Albasser, M. M., Dumont, J. R., Amin, E., Holmes, J. D., Horne, M. R., Pearce, J. M., et al. (2013). Association rules for rat spatial learning: The importance of the hippocampus for binding item identity with item location. Hippocampus, 23(12), 1162–1178. Alvarez, R. P., Biggs, A., Chen, G., Pine, D. S., & Grillon, C. (2008). Contextual fear conditioning in humans: Cortical-hippocampal and amygdala contributions. Journal of Neuroscience, 28(24), 6211–6219. Bach, D. R., Seymour, B., & Dolan, R. J. (2009). Neural activity associated with the passive prediction of ambiguity and risk for aversive events. Journal of Neuroscience, 29(6), 1648–1656. Benedek, M., & Kaernbach, C. (2010). A continuous measure of phasic electrodermal activity. Journal of Neuroscience Methods, 190(1), 80–91. Boucsein, W. (1992). Electrodermal activity. New York: Plenum University Press. 442 p. Boucsein, W., Fowles, D. C., Grimnes, S., Ben-Shakhar, G., Roth, W. T., & Dawson, M. E. (2012). Publication recommendations for electrodermal measurements. Psychophysiology, 49(8), 1017–1034. Büchel, C., Dolan, R. J., Armony, J. L., & Friston, K. J. (1999). Amygdala-hippocampal involvement in human aversive trace conditioning revealed through eventrelated functional magnetic resonance imaging. Journal of Neuroscience, 19(24), 10869–10876. Büchel, C., Morris, J., Dolan, R. J., & Friston, K. J. (1998). Brain systems mediating aversive conditioning: An event-related fMRI study. Neuron, 20(5), 947–957. Bussey, T. J., Warburton, E. C., Aggleton, J. P., & Muir, J. L. (1998). Fornix lesions can facilitate acquisition of the transverse patterning task: A challenge for ‘‘configural’’ theories of hippocampal function. Journal of Neuroscience, 18(4), 1622–1631. Carter, R. M., O’Doherty, J. P., Seymour, B., Koch, C., & Dolan, R. J. (2006). Contingency awareness in human aversive conditioning involves the middle frontal gyrus. Neuroimage, 29(3), 1007–1012. Dumont, J., Petrides, M., & Sziklas, V. (2007). Functional dissociation between fornix and hippocampus in spatial conditional learning. Hippocampus, 17(12), 1170–1179. Eichenbaum, H. (2004). Hippocampus: Cognitive processes and neural representations that underlie declarative memory. Neuron, 44(1), 109–120. Eickhoff, S. B., Stephan, K. E., Mohlberg, H., Grefkes, C., Fink, G. R., Amunts, K., et al. (2005). A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. Neuroimage, 25(4), 1325–1335. Etkin, A., Egner, T., & Kalisch, R. (2011). Emotional processing in anterior cingulate and medial prefrontal cortex. Trends in cognitive Sciences, 15(2), 85–93. Fanselow, M. S. (2010). From contextual fear to a dynamic view of memory systems. Trends in Cognitive Sciences, 14(1), 7–15. Fanselow, M. S., & Dong, H. W. (2010). Are the dorsal and the ventral hippocampus functionally distinct structures? Neuron, 65(1), 7–19. Friston, K. J., Buechel, C., Fink, G. R., Morris, J., Rolls, E., & Dolan, R. J. (1997). Psychophysiological and modulatory interactions in neuroimaging. Neuroimage, 6(3), 218–229. Gitelman, D. R., Penny, W. D., Ashburner, J., & Friston, K. J. (2003). Modeling regional and psychophysiologic interactions in fMRI: The importance of hemodynamic deconvolution. Neuroimage, 19(1), 200–207. Greene, C. M., Flannery, O., & Soto, D. (2014). Distinct parietal sites mediate the influences of mood, arousal and their interaction on human recognition memory. Cognitive, Affective, & Behavioral Neuroscience, 14(4), 1327–1339. Grillon, C., Baas, J. M., Cornwell, B., & Johnson, L. (2006). Context conditioning and behavioral avoidance in a virtual reality environment: Effect of predictability. Biological Psychiatry, 60(7), 52–759. Grillon, C., Baas, J. P., Lissek, S., Smith, K., & Milstein, J. (2004). Anxious responses to predictable and unpredictable aversive events. Behavioral Neuroscience, 118(5), 916–924. Hannula, D. E., Tranel, D., & Cohen, N. J. (2006). The long and the short of it: Relational memory impairments in amnesia, even at short lags. Journal of Neuroscience, 26(32), 8352–8359. Indovina, I., Robbins, T. W., Núnez-Elizalde, A. O., Dunn, B. D., & Bishop, S. J. (2011). Fear-conditioning mechanisms associated with trait vulnerability to anxiety in humans. Neuron, 69(3), 563–571. Iordanova, M. D., Burnett, D. J., Aggleton, J. P., Good, M., & Honey, R. C. (2009). The role of the hippocampus in mnemonic integration and retrieval: Complementary evidence from lesion and inactivation studies. European Journal of Neuroscience, 30(11), 2177–2189. Kriegeskorte, N., Simmons, W. K., Bellgowan, P. S., & Baker, C. I. (2009). Circular analysis in systems neuroscience. The dangers of double dipping. Nature Neuroscience, 12(5), 535–540.

Lang, S., Kroll, A., Lipinski, S. J., Wessa, M., Ridder, S., Christmann, C., et al. (2009). Context conditioning and extinction in humans: Differential contribution of the hippocampus, amygdala and prefrontal cortex. European Journal of Neuroscience, 29(4), 823–832. Leutgeb, S., Leutgeb, J. K., Barnes, C. A., Moser, E. I., McNaughton, B. L., & Moser, M. B. (2005). Independent codes for spatial and episodic memory in hippocampal neuronal ensembles. Science, 309(5734), 619–623. Lovibond, P. F., Liu, J. C., Weidemann, G., & Mitchell, C. J. (2011). Awareness is necessary for differential trace and delay eyeblink conditioning in humans. Biological Psychology, 87(3), 393–400. Maren, S., Aharonov, G., & Fanselow, M. S. (1997). Neurotoxic lesions of the dorsal hippocampus and Pavlovian fear conditioning in rats. Behavioural Brain Research, 88(2), 261–274. Maren, S., Phan, K. L., & Liberzon, I. (2013). The contextual brain: Implications for fear conditioning, extinction and psychopathology. Nature Reviews Neuroscience, 14(6), 417–428. Marschner, A., Kalisch, R., Vervliet, B., Vansteenwegen, D., & Büchel, C. (2008). Dissociable roles for the hippocampus and the amygdala in human cued versus context fear conditioning. Journal of Neuroscience, 28(36), 9030–9036. Moses, S. N., & Ryan, J. D. (2006). A comparison and evaluation of the predictions of relational and conjunctive accounts of hippocampal function. Hippocampus, 16(1), 43–65. Murty, V. P., Ritchey, M., Adcock, R. A., & LaBar, K. S. (2010). FMRI studies of successful emotional memory encoding: A quantitative meta-analysis. Neuropsychologia, 48(12), 3459–3469. Nadel, L., & Willner, J. (1980). Context and conditioning: A place for space. Physiology & Behavior, 8(2), 218–228. O’Doherty, J. P., Dayan, P., Schultz, J., Deichmann, R., Friston, K., & Dolan, R. J. (2004). Dissociable roles of ventral and dorsal striatum in instrumental conditioning. Science, 304(5669), 452–454. O’Doherty, J. P., Hampton, A., & Kim, H. (2007). Model-based fMRI and its application to reward learning and decision making. Annals of the New York Academy of Sciences, 1104(1), 35–53. Olson, I. R., Moore, K. S., Stark, M., & Chatterjee, A. (2006). Visual working memory is impaired when the medial temporal lobe is damaged. Journal of Cognitive Neuroscience, 18, 1087–1097. Pineles, S. L., Orr, M. R., & Orr, S. P. (2009). An alternative scoring method for skin conductance responding in a differential fear conditioning paradigm with a long-duration conditioned stimulus. Psychophysiology, 46(5), 984–995. Ploghaus, A., Tracey, I., Gati, J. S., Clare, S., Menon, R. S., Matthews, P. M., et al. (1999). Dissociating pain from its anticipation in the human brain. Science, 284(5422), 1979–1981. Pohlack, S. T., Nees, F., Liebscher, C., Cacciaglia, R., Diener, S. J., Ridder, S., et al. (2012a). Hippocampal but not amygdalar volume affects contextual fear conditioning in humans. Human Brain Mapping, 33(2), 478–488. Pohlack, S. T., Nees, F., Ruttorf, M., Schad, L. R., & Flor, H. (2012b). Activation of ventral striatum during aversive contextual conditioning in humans. Biological Psychology, 91(1), 74–80. Poppenk, J., Evensmoen, H. R., Moscovitch, M., & Nadel, L. (2013). Long-axis specialization of the human hippocampus. Trends in Cognitive Sciences, 17(5), 230–240. Quirk, G. J., Armony, J. L., & LeDoux, J. E. (1997). Fear conditioning enhances different temporal components of tone-evoked spike trains in auditory cortex and lateral amygdala. Neuron, 19(3), 613–624. Rudy, J. W. (2009). Context representations, context functions, and the parahippocampal–hippocampal system. Learning & Memory, 16(10), 573–585. Rudy, J. W., & O’Reilly, R. C. (2001). Conjunctive representations, the hippocampus, and contextual fear conditioning. Cognitive, Affective, & Behavioral Neuroscience, 1(1), 66–82. Rudy, J. W., & Sutherland, R. J. (1995). Configural association theory and the hippocampal formation: An appraisal and reconfiguration. Hippocampus, 5(5), 375–389. Saksida, L. M., Bussey, T. J., Buckmaster, C. A., & Murray, E. A. (2007). Impairment and facilitation of transverse patterning after lesions of the perirhinal cortex and hippocampus, respectively. Cerebral Cortex, 17(1), 108–115. Sanderson, D. J., Pearce, J. M., Kyd, R. J., & Aggleton, J. P. (2006). The importance of the rat hippocampus for learning the structure of visual arrays. European Journal of Neuroscience, 24(6), 1781–1788. Satpute, A. B., Mumford, J. A., Naliboff, B. D., & Poldrack, R. A. (2012). Human anterior and posterior hippocampus respond distinctly to state and trait anxiety. Emotion, 12(1), 58–68. Scharmüller, W, & Schienle, A. (2014). Functional brain connectivity during fear of pain: A comparison between dental phobics and controls (Epub ahead of print). Brain Connectivity. Sehlmeyer, C., Schöning, S., Zwitserlood, P., Pfleiderer, B., Kircher, T., Arolt, V., et al. (2009). Human fear conditioning and extinction in neuroimaging: A systematic review. PLoS ONE, 4(6), e5865. Shafer, A. T., & Dolcos, F. (2012). Neural correlates of opposing effects of emotional distraction on perception and episodic memory: An event-related FMRI investigation. Frontiers in Integrative Neuroscience, 6(70), 1–15. Simon, O., Kherif, F., Flandin, G., Poline, J. P., Rivière, D., Mangin, J. F., et al. (2004). Automatized clustering and functional geometry of human parietofrontal networks for language, space, and number. Neuroimage, 23(3), 1192–1202. Small, S. A., Schobel, S. A., Buxton, R. B., Witter, M. P., & Barnes, C. A. (2011). A pathophysiological framework of hippocampal dysfunction in ageing and disease. Nature Reviews Neuroscience, 12(10), 585–601.

C. Baeuchl et al. / Neurobiology of Learning and Memory 121 (2015) 1–11 Stewart, M. A., Stern, J. A., Winokur, G., & Fredman, S. (1961). An analysis of GSR conditioning. Psychological Review, 68(1), 60–67. Sutherland, R. J., & Rudy, J. W. (1989). Configural association theory: The role of the hippocampal formation in learning, memory, and amnesia. Psychobiology, 17(2), 129–144. Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N., et al. (2002). Automated anatomical labelling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject-brain. Neuroimage, 15(1), 273–289.

11

Wager, T. D., Rilling, J. K., Smith, E. E., Sokolik, A., Casey, K. L., Davidson, R. J., et al. (2004). Placebo-induced changes in fMRI in the anticipation and experience of pain. Science, 303(5661), 1162–1167. Watson, P. D., Voss, J. L., Warren, D. E., Tranel, D., & Cohen, N. J. (2013). Spatial reconstruction by patients with hippocampal damage is dominated by relational memory errors. Hippocampus, 23(7), 570–580. Wiltgen, B. J., Sanders, M. J., Anagnostaras, S. G., Sage, J. R., & Fanselow, M. S. (2006). Context fear learning in the absence of the hippocampus. Journal of Neuroscience, 26(20), 5484–5491.