Abstract

The dual-process theory assumes that contexts are encoded in an elemental and in a conjunctive representation. However, this theory was developed from animal studies, and we still have to explore if and how elemental and conjunctive representations contribute to, for example, contextual anxiety in humans. Therefore, 28 participants underwent differential context conditioning in a newly developed flip-book paradigm. Virtual rooms were presented similar to a flip-book, that is, as a stream of 49 consecutive screenshots creating the impression of walking through the rooms. This allowed registration of event-related brain potentials triggered by specific screenshots. During two acquisition phases, two rooms were shown in this way for six times each. In one room, the anxiety context (CTX+), mildly painful electric stimuli (unconditioned stimuli [USs]) were administered unpredictably after 12 distinct screenshots, which became threat elements, whereas 12 selected comparable screenshots became nonthreat elements (elemental representation); all screenshots represented the anxiety context (conjunctive representation). In the second room, the safety context (CTX−), no USs were applied; thus, all screenshots created the safety context whereby 12 preselected screenshots represented safety elements. Increased US expectancy ratings for threat versus nonthreat or safety elements reflected elemental representation. Conjunctive representation was evident in differential ratings (arousal and contingency) and increased P100 and early posterior negativity amplitudes for threat and nonthreat CTX+ versus safety CTX− screenshots. These differences disappeared during two test phases without US delivery indicating successful extinction. In summary, we revealed the first piece of evidence for the simultaneous contributions of elemental and conjunctive representation during context conditioning in humans.

INTRODUCTION

Particularly when confronted with threat, the context facilitates the interpretation of the situation and helps the organism to respond properly (Glenn, Risbrough, Simmons, Acheson, & Stout, 2018). The dual-process theory, developed on animal studies, assumes that contexts are processed in two ways: On the one hand, as suggested by the features view, a context is represented by its single elements, for example, the color of the wall, which each may become associated with an event. On the other hand, according to the conjunctive view, a context is encoded as a conjunctive representation based on the binding of its elements into one unique representation, and this representation may become associated with an event too (Rudy, 2009; Rudy, Huff, & Matus-Amat, 2004; Rudy & O’Reilly, 2001; Nadel & Willner, 1980). Evidence suggests that the elemental representation requires the neocortical system (Nadel & Willner, 1980) and the amygdala (Rudy et al., 2004; Phillips & LeDoux, 1992), whereas the conjunctive representation highly relies on the involvement of the hippocampus (Rudy et al., 2004).

The verification of the dual-process theory and the underlying neural systems in humans has been challenging. Recently, reinforcement learning was suggested to disentangle both representations (Duncan, Doll, Daw, & Shohamy, 2018). Most human studies used a differential context conditioning paradigm, that is, an aversive unconditioned stimulus (US) is administered unpredictably during the presentation of a neutral context. As the context is the best predictor of the US, it becomes an anxiety context (CTX+) eliciting anxiety (also called sustained fear). A second contextual stimulus is never paired with a US and therefore becomes a safety context (CTX−; Glotzbach-Schoon, Tadda, et al., 2013; Maren, Phan, & Liberzon, 2013; Rudy et al., 2004). Baeuchl, Meyer, Hoppstädter, Diener, and Flor (2015) defined contexts by the placement of single features (e.g., furniture) in a context and revealed that this configural information elicited anxiety responses in humans, including increased skin conductance level, arousal, and contingency ratings, as well as lower valence ratings in the context configuration, which was paired with the US compared with an unpaired configuration. Thus, results emphasize the importance of conjunctive context representation; however, the paradigm does not allow conclusions on the impact of single contextual elements. Stout et al. (2018) tried to overcome this problem by separating conjunctive and elemental learning in subsequent paradigms using fMRI. The conjunctive context learning paradigm revealed hippocampal but not amygdala activity, and the elemental learning paradigm indicated amygdala but not hippocampal activity (Stout et al., 2018). The configurations of elements, which predict a certain outcome, resulted in increased hippocampal activity and is related to the connectivity between hippocampus and striatum (Duncan et al., 2018).

The impaired processing of contexts associated with impaired hippocampal functioning may be involved in the etiology of posttraumatic stress disorder (PTSD; Liberzon & Abelson, 2016; Acheson, Gresack, & Risbrough, 2012). Interestingly, patients with PTSD had difficulties in reporting contingencies between an anxiety context and an aversive event and additionally had higher hippocampus activation required to learn the associations (Steiger, Nees, Wicking, Lang, & Flor, 2015). Therefore, the consideration of the dual representation of a context and the disentanglement of elemental and conjunctive representation is highly important.

To overcome this problem, we combined our virtual reality (VR) context conditioning paradigm (Glotzbach-Schoon, Tadda, et al., 2013) with the multiCS conditioning paradigm, in which several stimuli of one category are shown as CS, but only some of them are paired with an US (Steinberg, Bröckelmann, Rehbein, Dobel, & Junghöfer, 2013). At the same time, we assessed verbal responses, as well as ERPs via electroencephalography (EEG; Miskovic & Keil, 2012; Junghöfer, Bradley, Elbert, & Lang, 2001) with regard to elemental and conjunctive context representation.

More precisely, Bröckelmann et al. (2011) and Steinberg et al. (2013) developed a multiCS conditioning paradigm, which investigates physiological responses of few trial learning in humans. To this end, more than 100 different faces were presented and each was paired either with pleasant (CS−) or unpleasant (CS+) consequences while magnetoencephalography was recorded (Steinberg et al., 2013). All participants were unaware of CS–US contingencies after conditioning (Steinberg et al., 2013). Importantly, physiological data indicated an increase in neural activity over frontal and right occipito-parieto-temporal regions for CS+ compared with CS− stimuli already in very early stages (<100 msec) of stimulus processing (Steinberg et al., 2013).

This fast and distinct processing of very shortly presented visual stimuli has been investigated by Junghöfer et al. (2001). In their fleeting images paradigm, emotionally arousing and neutral pictures of the International Affective Picture System were serially presented alternating between high and low arousing content. Notably, EEG data showed that blocks of high compared with low arousing pictures elicited increased brain activation in the secondary visual cortex. Therefore, the fast presentation of serial stimuli—as demonstrated in Junghöfer et al.'s (2001) fleeting images paradigm—provides the possibility to measure early attention processes (Luck, Woodman, & Vogel, 2000) and affective stimulus processing (for a review, see Olofsson, Nordin, Sequeira, & Polich, 2008) with ERPs.

Particularly, the P100 is a positive voltage change and arises around 100 msec after stimulus onset in the occipital cortex (Hillyard & Münte, 1984). Attended stimuli elicit larger P100 amplitudes relative to ignored stimuli (Clark & Hillyard, 1996; Rugg, Milner, Lines, & Phalp, 1987). Additionally, the early posterior negativity (EPN) emerges around 150 msec and is most pronounced between 200 and 300 msec after stimulus onset at occipital and parietal scalp areas (Mühlberger et al., 2009; Schupp, Junghöfer, Weike, & Hamm, 2004; Schupp, Öhman, et al., 2004). It has been shown that affective pictures, for example, from the International Affective Picture System, or faces elicit larger EPN amplitudes compared with neutral ones (Mühlberger et al., 2009; Schupp, Junghöfer, et al., 2004; Schupp, Öhman, et al., 2004). Importantly, physical properties of the stimuli can also modulate the amplitudes of both P100 and EPN amplitudes (Bradley, Hamby, Löw, & Lang, 2007). The late positive potential (LPP) emerges between 400 and 1000 msec after stimulus onset in central and parietal cortex (Kastner, Flohr, Pauli, & Wieser, 2016; Bradley et al., 2007; Schupp, Junghöfer, et al., 2004). Motivational relevant, that is, emotional arousing, stimuli elicit increased LPP amplitudes compared with neutral stimuli (Bradley et al., 2007; Amrhein, Mühlberger, Pauli, & Wiedemann, 2004; Schupp et al., 2000).

In the current study, we for the first time investigated the interplay of elemental and conjunctive context representation in one paradigm by combining the multiCS conditioning and the fleeting images paradigms with our VR context-conditioning paradigm. In particular, we presented sequential screenshots of two virtual offices as a flip-book simulating a walk through the offices. During differential context conditioning, single screenshots of one context (the anxiety context, CTX+) were paired with the US and therefore called threat elements. Different screenshots of the same context, called nonthreat elements, were never associated with the US. Screenshots of the other context (safety context, CTX−) served as safety elements. Notably, single screenshots are elements of the context representing the distinct view of the participants at the time of US presentation.

We expected conjunctive representation to be reflected in context conditioning effects indicated by differential verbal and electrocortical responses to all screenshots depicting CTX+ (i.e., threat and nonthreat elements) compared with CTX− (safety elements) screenshots. Additional elemental representation should be reflected by differential responses to threat elements as compared with nonthreat elements.

METHODS

Participants

In total, 28 participants (17 women; mean age = 26.00 years, SD = 7.90 years) were included in the final analyses (Table 1). Thirteen additional participants were examined but had to be excluded because of previous experience with context conditioning (n = 1), quitting the experiment (n = 1), incomplete EEG data (n = 2), and too many artifacts (n = 9; see Data Recordings and Data Reduction of EEG section).

Table 1. 
Descriptive Statistics of the Sample
QuestionnaireMean (SD)
n (women) 28 (17) 
n context aware 24 
Age (SD26.00 (7.90) 
ASI (SD15.00 (7.17) 
STAI trait (SD38.32 (9.84) 
STAI state preexperiment (SD37.64 (8.36) 
STAI state postexperiment (SD39.07 (8.12) 
PANAS PA preexperiment (SD28.11 (5.82) 
PANAS PA postexperiment (SD24.00 (6.01) 
PANAS NA preexperiment (SD12.39 (2.51) 
PANAS NA postexperiment (SD12.64 (4.60) 
IPQ general presence (SD2.96 (1.84) 
IPQ spatial presence (SD2.88 (1.42) 
IPQ involvement (SD2.09 (1.21) 
IPQ experienced realism (SD2.06 (1.20) 
QuestionnaireMean (SD)
n (women) 28 (17) 
n context aware 24 
Age (SD26.00 (7.90) 
ASI (SD15.00 (7.17) 
STAI trait (SD38.32 (9.84) 
STAI state preexperiment (SD37.64 (8.36) 
STAI state postexperiment (SD39.07 (8.12) 
PANAS PA preexperiment (SD28.11 (5.82) 
PANAS PA postexperiment (SD24.00 (6.01) 
PANAS NA preexperiment (SD12.39 (2.51) 
PANAS NA postexperiment (SD12.64 (4.60) 
IPQ general presence (SD2.96 (1.84) 
IPQ spatial presence (SD2.88 (1.42) 
IPQ involvement (SD2.09 (1.21) 
IPQ experienced realism (SD2.06 (1.20) 

NA = negative affect; PA = positive affect.

Twenty-four participants were considered contingency aware as they were able to identify the anxiety context (see Procedure and Design section). As contingency awareness could influence cue as well as context conditioning (Andreatta, Glotzbach-Schoon, et al., 2015; Sevenster, Beckers, & Kindt, 2014), we also analyzed the 24 aware participants (17 women) only. As these results did not change, we report results for all 28 participants. The study was approved by the local ethics committee of Psychology.

Stimulus Material

Unconditioned Stimuli

Mildly painful electric shocks served as USs and were applied on all participants’ right inner forearm. The electric stimulus was delivered at a frequency of 50 Hz and a duration of 200 msec generated by a constant current stimulator (Digitimer DS7A, Digitimer Ltd). To guarantee that the stimulus was mildly painful for each participant, we determined the individual pain threshold. Accordingly, starting with 0 mA, we applied stimuli with increasing intensity in steps of 0.5 mA until the participant rated them on an 11-point Likert scale (0 = no sensation at all, 4 = pain perceptible, 10 = unbearable pain) with equal or higher than 4. Afterward, we decreased the intensity in steps of 0.5 mA until the participant rated the stimulation lower than 4. The increasing and decreasing procedure was repeated a second time. Then, the mean of the four threshold intensities of the increasing and decreasing procedures was calculated. To make sure that the stimulus during the experiment remained slightly above perceptible pain, 30% of the stimulation intensity was added to the calculated mean. This electric stimulus was finally applied and rated by the participant. The mean intensity was 1.43 mA (SD = 0.70), and the mean rating was 6.14 (SD = 1.21). The experimenter explained that all electric stimuli applied during the experiment will be adapted to this determined stimulation intensity. Participants were instructed that, by paying attention, they might be able to predict the electric stimuli.

Elements

The conditioned stimuli were 49 screenshots of two virtual offices, respectively. The offices contained comparable amounts of furniture and similar size of the floor plan, but individual features that rooms were easily distinguishable. Screenshots were recorded every 2 sec while a person was guided on predetermined paths through the contexts inspecting the rooms in first-person perspective. Paths started in a corridor in front of the respective context and ended in the corridor. Twelve screenshots (threat elements) of one context, the anxiety (CTX+), became associated with the US. Twelve different screenshots (nonthreat elements) depicting the same context were never associated with the US. Additionally, 12 screenshots of the second context, the safety context (CTX−), were chosen as safety elements. The screenshots were selected as soon as the whole room was recognizable and when participants moved through the room, but not at the very end of the trial when only the door and/or the corridor was depicted. Please note that the time interval between the delivery of USs during acquisition was always more than 7.5 sec. In a pilot study (see Genheimer, 2014), 14 participants (eight women, age: M = 24.7, SD = 3.97) rated the screenshots on 9-point Likert scales according to valence (1 = unpleasant, 9 = pleasant), arousal (1 = calm, 9 = intense), and complexity (1 = figure-ground-contrast, 9 = scenery). Brightness and entropy of the screenshots were measured with Adobe Photoshop CS4 (Version 11.0, Munich, Germany). Separated ANOVAs were calculated containing the within-subject factors Context (Room 1, Room 2) and Screenshot Cluster (Cluster 1, Cluster 2 indicating which of the 2 × 12 selected screenshots of each context were used as threat and nonthreat elements or as safety elements) for valence, arousal, complexity, brightness, and entropy, respectively. Statistics revealed no differences in the above-mentioned measures between the screenshot clusters (all ps > .061), indicating a reliable set of stimuli. Hence, for counterbalancing the stimulus material, four different experimental runs were created. Anxiety and safety contexts were permuted, as well as the stimulus clusters that represented threat and nonthreat elements as well as safety elements. Each participant was assigned to one experimental run. All participants saw the exact same stimuli.

Measures

Ratings

Participants rated the contexts regarding valence (How positive or negative was this room for you?) from 0 = very negative to 50 = neutral to 100 = very positive, arousal (How arousing was this room for you?) from 0 = no arousal to 100 = very high arousal, anxiety (How anxious did you feel in this room?) from 0 = no anxiety until 100 = very high anxiety, and contingency (How likely did you receive an electric shock in this room?) from 0% = very unlikely to 100% = very likely by moving a red arrow below the scale with the computer keyboard. Before each rating, a picture of the respective context was presented for 2 sec.

Additionally, we assessed contingency ratings for the three categories of screenshots (threat, nonthreat, and safety elements) with the following question presented on the computer screen: Have you received an electric shock at this place of the room earlier? Using the respective buttons on the keyboard, participants could select their answer on an 11-point Likert scale from 0 (surely no US) until 10 (surely US).

Electroencephalography

The EEG was recorded with 28 Ag–AgCl electrodes according to the international 10–20 system (Fp1, Fp2, F7, F3, Fz, F4, F8, FC5, FC1, FC2, FC6, T7, C3, Cz, C4, T8, CP5, CP1, CP2, CP6, P7, P3, Pz, P4, P8, O1, Oz, O2). The electrodes were fixed in a cap and positioned on the scalp (Acticap, Brain Products GmbH). Additionally, horizontal eye movements were registered by two electrodes placed on the left and right canthi, and vertical eye movements were registered by two electrodes placed centrally above and beneath the left eye. The electrode Cz was used as online reference and Fz as ground electrode. An electrolyte gel (EASYCAP GmbH) was used to facilitate the conductance of the electric signal from the scalp. We cautiously exfoliated the scalp to get impedances below 5 kΩ. This was controlled by Acticap control software (Version 1.2.2.0, Brain Products GmbH). The sampling rate was 1000 Hz. An online Notch filter of 50 Hz was applied. Data were recorded using Vision Recorder software (Version 1.20, Brain Products GmbH).

Procedure and Design

The experiment was programmed with Presentation software (Version 15.1, Neurobehavioral Systems, Inc.). All instructions, pictures, and rating scales were presented on a Powerwall, which is a large screen of 2 m in height and 3.22 m in length, allowing us to show contextual stimuli and furniture in approximately full size. During the experiment, participants sat about 2 m in front of the Powerwall. At the beginning, participants read the study instructions and gave their informed consent. The experimenter answered potentially raised questions. Afterward, questionnaires about demographical data, Anxiety Sensitivity Index (ASI; Alpers & Pauli, 2001; Reiss, Peterson, Gursky, & McNally, 1986), State–Trait Anxiety Inventory (STAI) state and trait (Laux, Glanzmann, Schaffner, & Spielberger, 1981; Spielberger, Gorsuch, & Edward, 1970), as well as Positive and Negative Affect Schedule (PANAS; Krohne, Egloff, Kohmann, & Tausch, 1996; Watson, Clark, & Tellegen, 1988), were completed. Then, the experimenter applied the EEG cap, the electrodes for electric stimulation, and the pain threshold procedure was explained and performed.

The entire experiment consisted of a habituation phase, two acquisition phases, two test phases, and postexperimental contingency ratings for elements (Figure 1). During habituation, participants saw each context once meaning 49 screenshots of one office followed by 49 screenshots of the other office. Each screenshot was presented for 1000 msec with an ISI with black screen of 300–700 msec. The series of all 49 screenshots of one context is referred to as context run. After habituation, participants rated valence, arousal, and anxiety of both contexts.

Figure 1. 

Experimental procedure. During habituation, the two contexts were presented once, that is, the 49 subsequent screenshots of each context were presented in predefined order for 1 sec each with an ISI of 300–700 msec. During each of the two acquisition phases, the two contexts were presented three times as in the habituation phase; however, 12 screenshots (threat elements, red frame) of the anxiety context (CTX+) were followed by mildly aversive electric stimuli (USs). Twelve comparable CTX+ screenshots (nonthreat elements, orange frame) and 12 screenshots of the safety context (CTX−; safety elements, green frame) served as control stimuli. The two test phases were identical with the acquisition phases, but without USs. Ratings were assessed after habituation, Acquisitions 1 and 2, and Tests 1 and 2. At the test phases, 12 threat, 12 nonthreat, and 12 safety screenshots were presented once for 2 sec in randomized order, and participants rated the likelihood of having received a US following this specific screenshot during the experiment.

Figure 1. 

Experimental procedure. During habituation, the two contexts were presented once, that is, the 49 subsequent screenshots of each context were presented in predefined order for 1 sec each with an ISI of 300–700 msec. During each of the two acquisition phases, the two contexts were presented three times as in the habituation phase; however, 12 screenshots (threat elements, red frame) of the anxiety context (CTX+) were followed by mildly aversive electric stimuli (USs). Twelve comparable CTX+ screenshots (nonthreat elements, orange frame) and 12 screenshots of the safety context (CTX−; safety elements, green frame) served as control stimuli. The two test phases were identical with the acquisition phases, but without USs. Ratings were assessed after habituation, Acquisitions 1 and 2, and Tests 1 and 2. At the test phases, 12 threat, 12 nonthreat, and 12 safety screenshots were presented once for 2 sec in randomized order, and participants rated the likelihood of having received a US following this specific screenshot during the experiment.

Participants were instructed that they would be able to predict the electric stimulus if they attentively followed the experiment; however, no further specification regarding the office or the screenshots was given. During the first acquisition phase (A1), participants saw the series of 49 screenshots of both contexts again, three times for each office (order: CTX+, CTX−, CTX+, CTX−, CTX−, CTX+). In one office (anxiety context, CTX+), 12 screenshots (threat elements) were paired once with the US delivered at the screenshot’s offset. Thus, in summary, 12 US were applied during A1, three to five USs per context run. A screenshot was labeled as threat element as soon as it was paired with the US. Twelve comparable screenshots of the CTX+ served as nonthreat elements, and 12 comparable screenshots of the CTX− constituted the safety elements. A1 ended with ratings of valence, arousal, anxiety, and contingency for each context.

The second acquisition phase (A2) followed the same procedure, however, with a different context run order: CTX−, CTX+, CTX+, CTX−, CTX+, CTX−. As in A1, the same 12 threat elements of CTX+ were followed by a US, the same 12 nonthreat elements of CTX+ and the same 12 safety elements of CTX− were used for analyses. For the assessment of contingency awareness, we asked participants after the acquisition in which room they received electric stimuli, that is, please answer verbally in which room electric stimuli were applied. Twenty-four participants were considered contingency aware as they correctly and distinctively described the anxiety context to the experimenter.

The two test phases (T1 and T2) were identical and similar to the acquisition phases, but with new context run orders (T1: CTX−, CTX+, CTX−, CTX+, CTX+, CTX−; T2: CTX+, CTX−, CTX−, CTX+, CTX−, CTX+). The US electrode was still attached to participants’ inner forearm, and no information about the US were provided.

Finally, the postexperimental contingency ratings were assessed by presenting the 12 threat, 12 nonthreat, and 12 safety elements for 2 sec in random order. Participants were asked to rate the likelihood that the US occurred after exactly this screenshot of this context earlier in the experiment.

Anxiety and safety context as well as threat, nonthreat, and safety elements were counterbalanced across participants. The experiment ended with the completion of the PANAS (Krohne et al., 1996; Watson et al., 1988), STAI state (Laux et al., 1981; Spielberger et al., 1970), and Igroup Presence Questionnaire (IPQ; Schubert, Friedmann, & Regenbrecht, 2001).

Data Recordings and Data Reduction of EEG

EEG data preprocessing was realized with BrainVision Analyzer 2.1 software (Brain Products, Inc.). First, an average reference was calculated by using all electrodes except the eye electrodes. A bandpass filter from 0.1 to 35Hz was applied. The signal was ocular corrected according to Gratton, Coles, and Donchin (1983), as implemented in BrainVision Analyzer. Afterward, temporal segments from 100 msec before until 1000 msec after each screenshot onset were extracted, baseline-corrected, and examined for artifacts; rejection criteria allowed a maximal voltage step of 50 μV, a minimal amplitude of −100 μV and a maximal amplitude of 100 μV (see Wieser, Pauli, Reicherts, & Mühlberger, 2010). Then, segments were averaged across each experimental conditions. Please note that participants were only included in the analyses, when more than 50% of all trials per condition could be used. The average number of analyzed trials was above 90% for each condition.

For P100 analysis, Wieser, Mühlberger, Kenntner-Mabiala, and Pauli (2006) suggest a peak latency in a time window of 90–150 msec after stimulus onset in participants between 30 and 40 years. Similarly, we performed a peak detection between 80 and 160 msec after stimulus onset. Peaks occurred on average between 94 and 105msec after stimulus onset. The mean amplitudes of these peaks of the occipital electrodes O1, Oz, and O2 were exported and averaged per condition. For the analysis of the EPN, we exported the mean amplitudes of the time window between 180 and 240 msec after stimulus onset. Wieser et al. (2006) also analyzed the electrode cluster P7, P8, O1, and O2 but used a slightly different time interval of 192–232 msec. An additional analysis with this time window revealed qualitatively similar EPN results confirming the validity of our findings. The Pz electrode was used for LPP analyses. Here, we exported the mean amplitudes for the time interval between 400 and 700msec for the early LPP as well as the time interval between 700 and 1000 msec for late LPP, respectively (Kastner et al., 2016; Schupp, Junghöfer, et al., 2004).

Statistical Analyses

IBM SPSS Statistics 23 (IBM Corporation) was used for the calculation of all analyses.

Ratings

Ratings of the contexts (valence, arousal, anxiety, and contingency) were analyzed with paired samples t tests for the habituation phase and with repeated-measures 2 × 2 ANOVAs with the within-subject factors Context (CTX+, CTX−) and Time (A1, A2 or T1, T2) for the acquisition and test phases. Postexperimental contingency ratings of the elements were analyzed by an ANOVA with the within-subject factor Element consisting of three levels (threat, nonthreat, safety). Please note that only 16 participants could be included in this analysis. For all other participants, the study ended directly after the test phase, and no ratings of the single elements were assessed.

Physiology

EEG findings are reported according to Picton et al. (2000). The habituation phase was analyzed separately for each ERP component with ANOVAs with the within-subject factor Element (threat, nonthreat, safety), for the EPN with the additional factor Hemisphere (left, right). The acquisition and the test phases were analyzed for P1 and LPPs with 2 × 3 ANOVAs with the within-subject factors Phase (A1, A2 or T1, T2) and Element (threat, nonthreat, safety) and for the EPN with a 2 × 2 × 3 ANOVA, including the additional factor Hemisphere (left, right). Please note that in A1 a screenshot did not become a threat element until it has been paired with a US in the previous context run. The first run contained four, the second run three, and the third run five USs. Subsequently, four threat elements in the second run and seven threat elements in the third run were analyzed. This resulted in 11 threat screenshots in A1, which were compared with a total of 36 nonthreat and 36 safety elements per participant. In A2, all threat elements have been paired with a US previously and therefore were composed of 36 screenshots.

All significant main and interaction effects were followed up by post hoc t tests. Greenhouse–Geisser corrections were applied in case of violating the sphericity assumption, and Greenhouse–Geisser Epsilon (GG-ε) is reported. The α level was set at .05. Partial eta square is reported for effect sizes.

RESULTS

Sample Characteristics

The analyses of all trait questionnaires including ASI (Alpers & Pauli, 2001; Reiss et al., 1986), the trait version of the STAI (Laux et al., 1981; Spielberger et al., 1970), and IPQ (Schubert et al., 2001) revealed unremarkable scores (Table 1). Please note that one participant did not fill in the second page of the IPQ; therefore, these values were interpolated on the basis of the grand mean of all other participants. STAI state (Laux et al., 1981; Spielberger et al., 1970) and negative affect (NA) (PANAS; Krohne et al., 1996) did not change over the course of the experiment (STAI: t(27) = 1.51, p = .142; NA: t(27) = 0.34, p = .733). The positive affect (PA) was reduced after the experiment compared with before, t(27) = 5.22, p < .001, which is in line with previous aversive conditioning experiments (Andreatta, Leombruni, Glotzbach-Schoon, Pauli, & Mühlberger, 2015).

Responses to Contextual Elements

ERP Responses to Contextual Elements

The analyses of the ERP components P100, EPN, early LPP, and late LPP during the habituation phase returned no significant element effects (all ps > .179), indicating an appropriate starting situation for further analyses.

P100

During acquisition, the analyses revealed a main effect of Phase, F(1, 27) = 10.88, p = .003, ηp2 = .287, indicating higher P100 amplitudes during A1 compared with A2. The main effect of Element, F(2, 54) = 6.73, p = .002, ηp2 = .199 (Figure 2A), derived from higher P100 amplitudes for threat compared with safety, t(27) = 3.50, p = .002, elements and for nonthreat compared with safety, t(27) = 2.77, p = .010, elements; threat and nonthreat elements elicited similar amplitudes, t(27) = 1.04, p = .309. The interaction of Phase × Element was not significant, F(2, 54) =1.56, p = .220, ηp2 = .054. These results indicate differential processing of CTX+ (both threat and nonthreat elements) versus CTX− screenshots, suggesting conjunctive context representation. During the test phase, neither the effect of phase nor of element nor the interaction Phase × Element reached significance (all ps > .069), which could indicate a weak extinction process.

Figure 2. 

Mean ERPs during acquisition. Depicted are P100 amplitudes averaged for the electrodes O1, Oz, and O2 (A) and EPN amplitudes averaged for the electrodes of the left hemisphere P7 and O1 (B) elicited by threat (black line), nonthreat (dark gray line), and safety (light gray line) elements.

Figure 2. 

Mean ERPs during acquisition. Depicted are P100 amplitudes averaged for the electrodes O1, Oz, and O2 (A) and EPN amplitudes averaged for the electrodes of the left hemisphere P7 and O1 (B) elicited by threat (black line), nonthreat (dark gray line), and safety (light gray line) elements.

Early posterior negativity

A main effect of Hemisphere, F(1, 27) = 4.20, p = .050, ηp2 = .135, indicated overall lower EPN amplitudes in the left compared with the right hemisphere. Additionally, a trend of a main effect of Element, F(2, 54) = 2.69, p = .077, ηp2 = .090, was revealed. The explorative post hoc tests showed that both threat, t(27) = 1.91, p = .067, and nonthreat, t(27) = 1.64, p = .112, elements elicited slightly more positive EPN amplitudes compared with safety elements. The comparison between threat and nonthreat elements returned no significance, t(27) = 0.84, p = .411. As revealed by further explorative analyses, this effect was driven by the left hemisphere. Thus, separated analyses for each hemispheres returned a main effect of Element in the left, F(2, 54) = 3.94, GG-ε = .814, p = .034, ηp2 = .127, but not in the right, F(2, 54) = 1.37, p = .262, ηp2 = .048, hemisphere. Post hoc t tests for the left hemisphere showed more positive EPN amplitudes for threat, t(27) = 2.13, p = .042, as well as for nonthreat, t(27) = 2.38, p = .025, elements compared with safety elements. EPN amplitudes elicited by threat and nonthreat elements did not differ, t(27) = 0.38, p = .704. Tests for the right hemisphere indicated a similar pattern regarding lowest amplitude for safety elements and most positive amplitude for threat elements, but not statistically significant (all ps > .162). These results suggest electrocortical differentiation between threat and nonthreat CTX+ versus CTX− elements speaking for conjunctive context representation. During the test phases, neither significant main effects of Phase, Hemisphere, and Element nor their interaction became significant (all ps > .082), also indicating slow extinction.

Early (400–700 msec) and late (700–1000 msec) LPP.

The analyses of the early LPP revealed no significant effect, neither main effect of Phase, nor Element, nor an interaction of both during acquisition (all ps > .492) and during test (all ps > .126). Similarly, the analyses of the late LPP component returned neither significant effect of Phase nor of Element nor the interaction (all ps > .633). Subsequently, the ANOVA for the test phase led to nonsignificant main effects of Phase, Element, and interactions of Phase × Element (all ps > .160).

Postexperimental Contingency Ratings for Elements

The main effect of Element, F(2, 30) = 40.66, GG-ε = .593, p < .001, ηp2 = .731, was followed up with post hoc t tests, which revealed higher contingency ratings for threat versus nonthreat elements, t(15) = 2.44, p = .028, or versus safety elements, t(15) = 6.48, p < .001, and for nonthreat compared with safety elements, t(15) = 6.77, p < .001. These results, on the one hand, indicate conjunctive representation as participants rated higher contingencies for all screenshots depicting CTX+ (both threat and nonthreat elements) compared with screenshots comprising CTX− (safety elements). Importantly, results, on the other hand, provide evidence for elemental representation as contingency ratings between threat and nonthreat elements differed, suggesting a very precise differentiation between reinforced and nonreinforced CTX+ elements. Please note that these effects were revealed after the test phases without US presentation, which suggests weak extinction (Figure 3).

Figure 3. 

Postexperimental contingency ratings for the 12 selected screenshots of each context representing elemental representation. Mean US contingency ratings and standard errors (SE) are depicted for CTX+ screenshots either paired with a US (threat elements, black bars) or not paired with a US during acquisition (nonthreat elements, dark gray bars) or matched CTX− control screenshots (safety elements, light gray bars). ***p < .001, *p < .05.

Figure 3. 

Postexperimental contingency ratings for the 12 selected screenshots of each context representing elemental representation. Mean US contingency ratings and standard errors (SE) are depicted for CTX+ screenshots either paired with a US (threat elements, black bars) or not paired with a US during acquisition (nonthreat elements, dark gray bars) or matched CTX− control screenshots (safety elements, light gray bars). ***p < .001, *p < .05.

Responses to Contexts

Contingency Ratings

Analysis of the acquisition phase revealed a significant main effect of Context, F(1, 27) = 95.75, p < .001, ηp2 = .780, and a significant interaction of Phase × Context, F(1, 27) = 7.42, p = .011, ηp2 = .216, indicating successful acquisition (see Figure 4A). Following up the interaction effect, post hoc t tests showed significantly higher contingency ratings for CTX+ compared with CTX− after A1, t(27) = 6.21, p < .001, and after A2, t(27) = 10.70, p < .001. The comparison of both acquisition phases for the contexts separately revealed marginally higher contingency ratings in A2 compared with A1 for CTX+, t(27) = 2.00, p = .056, and nonsignificant differences between A1 and A2 for CTX−, t(27) = 1.38, p = .179. For the test phase, the Phase × Context interaction reached significance too, F(1, 27) = 4.25, p = .049, ηp2 = .136, indicating higher contingency ratings for CTX+ compared with CTX− after T1, t(27) = 4.50, p < .001, as well as after T2, t(27) = 3.56, p = .001. These results indicate incomplete extinction during the test phases (Figure 4A).

Figure 4. 

Ratings of the contexts representing conjunctive representation. Lines (with standard errors) depict ratings of contingency (A), arousal (B), anxiety (C), and valence (D) for anxiety context (CTX+, black lines) and safety context (CTX−, gray lines). Arousal, anxiety, and valence ratings were assessed after habituation (Hab). Contingency, arousal, anxiety, and valence were reported after Acquisition 1 (A1) and Acquisition 2 (A2) and after Test 1 (T1) and Test 2 (T2). ***p < .001, **p < .01, *p < .05.

Figure 4. 

Ratings of the contexts representing conjunctive representation. Lines (with standard errors) depict ratings of contingency (A), arousal (B), anxiety (C), and valence (D) for anxiety context (CTX+, black lines) and safety context (CTX−, gray lines). Arousal, anxiety, and valence ratings were assessed after habituation (Hab). Contingency, arousal, anxiety, and valence were reported after Acquisition 1 (A1) and Acquisition 2 (A2) and after Test 1 (T1) and Test 2 (T2). ***p < .001, **p < .01, *p < .05.

Arousal, Anxiety, and Valence Ratings

The habituation phase (Hab) ratings reveal that participants rated no difference between both contexts regarding valence, arousal and anxiety (all ps > .056; see Figure 4), thus indicating that the contexts were suitable for the subsequent conditioning.

Arousal

Successful acquisition was indicated by a main effect of context, F(1, 27) = 28.92, p < .001, ηp2 = .517, with higher arousal ratings for CTX+ compared with CTX−. The main effect of phase and the interaction of Phase × Context were not significant (all ps > .440). Analysis for the test phase revealed slow extinction learning as verified by a significant interaction of Phase × Context, F(1, 27) = 15.70, p < .001, ηp2 = .368, and significantly higher arousal ratings for CTX+ versus CTX− after T1, t(27) = 5.32, p < .001, as well as after T2, t(27) = 2.31, p = .029 (Figure 4B).

The comparison between T1 and T2 for contexts separately showed significantly higher arousal in T1 compared with T2 in CTX+, t(27) = 3.68, p = .001, but no differences in CTX−, t(27) = 0.27, p = .791.

Anxiety.

Descriptively, the acquisition phase caused higher anxiety ratings for CTX+ compared with CTX− (see Figure 4C); however, the analyses revealed neither a significant effect of Phase, nor Context, nor of Phase × Context (all ps > .257). The ANOVA in the test phase revealed only a main effect of Phase, F(1, 27) = 11.27, p = .002, ηp2 = .295, demonstrating higher anxiety after T1 versus T2. Although descriptively, anxiety ratings were higher for CTX+ compared with CTX− after T1, but similar after T2, the interaction Phase × Context was not significant, F(1, 27) = 2.02, p = .167, ηp2 = .069.

Valence.

At acquisition, the valence ratings only revealed a significant main effect of Phase, F(1, 27) = 7.52, p = .011, ηp2 = .218, indicated by lower pleasantness in A1 compared with A2, which could either indicate habituation to the experimental setting or negative valence induced by aversive anxiety learning. Neither the effect of Context nor the interaction of Phase × Context reached significance (all ps > .171).

For the test phase, we revealed a similar pattern of effects: A main effect of Phase indicated lower valence in T1 compared with T2, F(1, 27) = 11.46, p = .002, ηp2 = .298, indicating a further increase in pleasant valence, likely because of the absence of electric stimuli during the test phase. However, the effect of context and the interaction of Phase × Context were not significant (all ps > .301; Figure 4D).

DISCUSSION

The goal of the current study was to find evidence for elemental and conjunctive representations of contexts in humans with a newly developed flip-book paradigm. Screenshots of two virtual contexts were presented sequentially simulating a first-person perspective movement through virtual environments. Such sequential presentation of these screenshots allowed the registration of verbal responses and ERPs triggered by specific screenshots and to disentangle the dual context representations. Considering these verbal as well as the electrocortical responses, we found evidence for the dual representation of contexts in humans (Rudy, 2009; Rudy et al., 2004; Nadel & Willner, 1980). First, evidence for an elemental representation of the threat context is reflected in the fact that participants were able to discriminate threat and nonthreatening elements of the anxiety context as indicated by higher contingency ratings for screenshots followed by threat compared with screenshots of the same context not followed by threat. Second, conjunctive representation of the threat context was revealed by higher arousal and contingency ratings as well as larger P100 and EPN amplitudes for both threat-related and threat-unrelated CTX+ elements compared with CTX− elements. Thus, the newly developed flip-book paradigm revealed clear evidence for elemental representation of the threat context on an explicit verbal level and for conjunctive representation on an explicit verbal and electrophysiological level.

The increased verbal and electrophysiological responses to all elements of the anxiety context compared with elements of the safety context corroborates and extends previous context conditioning studies (Glotzbach-Schoon, Andreatta, Mühlberger, & Pauli, 2013). Ratings revealed the comparability of our new flip-book paradigm with previous context conditioning studies (Andreatta, Leombruni, et al., 2015; Baas, Nugent, Lissek, Pine, & Grillon, 2004); successful context conditioning was consistently indicated by higher US expectancy and higher arousal ratings in CTX+ versus CTX−. This study also extends previous findings, as it is the first to reveal electrophysiological indicators of conjunctive representation of contexts. Specifically, we observed similar and increased P100 amplitudes triggered by both threat-related and threat-unrelated CTX+ screenshots compared with CTX− screenshots, whereas such differences were weaker for the EPN and not observed for the LPP. Thus, elements of a context were revealed in rather early ERP components and independent of a specific threat association emphasizing a conjunctive representation of contexts.

In detail, the P100 has been investigated in studies presenting pictures such as faces or checkerboards on a computer screen and is mainly interpreted as reflecting early attentional processes with higher amplitudes for attended compared with unattended stimuli (Wieser et al., 2010; Clark & Hillyard, 1996). The P100 amplitude may also be modulated by physical properties of the stimuli, which we think is unlikely here as stimulus complexity and presentation size were carefully controlled (Genheimer, 2014). Therefore, a plausible explanation for the P100 indicating conjunctive context representation is the higher arousal during acquisition in CTX+, which could have increased the overall attention for CTX+ compared with CTX−. Moreover, generalization effects have to be discussed. As threat and nonthreat elements depict the same context and therefore the same furniture and arrangements of the environment, the generalization of threat responses to all screenshots depicting CTX+ compared with CTX− might be obvious. However, all screenshots show the furniture in different angles and perspectives, which makes every screenshot a unique element as part of the entire context. Additionally, the timing is another important aspect, which represents a context (Maren et al., 2013). As threat elements and therefore the US administrations were distributed among the entire CTX+ phase, the second acquisition phase could have elicited increased attention when participants expected a US according to the context of time. Subsequently in the test phase without US delivery, the arousal level decreased in CTX+, and therefore, similar levels of attention were reflected in similar P100 amplitudes for all screenshots.

In line, the EPN was shown to be strongly modulated by the emotional content of a stimulus (Schupp, Junghöfer, et al., 2004). Here, the threat in the anxiety context could imply a generalized higher emotional relevance in comparison to the safety context resulting in higher EPN amplitudes. Our results may be best compared with Junghöfer et al. (2001) and Schupp, Junghöfer, et al. (2004), who found more negative EPN amplitudes elicited by high versus low arousing pictures as well as for both pleasant and unpleasant pictures as compared with neutral pictures, whereby pleasant pictures elicited even more negative EPN than unpleasant stimuli. In contrast to Schupp, Junghöfer, et al. (2004), the design of the current study did not include neutral stimuli as the elements belonged either to the anxiety or to the safety context. Importantly, safety elements were rated as slightly pleasant and clearly low arousing, possibly because they predicted safety (Gazendam, Kamphuis, & Kindt, 2013). On the contrary, threat and nonthreat elements were rated as unpleasant and high arousing because they were associated with the threat. We found more negative EPN amplitudes to CTX+ versus CTX− screenshots, but importantly to both threat-related and threat-unrelated CTX+ screenshots. Thus, we interpret this as electrophysiological evidence for successful context conditioning and conjunctive representation of the context. The left hemisphere localization of this effect is likely because we only included right-handed participants who received electric stimuli on their right forearm, and previous studies indicated enhanced stimulus processing in the contralateral hemisphere of the shock application site (Andreatta, Glotzbach-Schoon, et al., 2015). The EPN results in the left hemisphere might be explained by the right visual field advantage for the processing of objects (Garcea, Almeida, & Mahon, 2012). In line, Stegmann, Reicherts, Andreatta, Pauli, and Wieser (2019) found activation patterns of contextual visual stimuli slightly shifted to the left side of the cortex. Correspondingly, with regard to the attentional systems, the right hemisphere codes for stimuli located in both the left and the right visual field, whereas the left hemisphere more specifically codes for locations in the right visual field (Corbetta & Shulman, 2002). Therefore, greater variance in the right compared with the left hemisphere in the current data could be the reason for EPN effects in the left hemisphere only.

Based on a combined EEG and fMRI study revealing a modest negative correlation between EPN amplitude and amygdala as well as ACC activity (Sabatinelli, Keil, Frank, & Lang, 2013), we further speculate that, in the current experiment, both threat and nonthreat elements of CTX+ were associated with increased amygdala activity. A previous study, which revealed hippocampus but not amygdala activity for conjunctive representations, used pictures of contexts containing the same elements in a different arrangement. The associative learning process between an arrangement and an aversive event during conditioning only implicated conjunctive context learning (Stout et al., 2018). In contrast, our paradigm contained both elemental and conjunctive context information requiring amygdala and hippocampus activity at the same time. In line, previous research found ongoing amygdala and hippocampus activity during presentation of a threat context (Andreatta, Glotzbach-Schoon, et al., 2015; Rudy, 2009). Additionally, LeDoux and Pine (2016) emphasized the tight connection between hippocampus and amygdala during threat perception. Therefore, we think that our interpretation is plausible but definitely needs further confirmation.

In contrast to our hypothesis, we found no modulation of LPP amplitudes, that is, neither by elemental stimuli nor by contexts. One reason could be the short ISI. The 35-Hz low-pass filter might have affected the baseline for the following stimulus, and therefore, general reduction in LPP was found. Moreover, LPP amplitudes mainly reflect motivational classification with higher amplitudes indicating stronger motivational salience of stimuli (Kastner et al., 2016; Bradley et al., 2007; Schupp, Junghöfer, et al., 2004). Perhaps both contexts had similar motivational salience, the CTX+ of avoidance and the CTX− of approach, and therefore elicited comparable LPP increases. Future studies should incorporate a neutral context to be able to detect such effects.

Our finding that the threat versus the nonthreat elements of the anxiety context differed in contingency ratings but not in electrocortical responses stands at a first glance in contrast to previous multiCS conditioning studies. In multiCS conditioning studies (Steinberg et al., 2013), stimuli of one category (e.g., faces) are presented sequentially in a random order, and some are followed by a US and some are not. Such studies mainly found no difference between threat-related compared with threat-unrelated stimuli in CS–US contingency ratings despite differential electrocortical responses to threat-related stimuli. The main difference between the multiCS versus the flip-book paradigm is the mode of stimuli presentation during acquisition, that is, a random unrelated sequence of quite similar stimuli versus a sequential fixed order of contextually linked stimuli, respectively. Thus, the flip-book paradigm may have allowed participants to predict an upcoming threat element based on the temporal sequence of the screenshots, that is, the temporal context, and therefore, they may have been able to focus attention on details of the displayed elements and subsequently may have been able to memorize these screenshots on an explicit level. Alternatively, a figure-ground phenomenon may be considered, that is, elements that have been part of the background context may have stepped to the foreground because of their predictive meaning for the US (Rudy, 2009) and therefore were preferentially processed and memorized. However, we have to admit that both explanations assume a differential processing of threat versus nonthreat elements, which should be reflected in differential ERPs. Again, we speculate that the methodological specificities may play a crucial role why we did not find such ERP differences. In our flip-book study, both the threat and nonthreat elements belong to the same anxiety context, which is distinct from the safety context, and this may have dampened ERP differences within CTX+; in multiCS studies, there is no second context, which may increase ERP differences between elements. In addition, the number of averaged trials is limited in this study compared with multiCS studies, and therefore, ERP differences may be relatively small and covered by relatively large noise. Future studies should consider and explore these speculations.

Though a study by Duncan et al. (2018) suggested that people rely more on conjunctive rather than elemental learning, we have to consider that we even emphasized conjunctive representations of the context because of the preceding habituation phase. Here, we allowed participants to explore contexts and to develop a conjunctive representation even before conditioning. Indeed, animal studies demonstrated a hippocampus dependent preexposure facilitation effect (Fanselow, 1990, 2010), meaning that if such a representation exists already, only a small component of the context can activate the entire context representation via pattern completion or can distinguish the contexts via pattern separation (Rudy, 2009; Rudy et al., 2004; O'Reilly & Rudy, 2001; Rudy & O’Reilly, 2001). Therefore, we suspect that future flip-book studies without habituation phase will be more likely to find elemental representation effects on ERPs.

Finally, we have to discuss that the contingency ratings for elements were collected postexperimentally, after the test phases of extinction. This indicates that, on a verbal explicit level, extinction of contextual conditioning is rather slow. The contingency ratings of the contexts assessed after the test phases also indicate slow extinction learning on an explicit verbal level. The accurate contingency ratings emphasize the important role of elemental representations.

The flip-book paradigm provides insights into the ongoing stimulus processing by continuously eliciting ERPs during context presentation and at the same time disentangling responses to individual screenshots, which compose the whole context. The paradigm is based on fleeting imaging studies (Junghöfer et al., 2001, 2006) and provides a powerful and exact measure of the electrocortical responses to one single shortly presented stimulus without transmission of the signal to the subsequent stimulus. One could argue that screenshots of a context still contain conjunctive information and therefore might require conjunctive representation besides the elemental representation. As such, future studies could take advantage of VR technology and present single virtual elements of a context instead of screenshots to disentangle elemental from conjunctive representation.

In summary, this is the first human study that directly compared elemental and conjunctive representations of contexts integrating verbal and electrocortical responses. Results suggest that in humans both elemental and conjunctive context representations contribute to conditioned contextual anxiety. Evidence for elemental representation is based on enhanced contingency ratings for threat elements compared with nonthreat elements of the anxiety context. Evidence for conjunctive representation arises from increased contingency and arousal ratings as well as increased P100 amplitudes for threat-related and threat-unrelated CTX+ versus CTX− elements. We conclude that the flip-book paradigm considers conjunctive as well as elemental context representations in a well-controlled manner and is therefore suitable for further investigations to shed light into the electrocortical processing of contexts and elements. Future studies could use the flip-book paradigm and investigate requirements, dispositions, and individual differences that lead to elemental or conjunctive context representation. The treatment of patients with impaired hippocampal functioning or PTSD may profit from the understanding of underlying mechanisms of contextual processing.

Acknowledgments

The work was supported by the Collaborative Research Center “Fear, Anxiety, Anxiety Disorders,” SFB-TRR 58 project B01. The study is part of H. G.’s dissertation project. We cordially thank Katrin Freundorfer for her help in data collection.

Conflict of Interest: P. P. is shareholder of a commercial company that develops virtual environment research systems. No further competing financial interests are declared.

Reprint requests should be sent to Hannah Genheimer, Department of Psychology I (Biological Psychology, Clinical Psychology and Psychotherapy), University of Würzburg, Marcusstr. 9-11, 97070 Würzburg, Germany, or via e-mail: hannah.genheimer@uni-wuerzburg.de.

REFERENCES

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
,
674
685
.
Alpers
,
G.
, &
Pauli
,
P.
(
2001
).
Angstsensitivitäts-Index
.
Würzburg
:
Julius-Maximilians-Universität
.
Amrhein
,
C.
,
Mühlberger
,
A.
,
Pauli
,
P.
, &
Wiedemann
,
G.
(
2004
).
Modulation of event-related brain potentials during affective picture processing: A complement to startle reflex and skin conductance response?
International Journal of Psychophysiology
,
54
,
231
240
.
Andreatta
,
M.
,
Glotzbach-Schoon
,
E.
,
Mühlberger
,
A.
,
Schulz
,
S. M.
,
Wiemer
,
J.
, &
Pauli
,
P.
(
2015
).
Initial and sustained brain responses to contextual conditioned anxiety in humans
.
Cortex
,
63
,
352
363
.
Andreatta
,
M.
,
Leombruni
,
E.
,
Glotzbach-Schoon
,
E.
,
Pauli
,
P.
, &
Mühlberger
,
A.
(
2015
).
Generalization of contextual fear in humans
.
Behavior Therapy
,
46
,
583
596
.
Baas
,
J. M.
,
Nugent
,
M.
,
Lissek
,
S.
,
Pine
,
D. S.
, &
Grillon
,
C.
(
2004
).
Fear conditioning in virtual reality contexts: A new tool for the study of anxiety
.
Biological Psychiatry
,
55
,
1056
1060
.
Baeuchl
,
C.
,
Meyer
,
P.
,
Hoppstädter
,
M.
,
Diener
,
C.
, &
Flor
,
H.
(
2015
).
Contextual fear conditioning in humans using feature-identical contexts
.
Neurobiology of Learning and Memory
,
121
,
1
11
.
Bradley
,
M. M.
,
Hamby
,
S.
,
Löw
,
A.
, &
Lang
,
P. J.
(
2007
).
Brain potentials in perception: Picture complexity and emotional arousal
.
Psychophysiology
,
44
,
364
373
.
Bröckelmann
,
A.-K.
,
Steinberg
,
C.
,
Elling
,
L.
,
Zwanzger
,
P.
,
Pantev
,
C.
, &
Junghöfer
,
M.
(
2011
).
Emotion-associated tones attract enhanced attention at early auditory processing: Magnetoencephalographic correlates
.
Journal of Neuroscience
,
31
,
7801
7810
.
Clark
,
V. P.
, &
Hillyard
,
S. A.
(
1996
).
Spatial selective attention affects early extrastriate but not striate components of the visual evoked potential
.
Journal of Cognitive Neuroscience
,
8
,
387
402
.
Corbetta
,
M.
, &
Shulman
,
G. L.
(
2002
).
Control of goal-directed and stimulus-driven attention in the brain
.
Nature Reviews Neuroscience
,
3
,
201
215
.
Duncan
,
K.
,
Doll
,
B. B.
,
Daw
,
N. D.
, &
Shohamy
,
D.
(
2018
).
More than the sum of its parts: A role for the hippocampus in configural reinforcement learning
.
Neuron
,
98
,
645
657
.
Fanselow
,
M. S.
(
1990
).
Factors governing one-trial contextual conditioning
.
Animal Learning & Behavior
,
18
,
264
270
.
Fanselow
,
M. S.
(
2010
).
From contextual fear to a dynamic view of memory systems
.
Trends in Cognitive Sciences
,
14
,
7
15
.
Garcea
,
F. E.
,
Almeida
,
J.
, &
Mahon
,
B. Z.
(
2012
).
A right visual field advantage for visual processing of manipulable objects
.
Cognitive, Affective, & Behavioral Neuroscience
,
12
,
813
825
.
Gazendam
,
F. J.
,
Kamphuis
,
J. H.
, &
Kindt
,
M.
(
2013
).
Deficient safety learning characterizes high trait anxious individuals
.
Biological Psychology
,
92
,
342
352
.
Genheimer
,
H.
(
2014
).
Fear and anxiety in virtual reality: Investigations of cue and context conditioning in virtual environment
.
Weisbaden
:
Springer
.
Glenn
,
D. E.
,
Risbrough
,
V. B.
,
Simmons
,
A. N.
,
Acheson
,
D. T.
, &
Stout
,
D. M.
(
2018
).
The future of contextual fear learning for PTSD research: A methodological review of neuroimaging studies
.
Current Topics in Behavioral Neurosciences
,
38
,
207
228
.
Glotzbach-Schoon
,
E.
,
Andreatta
,
M.
,
Mühlberger
,
A.
, &
Pauli
,
P.
(
2013
).
Context conditioning in virtual reality as a model for pathological anxiety
.
e-Neuroforum
,
4
,
63
70
.
Glotzbach-Schoon
,
E.
,
Tadda
,
R.
,
Andreatta
,
M.
,
Tröger
,
C.
,
Ewald
,
H.
,
Grillon
,
C.
, et al
(
2013
).
Enhanced discrimination between threatening and safe contexts in high-anxious individuals
.
Biological Psychology
,
93
,
159
166
.
Gratton
,
G.
,
Coles
,
M. G. H.
, &
Donchin
,
E.
(
1983
).
A new method for off-line removal of ocular artifact
.
Electroencephalography and Clinical Neurophysiology
,
55
,
468
484
.
Hillyard
,
S. A.
, &
Münte
,
T. F.
(
1984
).
Selective attention to color and location: An analysis with event-related brain potentials
.
Perception & Psychophysics
,
36
,
185
198
.
Junghöfer
,
M.
,
Bradley
,
M. M.
,
Elbert
,
T. R.
, &
Lang
,
P. J.
(
2001
).
Fleeting images: A new look at early emotion discrimination
.
Psychophysiology
,
38
,
175
178
.
Junghöfer
,
M.
,
Sabatinelli
,
D.
,
Bradley
,
M. M.
,
Schupp
,
H. T.
,
Elbert
,
T. R.
, &
Lang
,
P. J.
(
2006
).
Fleeting images: Rapid affect discrimination in the visual cortex
.
NeuroReport
,
17
,
225
229
.
Kastner
,
A. K.
,
Flohr
,
E. L. R.
,
Pauli
,
P.
, &
Wieser
,
M. J.
(
2016
).
A scent of anxiety: Olfactory context conditioning and its influence on social cues
.
Chemical Senses
,
41
,
143
153
.
Krohne
,
H. W.
,
Egloff
,
B.
,
Kohmann
,
C. W.
, &
Tausch
,
A.
(
1996
).
Untersuchungen mit einer deutschen version der Positive and Negative Affect Schedule (PANAS)
.
Diagnostica
,
42
,
139
156
.
Laux
,
L.
,
Glanzmann
,
P.
,
Schaffner
,
P.
, &
Spielberger
,
C. D.
(
1981
).
Das state-trait angstinventar
.
Weinheim
:
Beltz
.
LeDoux
,
J. E.
, &
Pine
,
D. S.
(
2016
).
Using neuroscience to help understand fear and anxiety: A two-system framework
.
American Journal of Psychiatry
,
173
,
1083
1093
.
Liberzon
,
I.
, &
Abelson
,
J. L.
(
2016
).
Context processing and the neurobiology of post-traumatic stress disorder
.
Neuron
,
92
,
14
30
.
Luck
,
S. J.
,
Woodman
,
G. F.
, &
Vogel
,
E. K.
(
2000
).
Event-related potential studies of attention
.
Trends in Cognitive Sciences
,
4
,
432
440
.
Maren
,
S.
,
Phan
,
K. L.
, &
Liberzon
,
I.
(
2013
).
The contextual brain: Implications for fear conditioning, extinction and psychopathology
.
Nature Reviews Neuroscience
,
14
,
417
428
.
Miskovic
,
V.
, &
Keil
,
A.
(
2012
).
Acquired fears reflected in cortical sensory processing: A review of electrophysiological studies of human classical conditioning
.
Psychophysiology
,
49
,
1230
1241
.
Mühlberger
,
A.
,
Wieser
,
M. J.
,
Herrmann
,
M. J.
,
Weyers
,
P.
,
Tröger
,
C.
, &
Pauli
,
P.
(
2009
).
Early cortical processing of natural and artificial emotional faces differs between lower and higher socially anxious persons
.
Journal of Neural Transmission
,
116
,
735
746
.
Nadel
,
L.
, &
Willner
,
J.
(
1980
).
Context and conditioning: A place for space
.
Physiological Psychology
,
8
,
218
228
.
O'Reilly
,
R. C.
, &
Rudy
,
J. W.
(
2001
).
Conjunctive representations in learning and memory: Principles of cortical and hippocampal function
.
Psychological Review
,
108
,
311
345
.
Olofsson
,
J. K.
,
Nordin
,
S.
,
Sequeira
,
H.
, &
Polich
,
J.
(
2008
).
Affective picture processing: An integrative review of ERP findings
.
Biological Psychology
,
77
,
247
265
.
Phillips
,
R. G.
, &
LeDoux
,
J. E.
(
1992
).
Differential contribution of amygdala and hippocampus to cued and contextual fear conditioning
.
Behavioral Neuroscience
,
106
,
274
285
.
Picton
,
T. W.
,
Bentin
,
S.
,
Berg
,
P.
,
Donchin
,
E.
,
Hillyard
,
S. A.
,
Johnson
,
R.
, et al
(
2000
).
Guidelines for using human event-related potentials to study cognition: Recording standards and publication criteria
.
Psychophysiology
,
37
,
127
152
.
Reiss
,
S.
,
Peterson
,
R. A.
,
Gursky
,
D. M.
, &
McNally
,
R. J.
(
1986
).
Anxiety sensitivity, anxiety frequency and the prediction of fearfulness
.
Behaviour Research and Therapy
,
24
,
1
8
.
Rudy
,
J. W.
(
2009
).
Context representations, context functions, and the parahippocampal–hippocampal system
.
Learning & Memory
,
16
,
573
585
.
Rudy
,
J. W.
,
Huff
,
N. C.
, &
Matus-Amat
,
P.
(
2004
).
Understanding contextual fear conditioning: Insights from a two-process model
.
Neuroscience & Biobehavioral Reviews
,
28
,
675
685
.
Rudy
,
J. W.
, &
O’Reilly
,
R. C.
(
2001
).
Conjunctive representations, the hippocampus, and contextual fear conditioning
.
Cognitive, Affective, & Behavioral Neuroscience
,
1
,
66
82
.
Rugg
,
M. D.
,
Milner
,
A. D.
,
Lines
,
C. R.
, &
Phalp
,
R.
(
1987
).
Modulation of visual event-related potentials by spatial and non-spatial visual selective attention
.
Neuropsychologia
,
25
,
85
96
.
Sabatinelli
,
D.
,
Keil
,
A.
,
Frank
,
D. W.
, &
Lang
,
P. J.
(
2013
).
Emotional perception: Correspondence of early and late event-related potentials with cortical and subcortical functional MRI
.
Biological Psychology
,
92
,
513
519
.
Schubert
,
T.
,
Friedmann
,
F.
, &
Regenbrecht
,
H.
(
2001
).
The experience of presence: Factor analytic insights
.
Presence: Teleoperators and Virtual Environments
,
10
,
266
281
.
Schupp
,
H. T.
,
Cuthbert
,
B. N.
,
Bradley
,
M. M.
,
Cacioppo
,
J. T.
,
Ito
,
T.
, &
Lang
,
P. J.
(
2000
).
Affective picture processing: The late positive potential is modulated by motivational relevance
.
Psychophysiology
,
37
,
257
261
.
Schupp
,
H. T.
,
Junghöfer
,
M.
,
Weike
,
A. I.
, &
Hamm
,
A. O.
(
2004
).
The selective processing of briefly presented affective pictures: An ERP analysis
.
Psychophysiology
,
41
,
441
449
.
Schupp
,
H. T.
,
Öhman
,
A.
,
Junghöfer
,
M.
,
Weike
,
A. I.
,
Stockburger
,
J.
, &
Hamm
,
A. O.
(
2004
).
The facilitated processing of threatening faces: An ERP analysis
.
Emotion
,
4
,
189
200
.
Sevenster
,
D.
,
Beckers
,
T.
, &
Kindt
,
M.
(
2014
).
Fear conditioning of SCR but not the startle reflex requires conscious discrimination of threat and safety
.
Frontiers in Behavioral Neuroscience
,
8
,
32
.
Spielberger
,
C. D.
,
Gorsuch
,
S. L.
, &
Edward
,
L. R.
(
1970
).
STAI manual for the State–Trait Anxiety Inventory
.
Palo Alto, CA
:
Consulting Psychologist Press
.
Stegmann
,
Y.
,
Reicherts
,
P.
,
Andreatta
,
M.
,
Pauli
,
P.
, &
Wieser
,
M. J.
(
2019
).
The effect of trait anxiety on attentional mechanisms in combined context and cue conditioning and extinction learning
.
Scientific Reports
,
9
,
8855
.
Steiger
,
F.
,
Nees
,
F.
,
Wicking
,
M.
,
Lang
,
S.
, &
Flor
,
H.
(
2015
).
Behavioral and central correlates of contextual fear learning and contextual modulation of cued fear in posttraumatic stress disorder
.
International Journal of Psychophysiology
,
98
,
584
593
.
Steinberg
,
C.
,
Bröckelmann
,
A.-K.
,
Rehbein
,
M.
,
Dobel
,
C.
, &
Junghöfer
,
M.
(
2013
).
Rapid and highly resolving associative affective learning: Convergent electro- and magnetoencephalographic evidence from vision and audition
.
Biological Psychology
,
92
,
526
540
.
Stout
,
D. M.
,
Glenn
,
D. E.
,
Acheson
,
D. T.
,
Spadoni
,
A. D.
,
Risbrough
,
V. B.
, &
Simmons
,
A. N.
(
2018
).
Neural measures associated with configural threat acquisition
.
Neurobiology of Learning and Memory
,
150
,
99
106
.
Watson
,
D.
,
Clark
,
L. A.
, &
Tellegen
,
A.
(
1988
).
Development and validation of brief measures of positive and negative affect: The PANAS scales
.
Journal of Personality and Social Psychology
,
54
,
1063
1070
.
Wieser
,
M. J.
,
Mühlberger
,
A.
,
Kenntner-Mabiala
,
R.
, &
Pauli
,
P.
(
2006
).
Is emotion processing affected by advancing age? An event-related brain potential study
.
Brain Research
,
1096
,
138
147
.
Wieser
,
M. J.
,
Pauli
,
P.
,
Reicherts
,
P.
, &
Mühlberger
,
A.
(
2010
).
Don't look at me in anger! Enhanced processing of angry faces in anticipation of public speaking
.
Psychophysiology
,
47
,
271
280
.