Abstract

When paired with aversive events, visual conditioned stimuli (CS) provoke increased activations in visual cortex. It is unclear however whether these changes reflect cognitive processes such as expectancy of the aversive unconditioned stimulus (US), or implicit associative learning of the contingencies outside awareness. Here, we used the “gambler's fallacy” phenomenon to parametrically and inversely manipulate the expectancy of an US and the number of conditioning trials: Increasing the number of CS–US pairings was associated with participants expecting the US to be less likely and vice versa. Magnetocortical activity evoked by the CS in occipital and supplementary motor areas was linearly related to the associative strength (number of CS–US pairings), but decreased as a function of expectancy. These results suggest that the cortical facilitation of fear cue processing is determined by associative strength and previous exposure to learning contingencies rather than by the cognitive anticipation for the US.

Introduction

The ability to adapt as a consequence of past negative experiences is crucial for survival. Therefore, learning the contingencies between threat signals and the potential dangers that they predict is considered the most ubiquitous adaptive behavior across species (LeDoux 1996). Pavlovian fear conditioning is probably the best-studied laboratory model for this kind of learning. During fear conditioning a previously innocuous stimulus becomes an effective elicitor of a conditioned fear response after having been paired with an aversive event.

It has been consistently shown that the activation of primary sensory cortices related to conditioned stimulus (CS) modality increases after successful learning of the CS–US (unconditioned stimulus) relationship (Buchel et al. 1998; LaBar et al. 1998; Fischer et al. 2002; Cheng et al. 2003; Knight et al. 2004; Moratti and Keil 2005; Moratti et al. 2006). This has been interpreted in terms of short-term plasticity of neural local networks within primary sensory cortex resulting in efficient detection and processing of threat cues (Keil et al. 2007).

Top-down modulation associated with cognitive factors such as attention and expectancy can prompt enhanced processing of fear-relevant stimuli in visual cortex (Vuilleumier and Driver 2007). However, conditioned autonomic responses such as skin conductance (Hamm and Vaitl 1996; Hamm et al. 2003; Knight et al. 2003) and conditioned amygdala activations (Morris et al. 1998, 1999, 2001; Whalen et al. 1998; Williams et al. 2004) have been shown to be independent from conscious awareness of the CS–US contingencies (but see Pessoa et al. 2005 for methodological considerations). Direct connections between the amygdala and visual cortex suggest implicit learning mechanisms that could act on visual processing without conscious processes (Ohman et al. 2007). In addition, Moratti et al. (2006) have shown that expectancy of the US alone without the activation of the fear system during CS–US acquisition is not sufficient to selectively modulate visual cortex for threat, which also suggests a critical role of learning and experience. Given this evidence, it is unclear to what extent the enhanced visual processing of threat cues represents declarative attention processes driven by US expectancy or, alternatively, short-term plasticity changes in the visual system and its afferent structures due to implicit learning.

In the field of eyeblink conditioning, Perruchet (1985) introduced a paradigm for evaluating the relative importance of expectancy and conditioning strength. Sequences of 1, 2, 3, and 4 CS–US and 1, 2, 3, and 4 CS-alone trials were intermixed such that the US followed the CS in 50% of the trials. Subjects were asked to rate the probability of US occurrence between trials. Whereas the associative strength increased with sequences of 1–4 CS–US trials, subjects’ expectancy of US appearance decreased linearly; a phenomenon known as the gambler's fallacy. Thus, conditioned response (CR) probability increased with associative strength, although the US expectancy decreased (see also Clark et al. 2001).

The aim of the current study was to examine if the increase of visual cortex activation due to CS–US association is a function of US expectancy or pure associative strength. As the Perruchet paradigm uses a low number of trials, the usage of this design poses some methodological problems as neuroimaging methods normally depend on a high number of trials to achieve sufficient signal-to-noise ratios. Steady-state visual evoked fields (ssVEF) registered by Magnetoencephalography (MEG) represent neuromagnetic oscillatory brain responses with high signal-to-noise ratios even at a single trial level (Keil et al. 2008). ssVEFs can be recorded by presenting visual stimuli in a luminance-modulated mode at a fixed frequency (Regan 1989) and represent stimulus phase locked oscillations in neural networks at the same fundamental frequency as the driving stimulus (Moratti et al. 2007). Further, ssVEFs and/or steady-state visual evoked potentials (ssVEPs registered with electroencephalography) have been shown to be modulated in amplitude by attentional processes (Morgan et al. 1996; Müller and Hillyard 2000; Muller and Hubner 2002) and/or by emotional arousal (Kemp et al. 2002; Keil et al. 2003; Moratti et al. 2008). Thus, we considered ssVEFs as ideally suited for the use within the Perruchet paradigm to investigate the relative importance of expectancy and associative strength in amplitude enhancement in visual cortex during fear conditioning.

By centrally presenting a grating pattern luminance modulated at 12.5 Hz for 5 s, we measured ssVEFs by using MEG. In 50% of the trials, the grating was paired (delay conditioning) with a US consisting of a 95-dB acoustic white noise. To examine the effects of US-expectancy versus associative strength on visual processing, intermixed sequences of 1, 2, 3, and 4 CS–US pairs and CS-alone trials were presented (see above). Between each trial, subjects communicated their expectancy of US occurrence in the subsequent trial. If enhanced visual cortex activation depends on US-expectancy, ssVEF amplitude originating from the visual cortex should increase as a function of expectancy ratings. By contrast, if enhanced CS processing depends on associative strength alone (number of CS–US pairings) rather than US-expectancy, ssVEF amplitude in visual cortex should increase with increasing number of CS–US pairings, although US-expectancy decreases. Figure 1 depicts the hypothetical ssVEF amplitude variations under the assumptions of the expectancy and associative strength theory, respectively (cf. Perruchet 1985).

Figure 1.

Predicted ssVEF amplitudes for the different experimental conditions, shown for the strength (black line) and the expectancy (gray line) theory.

Figure 1.

Predicted ssVEF amplitudes for the different experimental conditions, shown for the strength (black line) and the expectancy (gray line) theory.

Materials and Methods

Participants

Fourteen right-handed subjects (8 females) participated in the study after having given written informed consent. The mean age of the sample was 26.2 years (range 22–57 years). All subjects had normal or corrected to normal vision and no family history of epilepsy. Subjects received 15€ or class credit for participation. The local ethics committee approved the study.

Stimuli and Experimental Design

The visual CS was a gray-shaded 45° grating (0.31 cycles/cm, square wave). The CS was projected on a screen in a magnetically shielded MEG chamber using a video projector (JVC DLA-G11E) and a mirror system. The CS subtended a visual angle of 8° both horizontally and vertically and was presented centrally. The US was a 95-dB sound pressure level (SPL) white noise with instantaneous onset delivered binaurally by an air tube system attached to a sound amplifier (4D Neuroimaging, ASG 1996). In each CS-alone trial the CS was presented 5 s with a sinusoidal luminance variation at 12.5 Hz in order to elicit an ssVEF. The CS–US trials only differed from the CS-alone trials in presenting the US at 4 s after CS onset for 1 s. The US coterminated with the CS at 5 s after CS onset.

The experimental design was modeled after Perruchet (1985) (see also Clark et al. 2001): The 156 conditioning trials were presented as runs of 1, 2, 3, or 4 CS-US pairings and as runs of 1, 2, 3, or 4 CS-alone trials. The trial transitions across and within runs were not different. Between each trial, subjects had to rate their US expectancy for the next trial. For each subject the run order was randomized but restricted to meet the condition described in Table 1. Within each order the probability of an US was independent of the run length. The US probability was 50%. Between CS offset and the US expectancy rating of the subjects, an interval of 2 s was introduced. Subjects’ response triggered an intertrial interval varying randomly between 4 and 8 s.

Table 1

Organization of trials

 Nonreinforcements
 
Reinforcements
 
Total 
Run length  
Number of runs 12 24 24 12 90 
Number of trials 12 18 24 24 24 24 18 12 156 
 Nonreinforcements
 
Reinforcements
 
Total 
Run length  
Number of runs 12 24 24 12 90 
Number of trials 12 18 24 24 24 24 18 12 156 

Procedure

After having given informed consent, the head shapes of the subjects and 5 index points (nasion, left and right periauricular points, and 2 additional positions at the forehead) were digitized to obtain the relative head position to the MEG sensors. Then, 2 Ag/AgCl electroocculogram (EOC) electrodes were attached near the left and right outer canthi and 2 above and below the right eye. An electrode at the right mastoid served as ground. Finally, subject's heads were placed in the MEG helmet.

After US expectancy rating training (see Supplementary Material Methods), the experimental session and MEG recording began. First, participants were shown 10 CS-alone trials under the instruction that they would not receive any US (habituation trials). Then, subjects were instructed that from now on the US would be presented in 50 % of the trials and that between each trial, they would have to evaluate their US expectancy for the upcoming trial. After the experiment, subjects were informed about the intended provocation of the gambler's fallacy and were paid or given class credits for participation.

Data Acquisition and Processing

The MEG was recorded continuously and digitized at a rate of 254.3 Hz, using a 148 channel whole head system (Magnes 2500 WHS, 4D Neuroimage, San Diego, CA). A bandpass filter of 0.1–50 Hz was applied online. The EOC was done with a Synamps amplifier (Neuroscan, El Paso, TX) using Ag/AgCl electrodes. The same sampling rate and online filter as with the MEG recording were applied.

The raw MEG data was filtered offline (0.1 Hz highpass with a 6 dB/octave slope and 30 Hz lowpass with a 48 dB/octave slope). Then, epochs around the CS onset with a pre-stimulus baseline of 0.5 s and a poststimulus interval of 4 s were extracted from the data and baseline corrected. The last second of the 12.5-Hz CS flicker was not analyzed as in CS–US trials the US could have introduced artifacts (e.g., startle response). All epochs contaminated with eye artifacts as measured by the EOC were discarded from analysis. Epochs containing amplitudes greater than 3 pico Tesla were discarded. Finally, all epochs were visually inspected for further artifacts (movements) and eliminated if necessary.

Extraction of ssVEF

To analyze changes of ssVEF amplitude in the visual system as a function of trial type (CS–US pairing vs. CS-alone) and run length (trial sequences of 1, 2, 3, and 4), the ssVEF of the first epoch after the run length and trial type of interest was extracted. For example, to assess the ssVEF under the lowest US expectancy (see Results) and highest conditioning strength, the first epoch after 4 consecutive CS–US parings was analyzed. As can be inferred from Table 1, this condition only occurred 3 times (number of runs), thus resulting in 3 epochs. After artifact rejection a minimum of 2 such epochs was required. In subjects with 3 epochs after artifact rejection, 2 epochs were randomly chosen.

By using a moving window procedure within a steady-state design reliable ssVEF signals can be extracted from even single epochs (Keil et al. 2008). Therefore, a 0.4-s window (containing 5 cycles of a 12.5 Hz oscillation) was shifted across the first 2 s of an epoch in steps of 0.08 s (one 12.5 Hz cycle). Doing so for 2 epochs resulted in 42 moving windows of 0.4-s lengths that were averaged afterwards. Such a segment contains 5 cycles of the ssVEF of the first 2 s of CS presentation in the time domain. The same was done for the subsequent 2 s of CS presentation.

For the run lengths of 1, 2, and 3 trials 24, 12, and 6 epochs were available, respectively (see Table 2). However, to keep the number of averages constant across run lengths avoiding different signal-to-noise ratios (to be comparable later), 2 epochs were randomly chosen out of 6, 12, and 24 available epochs. Thus, applying the above described moving window procedure an early and late segment containing 5 cycles of the 12.5-Hz ssVEF response based on 42 averages were obtained for each run length (1, 2, 3, and 4 sequences), trial type (CS–US, CS-alone) and subject.

Table 2

Brain regions containing peaks of linear contrast derived F values

 Fmax Fcrit (P < 0.01) MNI coordinates
 
   x y z 
Region (first 2 s) 
    Cuneus posterior right 28.6 13.4 15 −83 22 
    Cuneus anterior right 28.1 13.4 15 −69 22 
    Calcarine fissure right 27.1 13.4 14 −76 11 
    Occipital superior gyrus right 26.7 13.4 19 −87 18 
    Occipital middle gyrus right 25.9 13.4 28 −78 28 
Region (following 2 s) 
    SMA left 13.02 8.6 −12 −8 62 
    SMA right 12.3 8.6 13 −9 60 
 Fmax Fcrit (P < 0.01) MNI coordinates
 
   x y z 
Region (first 2 s) 
    Cuneus posterior right 28.6 13.4 15 −83 22 
    Cuneus anterior right 28.1 13.4 15 −69 22 
    Calcarine fissure right 27.1 13.4 14 −76 11 
    Occipital superior gyrus right 26.7 13.4 19 −87 18 
    Occipital middle gyrus right 25.9 13.4 28 −78 28 
Region (following 2 s) 
    SMA left 13.02 8.6 −12 −8 62 
    SMA right 12.3 8.6 13 −9 60 

Note: F values in these regions with their corresponding permutation based critical F values and MNI coordinates are provided.

The resulting ssVEF averages were transformed into the frequency domain by applying a Fast Fourier Transform. The real and imaginary parts of the 12.5-Hz Fourier component were extracted reflecting the stimulus locked ssVEF activity.

Source Reconstruction

The underlying current source density of the ssVEF was estimated for each segment (first 2 s and subsequent 2 s of CS presentation), run length (1, 2, 3, 4), trial type (CS–US and CS-alone), and subject at the 12.5-Hz stimulus driving frequency using a 12 minimum norm estimation (MNE) with standard Tikhonov regularization as implemented in BrainStorm (http://neuroimage.usc.edu/brainstorm/). The Montreal Neurological Institute (MNI) phantom brain (Collins et al. 1998) with 7204 surface dipoles served as brain model. This MNI dipole mesh (7204 nodes) was used to calculate the forward solution using a head model based on overlapping spheres (for each channel a local sphere was fitted to the underlying head shape points) (Huang et al. 1999). The MNE was calculated in the frequency domain by submitting the real and imaginary parts of the 12.5-Hz Fourier component to the MNE analysis, as described by Jensen and Vanni (2002), and by using the root square of the sum of squares of the 2 MNE transformed Fourier parts as an estimate of absolute source strength (see also Moratti et al. 2008).

Statistical Analysis

Based on earlier findings (Perruchet 1985; Clark et al. 2001) a strong linear relationship across run lengths of 4, 3, 2, and 1 CS-alone and 1, 2, 3, and 4 CS–US trials for US expectancy ratings was expected (see Fig. 1). Therefore, a linear F contrast as suggested by Rosenthal and Rosnow (1985) was fitted to these 8 conditions. This procedure can be conceptualized as a repeated measure ANOVA weighting each factor level mean by coefficients that describe a hypothesis-derived model. Here, the coefficients were chosen to form a linear sequence, reflecting the expected patterns under the competing hypotheses.

The aim of the present study was to test, whether the visual cortex ssVEF amplitude increased linearly as a function of expectancy or conditioning strength (see Fig. 1). Therefore, a linear F contrast across 4, 3, 2, and 1 CS-alone and 1, 2, 3, and 4 CS–US conditions (see Fig. 1) was modeled at each of the 7204 source locations of the MNI brain for the first and subsequent 2 s of CS presentation. The MNE ssVEF source strengths at the 12.5-Hz driving stimulus frequency served as dependent variables for this linear contrast across the 8 conditions.

To control the family wise error of 7204 comparisons, the critical F value corresponding to an alpha level of 0.01 was derived from a nonparametric permutation test (Karniski et al. 1994). At each draw the MNE amplitudes of the 8 conditions were randomly shuffled across conditions as the null hypothesis was that there were no differences between conditions. Then, a linear F contrast was calculated at each of the 7204 dipole locations of the MNI brain. The greatest F value of all 7204 F contrasts entered the permutation distribution. This procedure was repeated 1000 times (1000 draws) resulting in the permutation distribution under the null hypothesis. The critical F value indicating a significant linear contrast (P < 0.01) was derived from the 99% percentile of the permutation distribution. Source clusters that showed F values greater than the critical F value were considered as brain regions of linearly increasing or decreasing ssVEF amplitude modulations across the 8 conditions.

Simulation Study

One major problem of the design of the current study was to keep the number of averages during ssVEF extraction constant across conditions (see above). Therefore, 2 epochs of each condition that resulted in 42 moving window averages were randomly selected for further analysis. However, the obtained results could have been specific to the selected trials and choosing different epochs could have led to different findings. Therefore, we repeated the whole analysis 100 times with random selection (drawing without replacement) of different epoch duplets (details see Supplementary Material Methods). A linear contrast at a dipole location of the MNI surface was considered as observed greater than chance when at least 60 out of 100 repetitions resulted in a significant F value again (binominal exact test H0: P = 0.5; 60 number of successes out of 100: P < 0.05).

Results

Behavioral Data

The US-expectancy between each trial was measured to assess if subjects applied the “gambler's fallacy” reported in previous studies capitalizing the same experimental design (Perruchet 1985; Clark et al. 2001). In the present study, subjects showed a strong negative linear relationship between US expectancy and conditioning strength (F1,98 = 166, P < 0.0001). The US expectancy declined with increasing number of CS–US parings (Fig. 2).

Figure 2.

(A) Visual cortex enhancement as a function of experience. Significant linear contrast F values are shown, representing linear enhancement of ssVEF amplitude during the first 2 s of viewing the CS, as it increased with the number of previously experienced CS–US pairs. F values are mapped onto the smoothed MNI brain. Only the right hemisphere is shown. The right panel depicts the mean source strength across the occipital dipole cluster shown in the left panel, for the 8 experimental conditions (blue line). The red line in the right panel shows the US expectancy ratings of the subjects in the same conditions. Error bars indicate standard errors. (B) Subsequent engagement of supplementary motor cortex. When considering the third and fourth second of viewing the CS, significant linear contrast F values were observed in supplementary motor areas only. The right panels depict the mean source strength values across the dipole clusters in left and right SMA (blue lines). The color bars indicate the F value derived from the linear contrast analysis. Gray shadings represent gyri and sulci on the smoothed MNI brain. Error bars indicate standard errors.

Figure 2.

(A) Visual cortex enhancement as a function of experience. Significant linear contrast F values are shown, representing linear enhancement of ssVEF amplitude during the first 2 s of viewing the CS, as it increased with the number of previously experienced CS–US pairs. F values are mapped onto the smoothed MNI brain. Only the right hemisphere is shown. The right panel depicts the mean source strength across the occipital dipole cluster shown in the left panel, for the 8 experimental conditions (blue line). The red line in the right panel shows the US expectancy ratings of the subjects in the same conditions. Error bars indicate standard errors. (B) Subsequent engagement of supplementary motor cortex. When considering the third and fourth second of viewing the CS, significant linear contrast F values were observed in supplementary motor areas only. The right panels depict the mean source strength values across the dipole clusters in left and right SMA (blue lines). The color bars indicate the F value derived from the linear contrast analysis. Gray shadings represent gyri and sulci on the smoothed MNI brain. Error bars indicate standard errors.

Magnetocortical Data

The aim of the present study was to investigate how the ssVEF amplitude of a visual CS was modulated along increasing levels of conditioning strength (4, 3, 2, and 1 CS-alone; 1, 2, 3, and 4 CS–US trials: 8 conditions) and simultaneously decreasing US expectancy. Before applying source localization, 5 cycles of the ssVEF oscillation were extracted from the MEG data for each condition and for the first and subsequent 2 s after CS onset, respectively (see Methods). Figure S1 (see Supplementary Material) depicts 5 cycles of the ssVEF signal for each of the 4 sequences within the CS–US and CS-alone presentations.

In the first 2 s of CS flicker, a right hemisphere source cluster in the occipital lobe showed increasing amplitudes as a function of conditioning strength (Fig. 2; linear contrast: permutation based critical F = 13.4; see peak F values within significant source clusters in Table 2). During the following 2 s of CS presentation, no such amplitude modulation could be observed in occipital cortex. However, a weaker but still significant linear trend in the same direction was found in the left and right supplementary motor area (SMA) (Fig. 2; linear contrast: permutation based critical F = 8.6; see Table 2 for peaks of F values within the significant source clusters).

Simulation Study

As the above results are based on random drawings of trials to keep the signal-to-noise ratios constant across the 8 conditions (see Methods), the observed ssVEF amplitude modulations as a function of conditioning strength could have been specific for the selected trials. Therefore, the random trial selection was repeated 100 times to determine the number of significant linear contrasts (successes) in the source space for the first and subsequent 2 s of CS presentation. Figure 3 depicts the number of successes for the 2 time windows of the CS that exceeded 60 (binomial exact test H0: P = 0.5; 60 number of successes out of 100: P < 0.05).

Figure 3.

The sum of successful linear F contrast replications for the first (A) and subsequent (B) 2 s mapped onto the smoothed MNI brain is shown. In (A) only the right hemisphere is shown. The color bar indicates the sum of successful replications out of 100. Gray shadings represent gyri and sulci on the smoothed MNI brain.

Figure 3.

The sum of successful linear F contrast replications for the first (A) and subsequent (B) 2 s mapped onto the smoothed MNI brain is shown. In (A) only the right hemisphere is shown. The color bar indicates the sum of successful replications out of 100. Gray shadings represent gyri and sulci on the smoothed MNI brain.

In the first 2 s a cluster of greater than chance replications of the results were observed in the occipital lobe (max number of replications: 87, P < 0.0001), although less spatially extended than in the original results (Figs 2 and 3). In the subsequent 2 s greater than chance replications were observed in the left (max number of replications: 84, P < 0.0001) and right SMA (max number of replications: 78, P < 0.0001).

Discussion

In this study, magnetocortical activity in visual cortex was tracked during fear conditioning, with the CS-US contingencies manipulated in a way that allowed the associative strength and the US expectancy to be varied parametrically and inversely to each other (Perruchet 1985; Clark et al. 2001). The research question was whether magnetocortical activity in visual areas increases with expectancy of the aversive US or as a function of associative strength. The present results suggest that, during the first 2 s of visual CS presentation, occipital cortex activation increased with increasing number of CS–US pairings and was inversely related to US anticipation. This supports the notion that early visual facilitation of CS processing does not depend on declarative cognitive processes such as US-expectancy (Morris et al. 1999; Knight et al. 2004; Knight et al. 2003; Moratti et al. 2006). Further, ssVEF amplitudes in the SMAs of both hemispheres also increased with associative strength only. These areas have been reported to be involved in fear learning (e.g., Knight et al. 2004) and may well mediate motor preparation. However, increased ssVEF activity in visual cortex as a function of associative strength was only observed during the first half of CS presentation and not during the second time window before US presentation. The lack of this effect may be due to possible habituation effects during the presentation of the CS flicker. In addition, grating orientation is easily and rapidly identified and therefore, prolonged sustained attentive analysis of the motivationally relevant stimulus is not as adaptive as with complex visual stimuli (e.g., Moratti et al. 2004) or might reflect the fact that the grating is easily identified.

Our findings are in line with 2 behavioral studies (Perruchet 1985; Clark et al. 2001) using the gambler's fallacy paradigm to study expectancy and conditioning strength. These studies relied on the conditioned eyeblink paradigm instead of fear conditioning as utilized in the current report. Nevertheless, both eye blink probability and ssVEF amplitude modulations in occipital cortex and SMA during fear conditioning were modulated as a function of associative strength rather than US expectancy. This points to common underlying learning mechanisms across different experimental designs, affecting multiple dependent variables (behavioral and neural) in a similar manner.

It should be mentioned that the present study used a delay conditioning procedure as described in the original article by Perruchet (1985). Clark et al. (2001) replicated the finding of experience-based enhancement of the CR for delay conditioning. They also found that, in the case of trace conditioning, the eyeblink probability increased with US expectancy rather than experience, adding further evidence that trace and delay conditioning represent distinct learning processes. As a consequence, increase of magnetocortical activity with associative strength may be specific to delay conditioning. Future studies should address the issue if trace conditioning yields activity increases in occipital and frontal structures with increasing US expectancy.

As an alternative overall interpretation, the experimental effects on MNE amplitude curves (see Fig. 2) could also be interpreted as amplitude decreases rather than increases. Repeating visual stimuli consistently produces neural activity reduction in the visual cortex that has been associated with “sharpening” processes with respect to the neural representation of the stimulus (Gruber and Müller 2002; Gruber et al. 2004). Nevertheless, as an important difference however, this process has been observed with familiar stimuli like meaningful objects only (flowers, birds, etc.), whereas repetition of unfamiliar stimuli like meaningless patterns (like a gray grating used in the present study) should leads to an increase of neural activity under this persepective (Conrad et al. 2007). Such augmentation has been interpreted as a formation of a new neural network whereas the decrease is associated with “sharpening” processes in conceptual networks (Conrad et al. 2007). As 1) in the present study an unfamiliar stimulus was used, and 2) the number of repetitions of the CS was equal across the trial types (CS-alone, CS–US), it is unlikely that the observed MNE curves are a product of repetition effects. Given the vast and extensive and consistent literature reporting neural activity increases in primary sensory cortex during fear conditioning (Buchel et al. 1998; LaBar et al. 1998; Fischer et al. 2002; Cheng et al. 2003; Knight et al. 2004; Moratti and Keil 2005; Moratti et al. 2006), it is more likely that the ssVEF augmentation in visual cortex and SMA represent short-term plasticity changes of neural local networks that ensure efficient detection and processing of threat cues (Keil et al. 2007).

Nevertheless, the present experimental design cannot rule out, that increased expectancy of the acoustic US (a loud white noise) drove attention towards auditory and away from visual processing, reflected in an amplitude decrease in visual cortex during high US expectation. As the ssVEF activity is driven by the visual flicker, the current measure of neural activity cannot address this issue. However, this would not explain the activity increase in SMA during low US anticipation. Future studies should address this issue.

Given the spatial limitations of MEG for deep sources, we could not observe activity modulations in the amygdala, a core structure in fear conditioning (LeDoux 2000). Furthermore, because ssVEFs represent re-entrant activity reverberation in visual cortex and higher order structures such as the frontal cortex (Silberstein et al. 1990; Perlstein et al. 2003), we did not expect to observe ssVEF oscillations in the amygdala, although a modulatory influence of the amygdala on visual cortex is likely (Amaral 1986). Further, the MNE approach is more adequate for superficial sources and does not capture deep sources. However, as the original goal of the study was to characterize re-entrant cortical ssVEF oscillations in space and time, the MNE approach was considered adequate for this purpose.

In sum, the present study demonstrated that in areas of sensory processing and motor preparation neural activity elicited by a visual CS increases as a function of associative experience. Interestingly, increased occipital cortex and SMA activation was independent of subjective US expectancy as demonstrated by a dissociation of ssVEF amplitude augmentation and US expectancy ratings: When subjects were almost sure (expectancy rating 78%) that they would receive an US, their visual cortex and SMA activation was low. However, the more often the CS–US pairing was experienced, the greater was the ssVEF amplitude in the right occipital lobe and the SMA of both hemispheres. At the same time, US expectancy decreased. Obviously, the brain is able to facilitate sensory processing and motor preparation after CS–US association even when subjects consciously do not anticipate the US. This supports the view that enhanced neural activity in visual and SMA cortex after fear learning is a rather implicit process of associative experience and does not depend on US expectancy.

Supplementary Material

Supplementary material can be found at: http://www.cercor.oxfordjournals.org/

Funding

Deutsche Forschungsgemeinschaft (DFG, MO 1043/2-1); and Programa del Grupo de Biociencia de la Comunidad Autónoma de Madrid MADR.IB (M06094501).

We thank Leonie Koban for help in data acquisition. Conflict of Interest: None declared.

References

Amaral
DG
Amygdalohippocampal and amygdalocortical projections in the primate brain
Adv Exp Med Biol.
 , 
1986
, vol. 
203
 (pg. 
3
-
17
)
Buchel
C
Morris
J
Dolan
RJ
Friston
KJ
Brain systems mediating aversive conditioning: an event-related fMRI study
Neuron.
 , 
1998
, vol. 
20
 (pg. 
947
-
957
)
Cheng
DT
Knight
DC
Smith
CN
Stein
EA
Helmstetter
FJ
Functional MRI of human amygdala activity during Pavlovian fear conditioning: stimulus processing versus response expression
Behav Neurosci.
 , 
2003
, vol. 
117
 (pg. 
3
-
10
)
Clark
RE
Manns
JR
Squire
LR
Trace and delay eyeblink conditioning: contrasting phenomena of declarative and nondeclarative memory
Psychol Sci.
 , 
2001
, vol. 
12
 (pg. 
304
-
308
)
Collins
DL
Zijdenbos
AP
Kollokian
V
Sled
JG
Kabani
NJ
Holmes
CJ
Evans
AC
Design and construction of a realistic digital brain phantom
IEEE Trans Med Imaging.
 , 
1998
, vol. 
17
 (pg. 
463
-
468
)
Conrad
N
Giabbiconi
CM
Müller
MM
Gruber
T
Neural correlates of repetition priming of frequently presented objects: Insights from induced gamma band responses
Neurosci Lett.
 , 
2007
, vol. 
429
 (pg. 
126
-
130
)
Fischer
H
Andersson
JL
Furmark
T
Wik
G
Fredrikson
M
Right-sided human prefrontal brain activation during acquisition of conditioned fear
Emotion.
 , 
2002
, vol. 
2
 (pg. 
233
-
241
)
Gruber
T
Malinowski
P
Müller
MM
Modulation of oscillatory brain activity and evoked potentials in a repetition priming task in the human EEG
Eur J Neurosci.
 , 
2004
, vol. 
19
 (pg. 
1073
-
1082
)
Gruber
T
Müller
MM
Effects of picture repetition on induced gamma band responses, evoked potentials, and phase synchrony in the human EEG
Cogn Brain Res.
 , 
2002
, vol. 
13
 (pg. 
377
-
392
)
Hamm
AO
Vaitl
D
Affective learning: awareness and aversion
Psychophysiology.
 , 
1996
, vol. 
33
 (pg. 
698
-
710
)
Hamm
AO
Weike
AI
Schupp
HT
Treig
T
Dressel
A
Kessler
C
Affective blindsight: intact fear conditioning to a visual cue in a cortically blind patient
Brain
 , 
2003
, vol. 
126
 (pg. 
267
-
275
)
Huang
MX
Mosher
JC
Leahy
RM
A sensor-weighted overlapping-sphere head model and exhaustive head model comparison for MEG
Phys Med Biol.
 , 
1999
, vol. 
44
 (pg. 
423
-
440
)
Jensen
O
Vanni
S
A new method to identify multiple sources of oscillatory activity from magnetoencephalographic data
Neuroimage.
 , 
2002
, vol. 
15
 (pg. 
568
-
574
)
Karniski
W
Blair
RC
Snider
AD
An exact statistical method for comparing topographic maps, with any number of subjects and electrodes
Brain Topogr.
 , 
1994
, vol. 
6
 (pg. 
203
-
210
)
Keil
A
Gruber
T
Muller
MM
Moratti
S
Stolarova
M
Bradley
MM
Lang
PJ
Early modulation of visual perception by emotional arousal: evidence from steady-state visual evoked brain potentials
Cogn Affect Behav Neurosci.
 , 
2003
, vol. 
3
 (pg. 
195
-
206
)
Keil
A
Smith
JC
Wangelin
BC
Sabatinelli
D
Bradley
MM
Lang
PJ
Electrocortical and electrodermal responses covary as a function of emotional arousal: a single-trial analysis
Psychophysiology.
 , 
2008
, vol. 
45
 (pg. 
516
-
523
)
Keil
A
Stolarova
M
Moratti
S
Ray
WJ
Adaptation in human visual cortex as a mechanism for rapid discrimination of aversive stimuli
Neuroimage.
 , 
2007
, vol. 
36
 (pg. 
472
-
479
)
Kemp
AH
Gray
MA
Eide
P
Silberstein
RB
Nathan
PJ
Steady-state visually evoked potential topography during processing of emotional valence in healthy subjects
Neuroimage.
 , 
2002
, vol. 
17
 (pg. 
1684
-
1692
)
Knight
DC
Cheng
DT
Smith
CN
Stein
EA
Helmstetter
FJ
Neural substrates mediating human delay and trace fear conditioning
J Neurosci.
 , 
2004
, vol. 
24
 (pg. 
218
-
228
)
Knight
DC
Nguyen
HT
Bandettini
PA
Expression of conditional fear with and without awareness
Proc Natl Acad Sci USA.
 , 
2003
, vol. 
100
 (pg. 
15280
-
15283
)
LaBar
KS
Gatenby
JC
Gore
JC
LeDoux
JE
Phelps
EA
Human amygdala activation during conditioned fear acquisition and extinction: a mixed-trial fMRI study
Neuron.
 , 
1998
, vol. 
20
 (pg. 
937
-
945
)
LeDoux
JE
The emotional brain
 , 
1996
New York
Simon and Schuster
LeDoux
JE
Emotion circuits in the brain
Annu Rev Neurosci.
 , 
2000
, vol. 
23
 (pg. 
155
-
184
)
Moratti
S
Clementz
BA
Gao
Y
Ortiz
T
Keil
A
Neural mechanisms of evoked oscillations: Stability and interaction with transient events
Hum Brain Mapp.
 , 
2007
, vol. 
28
 (pg. 
1318
-
1333
)
Moratti
S
Keil
A
Cortical activation during Pavlovian fear conditioning depends on heart rate response patterns: an MEG study
Brain Res Cogn Brain Res.
 , 
2005
, vol. 
25
 (pg. 
459
-
471
)
Moratti
S
Keil
A
Miller
GA
Fear but not awareness predicts enhanced sensory processing in fear conditioning
Psychophysiology.
 , 
2006
, vol. 
43
 (pg. 
216
-
226
)
Moratti
S
Keil
A
Stolarova
M
Motivated attention in emotional picture processing is reflected by activity modulation in cortical attention networks
Neuroimage.
 , 
2004
, vol. 
21
 (pg. 
954
-
964
)
Moratti
S
Rubio
G
Campo
P
Keil
A
Ortiz
T
Hypofunction of right temporoparietal cortex during emotional arousal in depression
Arch Gen Psychiatry.
 , 
2008
, vol. 
65
 (pg. 
532
-
541
)
Morgan
ST
Hansen
JC
Hillyard
SA
Selective attention to stimulus location modulates the steady-state visual evoked potential
Proc Natl Acad Sci USA.
 , 
1996
, vol. 
93
 (pg. 
4770
-
4774
)
Morris
JS
DeGelder
B
Weiskrantz
L
Dolan
RJ
Differential extrageniculostriate and amygdala responses to presentation of emotional faces in a cortically blind field
Brain.
 , 
2001
, vol. 
124
 (pg. 
1241
-
1252
)
Morris
JS
Ohman
A
Dolan
RJ
Conscious and unconscious emotional learning in the human amygdala
Nature.
 , 
1998
, vol. 
393
 (pg. 
467
-
470
)
Morris
JS
Ohman
A
Dolan
RJ
A subcortical pathway to the right amygdala mediating “unseen” fear
Proc Natl Acad Sci USA.
 , 
1999
, vol. 
96
 (pg. 
1680
-
1685
)
Müller
MM
Hillyard
S
Concurrent recording of steady-state and transient event-related potentials as indices of visual-spatial selective attention
Clin Neurophysiol.
 , 
2000
, vol. 
111
 (pg. 
1544
-
1552
)
Muller
MM
Hubner
R
Can the spotlight of attention be shaped like a doughnut? Evidence from steady-state visual evoked potentials
Psychol Sci.
 , 
2002
, vol. 
13
 (pg. 
119
-
124
)
Ohman
A
Carlsson
K
Lundqvist
D
Ingvar
M
On the unconscious subcortical origin of human fear
Physiol Behav.
 , 
2007
, vol. 
92
 (pg. 
180
-
185
)
Perlstein
WM
Cole
MA
Larson
M
Kelly
K
Seignourel
P
Keil
A
Steady-state visual evoked potentials reveal frontally-mediated working memory activity in humans
Neurosci Lett.
 , 
2003
, vol. 
342
 (pg. 
191
-
195
)
Perruchet
P
A pitfall for the expectancy theory of human eyelid conditioning
Pavlov J Biol Sci.
 , 
1985
, vol. 
20
 (pg. 
163
-
170
)
Pessoa
L
Japee
S
Ungerleider
LG
Visual awareness and the detection of fearful faces
Emotion.
 , 
2005
, vol. 
5
 (pg. 
243
-
247
)
Regan
D
Human brain electrophysiology: evoked potentials and evoked magnetic fields in science and medicine
 , 
1989
New York
Elsevier
Rosenthal
R
Rosnow
R
Contrast analysis: focused comparisons in the analysis of variance
 , 
1985
Cambridge, UK
Cambridge University Press
Silberstein
RB
Schier
MA
Pipingas
A
Ciorciari
J
Wood
SR
Simpson
DG
Steady-state visually evoked potential topography associated with a visual vigilance task
Brain Topogr.
 , 
1990
, vol. 
3
 (pg. 
337
-
347
)
Vuilleumier
P
Driver
J
Modulation of visual processing by attention and emotion: windows on causal interactions between human brain regions
Philos Trans R Soc Lond B Biol Sci.
 , 
2007
, vol. 
362
 (pg. 
837
-
855
)
Whalen
PJ
Rauch
SL
Etcoff
NL
McInerney
SC
Lee
MB
Jenike
MA
Masked presentations of emotional facial expressions modulate amygdala activity without explicit knowledge
J Neurosci.
 , 
1998
, vol. 
18
 (pg. 
411
-
418
)
Williams
MA
Morris
AP
McGlone
F
Abbott
DF
Mattingley
JB
Amygdala responses to fearful and happy facial expressions under conditions of binocular suppression
J Neurosci.
 , 
2004
, vol. 
24
 (pg. 
2898
-
2904
)