Cortical activation during Pavlovian fear conditioning depends on heart rate response patterns: An MEG study

Abstract

In the present study, we examined stimulus-driven neuromagnetic activity in a delayed Pavlovian aversive conditioning paradigm using steady state visual evoked fields (SSVEF). Subjects showing an accelerative heart rate (HR) component to the CS+ during learning trials exhibited an increased activation in sensory and parietal cortex due to CS+ depiction in the extinction block. This was accompanied by a selective orientation response (OR) to the CS+ during extinction as indexed by HR deceleration. However, they did not show any differential cortical activation patterns during acquisition. In contrast, subjects not showing an accelerative HR component but rather unspecific HR changes during learning were characterized by greater activity in left orbito-frontal brain regions in the acquisition block but did not show differential SSVEF patterns during extinction. The results suggest that participants expressing different HR responses also differ in their stimulus-driven neuromagnetic response pattern to an aversively conditioned stimulus.

Introduction

Learning that certain events in an individual's environment signal potential threat or danger is an important ability to ensure survival of the organism. Pavlovian fear conditioning [69] as a laboratory model of the aforementioned ability represents a simple form of associative learning that is part of most mammalian defensive behavior systems [21].

The neural systems underlying fear conditioning have been elucidated by animal models as well as human imaging studies using functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) [13], [58]. The amygdala has been regarded as the key structure processing aversive stimuli via afferents from sensory thalamus [55], [84], enabling fast evaluation of noxious stimuli without complex computations of the cortex. As the amygdaloidal complex is highly interconnected with the temporal, orbito-frontal, and insular cortices [1], [2], [64], [74], the amygdala is anatomically well placed for stimulus association learning in various sensory domains. Given its fast access to sensory information and its neuroanatomical location, the amygdala has been assigned a central role in mediating synaptic changes at the cortical level [59] necessary for the association between the conditioned (CS) and the unconditioned stimulus (US). For example, Buchel and colleagues [14] demonstrated in an fMRI study that the amygdala activation was most prominent during acquisition of the conditioned response (CR) and LaBar and collaborators [50] showed a decline of amygdala activation over trials possibly reflecting that amygdala activation is only necessary until associations between the CS and US have been formed.

In addition to amygdala circuitry and highly interconnected with it, a distributed network of cortical areas seems to be involved in classical fear conditioning as well. For example, standard fear conditioning paradigms using auditory CS not only reported enhanced spike firing rates in neurons of the amygdala and hippocampus, but also of neurons in auditory cortex [5], [52], [75]. In the visual system of primates, the amygdala projects back to various stages in the ventral visual path and receives itself highly processed sensory information [3]. This neural architecture may be the basis of a mechanism, which enhances stimulus processing in visual sensory systems. In line with this notion, imaging studies in humans have shown greater activations for the reinforced visual CS (CS+) as compared to the nonreinforced CS (CS−) in temporal and occipital cortices [15], [39].

Additional support to the involvement of cortical areas in fear conditioning comes from cognitive theories of Pavlovian conditioning that emphasize anticipation or attentional processes [70]. For instance, Rescorla [77] has emphasized that the CS+ in a Pavlovian conditioning paradigm gains predictive value that enables the organism to anticipate an aversive event. The ability to obtain such a predictive value has been associated with the salience of a stimulus [57]. Thus, at a neurophysiological level, cortical systems mediating attentional and evaluative processes should come into play.

There is good evidence for the involvement of a widespread fronto-parietal cortical network in orienting to a stimulus and integrating its features [16], [54], [68], [80]. Human imaging studies of Pavlovian conditioning have consistently reported enhancement of cortical activity for visual reinforced CSs in brain regions like the frontal, temporal, parietal cortices, and the anterior cingulate [14], [22], [23], [39], [47], [50] possibly reflecting the involvement of a neural network mediating attention/emotion aspects of that stimulus. In a recent fMRI study, Armony and Dolan [4] demonstrated that a visual context stimulus modulated conditioned BOLD responses in auditory cortex associated with activity in parietal cortex, further illustrating the importance of cortical structures being part of a cortical attention network.

In the present work, we investigated changes of cortical processes over time using the steady state visual evoked field (SSVEF) technique. SSVEFs (a counterpart of the SSVEP in EEG research) reflect widely distributed functional networks oscillating coherently at the driving stimulus frequency and are sensitive to attentional and complex cognitive processes [61], [66], [71], [85].

As expectancy and attentional processes may be involved in Pavlovian conditioning as outlined above, the SSVEP/SSVEF paradigm seems to be an adequate tool in order to track ongoing neural activity associated with processes of aversive conditioning. A recent SSVEP study by Gray and coworkers [26] investigated cortical oscillatory activity during the anticipation of an electric shock and revealed the involvement of frontal, temporal, and occipital electrode sites during anticipatory anxiety. Recently, Moratti and collaborators [60] have identified a fronto-parietal network during visual affective stimulus processing using magnetoencephalography (MEG) and the SSVEF technique. This finding supports the involvement of networks traditionally related to attention processes in processing of emotional information. In the same vein, it has been demonstrated that cued spatial attention and motivated attention to affective stimuli possibly interact within the same neuronal networks of attention and stimulus processing [44]. This raises the question whether similar neurophysiological processes are associated with viewing arbitrary visual stimuli which are associated with aversive responses as a consequence of classical fear conditioning. There is, however, no study to date using the SSVEP/SSVEF technique in combination with a discriminative aversive conditioning design which allows the direct comparison between the stimulus-driven neuromagnetic response of a reinforced (CS+) and nonreinforced (CS−) stimulus during acquisition and extinction blocks.

Therefore, the aim of the present study was to examine cortical sources of the SSVEF as modulated by the predictive value of a visual CS during acquisition and extinction, using MEG. Since the SSVEF response is a stimulus-driven ongoing oscillatory response in cortical networks responsible for processing the visual CS+ and CS−, estimating the underlying sources should elucidate the involvement of relevant cortical structures in Pavlovian conditioning. Applying the minimum norm estimate (MNE) [34], [35] to the SSVEF response to determine the cortical sources of the neuromagnetic field, we hypothesized that after an association between the CS+ and the US has been established, the motivationally relevant CS+ will generate greater amplitudes in fronto-parietal and extrastriate cortex responsible for allocating attention and feature extraction, respectively. As the MEG has sufficient temporal resolution, we were interested at what time interval during CS depiction a differential activation pattern in the aforementioned cortical structures could be observed for the CS+ and CS−, respectively.

Heart rate (HR) was recorded to validate the effectiveness of the conditioning procedure. Further, we wanted to evaluate the brain responses of subjects who reacted with accelerative or decelerative HR change components in response to the CS+ during acquisition as has been observed by various authors [28], [32], [38]. Hodes et al. [38] hypothesized that HR accelerators learn a fear response whereas HR decelerators just learn that two events are related in time. Hamm and Vaitl [28] demonstrated that only subjects responding with an accelerative HR component displayed potentiated startle responses to the CS+. We wanted to explore if attentional and sensory cortical networks during Pavlovian conditioning are involved in HR accelerators and decelerators to the same extent. Subjective judgments of the US were obtained in order to assess the aversiveness of the US. Further, we collected questionnaire data in order to evaluate state/trait anxiety and depressiveness of the participants.