The topic of rapidly directing perceptual resources to relevant features in the field of view has been a concern of recent studies examining Event-Related Potential (ERP) correlates of emotional perception. For instance, Pourtois and colleagues (Pourtois et al., 2004
) employed a visual hemifield paradigm with covert orienting to emotional faces. For fearful compared to happy faces, these authors reported enhancement of the first detectable visual ERP deflection at 90 ms after stimulus onset, possibly originating in the striate cortex. They concluded that the emotional relevance of the stimuli might be associated with increased activation in the primary visual cortex, possibly due to an interaction with sub-cortical structures. Modulations of very early perceptual processing by affective scenes or faces are typically difficult to interpret because affective stimulus properties may be confounded with physical characteristics. Therefore, studies using conditioning approaches have been used to examine early affective perception and have suggested differential sensitivity to the CS+ and CS− at early and very early stages of visual analysis (Pizzagalli et al., 2003
). Following on such findings, Stolarova and co-workers (2006)
studied the temporal development of such early modulations across conditioning blocks. They presented grating patterns to the hemifields, paired with unpleasant or neutral pictures in a delayed classical conditioning paradigm, a procedure that was also used in the present study. This approach allows to control for the physical stimulus features of conditioned stimuli (i.e., grating patterns) and to explore the development of perceptual facilitation across blocks of trials, as a function of learning. As a main result, this study suggested greater amplitude of the visual ERP for CS+ stimuli versus CS− stimuli, occurring as early as 65 to 90 ms following stimulus onset. This difference between CS+ and CS−, although more widespread in the course of learning, did (i) not gradually increase in amplitude over blocks and (ii) is not indicative of differential synchrony as are measures of oscillatory activity. The present study therefore aimed to complement the ERP findings by examining the time course of large-scale oscillations during differential classical conditioning.
The present study focused on an early evoked oscillatory brain response in the upper beta/lower gamma range, which shows a high degree of phase consistency across trials and in the visual system tends to occur very early after onset of a stimulus (Tallon et al., 1995
). As compared to the so-called induced gamma band response, this early phase-locked response has been associated with somewhat lower peak frequencies, often lying in the 20 to 40 Hz range (e.g., Herrmann et al., 2004
; Tallon-Baudry et al., 1997
). Because it peaks around 70–100 ms after onset of a stimulus, this measure is suited to examine low-level responses of the human visual cortex. Even in this early time range, however, caution is warranted in terms of the interpretation of the neural origin of this early response. ERP work in combination with current source estimations has suggested that in time segments after 70–75 ms, there is evidence for extra-striate sources contributing to the measured signal (Foxe and Simpson, 2002
), which is in line with macaque data suggesting very rapid involvement of parietal and ventral cortex. As a consequence, modulations in the 70 to 90 ms range may reflect local network activity and re-entrant (i.e. top-down) modulation on a small scale, involving higher-order cortices to some degree as well. A related topic concerns the frequency range of this early phase-locked response as observed in our study. Most studies capitalizing on wavelet analysis of EEG epochs obtained in visual paradigms have reported somewhat higher peak frequencies (e.g., Herrmann et al., 1999
). Data on evoked oscillations in humans with hemi- or quarter-field stimulation are scarce however, and future work may address in more depth the role of different frequencies in early large-scale visual cortical processing. A previous study with affective pictures presented to the visual hemifields found early evoked oscillations in a frequency range similar to the present one (Keil et al., 2001
). Most interestingly, Axmacher and colleagues have recently suggested that synchronization at different frequencies may be related to different aspects of neuroplastic changes associated with memory formation (Axmacher et al., 2006
). It is thus conceivable that the modulations observed in the present study may not only reflect altered perceptual processing but also the mechanism underlying the reorganization occurring as a consequence of learning the contingencies.
In the domain of auditory classical conditioning, research has shown learning induced plasticity in the receptive fields of the primary auditory areas in animals (Diamond and Weinberger, 1984
; Weinberger, 2004
) and humans (Morris et al., 1998
). As a possible underlying mechanism for such fast experience dependent cortical reorganization, an increase in dopamine or acetylcholine release has been proposed, possibly leading to long-term potentiation and the strengthening of neural connectivity. A variety of studies (Li et al., 2004
) implicate plasticity of visual cortex in perceptual learning and that changes in phase synchrony are typically related to changes in perception (Elbert and Keil, 2000
). Adaptive changes of visual receptive fields, which are similar to the learning induced cortical adaptation in primary auditory areas, have repeatedly been reported (Gilbert et al., 2000
), but the relevance of these findings for rapid emotional perception has been unclear. As compared to other studies of stimulus repetition in the visual system, which typically report repetition suppression across blocks (see e.g., Grill-Spector et al., 2006
), we found increased amplitude and synchrony for the CS+ condition specifically, as long as the conditioning regime was maintained. This could be taken as evidence that networks involved in CS-processing quickly come to include additional sub-networks, a process similar to the one observed by Gruber and colleagues during rapid perceptual learning (Gruber et al., 2002
) or during repeated presentation of unfamiliar objects (Gruber and Müller, 2005
) Thus, learning to quickly respond to a stimulus associated with affective relevance may gradually increase the number of neurons in visual cortex that are coherently active in the presence of that stimulus. Caution is however warranted when interpreting synchrony maps derived from surface EEG. Although amplitude and phase are mathematically independent, it is conceivably that synchrony is more pronounced between source regions showing good signal-to-noise, i.e., high amplitude. Thus synchrony maps would nodd greatly add to amplitude data and would not indicate specific changes in synchrony. In the present data, there is however a pronounced difference between power and synchrony maps, which made us report synchrony findings in the present study. As can be seen in , power maps show a small but still clear frontal maximum throughout conditions, which is not at all present in the synchrony maps. It is thus unlikely that synchrony just mirrors amplitude
When comparing the time course of startle modulation and spectral changes, there is a gradual enhancement for the CS+ evoked oscillatory response, but a decrease in the overall startle amplitude, which might suggest lack of differential affective responding in the second acquisition block. As an alternative explanation, startle data tend to habituate relatively quickly (Bradley et al., 1993
) and it is likely that after 54 startle trials in the first conditioning block, the startle amplitude decreased and differential affective responding between stimulus categories was not obvious in startle recording during the second conditioning block. Self-report data (preference rankings for CS stimuli) solicited by each participant at the end of the second conditioning block showed highly significant dislike for the CS+ in almost all participants however, even in the absence of awareness for the experimental contingencies. This latter finding corroborates the assumption that differential conditioning was achieved across blocks, and was accompanied by selective increase of amplitude and synchrony.
The present work demonstrates that the initial sensory response in the human visual cortex is sensitive to simple features associated with emotionality and that the amplification of this response increases as a function of experience. This finding is associated with a specific enhancement in large-scale synchrony within the extended visual cortex. Based on previous experience and based on knowledge about context properties, the emotional brain may thus constantly adapt to key features of relevant stimuli. Continuous adaptation and optimization of the visual system may thus enable observers to efficiently react to potential threat stimuli at the earliest stages possible.