To our knowledge, this study examined, for the first time, a marker of intracortical inhibition in the DLPFC in specific phobias. We found reduced TMS-evoked N100 amplitudes in the DLPFC in the specific phobia compared to the control group at rest.

Furthermore, emotional processing when memorizing fearful compared with neutral facial expressions led to a reduction in intracortical inhibition in the DLPFC in the specific phobia group as reflected by the N100.

In contrast, there was no significant change in the N100 amplitude of the DLPFC during working memory-related processing in the 1-back task with a low working memory load compared with a rest condition.

4.1 TMS-evoked N100 in the DLPFC

The N100 in the DLPFC as part of the TEP has been discussed to occur as lateralized negative deflection at the stimulation site (Jarczok et al., 2021; Rocchi et al., 2021). We found an ipsilateral N100 at the stimulation site at electrodes F5 and F6 with significantly smaller amplitudes in the specific phobia compared with the control group. The N100 as a parameter for intracortical inhibition is likely to reflect GABA-B reactivity in the DLPFC (Premoli, Castellanos, et al., 2014; Premoli, Rivolta, et al., 2014; Rogasch et al., 2015). A reduced N100 could thus indicate a lower GABA-B-mediated inhibition in the DLPFC in specific phobias. Interestingly, there was a general group-specific reduction in the N100 in the absence of phobic stimuli. Both groups were free of any psychotropic drugs at the time of participation. Hence, medication can be excluded as a cause for the N100 reduction. Additionally, there were no comorbid anxiety disorders, major depression, or any other psychiatric disorder in both groups.

There is evidence for anxiety disorders, and thus specific phobias, to be characterized by higher levels of trait anxiety (Raymond et al., 2017). This implies that trait anxiety as a relatively stable predisposition over time (Vagg et al., 1980) may be associated with a general impairment of intracortical inhibition in specific phobias.

The finding of a reduced N100 in anxious subjects contributes to the literature on how trait anxiety modulates DLPFC activation (Bishop, 2009; Eysenck et al., 2007). Neuroimaging studies have found a threat-independent deficit of the DLPFC inhibiting interfering information in highly anxious subjects (Basten et al., 2011; Bishop et al., 2007; Derakshan & Eysenck, 2009). Furthermore, insufficient recruitment of the DLPFC, which is required for inhibitory control and was measured by a distractor inhibition task, has been suggested in high trait anxiety (Bishop, 2009). However, these studies examined general activation levels in the DLPFC and did not elucidate the exact mechanisms of how cortical processing is altered with respect to changes in excitability and/or inhibition. Our results indicate a dysfunctional top-down control of the DLPFC in specific phobias by showing reduced GABA-B-mediated intracortical inhibition in the left and right DLPFC at rest.

Another important finding of this study was a reduced N100 amplitude in specific phobias in response to fearful facial expressions compared to neutral ones. Thus, processing of fearful stimuli modulated intracortical inhibition levels in the DLPFC. We would like to point out that we cannot exclude that this is true to a similar extent for the control group, as the interaction CONDITION × GROUP did not reach statistical significance.

We presented emotional facial expressions in our 1-back task paradigm as this has been described as an adequate method to activate the amygdala, especially in anxious subjects (Breiter et al., 1996; Stein et al., 2007; Yang et al., 2002). Fearful facial expressions are a relevant biological stimulus indicating potential threat (Whalen, 1998).

A reduced inhibition in specific phobias could be related to a more active mirror system in anxious subjects viewing fearful compared with, for example, happy expressions (Rahko et al., 2010). Behavioral and neuroimaging data of previous studies suggest an impaired attentional control of the DLPFC during the confrontation with fearful facial expressions (Bishop et al., 2004). It was shown that this trait modulates neural activity in addition to state anxiety when processing fearful compared to neutral facial expressions (Bishop et al., 2007; Bishop, 2009). Higher state anxiety induced by fearful facial expressions was associated with lower activity in the DLPFC and higher amygdala response (Bishop et al., 2004). Our results show modulation of intracortical inhibition in the DLPFC, possibly due to an unspecific bottom-up arousal increase in the limbic system during the processing of fearful face expressions.

Moreover, we found the longest reaction times and highest error rates when memorizing angry facial expressions, but no significant N100 reduction. Therefore, longer processing time and effectiveness cannot be used to explain impaired DLPFC inhibition. The latter was specifically related to emotional processing of fearful facial expressions.

In addition, we found no significant N100 reduction in working memory-related processing at low load (1-back) in either group, highlighting that DLPFC inhibition in specific phobias was modulated by emotional processing of fear-relevant stimuli in the DLPFC and not by a low load 1-back task. However, it remains unclear whether a modulation effect would only become apparent at higher load. Studies comparing low with high perceptual load in anxiety found that anxiety did not affect DLPFC activity at high perceptual load (Bishop et al., 2007). Moreover, the lack of a modulating working memory effect could be due to the short time interval between stimulus offset and TMS pulse, as maintenance processes become more important after approximately 500 ms (Figueira et al., 2018). However, longer maintenance intervals could make the memory-related aspects of the task more difficult.

In our sample, we included different subtypes of specific phobias, with animal phobias being the largest subgroup. Since the emotion effect was even more pronounced when considering the animal phobia subgroup alone, we can conclude that our results of reduced N100 amplitudes in response to fearful expressions are valid for subjects with animal phobia.

Other subtypes were mixed and underrepresented with n = 7. Therefore, we cannot say with certainty that our results also apply to other nonanimal phobias. However, there was no significant difference in the N100 amplitude between the subtypes in any condition.

Accordingly, we found no evidence for significant differences between different specific phobia subtypes. Further studies are needed to confirm if the effect of an impaired inhibition in the DLPFC that we found for animal phobias is also valid for other subtypes of specific phobia (Lueken et al., 2011). For further details on the analysis of different subtypes see Supplementary Material A.

Since we had a moderate sample size (n = 46), we added confidence intervals for our calculated effect sizes (see Table C in Supplementary Material) to allow the reader to compare our observations with those of other related studies, and to facilitate the reproducibility of our results in future studies. A more detailed discussion on the power of our results can be found in Supplementary Material D.

Taken together, we found impaired intracortical inhibition in the DLPFC in the specific phobia group at rest, which was additionally modulated by the confrontation with fearful facial expressions. Thus, a generally impaired inhibitory function in the DLPFC was associated with specific phobias. In this respect, the existence of a categorical psychiatric diagnosis (independent of the subject's actual state) can be compared to a persistent anxiety trait in a dimensional model. Moreover, inhibition in specific phobias was modulated by processing of fearful facial expressions. This is possibly due to nonspecific bottom-up increases in limbic system activation during emotional processing of fearful faces, comparable with mild acute state anxiety.

4.2 Limitations

A limitation results from TMS, which was performed without neuronavigation. However, TMS was aligned to the localization described by Rusjan and colleagues (Rusjan et al., 2010), often used in TMS-EEG studies (Fitzgerald et al., 2009; Jarczok et al., 2021; Rogasch et al., 2015).

Furthermore, TMS was performed without auditory masking, thus auditory-evoked potential (AEPs) may have contributed to the TEP. AEPs are divided into three components: N1a, N1b, and N1c (Alain et al., 1997; Bender et al., 2006; Knight et al., 1988). However, AEPs do not occur at the stimulation site since lateralized parts like the N1c appear to have a contralateral topography with monaural stimulation as is to a limited extent the case with TMS (Hine & Debener, 2007; Langers et al., 2005). Thus, AEPs do not mimic a transcranial N100.