The impact of the left inferior frontal gyrus on fear extinction: A transcranial direct current stimulation study

Introduction: Fear extinction is a fundamental component of exposure-based therapies for anxiety-related disorders. The renewal of fear in a different context after extinction highlights the importance of contextual factors. In this study


Introduction
Anxiety is among the most prevalent and frequently relapsing emotional disorders, and affects approximately 284 million individuals worldwide [1,2].To treat this disorder, exposure-based behavioral therapies, based on the fear extinction model [3], have been extensively investigated and applied in clinical practice.However, the effectiveness of this treatment is often diminished by the renewal effect [4], which refers to the reappearance of the extinguished fear response when patients encounter the fear-inducing situations in real-life after treatment, due to contextual alterations.Therefore, investigating strategies to modulate the renewal effect of fear extinction might contribute to the development of more efficient treatment approaches.
Current evidence suggests the involvement of multiple brain regions in the process of extinction renewal [5].Considerable evidence has demonstrated that the hippocampus, medial prefrontal cortex (mPFC), and amygdala are primary regions regulating contextual fear-memory learning and retrieval [6,7].Specifically, studies have indicated that the hippocampus has a crucial role in context encoding in fear conditioning and in modulating context-dependent extinction memories in both animals and humans [8][9][10][11].Evidence suggests that repeated excitability-enhancing anodal transcranial direct current stimulation of the hippocampus enhanced performance in a contextual fear discrimination task in mice [12].The amygdala is well-documented for its critical role in the contextual control of fear expression, and activated during the renewal of fear [13,14].Furthermore, studies in animals and humans have reported that the activation of the mPFC is involved in the contextual encoding of extinction memories [15,16].Human fMRI studies suggest furthermore that the hippocampus modulates contextual encoding and the contextual retrieval of fear memories through its interactions with the amygdala [16,17].The interaction of the hippocampus and the mPFC is moreover suggested to be involved in the contextual retrieval of extinction memories [18].
Beyond these aforementioned regions, the left inferior frontal gyrus (LiFG) might be involved in fear extinction too.The LiFG has a crucial role for several cognitive functions.It is well known to be associated with language production and comprehension, and plays a role in speech production and articulation, as well as grammar and syntactic processing [19,20].In addition, this region is involved in controlling and regulating the content of thoughts and memories [21].It is able to consciously direct and regulate the content of thoughts by intentionally limiting the accessibility of memory, and it contributes to both, the encoding and involuntary recall of intrusive memories [22].Moreover, the LiFG is involved in emotion processing and the regulation of emotional responses.It contributes to the cognitive assessment of emotional stimuli and the management of negative emotions [23].In the field of fear extinction, this region shows larger activity in response to threat cues compared to safety cues [24], and might contribute to context-dependent renewal after extinction.Recent fMRI studies have demonstrated a prominent activation of the LiFG during context-specific appetitive extinction learning without fear components.Enhanced activity of the LiFG was associated with higher renewal levels, suggesting an engagement of this area in the contextual processing of extinction [25,26].Based on this evidence, the LiFG presumably plays an important role in the process of context-dependent renewal of extinguished memory content.However, research exploring the specific involvement of the LiFG in fear extinction and renewal is lacking.
Transcranial direct current stimulation (tDCS) is a well-introduced non-invasive brain stimulation tool to induce targeted alterations of cortical excitability, resulting in long-term potentiation and long-term depression (LTP/LTD)-like plasticity in the human brain, and has shown its potential to modulate various cognitive functions, including working memory, motor learning, and attention [27][28][29].Furthermore, tDCS has potential for the treatment of neurological disorders, such as epilepsy and Parkinson's disease [30,31].Recent studies have also demonstrated its potential to modulate fear extinction in humans [2,32,33].The objective of this study was to explore the causal role of the LiFG in the context-dependency of fear extinction learning by applying tDCS during extinction.We hypothesized that in case of the causal involvement of this area in the context-dependency of fear extinction learning, up-or downregulation of cortical excitability in this region will modulate fear renewal, with excitability-enhancing stimulation increasing the context-dependency of fear extinction learning and thereby increasing the renewal effect when participants are re-exposed to the original fearful context, and excitability-reducing stimulation producing the opposite effects.

Participants
192 healthy adults, aged 18-40 years, were recruited via online advertisements (97 females).All participants were right-handed according to the Edinburgh Handedness Inventory [34].All of them were either native German speakers or had at least a B1 level of German language proficiency.They were all naïve to the fear, and extinction learning procedure.Twelve participants were excluded (11 participants had to be excluded as they had null responses during the fear acquisition day, which was defined as a skin conductance response (SCR) amplitude of less than 0.01 μS in all trials [33], and one participant withdrew on the third day of the experiment due to a positive COVID-19 test result).The remaining 180 participants were randomly assigned to one of nine groups: they received anodal, cathodal, or sham tDCS during extinction, and three different context combinations were used in the different tDCS groups (AAA, ABA, and ABB; for a more detailed explanation see below).The sample size calculation was based on recommendations for non-invasive brain stimulation, due to the absence of relevant experiments providing specific data.Setting a medium effect size of f = 0.15, an α error of 0.05, and a power of 0.8, the resulting sample size was 17 participants per group.To compensate for dropouts, and outliers, 20 participants were enrolled per group.Data from three participants were removed from the analysis because their SCR were zero during all trials of the extinction phase.Thus, data recorded from 177 participants were included in the final analysis.The demographic information of the participants is shown in Table 1.
Participants were excluded if they fulfilled one of the following criteria: any history of or evidence for neurological diseases, metal implants in the head or neck, pacemaker or any implantable device, history of epilepsy or head injury, current or history of a psychiatric condition or mental illness, pregnancy, smoking, alcohol or drug abuse.Participants were asked to refrain from alcohol for at least 24 h and from caffeine for at least 2 h before each phase of the experiment.The study protocol was approved by the local ethics committee of the Leibniz Research Centre and complied with the Declaration of Helsinki.All participants gave informed consent prior to the experiment and they were provided with allowances to compensate for their time.

Fear conditioning paradigm
The employed fear conditioning paradigm was initially introduced by Milad et al. [35] and administered over three consecutive days (approximately 24-h intervals).Each day consisted of one phase: fear acquisition, extinction, or recall.Throughout the experiment, contexts were depicted through photographs of a desk lamp next to a computer on a table (context A, shown in Fig. 1) or a desk lamp on a bookshelf (context B, shown in Fig. 1).The conditioned stimulus (CS) was the color of the lamp's light (CS+: blue, CS-: yellow).The unconditioned stimulus (US) was an electric pulse applied over the tip of the right index finger through an electrode.The US followed the CS+ during acquisition, and the CS-was never followed by the US (Fig. 1).
The acquisition phase included 16 CS-and 16 CS+ trials.10 of the CS+ trials were followed by the US (reinforcement rate of 62.5 %).Context A was used in all conditions during the acquisition phase.Each trial consisted of 1s of presentation of the context alone, followed by 12 s of CS presentation.In reinforced trials, a 100 ms aversive US, consisting of four consecutive current pulses (duration of each pulse 500 μs with an inter-pulse interval of 33 ms) was applied.The US was applied 11.9 s after CS-onset, therefore co-terminating with the CS [36].Between visual stimulus presentations, a black background with a white cross in the middle was presented (duration of the intertrial interval (ITI) randomized between 23 and 26 s).This phase of the task lasted 20 min.
During extinction and recall, depending on the group, either context A or B was used.The trials included 8 CS+ and 8 CS-, and no US was paired with any of the CS+ trials in these phases.Trial duration and ITIs were the same as in acquisition, therefore the duration of these phases was exactly half of that of the acquisition (10 min each).Trial order within each phase were organized in a pseudorandom manner, adhering to three specific limitations: during the acquisition training, the first and the two last CS+ trials were followed by the US; no more than two consecutive trials included the same type of event, and the number of each type of event was identical for the first and second part of each phase.The order of trials was the same for all participants.
Skin conductance responses (SCRs): SCRs were obtained with the Biopac system during the whole task.The sampling rate was 1 kHz, the gain was 10 μS/V, and to prevent effects of high-frequency noise and low-frequency drifts, the SCRs data were band-pass filtered (0.5-10 Hz) (MP160, EDA100c-MRI module, BIOPAC Systems Inc., Goleta, CA).Two disposable standard snap electrodes (EDA BIOPAC EL509, BIOPAC Systems Inc., Goleta, CA) with the size of 27 mm width × 36 mm length × 1.5 mm thickness were positioned on the thenar and hypothenar   eminences of the non-dominant left hand.Participants were asked to keep their left hand as still as possible during the test.tDCS: tDCS was applied using a Starstim stimulator (NeuroElectrics, Spain), and the montage was defined based on modeling of electrical current distribution using SimNIBS [37].The dorsolateral prefrontal cortex (DLPFC) and ventromedial prefrontal cortex (vmPFC) are centrally involved in the regulation of fear extinction [18,38], and previous studies have shown that tackling these areas via non-invasive brain stimulation modulates fear extinction [39,40].To specifically assess the causal role of the LiFG in fear extinction, we thus developed a stimulation protocol via computational modelling which fulfilled the criteria of minimized electric field (EF) strength in the right and left DLPFC and vmPFC regions, while maximizing the EF in the LiFG area.The montage was implemented using four round rubber electrodes (each with a diameter of 2 cm) along with a layer of conductive paste as conductive contact medium with the scalp (TEN20 Paste, Weaver and Company, Aurora, CO, USA).The central electrode was placed 1 cm anterior to the midpoint of FC5 and FT7 (international 10-20 EEG system).Three return electrodes were placed over AF7, CP5, and FC3.An anesthetic cream (EMLA®, 2.5 % lidocaine, 2.5 % prilocaine) was applied to the four stimulation sites to facilitate blinding of the participants to tDCS [41].Throughout the extinction phase, 2 mA current intensity was gradually ramped up over 30 s, and maintained for 10 min, thus covering the whole extinction period, and then ramped down to zero over 30 s.For sham stimulation, current was ramped up to 2 mA over 30 s, maintained at 2 mA only for 30 s, and then ramped down again over 30 s.For the remaining sham stimulation session, current intensity remained at 0 mA.This protocol maintains participant blinding and does not cause neurophysiological effects beyond the stimulation duration with the target intensity, and thus does not have a relevant impact on performance [42,43].Participants were unaware of the type of stimulation.During the other two phases, tDCS electrodes were placed on the head to minimize the difference of personal experience of the participants across phases, though the device remained off for the whole time.
Questionnaires: Prior to the experiment, participants were required to complete two validated questionnaires (Depression Anxiety Stress Scale 21 (DASS-21) and Anxiety Sensitivity Index 3 (ASI-3)) to assess their recent psychological state.The DASS-21 was used to assess the presence and severity of depression, anxiety, and stress symptoms [44].It includes three dimensions (depression, anxiety, and stress), each with 7 items, resulting in 21 items.Each item was rated on a four-point Likert scale ranging from 0 (did not apply to me at all) to 3 (applied to me very much).Higher scores indicate that participants experienced more of these psychological symptoms during the week before the experimental phase.The ASI-3 is an 18-item self-report measure assessing sensitivity to anxiety-related symptoms and perceptions [45].It contains physical, cognitive, and social subscales, and each subscale contains 6 items.Each item was rated on a Likert scale ranging from 0 (very little) to 4 (very much).Higher scores indicate larger anxiety sensitivity.Furthermore, before each phase, the Positive and Negative Affect Scale (PANAS) was administered to assess the current emotional state of participants [46].The scores for positive and negative affect were separately assessed on a Likert scale ranging from 1 (very slightly) to 5 (extremely), with each affect being rated by 10 items.Higher scores indicate more prominent positive or negative affect.
After the acquisition phase, participants were asked if they received an electric pulse in the last CS+ trial.They were also required to estimate the intensity of the last US they received on a Likert scale ranging from 0 (not unpleasant) to 9 (very unpleasant).In addition, participants were required to assess the likelihood of US occurrence after each of the CS+ and CS-images on a scale ranging from 0 to 100 %.
After each phase, participants were asked to rate valence, arousal, and fear associated with the CS+ and CS-images.The valence was evaluated via a Likert scale ranging from one (very pleasant) to five (very unpleasant).Arousal and fear were measured using visual analogue scales ranging from 0 (very relaxed, not afraid) to 100 % (very excited, very afraid), respectively.
To assess blinding success and side-effects of tDCS, participants completed a questionnaire after each phase [47].Participants were required to first answer whether they received tDCS in this phase; in case of a positive response, the respective side-effect related questions needed to be responded, including the presence and severity of visual phenomena, itching, tingling, burning, and pain sensations during stimulation on a Likert scale ranging from 0 (none) to 5 (extremely strong).If participants answered that they received tDCS, they were also asked to fill in a 24-h post-stimulation side-effect questionnaire before the extinction and recall phases in which they were asked to rate the presence of skin redness, headache, fatigue, concentration difficulties, nervousness, and sleep problems, using the same six-point Likert scales [48].

Procedure
At the start of each phase, participants were positioned in front of a computer screen (size 43.2 × 28.8 cm), and the basic procedure of the experiment was introduced to them.Subsequently, tDCS electrodes, as well as electrodes for SCR recording, and US application were connected.After that, only on the first day, the individualized US intensity was determined using a constant current stimulator (DS7A, Digitimer Ltd., London, UK, maximum current output of 100 mA).The current amplitude was gradually increased, and subjects were asked to report their perception of the stimulus using a nine-point Likert scale (from 0, "no or very slightly" to 9, "completely intolerable"), until level 8, "highly uncomfortable, but not painful" was reached.This intensity was kept constant throughout the whole experiment.
At the start of the experiment, subjects were instructed to pay attention to a possible connection between the light color and the electrical pulse applied to the finger.In the beginning of the next two phases, they were told that any connection they noticed on the first day would remain unchanged [36].Presentation software (version 23.0, Neurobehavioral System Inc., Berkeley, CA) was used to create the experimental paradigm.The experiment was conducted in a dimly lit and soundproof laboratory.The temperature of the lab was kept in the range of 22-24 • C.

Calculations and statistical analysis
The SCR data were processed in Matlab software (Release 2021b, The MathWorks Inc., Natick, MA) using a semi-automated peak detection algorithm [49].The SCR amplitudes were defined as the highest trough-to-peak alteration within the time frame of 1-12.5 s following the CS onset.The detected peaks met the criteria of a minimum amplitude of 0.01 μS and a minimum rise time of 0.5s [50].To standardize SCR for individual differences, we performed the mean correction method for each raw SCR value [51].This involves calculating the average value of the entire SCR data for each individual and subtracting this mean SCR from each raw SCR data.
All statistical analyses were conducted with IBM SPSS Statistics software (IBM Corp., IBM SPSS Statistics for Windows, Version 29.0.Armonk, NY, United States).Chi-square test and one-way ANOVAs were conducted to evaluate the demographics and psychological assessments (DASS-21, ASI-3) to check for equivalence between groups.The PANAS was analyzed via a mixed model ANOVA, with Session (1, before acquisition; 2, before extinction; 3, before recall) and Type (positive and negative affect) as within-subject factors, and tDCS (anodal, sham, cathodal) and Context (AAA, ABA, ABB) as between-subject factors.In all ANOVAs, if Mauchly's sphericity test was significant, a Greenhouse-Geisser correction was used.Post-hoc least significant difference (LSD) comparisons were performed in case of significant outcomes of the ANOVA.Statistical significance was set at p ≤ 0.05 for all tests.
For fear acquisition, the ratings of the last US and the guessed likelihood of US occurrence after CS presentation were analyzed via twoway ANOVAs, three types of tDCS and three kinds of Context as between subject factors, and rating scores as a dependent variable.Mixed model ANOVAs were performed to assess the data from the questionnaires for stimulus ratings (valence, arousal, fear).The between-subject factors were tDCS (anodal, cathodal, sham), and three kinds of Context (AAA, ABA, ABB), whereas Phase (after acquisition, after extinction, after recall) and Stimulus (CS+, CS-) served as within-subject factors.
Since more than 95 % of participants responded that they did not receive tDCS during the first and third days, data from the side-effect questionnaires were only analyzed for the second day.A Chi-square test was used to explore blinding efficacy.The adverse effects of tDCS were analyzed separately via two-way ANOVAs with three types of tDCS and three kinds of Context as the independent variables and rating scores for each side-effect as the dependent variable.

Demographics
Table 1 contains the demographic information of the different groups, including age, gender, education, and US intensity.No statistically significant differences between the groups emerged with respect to these variables as indicated by the results of the chi-square test and oneway ANOVAs (Table 1).

Skin conductance responses
In this section, we first report the analysis results for the absolute SCRs values, followed by the results of the analysis of differential SCRs (CS+ -CS-).

Fear acquisition training (day 1):
The mixed model ANOVA revealed a significant main effect of Stimulus (F (1, 168) = 134.119,p = 0.001).The CS+ (mean = 1.069,SE = 0.069) elicited a significantly larger response compared to the CS-(mean = 0.483, SE = 0.035).A significant main effect of Trial was also observed (F (9.9, 1667.168)= 5.072, p = 0.001).Furthermore, the Stimulus × Trial interaction was significant (F (9.7, 1629.819)= 5.674, p = 0.001).No other significant main effects or interactions emerged.Post hoc tests showed significantly larger SCR responses to the CS+ compared to the CS-from trial 2 to trial 16 (all p < 0.001), indicating that the participants learned that the CS+ is followed by an electric pulse (Table 2, Table S1 and Fig. 2).These results indicate that the participants learned that the CS+, but not the CS-predicts the US.

Relative stimulus-dependent SCRs (CS+ -CS-)
Acquisition: The mixed model ANOVA showed a significant main effect of Trial (F (9.8, 1645.378)= 5.715, p = 0.001).Post hoc test demonstrated that the difference between CS+ and CS-was significantly lower in the 1st trial compared to trials 3, 5, 6, 7, 9 and up to the last trial, suggesting an increase in differential responses during the fear acquisition.No other significant main effects or interactions were found (Table 3, Table S4 and Fig. 5).
Extinction Recall: The mixed model ANOVA revealed a significant main effect of tDCS (F (2, 168) = 3.573, p = 0.030).The response difference between CS+ and CS-was larger in the anodal condition compared to the cathodal condition (p = 0.009), while there was no significant difference compared to the sham condition (p = 0.11).This result demonstrates that anodal tDCS led to preservation of fear compared to the cathodal stimulation condition.Furthermore, a significant Trial × tDCS interaction emerged (F (11.4,959.258) = 1.789, p = 0.049).Post-hoc comparisons revealed that in the 1st and 8th trials, the difference between CS+ and CS-elicited SCRs in the cathodal condition was significantly smaller compared to the anodal and sham tDCS conditions (p < 0.05).In addition, for the 2nd trial, the anodal condition showed a larger difference compared to the cathodal (p = 0.028) and sham (p = 0.036) conditions (for more details refer to Table 3, Table S6 and Fig. 5).
The results of the other questionnaires (Psychological assessments, the US rating, PANAS, valence, arousal and fear ratings, tDCS blinding and side-effects) are provided in the supplementary material.

Discussion
In this study, we investigated the influence of applying tDCS over the LiFG during the extinction phase of fear extinction learning in different context combinations (AAA, ABA, or ABB with respect to acquisition, extinction, and extinction recall) to explore the involvement of this area in context encoding.Towards this objective, a Pavlovian fear conditioning paradigm was applied, and physiological responses (SCRs) and subjective ratings were recorded.
Overall, our observations did not support a causal role of the LiFG in the context-dependency of fear extinction learning.For SCR, anodal tDCS led however to context-independent augmented fear responses during fear extinction and recall, implying a general impairment of fear extinction learning caused by excitability enhancement of this area.This increased response persisted throughout the recall phase together with an enhanced fear response to the safety cue, i.e. during the extinction recall phase, anodal tDCS resulted in larger responses to both CS+ and CS-compared to the sham condition.On the other hand, cathodal tDCS led to reduced differences between responses to the CS+ and CS-during  the extinction recall phase compared to anodal and sham stimulation.This implies that cathodal tDCS prevents the return of fear and preserves the extinction effects during the recall phase.These observations, besides the absence of a significant effect of context, suggest that though the LiFG is involved in the fear network, modulating its excitability may not be associated with the context-dependency of fear extinction learning.
An involvement of the LiFG in the human fear network has been previously shown in several studies.An fMRI study has shown that a painful US -similar to the one used in our study -leads to a larger activity of the LiFG towards the CS+ compared to the CS-during fear conditioning [24].In another fMRI study, during fear acquisition in healthy humans, larger BOLD responses in the LiFG were evoked in the unpaired CS+ than in the CS-condition [52].Further correlational evidence for the involvement of the LiFG in the fear network originates from a study showing increased activity of the LiFG during renewal in both healthy controls, and PTSD patients [53].These studies together are supportive for a role of this region in the fear network in healthy participants during fear acquisition, and renewal.
Also, the deterioration of fear extinction learning by anodal stimulation over the LiFG is consistent with findings of prior research.Panic disorder (PD) patients showed increased LiFG activation in fear conditioning and extinction recall compared to healthy controls [54][55][56].Cognitive behavioral therapy, which is based on the principles of extinction [57], decreased LiFG activity in respective patients [55].Furthermore, PD patients showed increased connectivity between the LiFG and regions of the fear network (including amygdala, hippocampus, anterior cingulate cortex, medial and lateral PFC), compared to healthy individuals [55].This suggests an increased association between the LiFG and the fear network in PD, in a way that might trigger emotional responses related to fear network activity more easily in these patients.Even this sustained activation of the connection attenuated the recall of extinction memories in PD patients [56].Increasing the activity of the LiFG by anodal tDCS in our study might therefore have increased the connectivity with the fear network, leading to increased fear responses and deterioration of fear extinction learning.Furtheremore, a positive correlation between left amygdala and inferior frontal gyrus (BA47) activation has been observed during the late fear acquisition [16], which might lead to larger amygdala activity and therefore larger SCRs to the CS+, as a result of increased activity of the LiFG via anodal tDCS.
Alternatively, the prediction error (PE), which controls learning processes based on the Rescorla-Wagner rule [58], is defined as the difference between the actual and expected intensity of the US in the fear conditioning and extinction paradigm [59].A negative PE signal is necessary for successful extinction [60,61], and increased fMRI activation of the LiFG in a fear conditioning task has been observed to be associated with a positive PE [60].Application of anodal tDCS over LiFG during extinction might have increased the positive PE and thereby fear responses to the threat cues.
During the extinction recall phase, larger fear responses to threat cues were observed in the anodal compared to the other two conditions, and the anodal tDCS condition resulted in a larger fear response compared to the sham condition at the first exposure to safety cues.The amygdala is activated during processing of threat cues and deactivated during processing of safety cues [62][63][64].Prior research has demonstrated that the connection of the LiFG with the amygdala is involved in fear conditioning [16], and its activation increases the return of fear [65].We speculate that the increased activity of the LiFG during extinction learning induced by anodal tDCS disrupt the regular extinction-related fear network modulation through its connections with the amygdala, thus impairing the efficiency of extinction learning, and leading to the recovery of fear in the following recall phase.
On the other hand, excitability-decreasing cathodal tDCS applied during extinction learning did not result in an improved extinction learning effect.However, the differences of fear responses to fear and safety cues during the recall phase were smaller compared to other conditions.This outcome might be associated with the engagement of the LiFG in the neural network of memory control [21].In a trauma film viewing test, healthy individuals who experienced intrusive recall showed increased activation of the LiFG compared to those without returning intrusive memories [66].Subsequent investigations revealed that the LiFG is not solely involved in traumatic memory encoding, but also in the involuntary recall of intrusive memories [67].Thus, we speculate that decreasing LiFG activation by cathodal tDCS during extinction learning interferes with the functioning of the neural network of memory control, thereby suppressing fear-related memory processing and retrieval.In addition, no better extinction effect was shown during cathodal stimulation compared to the sham condition, likely because extinction learning is the formation of new not fear-related memories, whereas the LiFG may be engaged in the encoding and recall of negative emotions [23,68].Alternatively, cathodal tDCS may improve the consolidation of newly established memories during the extinction learning phase, which is suggested by its improving effects on retention of extinction memory during recall.This might have been accomplished by an improved signal to noise ratio via excitability-diminishing cathodal tDCS, which might reduce activation of fear memories during extinction.A related mechanism has been shown for the effects of cathodal stimulation on motor memory formation [69].
In contrast to the observed vegetative responses, valence, arousal, and fear ratings showed no significant differences between conditions (see the supplemental materials, section valence, arousal, and fear ratings for details).One potential reason for this is that the questionnaires are largely influenced by subjective perceptions [70].In contrast, SCR as Fig. 5. Differential SCRs for each trial during three phases, calculated as the difference between response to the CS+ compared to the CS-.The color of the lines shows the tDCS condition, and error bars show the standard error of the mean.The post hoc test results revealed that during the extinction recall phase, the difference between SCRs with respect to the CS+ and CS-was smaller in the cathodal compared to the anodal and sham conditions in the 1st and 8th trials (p < 0.05, hash symbol), whereas the anodal condition showed a larger difference compared to the other two conditions in the 2nd recall trial (p < 0.05, asterisk).(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)a widely employed index of fear conditioning, measures sympathetic autonomic system activity, an indirect index of fear [71].It offers insights into autonomous emotional regulation, excluding conscious processing.Furthermore, the amygdala plays a critical role in regulating vegetative aspects of the emotional response, encompassing autonomic functions [72].The modulation of the LiFG may primarily evoke autonomic emotional responses through its connectivity with the amygdala.
The initial hypothesis of an involvement of the LiFG in the context dependency of fear extinction learning was based on the results obtained from an appetitive learning task [25], which differs from the fear conditioning/extinction paradigm explored in the present study.It is possible that the neural mechanisms involved in the control and regulation of these two paradigms are different [73].Furthermore, in the respective study by Lissek et al. [25], the acquisition, extinction, and recall phases were completed on the same day, while in our study, these phases were conducted on three consecutive days to avoid an extinction deficit due to the administration of the extinction protocol soon after acquisition [74].Previous evidence has shown that in a contextual fear conditioning paradigm, immediate extinction following acquisition leads to a stronger renewal effect during recall compared to delayed extinction after 24 h [75].Furthermore, our experimental paradigm was relatively simple compared to the study of Lissek and co-workers, which included up to 16 types of stimuli compared to the two types of stimuli employed in the present protocol.Another point that is worth further exploration is the duration of context presentation.In our paradigm, the context alone was presented only for 1 s, which might have led to focusing more on the CS rather than the context.It therefore might be valuable in future studies to ask the participants if they noticed any change of context, as attention towards the context could play a significant role on its effects [76].It might also be useful to extend the duration of context-alone presentation to increase attention to the context and explore the impact of this factor on fear renewal.

Limitations and future directions
Some limitations of this study have to be taken into account.First, our findings are primarily based on psychophysiological measures (SCRs) and questionnaire assessments.Though SCR is the most widely employed measure of conditioned fear responses, it has some drawbacks, such as sensitivity to movement and temperature [77,78].Incorporating additional measures, such as pupillary response, electroencephalography, and heart rate could provide a more comprehensive and reliable reservoir of physiological outcome measures [79,80].
In addition, despite the relatively large scale of our overall dataset, regarding the context and tDCS polarity factors, the number of participants per group remained limited.This makes the findings preliminary and emphasizes the necessity for replication with larger samples in future studies.Moreover, this paradigm might be somewhat weak for assessing context-dependent fear learning.The reason for this could be similarities in the context pictures themselves, or perhaps the context and cue are not coded separately when presented this way.Furthermore, using neuroimaging techniques such as fMRI in future studies to look into the modulation of brain activity and also functional connectivity, specifically as a result of tDCS application, could provide deeper insights into the underlying mechanisms.

Conclusion
This study provides evidence that tDCS targeting the LiFG can modulate fear extinction, and recall.Anodal tDCS deteriorated fear extinction learning, and maintained fear memory during the extinction recall phase.In contrast, cathodal tDCS enhanced extinction retention.The effects of tDCS were context-independent.The results of the present study, though it did not approve an association of the LiFG with contextdependency of fear extinction learning, and recall, suggest that the LiFG may have a relevant role in the fear network and is involved in the processing and consolidation of fear memories.However, for a more precise understanding of the exact role of the LiFG, further investigations incorporating neuroimaging techniques and the combined application of neuroimaging and brain stimulation are imperative.These findings provide further mechanistic information about the effects and promising potential of tDCS, which might be exploited for the treatment of neurological and psychiatric disorders.
each group, participants received one type of tDCS and were exposed to one context/context combination in the different experimental phases.A/s/c indicates anodal/sham/cathodal tDCS; AAA/ABA/ABB represents the context in fear extinction task (A = context A, B = context B, context order: acquisition/extinction/ extinction recall).For instance, a_AAA stands for anodal stimulation in the context AAA; s_ABA for sham stimulation in context ABA.M = Male; F=Female; US intensity refers to the intensity of the US in milliamps.DASS21 = Depression Anxiety Stress Scale 21; ASI-3 = Anxiety Sensitivity Index 3; p=p-value, df = degree of freedom, F --F-value, NA = not applicable.Gender was analyzed with a chi-square test, and the chi square value was 1.128.Values are given as mean ( ± standard deviation).No significant differences between groups were observed for any of the measures.

Fig. 1 .
Fig. 1.Experimental design.On Day 1, participants performed the fear acquisition phase (16 CS+; 16 CS-; 10 US) in context A. On Day 2, participants randomly received anodal, cathodal, or sham tDCS during the extinction learning phase (8 CS+; 8 CS-; no US) in context A, or context B. On Day 3, participants underwent the recall phase (8 CS+; 8 CS-; no US) in context A, or context B. tDCS: transcranial Direct Current Stimulation; CS+: reinforced conditioned stimulus; CS-: not reinforced conditioned stimulus; US: unconditioned stimulus; Questionnaire: The Positive and Negative Affect Scale was conducted prior to each phase.After each phase, valence, arousal, and fear rating data were obtained via respective questionnaires.

Fig. 2 .Fig. 3 .
Fig. 2. Shown are the absolute SCR values in the acquisition phase separately for different stimuli (CS+ and CS-) in each context.The color of the lines shows the tDCS condition and error bars show standard error of the mean.The mixed model ANOVA results showed no significant differences between stimulation types and context combinations (p > 0.05).(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) S1.B.

Fig. 4 .
Fig. 4. SCRs over the eight trials in the extinction recall phase, shown separately for different stimulus types (a: CS+ and b: CS-).The color of the lines shows the tDCS condition, and error bars show standard error of the mean.As revealed by the post hoc tests, a) for the anodal condition, SCRs in response to the CS+ were significantly larger compared to the cathodal condition at the 1st and 2nd trials (p < 0.05, black asterisks), and compared to the sham condition at the 2nd and 4th trials (p < 0.05, red asterisks), b) for the CS-, the anodal condition yielded a significantly larger response compared to the sham condition at the 1st trial (p = 0.032) and the cathodal condition showed a significantly larger response compared to the other two conditions at the 8th trial (p < 0.05, the hash symbol).(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Table 1
Demographic data of participants.Comparisons between groups were conducted using one-way ANOVAs for psychological assessment variables.For gender, a chisquare test was performed.

Table 2
Skin conductance responses.Results of the mixed model ANOVA for absolute values in the acquisition, extinction, and recall phases.
a Significant results at p < 0.05.Y.Ma et al.

Table 3
Results of the mixed model ANOVA for acquisition, extinction and recall phases, conducted for the dependent variable C+/C-difference.
Y. Ma et al.