Most conceptions of the relationship between mood and emotions suggest that moods may potentiate matching emotional reactions (for example, irritable mood facilitates angry reactions [1]). Depressed individuals show more attention towards negative, anxiogenic stimuli [2] which has also been found to be a risk factor for developing major depressive disorder (MDD) [3]. Importantly, functional magnetic resonance imaging (fMRI) studies have demonstrated activation anomalies in both MDD patients and in patients at risk for depression during presentation of fearful images [4, 5].

Interestingly, similar results have been found in healthy individuals with family history of depression (FHP) when compared to healthy individuals without any family history of the disease (FHN). FHP subjects exhibit impairment in emotion recognition [6] and have been shown to have higher amygdala and nucleus accumbens activation in response to the presentation of fearful faces when compared to age-matched FHN controls, in line with previous findings showing that FHP subjects have significantly elevated waking salivary cortisol when compared to FHN subjects [7]. However, when face viewing is accompanied by a constrained attention task (that is, having to rate nose width on the face and subjective fear while viewing the face), the differences between FHP and FHN subjects disappear whilst prefrontal activity increases [8]. This suggests that FHP subjects may be able to normalize emotion-related neural functions by focusing their attention and that face-viewing with unconstrained attention may leave room for aberrant psychological processes associated with the risk for developing MDD [8]. However, both behavioral and event related potential (ERP) studies have identified subtle deficits in selective attention among FHP individuals that may affect their ability to adequately regulate emotion under stressful circumstances [9].

In order to investigate the interplay between cognitive and emotion processing in both FHP and FHN participants, a masked emotion task combined with a cognitive task known to elicit cognitive processing bias in MDD might be useful to reveal performance differences between these two groups. In this context, neuropsychological studies in a task called 'dot-probe' suggest that depression is associated with an attentional bias towards negative information [10] and that effortful cognitive control of negative emotions can reduce the bias towards fearful stimuli [11]. Neuroimaging dot-probe studies suggest that unmasked fearful faces facilitate visual processing [12, 13] and that the amygdala modulates fear responses in the occipital cortex [14]. Further, previous fMRI studies on participants performing in the dot-probe task (for example, [12]) have shown that the amygdala directs spatial attention to backward masked fearful faces through a network of brain structures that include the left anterior cingulate cortex (CC), right superior temporal sulcus and right lingual gyrus [15–17]. Other research in the dot-probe task has shown that individuals with MDD cannot avoid attending negative information in their environment [18] and FHP individuals attend selectively to sad faces [19]. Importantly, there is evidence that effortful control modulates the relationship between negative affectivity and attentional bias in the dot-probe task, with low levels of effortful control and high levels of negative affectivity predicting a preference for threat stimuli [11]. With respect to the fact that, when viewing subliminal masked stimuli, participants do not focus their attention on the masked stimuli, the dot-probe task is highly interesting because it might elicit activity associated with vulnerability to MDD [8]. However, to the best of our knowledge, little or no research has investigated putative functional anomalies in the brain showing during performance in this task in either MDD or FHP individuals.

In the present study, we hypothesized that patients with MDD compared to HC subjects, but also FHP subjects compared to FHN subjects, exhibit differences in emotional processing of fearful versus neutral stimuli when attention is biased during performance in a dot-probe task. Based on previous fMRI findings (for example, [17]) in a similar behavioral task [8, 20–24], we selected the CC, amygdala, insula and prefrontal cortex as primary regions where we expected significant differences between groups to appear.

While being presented with masked fearful stimuli, our participants showed significant differences in areas that are thought to play a key role in emotion processing, namely the CC and insula. Further, our results suggest a link between family history of MDD and functional anomalies in the Heschl's gyrus.

MDD patients showed reduced activity in the left middle CC when compared to the HC group, adding to previous findings suggesting an important role of CC anomalies in the diagnosis of anxiety disorders and/or depression [34]. In particular, this effect was observed when fearful facial expressions elicited stronger activation, in line with earlier fMRI research suggesting a role for the CC in orienting spatial attention to crude threat signals [8, 17]. In this study, we found no effect in the amygdalae for all participants, suggesting a more direct involvement of the CC in attention recruitment during performance in the dot-probe task. This result might also agree with previous findings showing an involvement of the CC in shaping emotional expectancy in both healthy individuals and patients with MDD [35]. Interestingly, we found no effect in the prefrontal cortex, which contrasts with previous research showing prefrontal anomalies in both MDD and healthy participants with family history of MDD [8, 24, 36]. Our findings might suggest that our version of the dot-probe task is not sensitive to prefrontal activation anomalies in either MDD or FHP subjects, in line with previous research on the dot-probe task showing the involvement of the anterior CC, amygdalae, temporal and occipital cortices [12, 13]. Although these results might need replication, an important consideration is to be made when comparing our data to previous findings in similar experimental contexts: while in the study of Monk et al. [8] participants were required to consciously shift their attention towards a specific feature of the stimulus presented (that is, participants were asked to rate the size of the nose in a given face), in the dot-probe task, stimulus perception is subliminal (emotional stimuli are masked by neutral stimuli). This might have an effect on how attention is recruited and might explain why, in our study, we detected no prefrontal effect. Previous research [15, 17, 37] has shown that the CC plays a key role in directing attention when an individual is not conscious of an emotional facial stimulus being presented. Further, recent ERP findings suggest that backward masked fearful face-elicited spatial attention facilitates behavior and modulates the early stage of facial processing [38].

Interestingly, when compared to the FHN group, the FHP group exhibited activation differences in the right Heschl's gyrus. In this area, FHP participants had greater activation for contrast F > N and smaller activation for contrast F < N. The Heschl's gyrus is a subregion of the superior temporal gyrus that, apart from being functionally involved in auditory processing, plays an important role in emotional processing, theory of mind and empathy [39, 40]. Volumetric reductions in this area have been found in MDD patients, even after recovery from the disease [41]. Moreover, similar results have also been shown in bipolar disorder patients [42]. Our results implicate activation differences in superior temporoparietal areas between individuals with and without family history of MDD during exposure to fearful facial expressions. As only the right hemisphere was involved, our findings might also suggest a lateralization effect. This is perhaps in line with previous fMRI research suggesting a role of the right Heschl's gyrus during exposure to emotional (auditory) stimuli [43] and showing that the activation of auditory processing regions specialized for language, like the Heschl's gyrus, can be detected during performance in tasks requiring visual perception of the human face [44]. This might support the belief that this cortical area plays a role in acquired dynamic audiovisual integration mechanisms in the left superior temporal sulcus [44]. In this context, our results suggest a non-task specific role of the Heschl's gyrus in facial emotion processing, which is perhaps lost in MDD.

It is certainly interesting that MDD patients and FHP participants showed activation anomalies in different cortical areas, when compared to FHN participants. However, in the present study, these two groups consisted of unrelated individuals and whether MDD affects functional aberrances already detected before its onset in FHP subjects should be determined by future longitudinal studies.

The present study has a number of limitations. The subject sample was probably too small to reveal behavioral differences across groups. Additionally, the total number of HC participants was almost double than the number of MDD patients. This surely had an effect on our results. For example, our raw data suggested that MDD patients made considerably more errors than the HC group, although this could not be supported by statistical significance. Increasing the participant sample and having a comparable number of HC versus MDD participants would probably have yielded more definitive results. Further research in a larger sample of participants is also needed to confirm our RT analysis and comparisons (the RTs of some participants were lost due to a system failure).

Importantly, in the present study, we did not include images displaying faces conveying positive (happy) emotions. For this reason, we cannot rule out that our fMRI findings simply reflected brain activation associated with the presentation of emotional stimuli. In this regard, further fMRI research should aim at comparing brain activation relative to both happy and fearful faces. Participants were asked after scanning whether they could recognize subliminal images and confirmed that they did not detect them. Employing a detection task within the session would have been difficult, because participants already had to respond to the dots they saw after the shortly presented face images (100 ms). Further, future studies should also investigate correlations between behavioral and MRI data. Quite a substantial limitation of the present study is also represented by the inclusion of MDD patients with differences in medication which, as shown by previous fMRI research (for example, [23, 45]), can affect brain activation. Finally, it also possible that the outcome of this research was affected by our recruitment method. We selected FHP participants as first-degree relatives of patients with well-known recurrent depression, but who did not necessarily take part in the study. As all the MDD patients recruited were assessed by the same psychiatrists, selecting relatives of MDD patients involved in the study might have contributed to ascertain family history of the disease in FHP participants. On the other hand, this would have introduced a genetic bias, whose selective effect on MRI data should be investigated in future research.

Our results suggest that, in the dot-probe task, FHP subjects exhibit altered activity in the right Heschl's gyrus associated with subliminal presentation of fearful stimuli, indicating that lateralized alteration in the functionality of this cortical area could be associated with a higher risk of becoming depressed, although this should be confirmed by longitudinal studies on a larger population sample. Moreover, in individuals with MDD, the CC might mediate a preference for negative emotions as delivered by subliminally presented human faces. Further research is surely needed to explore the correlation between cortical and/or subcortical anomalies and behavioral responses in a similar experimental setting and to investigate putative therapeutic effects of psycho- and pharmacotherapy on the activation anomalies we detected.