The number of neuroimaging studies in emotional similarity research is limited. To our knowledge, no neuroimaging studies have investigated neural differences in explicit judgments of similarity among the prevalent stimuli in research of emotional cognition, namely, natural, complex neutral and emotional picture scenes. Only a handful of studies have combined behavioural measures of similarity with neural data by using RSA. The results of these studies might help in understand the brain regions, which code the similarity among emotional stimuli. In these studies, during the fMRI scan participants were asked to attend to pictures while performing non-emotional rating tasks (e.g., ratings of indoor versus outdoor scenes). This was combined with behavioural judgements of similarity among the experimental stimuli. They found that brain activity patterns in regions involved in emotional processing, such as the insula and the ventromedial prefrontal cortex (VMPFC), represent the similarity structure between emotional and neutral stimuli observed at behavioural level (Chavez and Heatherton
2015; Levine et al.
2018).
Additional, indirect evidence about what make two emotional stimuli similar to each other at neural level is gleaned from neuroimaging investigations of emotional processing and categorisation. These mainly aimed at investigating how the brain codes the relationship between specific emotions, supporting either a categorical (Ekman and Friesen
1976), a dimensional (Russell and Pratt
1980), or a constructionist view (Barrett
2017). In these studies, participants were asked either to passively look at images, to attend to the feelings they evoke, to rate the valence and the arousal of these feelings, or to rate the valence and arousal of the picture and categorise it according to emotional labels (Costa et al.
2014; Ohira et al.
2006; Machajdik et al.
2010; Baucom et al.
2012; Sakaki et al.
2012; Yuen et al.
2012; Edmiston et al.
2013; den Stock et al.
2014; Motzkin et al.
2015; Hrybouski et al.
2016). The results of these studies were discrepant, probably because of the different perspectives of emotions adopted and methods used to elicit the emotions (Wager et al.
2015). In particular, locationist studies attempted to discover the unique brain feature associated with each emotional category, by adopting a one (brain region)-to-one (emotion) approach. For example, fear has been consistently localised in the amygdala (LaBar et al.
1998; LeDoux
2007; Öhman
2009), disgust in the anterior insula (Calder
2003; Wicker et al.
2003; Jabbi et al.
2008), sadness in the anterior cingulate cortex (Phan et al.
2002; Murphy et al.
2003), anger in the orbitofrontal cortex (Murphy et al.
2003; Vytal and Hamann
2010), and happiness in the dorsomedial prefrontal cortex (DMPFC) (Lindquist et al.
2012). As highlighted by Lindquist et al. (
2012), supports for a locationist account would be found if instances of an emotion category (e.g., fear) are consistently and specifically associated with increased activity in a brain region (or in a set of regions within a network) across multiple published studies. However, first, many studies showed that the aforementioned regions are associated with multiple categories of emotions (Lindquist et al.
2012), and during many other sensory, perceptual and cognitive functions (Yarkoni et al.
2011; LeDoux
2012). Second, it is not clear whether the findings from the locationist literature are reliable enough or consistent across studies (Wager et al.
2015). For these reasons, a psychological constructionist approach to emotion is preferable. According to this perspective, emotions are ‘situated conceptualisations’, that is, subjective interpretations of what is happening around us. Emotions arise from the interaction among many brain regions, interconnected in large-scale networks, according to a many-to-one approach. These brain regions are implicated not only in emotional processing, but also in more ‘cognitive’ functions, such as conceptualization (simulation of previous experiences), language (representation and retrieval of semantic concepts), and executive attention (attention and working memory).
However, this represents only indirect evidence of the neurobiological underpinnings of emotional similarity. The neural mechanism that allows emotion to influence overall perceptions of similarity is still unknown, as are putative neural differences during explicit judgements of similarity between natural, complex neutral and emotional events.