Proactive and reactive control during social information processing in neutral and interpersonal contexts

Abstract

The ability to behave flexibility and in adaptation to the environment are among the repertoire of human skills. We are constantly exposed to numerous contexts and sources of information, to which we need to provide a suitable response. This is especially necessary when we face situations of certain complexity, such as when routine procedures lead to divergent responses (Norman & Shallice, 1980). This set of skills, better known as cognitive control, allows us to navigate changing and challenging situations. The studies on control mechanisms agree on the importance of frontoparietal regions that implement adjustments in high demand scenarios, which have been jointly named as the Multiple Demand Network (MD; Duncan, 2010). These regions contribute when it is necessary to prepare a task set (proactive control) as well as to provide responses at stimuli presentation (reactive control, Braver, 2012). Further, Dosenbach et al. (2008) proposed a distinction of control mechanisms in different networks with complementary profiles, initiated during effortful tasks. Thus, a frontoparietal network would act transiently in response to events, whereas a cinguloopercular one would sustain an active and stable representation of task context. Although the work on control mechanisms is extensive, most studies focus on stimuli or paradigms rather simple (but see Egner et al., 2008; González-García et al., 2017; Palenciano et al., 2019a,b) or lacking social context. As social beings, humans are exposed constantly to interactions with other people, where multiple sources of personal information generate expectations about them. This suggests that these phenomena are tied to some effort. In fact, several studies have explored strategic behaviour during social interactions (e.g. Ruz et al., 2013), since even if our interpersonal decisions are followed by a sense of rationality, these are biased by evaluative information about others (Díaz-Gutiérrez et al., 2017), even if it is unrelated to their behaviour (Alguacil et al., 2015; Tortosa et al., 2013). With all of the above, this thesis has aimed to study the role of control mechanisms in the maintenance and representation of social information and its interference in neutral and interpersonal contexts. To fulfil this goal, we have carried out three experiments combining with uni- and multivariate analyses on fMRI data. First, we aimed to examine, in a neutral context, how the representation of social stimuli varied depending on when it is necessary to implement such information. To do this, we employed a task-switching paradigm adapted to fMRI, where participants had to make sequential categorization judgements on human faces (race, gender, emotion) and needed to maintain these task sets in a currently-active or delayed manner. In addition, we used MPVA to study fine-grained differences in the neural representation of social dimensions, depending on when they were relevant. Results show that intended task sets are represented differently from the currently-active task. Thus, lateral orbitofrontal cortex codes the current task only, whereas lateral prefrontal cortex represents intended task sets. This finding supports previous studies indicating that this region contains specific information about intended abstract goals (Momennejad & Haynes, 2013). Moreover, a remarkable finding is that a region located in ventromedial prefrontal cortex/orbitofrontal cortex contains information about relevant task sets regardless of when they need to be implemented (now or later on), in line with a recent proposal whereby this region would serve as a map for cognitive states (Schuck et al., 2016). Given the relevance of social information, and in particular, human faces to predict people’s behaviour, we decided to examine control mechanisms in an interpersonal context. Here, we adapted a Trust Game (Camerer, 2003) where participants needed to decide to cooperate or not with a series of partners and the final outcome from the interaction depended on reciprocation between the participant and the partners in the game. Further, different dimensions could predict partners behaviour, so we could compare the role of control when this information was social (identity or emotional expression) or non-social (colour of a frame). On top of that, we employed a mixed design (Petersen & Dubis, 2012; Visscher et al., 2003) to study sustained and transient mechanisms in charge of representing social and non-social relevant dimensions and their conflict. Results show general behavioural interference effects between the different dimensions. Also, conflict between the social dimensions elicited transient activation in frontoparietal regions, which highlights the relevance of social information in interpersonal decisions. Last, we examined control mechanisms in another interpersonal context in which social information came from descriptions about the partners’ moral traits. This way, we adapted a different paradigm, an Ultimatum Game (Güth, 1982). In this case, we wanted to explore the neural mechanisms underlying the representation of social information and the generation of expectations about others. Likewise, this paradigm allowed us to examine the interaction between such predictions and partners’ actual behaviour. The findings from this study showed how social expectations modulate decisions, even if they are not predictive of people’s behaviour. Moreover, a set of frontal regions represented the valence of participants’ expectations. Importantly, the better these areas represented this information, the more biased were participants’ decisions by it. Further, the influence of social information was manifested when participants’ expectations did not match their partners’ behaviour, which triggered frontal activation in control-related areas. Overall, our findings highlight the versatility of control mechanisms. These underlie the preparation for future tasks, as well as the maintenance of social expectations. Further, the representation of social information affects performance and interpersonal decisions making. In this way, reactive mechanisms are required to implement appropriate adjustments when different sources of information lead to divergent responses or when social information induces predictions about others that are not met.