Affective and cooperative social interactions modulate effective connectivity within and between the mirror and mentalizing systems

Abstract

Decoding the meaning of others' actions, a crucial step for social cognition, involves different neural mechanisms. While the "mirror" and "mentalizing" systems have been associated with, respectively, the processing of biological actions versus more abstract information, their respective contribution to intention understanding is debated. Processing social interactions seems to recruit both neural systems, with a different weight depending on cues emphasizing either shared action goals or shared mental states. We have previously shown that observing cooperative and affective social interactions elicits stronger activity in key nodes of, respectively, the mirror (left posterior superior temporal sulcus (pSTS), superior parietal cortex (SPL), and ventral/dorsal premotor cortex (vPMC/dPMC)) and mentalizing (ventromedial prefrontal cortex (vmPFC)) systems. To unveil their causal organization, we investigated the effective connectivity underlying the observation of human social interactions expressing increasing cooperativity (involving left pSTS, SPL, and vPMC) versus affectivity (vmPFC) via dynamic causal modeling in 36 healthy human subjects. We found strong evidence for a model including the pSTS and vPMC as input nodes for the observed interactions. The extrinsic connectivity of this model undergoes oppositely valenced modulations, with cooperativity promoting positive modulations of connectivity between pSTS and both SPL (forward) and vPMC (mainly backward), and affectivity promoting reciprocal positive modulations of connectivity between pSTS and vmPFC (mainly backward). Alongside fMRI data, such divergent effective connectivity suggests that different dimensions underlying the processing of social interactions recruit distinct, although strongly interconnected, neural pathways associated with, respectively, the bottom-up visuomotor processing of motor intentions, and the top-down attribution of affective/mental states.

Keywords: dynamic causal modeling; effective connectivity; intention understanding; mentalizing system; mirror neuron system; social cognition; social interaction.

Figure 1

Experimental stimuli. Representative examples of color pictures depicting cooperative (left) and affective (right) social interactions (see also Canessa et al., 2012; Proverbio et al., 2011). Reproduced with permission of the copyright owner

Figure 2

Brain activity highlighted by a parametric coding of social interactions. The brain regions activated by the observation of social interactions regardless of their purpose (a), and those in which activity is more strongly related to the degree of cooperativity than affectivity (b) or affectivity than cooperativity (c). All the reported activations survived a statistical threshold of p < .05 corrected for multiple comparisons based on cluster extent (Chumbley & Friston, 2009)

Figure 3

Modulation of forward and backward effective connectivity by the degree of perceived affectivity versus cooperativity. (a) The endogenous connectivity architecture of the winning reduced model after random‐effect analyses at the parameter level (see Table 3 for the values of connectivity strength). Red and blue arrows depict excitatory and inhibitory endogenous connections, respectively. (b,c) The positive (red arrows) or negative (blue arrows) modulation of endogenous connectivity by the degree of cooperativity (b) or affectivity (c) expressed by observed social interactions. The straight dashed lines depict the driving input (i.e., “Observation of social interactions”) entering the system both in pSTS and vPMC. (d) The effective connections showing a significantly different modulation by the degree of cooperativity versus affectivity in the winning model (see Table 3 for the values of modulation strength). The reciprocal effective connections between pSTS and vPMC, and the connections from vPMC and pSTS to SPL, were more strongly upregulated by cooperativity (positive modulation) than affectivity (negative modulation), while the opposite occurs in pSTS‐vmPFC reciprocal effective connectivity (see also (f)). Red and blue letters denote, respectively, positive and negative modulations of endogenous connectivity by the degree of cooperativity (“C”; “c”) or affectivity (“A”; “a”), with letter‐size representing their relative effect (e.g., Ca = larger modulation by cooperativity than affectivity). (e) The forward and backward effective connections between pSTS and the other three network nodes reported in subsequent panels. As shown in (f), perceived affectivity and cooperativity exerted oppositely valenced modulations on pSTS‐vmPFC (green in (e–g); positively modulated by affectivity), and pSTS‐SPL (red) and pSTS‐vPMC (blue) (positively modulated by cooperativity) reciprocal effective connectivity. A breakdown of this graph into forward and backward connections (g,h) additionally shows that these connections are also subject to oppositely directed modulations by cooperativity and affectivity, that is, stronger modulation by cooperativity on forward excitatory pSTS‐SPL and backward inhibitory vPMC‐pSTS connectivity, and by affectivity on backward inhibitory vmPFC‐pSTS connectivity