Monday, November 19, 2012

Toward an integrative account of social cognition: Marrying theory of mind and interactionism to study the interplay of Type 1 and Type 2 processes

The concept of Type 1 and Type 2 brain processes (System 1 is fast, intuitive, and emotional; System 2 is slower, more deliberative, and more logical) recently entered public awareness through Daniel Kahneman's outstanding book, Thinking, Fast and Slow.

In trying to better conceptualize how the brain does social cognition, this research article argues that interactionism (an approach that draws on phenomenological, enactive, and embodied approaches to
cognition) captures aspects of social cognition mostly related to Type 1 processes, while theory of mind (ToM - an approach that suggests we understand other people’s behavior by attributing mental states to them) is more focused on those based on Type 2 processes. It makes sense - real life social interactions are rarely based on either Type 1 or Type 2 processes alone.

The whole paper is available online and as a PDF - fully open access, as are all Frontiers journals.

Toward an integrative account of social cognition: Marrying theory of mind and interactionism to study the interplay of Type 1 and Type 2 processes

  • 1Department of Philosophy, Institute of Philosophy and Semiotics, University of Tartu, Tartu, Estonia
  • 2Center for Subjectivity Research, University of Copenhagen, Copenhagen, Denmark
  • 3Decision Neuroscience Lab, Department of Psychology, Stanford University, Stanford, CA, USA
Traditional theory of mind (ToM) accounts for social cognition have been at the basis of most studies in the social cognitive neurosciences. However, in recent years, the need to go beyond traditional ToM accounts for understanding real life social interactions has become all the more pressing. At the same time it remains unclear whether alternative accounts, such as interactionism, can yield a sufficient description and explanation of social interactions. We argue that instead of considering ToM and interactionism as mutually exclusive opponents, they should be integrated into a more comprehensive account of social cognition. We draw on dual process models of social cognition that contrast two different types of social cognitive processing. The first type (labeled Type 1) refers to process es that are fast, efficient, stimulus-driven, and relatively inflexible. The second type (labeled Type 2) refers to processes that are relatively slow, cognitively laborious, flexible, and may involve conscious control. We argue that while interactionism captures aspects of social cognition mostly related to Type 1 processes, ToM is more focused on those based on Type 2 processes. We suggest that real life social interactions are rarely based on either Type 1 or Type 2 processes alone. On the contrary, we propose that in most cases both types of processes are simultaneously involved and that social behavior may be sustained by the interplay between these two types of processes. Finally, we discuss how the new integrative framework can guide experimental research on social interaction.

Full Citation: 
Bohl V and van den Bos W (2012) Toward an integrative account of social cognition: marrying theory of mind and interactionism to study the interplay of Type 1 and Type 2 processes. Frontiers in Human Neuroscience, 6:274. doi: 10.3389/fnhum.2012.00274

Here is a little bit of the paper, the section in which the authors look at an integrative approach that incorporates both ToM and interactionism:


Different dual process accounts for social understanding have been proposed by several authors over the past 10 years (e.g., Lieberman et al., 2002; Kilner et al., 2007; Keysers and Gazzola, 2007; De Lange et al., 2008). However, there is no consensus on whether and how these two types of processes cooperate and inform each other. Here we will review different positions commonly found in the literature, and argue that real life social interactions are rarely based on either Type 1 or Type 2 processes alone. On the contrary, we expect that in many cases of everyday social encounters both processes are simultaneously involved and that social behavior may be sustained by the interaction between these two types of processes.

Some authors have suggested that the MNS and ToMS constitute independent networks and have complementary roles in social cognition (e.g., Saxe and Wexler, 2005; Jacob and Jeannerod, 2005; van Overwalle and Baetens, 2009). The general idea is that in most social situations we can rely on the set of low-level processes to support fluent and effortless social interactions. However, “[t]here seems [...] to be a transition from the mirror to the mentalizing system [...] when perceived body motions are contextually inconsistent, implausible, or pretended” (van Overwalle and Baetens, 2009). In line with interactionism, as presented by Shaun Gallagher (2008a), this means that the brain switches from MNS to ToMS when the situation appears puzzling.

Furthermore, van Overwalle and Baetens (2009) argue that current neuroscientific data supports the hypothesis that both systems are operating in isolation because most studies in their meta-analysis showed the involvement of only one of the two systems. Indeed, many studies have shown that in situations that call for an observational stance, for instance while reading a novel, we rely solely on the ToMS (Amodio and Frith, 2006). Similarly, there are many studies that have shown the sole involvement of the MNS in tasks involving imitating or passively watching simple movements. However, this only indicates that there are contexts in which these systems are able to operate in isolation. A possible explanation for the pattern of results reported in van Overwalle and Baetens’ meta-analyses is that the experimental paradigms, and particularly the lack of real interactions, render MNS and ToMS to operate in isolation (see also Schilbach et al., forthcoming).

A more common hypothesis in dual process models in general is that the Type 1 processes inform and support the Type 2 processes (e.g., Blakemore and Decety, 2001). This line of reasoning follows the logic of classic cognitivist models in which perceptual processes are temporally primary, and their outputs feed, uni-directionally, into cognitive processes. As such, the direct grasp of motor or proximal intentions (to switch a button) are used on a higher level to infer distal intentions (e.g., to turn on the light to read a book).

In line with this idea, some authors have suggested that the Type 1 processes associated with the MNS are a prerequisite for the ToM processes (e.g., Ohnishi et al., 2004). Given the neuroscientific evidence indicating that ToMS has often been found to operate in isolation from MNS, a strong version of this claim seems unlikely to be true. However, this does not refute weaker versions of this claim suggesting that Type 1 processes are an ontogenetic and/or phylogenetic prerequisite for Type 2 processes (Gopnik and Meltzoff, 1994; De Waal and Ferrari, 2010).

More recently, evidence has emerged in separate fields of research to support the idea that higher order (i.e., Type 2) representations regulate or bias perceptual (i.e., Type 1) processes. Several studies have already shown that higher order intentions can directly affect visual perception (Allport, 1987), for instance how we perceive tools is influenced by our intention to use them (Witt et al., 2005). More importantly, Teufel and colleagues (Teufel et al., 2009, 2010) have shown that mental state attribution (in this case the belief about whether the other person could see or not) also influences gaze-perception and automatic gaze-following. Based on these results, the authors hypothesize that perceptual modulation is the result of direct top-down modulation by the ToMS (Teufel et al., 2010). Additionally, others have suggested that the ToMS may also directly modulate the MNS (Liepelt and Brass, 2010; Ondobaka et al., 2012). For instance, Ondobaka et al. have shown that the motor congruence effect, or the fact that performing a movement is facilitated when another person is performing the same movement, is modulated by higher order mental state attributions. That is, they found that the movement congruency effect was present only when participants acted with the same higher order intention as the co-actor. So in a similar fashion as our own proximal intentions (to pick up the cup) are often superseded and influenced by distal intentions (to get something to drink), our understanding of others also involves the top-down influence of higher order intentions. These findings are also in line with Jacob’s alternative interpretation of the function of MNS which suggests that the MNS by itself is responsible not for the ascriptions of even the most simple intentions, but rather for computing “the motor commands from a representation of the agents prior intention” (Jacob, 2008); the intentions are thought to be computed by other means. On the other hand, Jacob also states that “the application of the concept of grasping triggered by the perception of an act of grasping would inferentially give rise to the related concept of, e.g., drinking” (Jacob, 2009), thus subscribing to the idea that the reciprocal interaction between these different systems may be essential in social understanding. In sum, these experiments also suggest that the engagement and top down influence of the putative ToMS is not only engaged when there is some incongruence in the social environment, but may also operate on whatever knowledge is available in order to make social interactions run more efficiently.

Taken together, the social neuroscience literature suggests that intentions are processed (and understood) at different levels of abstraction by different types of processes and that there are reciprocal interactions between these types of processing. Inspired by the predictive coding theory of the MNS (Kilner et al., 2007) we hypothesize that the MNS and the ToMS often operate in parallel; both function to enable predictions of behavior, which in turn facilitates social interactions. Furthermore, as is also suggested by dual process models, it is the interaction between different types of social cognitive processes that allows social behavior to be fast and effective, and at the same time flexible and context-dependent (Kilner et al., 2007; Evans, 2010).

Thus, although we fully agree that the “dynamic interaction between [distal] and [proximal] intentions modulates the processing of the observed actions of other people” (Ondobaka et al., 2012, p. 34), we also suggest that the processing of observing actions of others contributes to social interactions. Mindreading as implemented by the ToMs is therefore also in the service of interpersonal interaction (Jacob and Jeannerod, 2005). In this sense the view marries the ToM assumption that mindreading is important for social cognition with the interactionist assumption that social cognition is first of all social interaction.

From this integrative account, two major conclusions for the interdisciplinary study of social cognition follow. Firstly, it follows that although it is possible to study processes emphasized by ToM and interactionism in isolation from each other, if we want to know how social cognition functions in real-life, we should rather study how these different processes influence one another. Secondly, in order to study the interaction between Type 2 (ToM) and Type 1 (interactionism) processes, we need to study actual social interactions and how these unfold over time. Figure 1 illustrates how ToM and interactionism can account for different aspects of social cognition on different levels of description and explanation. The arrows represent connections that we have outlined in the current paper: (1) the possible reciprocal influence between Type 1 and Type 2 processes on the sub-personal level; (2) the influence of supra-individual factors on Type 1 processes; (3) the correlations between Type 1 processes and personal level descriptions; (4) the correlations between Type 2 processes and personal level descriptions19 . In the next section we consider several methodological implications for social cognitive neurosciences.

No comments: