The “mindreading is prior” model proposed by Carruthers postulates that one mechanism (mindreading) is all that is needed to understand the mental states of oneself and others. Although we agree that shared mechanisms can implement the computations underlying both self- and other-mentalizing, we question whether all types of mentalizing use this one mechanism in an identical fashion. We present evidence from functional neuroimaging and research suggesting that mentalizing differs when the target individuals are of differing similarity to oneself.
Because the “mindreading is prior” model asserts that one mechanism is used indifferently for self- and other-mentalizing, this generates three predictions that all need to be satisfied in the brain. First, there should be an identical neural implementation of mentalizing for self and other. Thus, the same brain regions that code for self-mentalizing computations should also be recruited for other-mentalizing. Second, such identical neural implementation should occur to a similar degree for both self and other. Thus, activation in shared regions should be equal in self- and other-mentalizing. Finally, there should be no other areas in the brain that are recruited specifically for just self- or just other-mentalizing.
In our own recent work (Lombardo et al., submitted) we find evidence that while predictions 1 and 3 are satisfied, prediction 2 is not. We find that although the ventromedial prefrontal cortex (vMPFC), posterior cingulate/precuneus (PCC), and right temporo-parietal junction (RTPJ) are all recruited for mentalizing about both self and other, they do so in a target-biased fashion. The vMPFC is recruited more for self-mentalizing than other-mentalizing, whereas PCC and RTPJ are recruited more for other- than self-mentalizing. Thus, while it is the case the identical neural mechanisms are implementing self- and other-mentalizing computations, they do not do so in an equal fashion. Some of these shared neural representations are tailored more for handling self-mentalizing, whereas other shared representations are tuned in to handle other-mentalizing. The “mindreading is prior” model is silent about how one unitary mindreading mechanism can be tuned preferentially for mentalizing about self more than other, or vice versa.
In addition to these neuroimaging results, there is behavioral evidence in individuals with autism spectrum conditions (ASC) that the magnitude of impairment in self- and other-mentalizing is unequal (Lombardo et al. Reference Lombardo, Barnes, Wheelwright and Baron-Cohen2007). In the domain of emotion understanding, on measures of other-mentalizing involving reading complex emotions (Reading the Mind in the Eyes Test; Baron-Cohen et al. Reference Baron-Cohen, Wheelwright, Hill, Raste and Plumb2001), adults with ASC show less impairment (Cohen's d=0.61) than on measures assessing understanding one's own emotions (Toronto Alexithymia Scale; Cohen's d=1.40). Thus, although impairment may exist in both self- and other-mentalizing in ASC, the impairments are unequal in magnitude. This evidence presents a paradox: How can one unitary mechanism working identically for self and other affect self and other in a differential manner?
The “mindreading is prior” model is also silent about what happens when an individual mentalizes about an individual of varying similarity to oneself. Carruthers treats the “other” as completely distinct from oneself, and he bases his theory on such a “monolithic other.” However, social psychological research suggests that the mechanisms for mentalizing about others functions differently depending the degree to which an individual perceives the other person to be similar or dissimilar to oneself. When another is perceived to be dissimilar to oneself, mentalizing responses follow a pattern similar to stereotyping; we expect dissimilar individuals to share mental states of individuals from the salient group that they belong to. However, when another individual is similar to oneself, social inference proceeds in a congruent manner to what we ourselves think or feel (Ames Reference Ames2004).
A similar distinction can be seen within neuroimaging research on this topic. Jason Mitchell and colleagues (Jenkins et al. Reference Jenkins, Macrae and Mitchell2008; Mitchell et al. Reference Mitchell, Macrae and Banaji2006) present elegant work showing that similar neural responses occur in vMPFC during self-mentalizing and mentalizing about similar others, but not during mentalizing about dissimilar others. Mentalizing about dissimilar others involves processing in the dorsomedial prefrontal cortex (dMPFC) rather than the vMPFC. Furthermore, when stereotyping is applied in the context of mentalizing about others, areas involved in semantic retrieval and selection such as the ventrolateral prefrontal cortex (VLPFC) are recruited, but not the vMPFC or dMPFC (Mitchell et al., in press).
Thus, behavioral and neural distinctions can be made about the mechanism underlying mentalizing about others of differing similarity to oneself. In the case of similar others, the regions involved in both self- and other-mentalizing may overlap considerably. However, different regions appear to be at work when making inferences about the mental states of dissimilar others. How does the “mindreading is prior” model account for such differences in the mindreading mechanism, solely based on the perceived similarity of others to oneself?
Finally, in our own work, we find that even in ASC the mechanism underlying mentalizing about self and similar others may be different to mentalizing about dissimilar others (Lombardo et al. Reference Lombardo, Barnes, Wheelwright and Baron-Cohen2007). We asked participants to make judgments about personality traits (e.g., honest, caring, anxious) in relation to themselves, a similar close other, and a dissimilar non-close other. In a control task, they were asked to simply count the syllables in personality trait words. In a surprise recognition memory test, we found that individuals with ASC had no impairment in memory for words encoded by syllable counting or in relation to a dissimilar other. However, when looking at memory for self-encoded or similar other-encoded traits, we found substantial impairment. Individuals with ASC had an impairment in processing linked to thinking about themselves and similar others, but no deficit in regards to a dissimilar other. Such dissociations for similar and dissimilar others imply the mindreading mechanism may conceal a greater complexity than is suggested by Carruthers' model.
