Third-person self-talk facilitates emotion regulation without engaging cognitive control: Converging evidence from ERP and fMRI

Does silently talking to yourself in the third-person constitute a relatively effortless form of self control? We hypothesized that it does under the premise that third-person self-talk leads people to think about the self similar to how they think about others, which provides them with the psychological distance needed to facilitate self control. We tested this prediction by asking participants to reflect on feelings elicited by viewing aversive images (Study 1) and recalling negative autobiographical memories (Study 2) using either “I” or their name while measuring neural activity via ERPs (Study 1) and fMRI (Study 2). Study 1 demonstrated that third-person self-talk reduced an ERP marker of self-referential emotional reactivity (i.e., late positive potential) within the first second of viewing aversive images without enhancing an ERP marker of cognitive control (i.e., stimulus preceding negativity). Conceptually replicating these results, Study 2 demonstrated that third-person self-talk was linked with reduced levels of activation in an a priori defined fMRI marker of self-referential processing (i.e., medial prefrontal cortex) when participants reflected on negative memories without eliciting increased levels of activity in a priori defined fMRI markers of cognitive control. Together, these results suggest that third-person self-talk may constitute a relatively effortless form of self-control.


Supplemental Results for Study 1
displays the morphology and topography of the LPP with respect to its modulation by Time and Valence.
The Time X Anterior/Posterior X Superior/Inferior interaction (F (1,28) = 10.77, p < .01, η 2 p = .28) showed that the LPP was larger at superior sites in the anterior region at both time windows, but was relatively larger at superior sites later in the time window at posterior sites.
The Time X Laterality effect (F (1,28) = 9.82, p < .01, η 2 p = .26) indicated that the LPP was larger at right locations earlier but larger at left locations later in time. The Time X Superior/Inferior X Laterality interaction (F (1,28) = 10.84, p < .01, η 2 p = .28) showed that the LPP was larger at right locations early at inferior sites and no difference between superior sites, but later the LPP was larger at left locations at superior sites more so than at inferior sites. The Laterality X Self-Talk Strategy (F (1,28) = 5.68, p < .05, η 2 p = .17) effect indicated that the LPP was larger at left than right sites for first-person blocks but that the LPP was larger at right than left locations in third-person blocks; neither simple effect was significant, however (Fs < 1). Together, the results indicated the typical superior-posterior distribution of the LPP, especially with respect to its modulation by valence. As indicated in the main text, however, there was no significant effect of Self-Talk Strategy (F(1,28) < 1, p = .36, η 2 p = .03) in the early LPP time window.

LPP 1-6s
The main effect of Superior/Inferior (F (1,28) = 12.14, p < .01, η 2 p = .30) was significant such that the LPP was larger at superior sites. The Time X Anterior/Posterior interaction (F (1,28) = 9.10, p < .001, η 2 p = .25) showed that the LPP was larger at posterior sites early and anterior sites later. The Time X Anterior/Posterior X Valence interaction (F (1,28) = 4.60, p < .01, η 2 p = .14) revealed that the valence effect (negative -neutral) begins more posterior and then becomes distributed and more anterior over time, consistent with past results 6 . The  Figure S2 for a visual depiction of the broad scalp distribution of this interaction).

SPN
As indicated in the main text, and consistent with our prediction, results revealed no main effect of Self-Talk Strategy. Figure S3 depicts the scalp distribution of this null effect.

Brain-Behavior Relationships
Finally, we evaluated the relationship between LPP modulation by Type of Self Talk Instruction and participants' ratings of compliance with the self-talk instructions.
Our analysis showed that the degree to which participants showed a larger LPP emotion effect (negative-neutral) in the First-Person block than in the Third-Person block was associated with reports of greater use of the appropriate pronouns across both blocks (r = .44, p < .05). That is, the expected blunting of the emotion effect in the Third-Person (compared to the First-Person) block scaled with the degree to which participants effectively employed the different personal pronouns in each block.

Supplemental Results for Study 2 Brain Behavior Correlations
We examined the relationship between negative affect and reduced activation in brain areas associated with self-referential processing by (a) extracting average activation for each participant from the functional region of interest from meta-analytic findings (Araujo, Kaplan, & Damasio, 2013), and then (b) correlating this with the difference between each subject's selfreported ratings of emotion in the I trials vs Name trials.
As noted in the text (see Footnote X), initial analyses of the distribution of these difference scores revealed two extreme values corresponding to the I > Name extracted beta values (i.e., scores that were greater than two standard deviations than the sample mean).
Subsequent analyses indicated that these two values exerted a relatively large amount of influence on the linear regression, Cook's Distance = 0.28, 0.26. Thus, we report the analyses below both with and without these values included, consistent with the reports we result in the main text.
Analyses involved all participants indicated that the correlation between the activation extracted from the left medial prefrontal cortex/anterior cingulate cortex correlated negatively with the drop in self-report negative affect that participants reported experiencing on Name trials compared to I trials (i.e., I distress scores -Name distress scores ) for both the functional ROI, R 2 = .08, B = -1.25, SE = .63, t(48) = -2.00, p =.05, and the AAL ROI from whole-brain analysis, R 2 = .08, B = -1.26, SE = .60, t(48) = -2.10, p < .05. However, subsequent analyses that did not include the two extreme values above revealed no significant brain-behavior correlation for either the functional    Voltage map depicting the topographical distribution of the null main effect of Self-Talk Strategy, averaged across Valence, on the SPN across early and late time windows (300-3000ms total). Darker shades of blue would indicate a larger SPN for Third-Person compared to First-Person Self-Talk, but only small, non-significant differences were observed (shown in light blue).