Remembering what could have happened: neural correlates of episodic counterfactual thinking

Abstract

Recent evidence suggests that our capacities to remember the past and to imagine what might happen in the future largely depend on the same core brain network that includes the middle temporal lobe, the posterior cingulate/retrosplenial cortex, the inferior parietal lobe, the medial prefrontal cortex, and the lateral temporal cortex. However, the extent to which regions of this core brain network are also responsible for our capacity to think about what could have happened in our past, yet did not occur (i.e., episodic counterfactual thinking), is still unknown. The present study examined this issue. Using a variation of the experimental recombination paradigm (Addis, Pan, Vu, Laiser, & Schacter, 2009. Neuropsychologia. 47: 2222-2238), participants were asked both to remember personal past events and to envision alternative outcomes to such events while undergoing functional magnetic resonance imaging. Three sets of analyses were performed on the imaging data in order to investigate two related issues. First, a mean-centered spatiotemporal partial least square (PLS) analysis identified a pattern of brain activity across regions of the core network that was common to episodic memory and episodic counterfactual thinking. Second, a non-rotated PLS analysis identified two different patterns of brain activity for likely and unlikely episodic counterfactual thoughts, with the former showing significant overlap with the set of regions engaged during episodic recollection. Finally, a parametric modulation was conducted to explore the differential engagement of brain regions during counterfactual thinking, revealing that areas such as the parahippocampal gyrus and the right hippocampus were modulated by the subjective likelihood of counterfactual simulations. These results suggest that episodic counterfactual thinking engages regions that form the core brain network, and also that the subjective likelihood of our counterfactual thoughts modulates the engagement of different areas within this set of regions.

Keywords: Counterfactual thinking; Episodic memory; FMRI; Imagination; Mental time travel; Partial least squares.

Figure 1

Experimental design. A screen with three episodic components (i.e., context, action/event, and outcome) and the specific condition as title (e.g., Positive) was presented to the participants for 10s. Participants were asked to press a button as soon as the construction period had finished (aproximately 4 to 5 seconds; see results), and were instructed to elaborate on their simulation for the remaing time. After a fixation cross, screens with the ratings of emotion and vividness, and then likelihood and regret/relief, were presented for 6 seconds each. Participants used a five-button MRI compatible response box to record their ratings. A final fixation cross indicated the end of the trial and beginning of next trial.

Figure 2

Latent variable 1 (LV 1). (A) Plot of brain scores with confidence intervals. Of note, confidence intervals are asymmetrical as a result of rescaling each condition’s distribution during bootstraping. (B) Plot of temporal brain scores indicating weighted average of activation across all voxels in all subjects during the length of the task (TRs = 2 seconds) across all voxels in all subjects. (C) Brain regions associated with the experimental conditions at selected TRs. All regions are shown at a threshold of p <.001. (D) Plots indicating percent signal change of peak voxels in right parahippocampal gyrus (BA 28; xyz = 24 −22 −8), left posterior cingulate cortex (BA 23; xyz = −4 −32 22), and left superior parietal lobule (BA 7; xyz = −30 −70 52).

Figure 3

Latent variable 2 (LV 2). (A) Plot of brain scores with confidence intervals. Of note, confidence intervals are asymmetrical as a result of rescaling each condition’s distribution during bootstraping. (B) Plot of temporal brain scores indicating weighted average of activation across all voxels in all subjects during the length of the task (TRs = 2 seconds) across all voxels in all subjects. (C) Brain regions associated with the experimental conditions at selected TRs. Brain regions associated with remember and likely counterfactuals are shown in yellow, while those associated with unlikely counterfactuals and the control condition are shown in cyan All regions are shown at a threshold of p <.001. (D) Plots indicating percent signal change of peak voxels in left parahippocampal gyrus (BA 19; xyz = −18 −48 −10), right parahippocampal gyrus (BA 19; xyz = 26 −52 −4), and left superior frontal gyrus (BA 10; xyz = −10 60 26).