Abstract
It has been hypothesized that our capacity to simulate hypothetical episodes is based on an episodic memory system that provides access to stored episodic details and the constructive processes to recombine these details into novel episodes. Key evidence for this constructive episodic simulation hypothesis has been provided by a number of neuroimaging experiments, which indicate that both episodic simulation and episodic memory are supported by the same core network of brain regions. Here, we employed Activation Likelihood Estimation (ALE) to quantitatively summarize the results of those experiments. Specifically, we only included experiments that formally tested for spatial overlap of episodic memory and episodic simulation (e.g., by employing a conjunction approach), thus providing strict evidence for the core-network. At the same time, we included studies that examined the simulation of a variety of different hypothetical episodes (i.e., possible personal future, fictitious, and counterfactual past episodes), thus increasing the generalizability of our results. Our meta-analysis revealed the expected convergence of activation within the medial temporal lobes, large parts of the medial surface as well as parts of the lateral temporal and inferior posterior parietal cortices. In addition, episodic simulation and memory both consistently recruited a few other regions including parts of the bilateral dorsolateral prefrontal cortex. All of the identified regions overlapped with the default network. Episodic simulation - compared with episodic memory - presumably further requires stronger engagement of some parts of the core network (e.g., for the retrieval of disparate episodic details) and also recruitment of additional brain regions (e.g., for the novel recombination). A second ALE analysis indeed identified regions that were consistently more strongly engaged during episodic simulation than episodic memory. These included parts of the core network in the left dorsolateral prefrontal cortex and inferior posterior parietal lobe in addition to auxiliary regions that were broadly distributed across the default and fronto-parietal control networks. Together, the analyses quantitatively specify the set of regions engaged during episodic simulation, thus providing a foundation for further inquiries into their interactions and specialized contributions.
