Schacter, D. L., Addis, D. R., Hassabis, D., Martin, V. C., Spreng, R. N., & Szpunar, K. K. (2012). The Future of Memory: Remembering, Imagining, and the Brain. Neuron, 76(4), 677–694. doi:10.1016/j.neuron.2012.11.001

Summary

In this paper, Schacter et al. review neuroscientific evidence for the relationship between recalling episoding memories and constructing imagined events. Broadly, the conclusion is that these two processes are very similar and rely on a lot of the same mechanisms, but also have measureable differences. They make four main points:

  1. Remembered events are always related to the past, but imagined events could be set in the past, present, or future. Thus, remembering vs. imagining shouldn’t be interpreted as past vs. future.
  2. There are differences between the processes involved in remembering and imagining. For example, imagining requires synthesizing an entirely new event and possibly concepts, while remembering—while perhaps a constructive process—need only retrieve existing pieces of information. Remembering the past has also been associated with a larger number and more vivid sensory details
  3. In terms of specific component processes, it seems that regions like the MPFC (medial prefrontal cortex) and posterior cingulate are involved in simulations of the self and social scenarios; regions like the medial temporal lobe and retrosplenial cortex are more involved in memory-based scene construction. The hippocampus is also involved in imagining and remembering, though its specific role isn’t as clear; there is some evidence that it’s involved specifically with retrieving relevant details, combining them into something coherent, and then storing the result.
  4. The default network is involved in both remembering and imagining and includes regions like the medial temporal lobe, frontal lobe, posterior cingulate cortex, retrosplenial cortex, and lateral parietal and temporal areas. This network can be coupled with other networks to support goal-directed simulations (particularly those that are autogbiographical).

In the conclusion, Schacter et al. suggest that “a key function of memory is to provide a basis for predicting the future via imagined scenarios and that the ability to flexibly recombine elements of past experience into simulations of novel future events is therefore an adaptive process”.

Takeaways

I think the most important takeaway for me is that mental simulation of (at least episodic) events is closely tied to the ability to retrieve existing memories and somehow recombine them. The interesting question for me, though, is how this recombining is actually accomplished, and the way it is different between reconstructing or reinterpreting memories and creating new simulations. One hypothesis might be that when either remembering or imagining, people first construct or retrieve some high-level narrative structure of the event. Then, they fill in the details, either by retrieving the specific details that were stored (in the case of remembering), or by sampling specific details (in the case of simulation). This might be one way to explain why simulations tend to have fewer details, too: for remembering, you perhaps attempt to retrieve all the stuff the was stored, while for simulating, you only sample details until the simulation is “coherent enough”. For some definition of “coherent” and “enough”, this might not need to include things like small perceptual or sensory details.

On a different note, this paper illustrates a different way in which “simulation” is used—here, they are using simulation to refer to imagined episodic events. This is in contrast to the simulation of mental imagery, in which simulation tends to mean something more along the lines of imagined objects or actions. Of course, to simulating an episodic event probably requires being able to simulate objects and actions, but it additionally requires being able to piece together multiple things into a coherent, high-level picture. Schacter et al. actually touch on this a bit in the paper (pg. 681):

Scene construction entails retrieving and intergrating perceptual, semantic, and contextual information into a coherent spatial context. Scene construction is held to be more complex than “simple” visual imagery for individual objects because it relies on binding together disparate types of information into a coherent whole, and likely involves processes mediated by several regions within the default network, most notably the medial temporal lobe. Scene construction is thought to be a critical component of both memory and imagination as mental simulations, whether of the past, future, or purely fictional, because they are all usually framed within a spatial context.