June 15, 2009
It constitutes mental time travel, with which we can recollect, in vivid detail, events that took place many years ago.
We have known, for the best part of a century, that memory is reconstructive rather than reproductive. That is, recollection involves piecing together specific details of the event, and mixing these with our own biases and beliefs.
While not being completely accurate, our
memories are, in most cases, reliable enough.
But a new study, due to be published in
the September issue of the journal Neuropsychologia, now
confirms that these simulations are indeed novel constructions, and
also shows that remembering actual experiences and imagining
possible future events depend on distinct subsystems within the
common core network.
Thus, it is suggested that fragments of past experiences provide a source of details which can be flexibly recombined to simulate future events.
There is, however, no direct evidence
for this; it is possible that imaging future events involves merely
recasting past experiences, such that simulating the future involves
retrieving a single memory of the past and projecting it forward in
time, and the studies carried out to date have not yet distinguished
between these possibilities.
Interestingly, activity in this network is greater during simulation of future events than during remembering.
This may reflect processes that do not
occur during remembering, such as the recombination of memory
fragments or, if imagining the future involves merely recasting the
past, the addition of a new "timestamp" to the memory of a past
23 participants - all of them college
students - were first asked to complete a spreadsheet detailing
memories of nearly 200 personal events they had experienced within
the past 5 years, each of which had occurred at a specific time and
in a specific place. A few days later, they underwent a brain
scanning session, during which they were asked to recall some of the
events that had actually happened and also to imagine past and
Each cue consisted of details of a person, object and place taken from multiple episodes provided in the first session, and the participants were instructed to imagine these details in a single novel episode.
In this way, the participants were
prevented from recasting past experiences as future events; they
also confirmed, in descriptions provided after the scanning
sessions, that they were able to combine the separate memory
elements into coherent representations of imagined future events.
But closer examination revealed that each was associated with a distinct subsystem within the network.
For example, extensive regions of the medial prefrontal cortex, parietal lobe and the anterior portion of the hippocampus were activated during imaging future events, but not during retrieval of memories. On the other hand, remembering, but not imagining, led to activation of parts of the visual cortex, likely reflecting the imagery associated with memory retrieval.
Furthermore, each subsystem was found to
be activated over a different timescale, with the imagining
subsystem becoming active earlier (at between 2-4 seconds) and
peaking for longer, than the remembering subsystem.