Memory depends on how well you can forget...

Benedict Carey reviews (PDF here) new work on memory (PDF here) by a Stanford group that is the first to record visual images of people’s brains as they suppress distracting memories:

Blocking out a distracting memory is something like ignoring an old (and perhaps distracting) acquaintance... it that much harder to reconnect the next time around...recent studies suggest that the brain plays favorites with memories in exactly this way, snubbing some to better capture others. A lightning memory... is not so much a matter of capacity as it is of ruthless pruning

The experiments:

...had 20 young men and women, mostly Stanford students, view in quick succession a list of 240 word pairs. These included 40 capitalized words, each paired with six related, lower-case words: For example, “ATTIC-dust,” “ATTIC-junk,” and so on....After studying the pairs, the participants were instructed to memorize three selected pairs from each of 20 capitalized words. In effect, this forced them to flag individual pairs, like ATTIC-dust, while trying to tune out very similar, distracting ones, like ATTIC-junk, for half of the total list of pairs they saw...They were told not to memorize any pairs from the other half of the list.....while participants were having their brains scanned by an M.R.I. machine, each person’s memory was tested several times, and scores ranged from about 30 percent accuracy to 80 percent. ...how well each person suppressed the distracting word pairs was also measured by comparing recall of those pairs with recall of the half of the list that was studied at first but later ignored.

from Kulh et al.'s abstract:

Functional magnetic resonance imaging during selective retrieval showed that repeated retrieval of target memories was accompanied by dynamic reductions in the engagement of functionally coupled cognitive control mechanisms that detect (anterior cingulate cortex) and resolve (dorsolateral and ventrolateral prefrontal cortex) mnemonic competition. Strikingly, regression analyses revealed that this prefrontal disengagement tracked the extent to which competing memories were forgotten; greater forgetting of competing memories was associated with a greater decline in demands on prefrontal cortex during target remembering. These findings indicate that, although forgetting can be frustrating, memory might be adaptive because forgetting confers neural processing benefits.

Figure Legend: A whole-brain regression analysis showed that memory suppression at test was predicted by repetition-related activation decreases in (a) left dorsal ACC (anterior cingulate) (b) right anterior VLPFC (ventrolateral prefrontal cortex) In the ACC and right VLPFC, high suppressors showed reliable decreases in activation from first to third retrieval practice trials, whereas low suppressors did not. High suppressors showed greater initial ACC activation than low suppressors, but comparable initial right VLPFC activation. (c) A whole-brain regression analysis showed that the repetition-related activation decline in the right DLPFC covaried with that in the ACC, showing a functional coupling between ACC and right DLPFC that specifically relates to changes in demands on cognitive control across retrieval repetitions.