For learning and memory to work effectively, we face a series of challenges: first we need to encode information into memory in a way that distinguishes it from all the other memories we have acquired previously; we then have to store it somewhere in the brain durably for perhaps many years; finally, for our memories to be of any use to us, we have to be able to precisely retrieve the correct stored information when we need it. Evidence suggests that encoding and retrieval processes can be supported by cognitive control processes such as elaboration, cue specification, and monitoring, which help them to be accomplished efficiently, whereas storage operations may occur largely automatically.
The important role of the hippocampus and its surrounding structures for memory storage has been established for many years due to studies of amnesic patients with hippocampal damage such as HM. With the advent of neuroimaging, roles in memory have also been discovered for different regions of prefrontal cortex. Distinct contributions have been proposed for three main prefrontal regions in the cognitive control of encoding and retrieving information: ventrolateral prefrontal cortex (VLPFC), dorsolateral prefrontal cortex (DLPFC) and anterior prefrontal cortex (APFC).
Many cognitive studies have demonstrated that the level of processing a stimulus receives during encoding (thinking about its meaning and relating it to associated information and previous knowledge) has a considerable effect on its memorability. An early functional neuroimaging experiment sought to identify the brain regions involved in such elaborative encoding that helps make our memories distinct from one another, increasing the likelihood of subsequent retention. Participants were asked to learn lists of words using either a visual-perceptual strategy (“does the word appear in upper- or lowercase?”) or a meaning-based elaboration strategy (“does the word relate to an abstract or concrete concept?”). Greater brain activity for the deeper, meaning-based encoding task was observed in left VLPFC, suggesting that this region may be important for successful encoding. This link was confirmed in another study by sorting brain activity at encoding on a trial-by-trial basis depending on whether stimuli were subsequently remembered or forgotten. The activity level in left VLPFC, among other areas, during encoding accurately predicted the likelihood of later successful memory.
Cognitive models of memory retrieval suggest the involvement of a number of distinct processing stages:
The first process has been termed cue specification, and is a ‘pre-retrieval’ process that involves thinking about what you’re going to try and remember so that you can guide your retrieval search and establish criteria to determine whether the correct information is recovered. Impaired cue specification has been proposed as one possible account for confabulation, in which some patients with frontal lobe lesions produce false memories and narrative accounts of imaginary personal experiences that they believe to be true. This form of false memory may occur because the retrieval cue is poorly specified, leading incorrect and inappropriate memories to be generated. These can sometimes be mundane, such as mistakenly believing one had steak for dinner last night, but also more bizarre, such as remembering flying in a spaceship.
One study that used functional neuroimaging to study cue specification isolated pre-retrieval processes by starting each trial of a memory test with an instruction as to the kind of detail that would need to be retrieved on that trial. On some trials, the instruction was then followed by an item to be remembered, whereas on other trials, the instruction was followed instead by a non-memory task. Activity in left VLPFC and APFC was seen when retrieval instructions were presented, regardless of whether a memory search was actually undertaken, suggesting a role in pre-retrieval cue specification processes.
Once information has been retrieved from memory, it must be maintained online in working memory while it is evaluated against the criteria established during the cue specification stage (e.g., “does the event I’ve retrieved date from around the time of the memory I was searching for?”). Functional neuroimaging studies suggest that the VLPFC is involved in the short-term maintenance of information retrieved from long-term memory. This maintained information is subject to monitoring and evaluation against criteria for success, so that a decision can be made about whether the sought-after information has been correctly retrieved, or whether the retrieval cue should be modified and another memory search undertaken. Several neuroimaging studies have associated right DLPFC with post-retrieval monitoring processes. For example, greater DLPFC activity has been observed during free recall of words, which requires an internally generated structure that makes critical demands on monitoring operations, than when recall is guided by some kind of category cue provided by the experimenter. Similarly, DLPFC activity is seen to a greater extent during recognition responses that are made with low confidence, suggesting the involvement of greater monitoring.
Another decision we might make about information retrieved from memory involves determining its source, such as distinguishing real experiences from those we might have imagined or been told about. Such reality monitoring processes appear to recruit APFC among other areas. Disruption to reality monitoring processes may underlie the delusions and hallucinations that are often seen in psychiatric disorders such as schizophrenia. For example, you might imagine a voice conveying a message, but misattribute the voice as real, coming from another person. Consistent with this view, when healthy volunteers perform a memory task that involves discriminating between words that had previously been perceived vs. imagined, activity is observed in the same brain regions as are dysfunctional in schizophrenia, supporting the suggestion that reduced ability to discriminate internally from externally generated information may account for some forms of hallucinations.