The Journal of Neuroscience, May 15, 2003 • 23(10):3999–4004 • 3999 Functional–Anatomic Correlates of Control Processes in Memory Randy L. Buckner Department of Psychology, Howard Hughes Medical Institute at Washington University, St. Louis, Missouri 63130 Introduction when memory retrieval requires flexible use of control processes A central role for control processes in long-term memory is ap- (Milner et al., 1985; Schacter, 1987; Shimmamura et al., 1991). In parent when one considers that there is far more information many cases, patients will fail to recall, or sometimes even recog- available in memory than can be accessed at any single moment nize, information (for review, see Wheeler et al., 1995) or inap- (Tulving and Pearlstone, 1966; Koriat, 2000). Much as selective propriately estimate the source or timing, or both, of a past epi- attention operates to focus processing on objects in the environ- sode (for review, see Schacter, 1987). In rare cases, a patient may ment, related control processes are required to constrain and make up episodes from the past while believing they really hap- select from information stored in memory. pened, a phenomenon called “confabulation” (Moscovitch, Neuroimaging studies have expanded our understanding of 1989; Burgess and Shallice, 1996). Unlike purer forms of amnesia, control processes in long-term memory in three important ways. which result from medial temporal and diencephalic lesions First, networks of regions, prominently including specific regions (Scoville and Milner, 1957; Squire, 1992; Cohen and Eichen- in prefrontal cortex, contribute to control processes during baum, 1993), the deficits after frontal damage align more to im- memory retrieval. Dissociations among regions suggest a func- proper execution of retrieval strategies (Incisa della Rocchetta tional–anatomic gradient with posterior frontal regions contrib- and Milner, 1993). Convergent with these patient observations, uting to domain-specific control processes (e.g., verbal versus acts of retrieval in healthy individuals, studied with neuroimag- nonverbal) and anterior regions contributing to abstracted con- ing methods, almost always show activation of frontal cortex trol processes that generalize across informational domains. Sec- (among other regions). Frontal activation has been associated ond, through convergent exploration of other cognitive task with retrieval of words, sentences, faces, pictures, and in quite forms such as working memory, findings suggest that frontally different retrieval task contexts (for review, see Tulving et al., mediated control processes are important to a wide range of task 1994; Buckner, 1996; Fletcher et al., 1997; Cabeza and Nyberg, contexts that include, but extend beyond, long-term memory 2000). retrieval. Finally, study of these networks in aging has identified Expanding beyond these general observations, functional im- two candidate mechanisms by which their disruption may con- aging results have placed constraints on both the anatomic spec- tribute to, or compensate for, cognitive decline. Evidence for ificity of regions within frontal cortex that associate with memory these three sets of findings are presented as examples of how and also their dissociated functional roles. Anatomically, regions neuroimaging methods have contributed to explorations of that are activated include those along the left inferior frontal higher-level cognition. gyrus (particularly for verbal materials) extending anteriorally into middle frontal gyrus [near Brodmann’s areas (BA) 44, 45, Specific frontal regions contribute to control processes and 47]. Regions near the frontal poles (near BA 10) have also Control processes are central to memory retrieval. Defined been observed consistently in imaging studies (Fig. 1). broadly, control processes are intentional processes initiated by a The functional roles of these several regions can be dissociated subject to solve a task. Control processes have certain properties; on multiple dimensions that obey a roughly posterior to anterior they require sequential steps of processing, are capacity limited, gradient [see also Petrides et al. (1995)]. Posterior frontal regions and tend to operate in situations in which the task goal cannot be tend to show lateralization depending on whether verbal or non- met through automated stimulus-response mappings (Schneider verbal materials are being retrieved, suggesting domain specific- and Shiffrin, 1977). The need for control processes in memory ity (Wagner et al., 1998; McDermott et al., 1999). By contrast, arises because the vast amount of information stored in memory more anterior regions tend to exhibit complex properties depen- far exceeds the specific information that is required at any given dent on higher-order task processes. For example, activation pat- moment. Control processes are therefore required to search, se- terns in anterior regions near BA 44/45 and 47 suggest a role in lect, and monitor information during acts of retrieval. dynamic access and selection during retrieval. Several studies, Damage to frontal cortex in humans can cause difficulties conducted in various contexts that involve retrieving and catego- rizing words, illustrate this point. This work was supported by the National Institutes of Health (MH57506), the James S. McDonnell Foundation Petersen and colleagues (1989) noted robust activation in left (99-63/9900003),theAlzheimer’sAssociation,andtheHowardHughesMedicalInstitute.IthankStevePetersenfor anterior prefrontal cortex during retrieval of verbs cued by nouns valuablediscussionandCindyLustig,MarkWheeler,AnthonyFotenos,LuigiMaccotta,DeniseHead,BrianGold,Jim (e.g., retrieving “walk” in response to “dog”) but considerably Kelly, and Marc Raichle for comments. less activation when subjects simply read the nouns. Controlled Correspondence should be addressed to Dr. Randy L. Buckner, Department of Psychology, Howard Hughes Med- processes are required during verb retrieval in this context be- ical Institute at Washington University, One Brookings Drive, Campus Box 1125, St. Louis, MO 63130. E-mail: [email protected]. cause the target word is not the automatic associate of the pre- Copyright © 2003 Society for Neuroscience 0270-6474/03/233999-06$15.00/0 sented noun; it must be sought and selected from many possible 4000 • J. Neurosci., May 15, 2003 • 23(10):3999–4004 Buckner • Control Processes in Memory gions, through poorly understood mechanisms, appear to con- tribute to the dynamic selection of representations during long- term memory retrieval, possibly through interactions with posterior regions in temporal and parietal cortex that serve as storage sites. Because remembering requires more detailed, or weakly associated, information to be retrieved, frontal participa- tion in control processes increases. Convergent with this idea, when these (or nearby) frontal regions are damaged, patients can have difficulties selecting appropriate representations. Patients with left frontal lesions, for example, are impaired on free recall in a manner that is consistent with impaired use of retrieval strate- gies (Incisa della Rocchetta and Milner, 1993). More broadly, patients with frontal lesions can show source memory impair- Figure 1. Frontal regions activated during memory retrieval show a roughly posterior to ment (Schacter, 1987) and also retrieval impairment on a wide anterior gradient of functional specialization. Posterior regions (red) tend to modulate on the range of other task forms, with the greatest deficits found during basis of retrieved content, and more anterior regions (blue) modulate based on the level and free recall in which minimal cues are available, and the most typeofcontrolledprocessingdemand.Frontal-polarregions(green)exhibitcomplexproperties modest deficits found during simple recognition tests in which that are content independent and associated with high-level task goals (see text). complete retrieval cues are provided (Wheeler et al., 1995). alternatives within memory. Single word reading, by contrast, is Frontal-polar regions contribute to control processes an automated task that requires minimal controlled processing associated with ongoing task modes because the association of a word’s sound to its visual form is Perhaps the most novel and perplexing finding to emerge from overlearned through extensive experience. When verb retrieval is neuroimaging research on memory retrieval is that frontal-polar practiced, allowing for rapid but more stereotyped retrieval, ac- regions, often right lateralized, are differentially activated during tivation in these frontal regions diminishes substantially (Raichle episodes of remembering (for review, see Tulving et al., 1994; et al., 1994). As another example, categorizing a word within an Buckner, 1996). Other forms of retrieval show considerably less arbitrary grouping (e.g., abstract versus concrete), which requires activation. For example, frontal-polar regions activate minimally retrieval of its context-relevant meaning, also encourages consid- when a subject retrieves any word beginning with a certain stem erable recruitment (Demb et al., 1995). By contrast, classifying cue (e.g., generating “street” in response to “str ”), but if the the alphabetic order of the letters in a word, which is a time- task requires using the stem cue to remember from a specific consuming task but nonetheless based on invariant informa- earlier study list of words, robust activation is observed (Squire et tional properties, does