February 19, 2007
The potential role of the parietal lobe in episodic memory and other cognitive functions
Although episodic memory has commonly been thought to depend on the medial temporal lobe (MTL) and frontal cortex (FC) (refer to Box 1), imaging studies have also consistently demonstrated activations in an additional region, the parietal lobe (PL), during episodic memory retrieval (Cabeza & Nyberg, 2000; Naghavi & Nyberg, 2005; Wagner et al., 2005). This phenomenon was first observed in studies using event-related potentials (ERP), which are brief changes in the brain’s electrical activity (or its electroencephalography signal) in response to a discrete sensory stimulus (Rugg & Allan, 2000). As discussed below, parietal activation has consistently been observed in recognition memory studies using ERP. Activation in the PL, however, has also been observed for a variety of other cognitive functions, leading many scientists to suggest that the PL could be part of an attentional/consciousness network that supports a variety of cognitive processes, including memory function.
ERP waveforms are separated into different components on the basis of scalp distributions and sensitivity to experimental manipulations (Rugg, 1995): waveforms with differentscientists have sought to explore possible parietal contributions to episodic memory retrieval and a collection of other cognitive functionsdistributions are unlikely to have common sources, and features that differ in sensitivity to experimental manipulations are likely to reflect distinct functional processes. Based on these criteria, several ERP components that are sensitive to cognitive processes have been specified. The P300 (also known as P3) is an ERP component that is relevant to investigations on the potential role of the PL for various cognitive functions; it is a positive-directed wave that commonly has maximum amplitude over central-parietal scalp regions. Depending on the experimental condition, deflections in the P3 can peak anywhere between 300 and 600 or more milliseconds post-stimulus (thus the “300” in its name).
Using ERPs, differential regional activity has been demonstrated when previously encountered items are identified as old (hits) compared to when new items are correctly identified as such (correct rejections). Hits, compared to correct rejections (CRs), have consistently elicited the aforementioned P3, which has been referred to as the left parietal ERP old/new effect (Rugg & Allan, 2000). The P3 was initially interpreted in terms of its possible functional properties (for review see Rugg, 1995), not in the context of recognition memory. Past studies that used ERP, such as the one conducted by Paller et al. (1987), sought to investigate possible functions of the P3 observed during memory encoding. In an important study by Smith and Halgren (1989), however, the P3 observed during retrieval was interpreted in terms of a model for recognition memory.
Smith and Halgren (1989) recorded the ERPs of epileptic patients, who had undergone a left or right anterior temporal lobectomy (ATL), and normal controls while they performed a recognition memory test. Subjects were not given any study instructions, but were told to expect a subsequent recognition test. During the following test phase, subjects responded to words they recognized from the study phase by pressing a button. Out of the three groups, only the left-ATL group did not demonstrate a reliable P3 during the test phase. The left-ATL group also had mild impairment on recognition compared to the right-ATL and normal control groups.
Smith and Halgren (1989) acknowledged that the lack of a reliable P3 in the left-ATL group, in combination with lower recognition scores (hits–false positives), could reflect a disruption to processes that are involved in recognition decisions. They also suggested, however, that normal subjects who performed at the accuracy level of the left-ATL patients might not have a reliable P3 either. To test this possibility, the five left-ATL patients with the highest recognition scores were compared to the five subjects who had the lowest recognition scores from each of the right-ATL and normal control groups. While behavioural differences were not found between these subgroups, the same P3 differences that were previously observed still remained.
To explain this dissociation between performance on the recognition test and reliability of the P3, Smith and Halgren (1989) suggested that the P3 reflected processes specific to recollection, the conscious recovery of contextual information for a previously experienced event. Moreover, they proposed that ERPs did not reflect differences in familiarity, even in cases when such differences allowed for correct recognition responses. Along these lines, Smith and Halgren suggested that their left-ATL patients were unimpaired on recognition judgments based on familiarity, but that they had difficulties with recollection.
Later studies sought to investigate whether the P3 differentially reflected recognition based on recollection or familiarity (Gardiner, 1988; Rugg & Doyle, 1992; Paller & Kutas, 1992). According to Rugg (1995), however, whether the P3 is more closely associated with the recollective or familiarity components of recognition memory has not been settled. Nonetheless, these later studies suggested a necessary condition for the emergence of the P3 during memory retrieval: conscious awareness for the fact that “old” items have recently been experienced. It is uncertain, however, whether the P3 reflects processes that are contingent upon this conscious awareness or processes that contribute to it.
In connection with these earlier ERP studies, an emerging collection of functional The MTL, FC, and possibly parietal regions probably do not operate in isolation from one another, but instead interact to support memory encoding and retrieval.neuroimaging studies has revealed parietal activation during episodic memory retrieval (Cabeza & Nyberg, 2000; Naghavi & Nyberg, 2005; Wagner et al., 2005). In studies that have used event-related functional magnetic resonance imaging (fMRI), greater activation has been observed for hits compared to CRs in the posterior parietal cortex (PPC) and retrosplenial cingulate (Wagner et al., 2005). Based on findings of this nature, scientists have sought to explore possible parietal contributions to episodic memory retrieval and a collection of other cognitive functions.
Further to the aforementioned left parietal ERP old/new effect, recent fMRI studies have led to specific hypotheses about what exactly the memory-related influences on parietal activation might be and how different memory operations might be related to different parietal regions. Wagner et al. (2005) have reported findings from their own lab in a small meta-analysis, which demonstrated activity in the PPC when information was perceived as being old, sometimes even when this belief was in error. For example, in a study by Kahn et. al (2004), subjects identified words they had seen during the study phase on a subsequent recognition test. During the test phase, activations were observed in the left inferior parietal cortex during false alarms, which suggest that activations in this region reflected perceived recognition.
Imaging studies have also revealed increased activity in the PPC when recollection of event details supplemented recognition. The subjective experiences of recollection and familiarity are often indexed by remember and know judgments, respectively (Tulving, 1985; Gardiner, 1988). Two regions in the left PPC, specifically the lateral and posterior intraparietal sulcus (IPS), responded preferentially to remember judgments on a recognition memory test. As such, these regions were identified as recollection-sensitive (Wheeler & Buckner, 2004). Moreover, a region along the bank of the IPS, which showed similarly increased activity for remember and know responses compared to CRs, was identified as being familiarity-sensitive.
The PPC has also demonstrated increased activations based on the type of information the subject tries to remember. For example, Dobbins and Wagner (2005) revealed increased activations in the left PPC during source recollection compared to novelty detection attempts. Greater activation was observed in the left PPC during attempts to recover source recollection for perceptual, as well as conceptual, episodic details compared to attempts for item novelty detection. These studies by Wagner et al. demonstrate that different parietal regions are differentially involved in various processes related to retrieval.
The MTL, FC, and possibly parietal regions probably not do operate in isolation from one another, but instead interact to support memory encoding and retrieval. Activations in all of these regions have been observed consistently during episodic memory retrieval (Naghavi & Nyberg, 2005). Correlated activation in the FL and the PL, however, has also been observed during other cognitive functions. In their meta-analysis of Positron Emission Tomography (PET) and fMRI studies, Cabeza and Nyberg (2000) found consistently correlated activations in the FL and PL during attention and working memory tasks. These findings are supported by a later meta-analysis of PET and fMRI studies by Naghavi & Nyberg (2005), who found that correlated activations were most pronounced in the dorsolateral prefrontal cortex (PFC) and the bilateral parietal cortex for attention and working memory, as well as visual awareness and episodic memory retrieval.
Many scientists have suggested that the correlated FC-PL activation reflects working memory and/or attentional processes. Rees and Lavie (2001) proposed that interactionsit could be that an attentional/consciousness network commonly serves a multitude of cognitive functions between fronto-parietal areas and modality-specific posterior regions support both visual attention and visual awareness. Cabeza et al. (2003) suggested that correlated activity in a fronto-parietal-cingulate-thalamic network reflects general attentional processes during episodic retrieval and visual attention. Wagner et al. (2005) have suggested that the PPC is involved in a network supporting attention, wherein the PPC might be involved in maintaining attention on internal mnemonic representations or shifting attention to it. Underlying all of these proposals is the general idea that an attentional/consciousness network supports a variety of cognitive processes, including memory functions. Although various brain regions may be involved in this hypothetical network, it seems that the frontal and parietal regions are thought to be essential components.
A recent study by Rossi et al. (2006) supports the notion of a hypothetical attentional/consciousness network. Targeting the IPS, a region that has been consistently activated during episodic memory retrieval in imaging studies (Wagner, 2005), Rossi et al. compared the effects of event-related repetitive transcranial magnetic stimulation (rTMS) to the right or left IPS regions, during a visuospatial episodic memory task, with those obtained in a matched sample group that received rTMS to the right or left dorsolateral PFC during the same task. The memory task required subjects to judge coloured images as “old” or “new” on a recognition task, depending on whether these images were presented in the preceding study phase.
Unlike that of the dorsolateral PFC-rTMS, the consequences of parietal cortex (PC)-rTMS on encoding and retrieval were negligible. When Rossi et al. tested 12 additional subjects to examine whether the rTMS train impaired visuospatial attention, they observed that, compared to sham stimulation, both right and left IPS-rTMS slowed reaction times, with longer reaction times for right-rTMS compared to left-rTMS.
Based on their results Rossi et al. (2006) suggested that the IPS activations demonstrated by imaging studies during episodic memory tasks do not reflect causal contributions to memory encoding and/or retrieval. Instead, these parietal activations were suggested to reflect contributions to a broad attentional network. Rossi et al. also claimed that the wide-spread attentional processes implicated for episodic memory retrieval could translate to the encoding phase: they proposed that episodic memory encoding requires attention to the presented items, as well as successive elaboration.
The notion of an attentional/consciousness network supporting multiple cognitive functions is in line with the notion of a global workspace, as proposed by Baars (1998, 2002). Global workspace is a mental capacity whereby the activities of the brain are focused on one dominant content, if even only momentarily. In support of this proposed mental capacity, daydreamers can testify to the hopelessness of writing a paper while indulging in a mental getaway. Baars also suggested that when perception is conscious, the corresponding information can be distributed amongst the specialized networks within the brain, as oppose to being restricted to sensory regions. When perception is unconscious, however, information processing is limited to sensory regions. In this way, problem solving, coordination and control could take place through a ‘central information exchange’, whereby some brain region(s) may distribute information to the rest of the brain. As many cognitive functions, such as working memory, are reliant on conscious elements that can be accessed through selective attention (Naghavi & Nyberg, 2005), it could be that an attentional/consciousness network commonly serves a multitude of such cognitive functions, and in this way it could be that such a network serves as a mechanism for the abovementioned distribution of conscious information throughout the brain. It will be interesting to see what future studies on this topic will find.