Brain, Vol. 122, No. 10, 1839-1850,
October 1999
© 1999 Oxford University Press
The functional neuroanatomy of comprehension and memory: the importance of prior knowledge
1 Wellcome Department of Cognitive Neurology, Institute of Neurology, University College London and 2 Department and Centre for Neuroscience, The University of Edinburgh, UK
Correspondence to:
Dr Eleanor Maguire, Wellcome Department of Cognitive Neurology, Institute of Neurology, University College London, 12 Queen Square, London WC1N 3BG UK E-mail: e.maguire{at}fil.ion.ucl.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Stories are a common way in which humans convey and acquire new information. Their effectiveness and memorability require that they be understood which, in turn, depends on two factors—whether the story makes sense and the prior knowledge that the listener brings to bear. Comprehension requires the linking of related pieces of information, some provided within the story and some by the listener, in a process establishing coherence. In this study, we examined brain activations associated with story processing. During PET scanning, passages of prose were read twice to subjects during successive scans with the requirement to remember them. These were either standard stories that were readily comprehensible, or unusual stories for which the global theme was very difficult to extract without prior knowledge of the mental framework. This was manipulated by the provision of relevant, irrelevant or no visual cues shortly before the story. Ratings of comprehension provided by the subjects just after each scan confirmed that standard stories were more comprehensible than the unusual stories, as were unusual stories with a mental framework compared with those without. PET results showed activation of anterior and ventral parts of the medial parietal/posterior cingulate cortex in association with hearing unusual stories when subjects were given prior knowledge of what it might be about. Medial ventral orbitofrontal cortex and left temporal pole activations were found to be associated with more general aspects of comprehension. Medial parietal cortex (precuneus) and left prefrontal cortex were associated with story repetition. We suggest that while the temporal pole is involved in the linking of propositions to build a narrative, the anterior medial parietal/posterior cingulate cortex is concerned with linking this information with prior knowledge. All of this occurs in the context of a general memory processing/retrieval system that includes the posterior parietal (precuneus) and prefrontal cortex. Knowledge of how distinct brain regions contribute differentially to aspects of comprehension and memory has implications for understanding how these processes break down in conditions of brain injury or disease.
PET; comprehension; prior knowledge; medial parietal
BA = Brodmann area
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
The exchange of ideas, facts and feelings is at the core of our conscious lives. The loss of the ability to comprehend or remember what others communicate is a devastating hallmark of conditions such as Alzheimer's disease. Much communication involves stories, conversations and discussions reflecting various forms of connected discourse. As Graesser and colleagues observe (Graesser et al., 1997
), discourse is part of the fabric of virtually all cognitive functions including perception, memory and problem solving. Both Johnson-Laird and Van Dijk and Kintsch distinguish three levels of discourse representation: the `phonemic' (Johnson-Laird, 1983) or `surface code' (van Dijk and Kintsch, 1983), which preserves the exact wording and syntax of clauses; the `propositional representation' or `textbase', which contains explicit propositions and certain inferences needed to establish the meaning of individual sentences (`local coherence'); and the `mental' or `situation model', which embodies the overall theme of the discourse (`global coherence'). The latter is constructed inferentially through interactions between the information conveyed in the story, and the prior knowledge of the listener.
A number of experimental studies have independently manipulated the coherence of passages of text and the level of prior knowledge of the listener/reader to examine the effects on comprehension and memory. In general, it has been found that `local coherence' is dominant when there is low background knowledge: the more coherent the text, the better the comprehension and recall (e.g. McNamara and Kintsch, 1996; McNamara et al., 1996; Voss and Silifies, 1996). Other studies (e.g. Dooling and Lachman, 1971; Bransford and Johnson, 1972; McNamara et al., 1996) have shown that in the case of texts with high local but low global coherence, the possession of a relevant (and activated) framework of prior knowledge also results in increased comprehension and recall. For instance, Bransford and Johnson read a story to subjects and asked them to rate their comprehension of its overall theme as well as testing their recall (Bransford and Johnson, 1972). This story comprised sentences that were all grammatically and semantically correct (i.e. had high local coherence), but the meaning of each sentence was ambiguous and did not seem to have a semantic relationship to previous sentences in the story (i.e. low global coherence). Subjects who were shown an explanatory picture relevant to the theme just before they heard the story rated comprehension of it higher and recalled it better than those who were not given relevant prior knowledge. The use of a picture was intended to mimic the more everyday situation of people bringing their own knowledge to bear in the course of conversation. Bransford outlines later studies illustrating the same general principle (Bransford, 1979).
In the clinical context, deficits in comprehension, often accompanied by memory deficits, are clearly apparent in dementing diseases with widespread cerebral involvement. More circumscribed memory deficits occur with medial temporal-lobe damage (Vargha-Khadem et al., 1997), but often in the context of preserved immediate comprehension. The question naturally arises as to what brain areas are necessary for integrating incoming information with what is already known, making it comprehensible and ready for committal to memory. Neuropsychological studies have reported deficits following right-brain-damage specifically in the comprehension of narratives (Gardner et al., 1983) and semantic integration (Beeman, 1993). However, Leonard and Baum suggest that the apparent impairments observed in right-brain-damage patients may result from increased processing demands rather than a specific deficit in the use of contextual information (Leonard and Baum, 1998).
While the functional neuroanatomy of phonology, lexicon and syntax has been scrutinized closely, there have been fewer neuroimaging studies examining comprehension and memory for passages of text in normal volunteers. Mazoyer et al. and Fletcher and colleagues, using PET to image blood flow during story comprehension, found increased activation of the temporal poles bilaterally and the left superior temporal gyrus (Mazoyer et al., 1993; Fletcher et al., 1995a), and of the posterior cingulate cortex (Fletcher et al., 1995a) during globally coherent stories compared with control tasks. However, no functional neuroimaging work has specifically examined the effects of story coherence and prior knowledge on brain activation. While St George and colleagues found that the N400 component of the event-related brain potential is responsive to global coherence (St George et al., 1994), the specific neural correlates of situation model development and global coherence remain to be clearly specified.
The present study investigated the functional neuroanatomy of the auditory processing of discourse by manipulating coherence and prior knowledge. Subjects listened to six stories, each repeated twice during the course of a series of 12 scans. They rated comprehension of a story immediately after each scan and were asked to recall as much of it as possible. The varying experimental conditions were designed to address specific questions as described below.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Subjects
Prior to scanning, potential subjects were interviewed and only those with healthy general medical, neurological and psychiatric profiles were invited to participate. Thirteen right-handed male subjects (mean age 31.9 years, range 25–43 years) took part in the study. All subjects gave informed written consent. Ethical approval was obtained from the joint National Hospital for Neurology and Neurosurgery/Institute of Neurology medical ethics committee. Permission to administer H215O was obtained from the Administration of Radioactive Substances Advisory Committee of the UK.
Experimental design
Each subject had 12 PET scans. There were six conditions with two scans per condition. For the duration of each 90 s per scan, subjects listened with their eyes closed to a single story and were asked to remember it. Stories were presented via earphones and read in a male voice. Subjects heard six stories in total during the scanning session, one for each condition, each story heard twice, during successive scans (see Table 1). Stories were randomized across conditions and the order of conditions randomized across subjects. No other visual or auditory cues were given during scanning itself.
|
There were two kinds of story: (i) `standard' stories that were straightforward to understand and possessed clear local and global coherence (conditions 5 and 6 in Table 1
); these were designed to be comprehensible without the provision of explicit prior knowledge in the form of pictures shown by the experimenter. (ii) `Unusual' stories that were difficult to comprehend. These had local coherence but low global coherence (conditions 1–4 in Table 1). The unusual stories were grammatically correct, but the text was designed to be confusing such that one could not understand the overall theme of the story without prior knowledge. The unusual stories used were taken from the original Bransford and Johnson study (Bransford and Johnson, 1972), with an additional story adapted from Dooling and Lachman (Dooling and Lachman, 1971), and the remainder were constructed along similar lines by the authors. One of the standard stories was taken from Fancher (Fancher, 1985) (used also in St George et al., 1994) and the other constructed by the authors (see Appendix 1 for examples of stories). Prior knowledge was manipulated in the following manner: before hearing a story (i.e. just before scanning), a subject was shown: (i) no picture; (ii) an irrelevant picture; or (iii) a relevant picture that rendered the subsequent story comprehensible. Subjects were made aware of the relevance (or not) of any pictures shown by the experimenter. Before the next scan, subjects either saw the same relevant or irrelevant picture or, for the first time, saw a relevant or irrelevant picture before hearing the same story again. Given this design, it was possible to ask a number of questions. Specifically, the design allows direct comparison of discourse processing with and without prior knowledge (e.g. C4 versus C2), an analysis of the effect of changes in prior knowledge status across two exposures to the same story (e.g. C3 versus C1), and an examination of the effects of stimulus repetition per se by repeating stories with no change in the status of prior knowledge (e.g. C2, C5, C6).
Immediately after each scan, subjects were asked to rate the comprehensibility of a story along a seven-point scale (as in Bransford and Johnson, 1972), where a rating of 1 represented a complete lack of understanding and 7 implied the story was fully comprehensible. Subjects were then asked to recall out loud as much of the story as possible, with what they said being recorded for later scoring. Memory for the stories was scored in terms of the number of `idea units' recalled, defined by Bransford and Johnson as corresponding to `either individual sentences, basic semantic propositions, or phrases' (Bransford and Johnson, 1972). Stories had between 14 and 19 idea units (see Appendix 1 for partitioning of a story into idea units). These idea unit scores were normalized to yield a percentage memory index. Subjects were debriefed after scanning by asking them to describe the strategies they had used to comprehend and remember the story.
PET scanning and data analysis
PET scans were obtained using a Siemens/CPS ECAT EXACT HR+ (model 962) PET scanner. Scanning was performed with septa retracted, in 3D mode. The field of view of 15.5 cm in the axial extent allowed the whole brain to be studied. Volunteers received an H215O intravenous bolus (330 MBq) infused over 20 s followed by a 20-s saline flush through a forearm cannula. Data were acquired in a 90-s scan frame. There were 12 successive administrations of H215O, 8 min apart. The integrated radioactivity counts that accumulated over the 90-s acquisition period, corrected for background, were used as an index of regional cerebral blood flow. Attenuation correction was computed with a transmission scan prior to emission scan acquisition. Images were reconstructed into 128 x 128 pixels in 63 planes with an inplane resolution of 6.5 mm. In addition, high resolution MRI scans were obtained with a 2.0 Tesla Vision system (Siemens GmbH, Erlangen, Germany) using a T1-weighted 3D gradient echo sequence. The image dimensions were 256 x 256 x 256 voxels. The voxel size was 1 x 1 x 2 mm. Images were analysed using Statistical Parametric Mapping (SPM96; Wellcome Department of Cognitive Neurology, London, UK; http://www.fil.ion.ucl.ac.uk) executed in MATLAB (Mathworks Inc., Sherborn, Mass., USA). All scans were automatically realigned to the first scan. The structural MRI scans were coregistered to the mean PET image for each subject and then stereotactically transformed to a standard MRI template (from the Montreal Neurological Institute; Evans et al., 1993) in stereotactic space (Talairach and Tournoux, 1988), and the same transformation matrix subsequently applied to the PET images (Friston et al., 1995). Images were smoothed using an isotropic Gaussian kernel of 16 mm (full width at half maximum). Data were analysed using a randomized block design with global brain activity as a subject-specific covariate of no interest. Areas of significant change in brain activity were determined using appropriately weighted contrasts between the task-specific scans and the t statistic. The resulting sets of t values constituted the statistical parametric map. Significance levels were set at P < 0.01.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Comprehension and memory
Subjects' ratings of the comprehensibility of the stories are displayed in Fig. 1A
. Comprehension varied from the low mean of 2.85 (C3 exposure 1) to the maximum of 7 (C5 exposure 2, C6 both exposures). Standard stories with global coherence (C5, C6) were rated as fully comprehensible, even after one exposure. The unusual story, where prior knowledge was present from the outset, was also rated highly (C4) but was rated as hard to understand when no prior knowledge was available (C1, C2), with a marginal increase in comprehension following repetition. Interestingly, when subjects were given prior knowledge between first and second exposures to a story (C3), comprehension ratings increased significantly once the global theme of the story was known. These observations were confirmed by separate analyses of variance of the comprehension ratings for the first and second hearings, and by a comparison of these two hearings. In each case these revealed a significant overall difference between conditions [first hearing: F(5,60) = 32.1, P < 0.0001; second hearing: F(5,60) = 14.86, P < 0.0001; interaction between first versus second hearing and conditions: F(5,60) = 15.42, P < 0.0001]. Breakdown of the main effects of conditions using Newman–Keuls tests revealed that upon first exposure, unusual stories were well understood if preceded by a relevant picture (C4), but not otherwise (i.e. C4 was rated higher than C1, C2 and C3; P < 0.01). Upon hearing these stories a second time, they became comprehensible if now preceded by a relevant picture for the first time, but remained poorly understood if they were again preceded by an irrelevant picture (i.e. C3 was now rated higher than C1 and C2; P < 0.01). Comprehension ratings for C3 were significantly higher on second hearing than on the first (P < 0.01). The two standard stories (C5 and C6) were always reported as highly comprehensible with or without a preceding picture cue.
|
Subjects' memory for the key ideas of the stories varied from a mean of 37% idea units recalled (C3 exposure 1) through to 83% (C6 exposure 2). Fig. 1B
shows memory performance across all conditions for both the first and second exposures. For each condition, memory scores were higher after the second exposure than following initial exposure. This effect is clearly apparent in Fig. 1B with all the open bars being higher than the black bars. Standard stories were also remembered better than the unusual stories. Where subjects had prior knowledge of an unusual story at the outset (C4), memory scores showed a trend towards being higher than when no prior knowledge was made available. Three analyses of variance of memory scores were conducted: following first exposure to a given story, following the second exposure, and comparing the first and second exposure. These showed significant differences in memory between conditions [first exposure: F(5,60) = 6.31, P < 0.0001; second exposure: F(5,60) = 7.20, P < 0.0001] and better memory after the second than after the first exposure [F(1,12) = 91.35, P < 0.0001]. However, breakdown of the main effects of conditions using Newman–Keuls tests revealed only a non-significant trend towards better memory when a relevant picture was shown before a story compared with conditions with either no or an irrelevant picture. The only significant effect was better memory for the readily comprehensible standard stories (C5 and C6), the best performance being seen when these stories were preceded by a relevant picture (C6).
PET results
Brain regions activated when a mental framework was provided by a relevant picture
Our primary interest was in the influence of prior knowledge on the neural substrates of story comprehension and memory. Accordingly, we initially identified brain regions that were differentially activated by the presence and relevance of the picture shown prior to scanning, beginning with the first time each story was heard (left side of Table 1). Details of these activations are outlined in Table 2. Significant activation was seen in anterior medial parietal/posterior cingulate cortex [Brodmann area (BA) 31] during unusual stories preceded by a relevant picture compared to such stories preceded by an irrelevant picture (C4 – C2; see Fig. 2A), or no picture [C4 – (C1 + C3)]. This anterior medial parietal/posterior cingulate activation could either be associated with increased comprehension, or could be due to the presence of a relevant picture to which a story could be related. It does not seem to be due to increased comprehension alone because comparison of scans taken during the standard and unusual stories that had not been preceded by a picture [C5 – (C1 + C3)] failed to show this effect, even though there was a striking difference in comprehensibility. Instead this latter comparison revealed activation of the left temporal pole (BA 38) and ventromedial orbitofrontal region (BA 11). Moreover, the mere presentation of any picture before a story (e.g. an irrelevant picture) did not reveal any differences in the medial parietal region [C2 – (C1 + C3)], but did show effects in the right inferior parietal cortex (BA 40).
|
|
Brain activations associated with story repetition
A second major pattern of activation in this experiment was associated with repetition of stories (Table 3
). Examination of the main effect of repetition, i.e. all second hearings compared with all first hearings of the stories, revealed activation in the left middle frontal gyrus (BA 10; see Fig. 2B), the anterior medial parietal/posterior cingulate region and, interestingly, in another area in the medial parietal cortex (BA 7, the precuneus) but one noticeably more posterior than that discussed in the above section (see Fig. 2C). The vector distance between the two medial parietal areas was 25 mm, clearly exceeding the 16 mm full width at half maximum used in smoothing the images, confirming their separability. As two of the conditions changed between first and second hearings of the stories (C1 and C3), with the introduction of picture cues prior to second hearing, the effect of story repetition alone was examined by comparing scans during the second hearing with those of the first in the remaining conditions (C2, C4, C5 and C6). The resulting activations were also in the left frontal, anterior medial parietal/posterior cingulate, and posterior medial parietal cortices.
|
Subject-generated mental frameworks
The debriefing of subjects revealed an intriguing strategy in which they consistently engaged. They reported that when listening to an unusual story for the second time, and still without any relevant visual cue (C1 and C2), they would try to place the incoming information into a mental framework of their own making. They had devised this framework after the first exposure to the confusing text in the 8-min interval between the first and second hearings. All the while, subjects were still unsure of the appropriateness of this devised framework and so declared low comprehension ratings for the overall story theme when asked to do so. But they were in a position to make the mental effort of integrating the second hearing of the story into this framework even though the experimenter-manipulated conditions of scanning were unchanged.
We reasoned that if generating a mental framework for themselves was equivalent to being presented with the relevant visual cue, then comparison of conditions 1, 2, 3 and 4 during the second hearing with conditions 1, 2 and 3 of the first hearing might also reveal anterior medial parietal/posterior cingulate activation, in addition to the pattern of areas activated in response to story repetition per se. This prediction was upheld (Table 4). Following on from this finding, a further prediction can be made. This is that comparison of conditions in which subjects use information within pictures provided by the experimenter and a mental framework of their own generation might fail to show differential medial parietal/posterior cingulate activation as the activations here would subtract out. The comparisons of interest were during second exposure (right side of Table 1) when the unusual stories were either preceded by a relevant picture for the first time (C3) or by an irrelevant picture (C1). A mental framework into which to fit the information would be present in each case [i.e. one provided by the experimenter (C3) and one devised by the subject (C1)]. The comparison shows temporal pole (BA 38) activation but not anterior medial parietal/posterior cingulate activity. This is in contrast to the same comparison during first exposure to stories [left side of Table 1: C4 – C2].
|
Relationship between blood flow, comprehension and memory
Finally, we considered whether regional brain activation varied as a function of subjects' reported comprehension and memory of the stories. Increasing activity in anterior medial parietal/posterior cingulate (BA 31) and ventromedial orbitofrontal (BA 11) cortex (see Fig. 2D
) was correlated to improving comprehension. When memory performance was used as the covariate, a correlation was found with activation in the left superior frontal cortex (BA 10) (Table 5).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Discourse is crucial to humans for conveying and acquiring new information. To understand a story or a passage of text, or to follow a conversation requires the linking of related pieces of information in a process establishing coherence. Factors affecting the development and maintenance of coherence are manifold, but two known important influences are the local and global coherence of the text and the prior knowledge of the listener or reader. In this study, we identified brain activations associated with auditory discourse processing (in the form of stories) by manipulating differences in global coherence and prior knowledge. The main finding was that an area of anterior medial parietal/posterior cingulate cortex was differentially active when subjects attempted to link what they were hearing with the prior knowledge that they brought to bear on the task.
The more detailed picture of the neuroanatomy of discourse processing revealed five principal regions of increased brain activation: two in medial parietal cortex, two within prefrontal cortex, and another in the left temporal pole. The circumstances in which differential activation of these regions is observed gives insight into the effects of prior knowledge and coherence, and into the distributed neuroanatomy of comprehension and memory processing.
Comprehension and memory
Considering the cognitive aspects of the experiment first, we employed a similar method to that of Bransford and Johnson (Bransford and Johnson, 1972). They found that subjects who had prior knowledge of the global theme of a low global coherence story rated comprehension as significantly higher than those who had not been exposed to prior knowledge. The effect of repetition of the story alone, without knowledge of the theme, increased comprehension ratings only marginally. Recall scores for idea units in the story followed a similar pattern. The original data of Bransford and Johnson were based on a single low coherence story (Bransford and Johnson, 1972). Our data, based on several stories, echoes very closely the original comprehension findings and clearly illustrates the powerful effect of prior knowledge and coherence on comprehension ratings. While Bransford and Johnson also found a significant increase in memory scores when subjects had prior knowledge of the low coherent story compared with when they did not (Bransford and Johnson, 1972), no significant differences were observed in the present study, except between standard and unusual stories, and first and second hearings. It is apparent from the recordings of memory recall performance for the unusual stories, that once comprehension was achieved (e.g. C3 exposure 2), an increase in the number of idea units subsequently recalled did not necessarily occur. It appears that once the theme of a story was grasped and emerged to the fore, the number of details freely recalled became less critical and faded into the background, with details being of primary importance in the effort to reach comprehension. A forced choice recognition memory test following each story exposure may have been more sensitive in revealing differences in memory scores between the unusual stories compared with the free recall method employed here, and/or a delay in recognition or recall testing until later in the interscan interval.
Anterior medial parietal/posterior cingulate cortex
We now turn to the data on functional activation. While discourse retrieval produced activation of posterior medial parietal and prefrontal regions, areas typically associated with general memory retrieval, this study has revealed separable medial parietal and prefrontal activations associated with distinctly different cognitive functions.
The effect of being given prior knowledge of the appropriate global theme of a story (particularly before its first exposure) was associated with a separable more anterior and ventral activation of the medial parietal/posterior cingulate cortex than the one noted for repetition of stories. Clearly, the medial parietal region is a large cortical area and presumably has many associated functions, yet relatively little is known about its functional segregation. Two previous studies have noted similar dissociations within medial parietal cortex to those found in the present study (Buckner et al., 1996; Fletcher et al., 1998). In commenting on their finding, Buckner and colleagues indicated that the precise function of the more anterior part of the medial parietal region was as yet unspecified. Here we find that it was activated when subjects had a mental framework into which to place incoming information (Bransford, 1979; Johnson-Laird, 1983; van Dijk and Kintsch, 1983). Having a relevant visual cue to call to mind when listening to a story for the first time—even if discourse only had local coherence—activated this area relative to situations when there was no mental framework. This activation was not merely due to a difference in comprehension, as it was not apparent when a standard story, heard for the first time and not preceded by a visual cue, was compared with the first hearing of an unusual story without visual cues. Clinical evidence supports the importance of the posterior retrosplenial cortex in memory (Valenstein et al., 1987; Rudge and Warrington, 1991), and attenuation of activation in this region was observed during a distracting task that interfered with memory encoding in a previous imaging study (Fletcher et al., 1995b). Our results now suggest that one important role for the posterior cingulate region in the comprehension–memory process is in the linking of incoming information with a repository of activated knowledge to form a coherent representation of discourse.
This interpretation is complemented by comparison of unusual stories without relevant visual cues during their second hearing with similar stories during their first hearing which revealed activation in this anterior medial parietal region. It would seem that such is the active nature of discourse processing that, by the second hearing of a story, subjects worked out mental frameworks of their own when prior knowledge was absent. This phenomenon reflects the active attempt by subjects to construct a mental model of the situation described from the successive sentences of a novel story (Johnson-Laird, 1983). This mental model is used implicitly to make inferences as part of the effort to understand meaning that is an entirely natural aspect of participation in this kind of experiment. Debriefing of the subjects revealed that, when listening to an unusual story for the second time, but still without any relevant visual cue (C1 and C2), they would often report that they had tried to place the incoming information into a mental framework of their own making. This framework was devised following the first exposure to the confusing text in the interval between the first and second hearings. Graesser and colleagues note that a general underlying principle of discourse processing is that people strive to achieve a global level of understanding and actively operate on what they hear/read to explain why information is mentioned (Graesser et al., 1997). Clearly the anterior medial parietal/posterior cingulate cortex is engaged by this process. Although observed to deactivate in several imaging studies (e.g. Buckner et al., 1996), its activation was found in a previous study of story comprehension where standard (i.e. readily comprehensible) stories were compared with unlinked sentences (Fletcher et al., 1995a). In that case, it was concluded that activation in this area was not due to a role in memory encoding alone, as the unlinked sentences would arguably have placed a greater load on memory; a similar argument can be made in the present case.
Activation of this area may not be restricted to processing stories or text. The posterior cingulate cortex has also been activated in spatial navigation tasks (e.g. Aguirre et al., 1996). Topographical learning, as with story comprehension, also necessitates the integration of incoming information to form a coherent representation of a route. Indeed, the `mental models' approach of Johnson-Laird explicitly asserts that such models have their origin in the evolution of perceptual ability (Johnson-Laird, 1983), and are thus as relevant to the perception of space as to the comprehension of discourse. This commonality across domains strengthens the view that this anterior medial parietal/posterior cingulate region is concerned with the incorporation online of information into an accumulating structure of which background or prior knowledge is a fundamental component. Subjects in the present study were made aware whether visual cues were relevant or not. Another interesting way to test the function of this anterior medial parietal region would be to compare subjects during exposure to unusual stories if they did not know whether the visual cue seen prior to exposure was relevant or not. The findings of the present study would predict that this region would be active as subjects tried to fit the discourse with a given cue in an effort to achieve global coherence (even if this mental effort is actually unsuccessful).
Posterior medial parietal cortex (precuneus)
The design of the study included, out of necessity, the factor of repeated exposure to stimuli. Story repetition alone, irrespective of the type of story or the status of prior knowledge, was associated with activation of the posterior medial parietal cortex (precuneus). Activation of the precuneus has often been observed in functional neuroimaging studies of memory retrieval (e.g. Squire et al., 1992; Shallice et al., 1994; Fletcher et al., 1996). It has been suggested that the precuneus region may play a role in imagery during memory retrieval (Fletcher et al., 1995c), although other studies have found activation of this region when imagery was not a factor (Buckner et al., 1996; Krause et al., 1998). However, it is unlikely that the precuneus activation was due to imagery per se in the present study, since the stories would have been likely to elicit imagery even during the first hearing and not just during the second as the differential activation during second hearing requires. An alternative imagery proposal might be that this activation was due to subjects imagining the visual cues seen prior to a scan, as more of these were present prior to scanning of the second exposure to stories (see Table 1). Perhaps subjects would have always called these to mind, although this is unlikely as these were, in many cases, irrelevant pictures (and declared to be by the experimenter).
Buckner and colleagues have instead proposed that posterior medial parietal region could be associated with retrieval effort (Buckner et al., 1996). The posterior medial parietal activation seen here is in keeping with this proposed role in episodic memory, as our subjects recognized the story content on second hearing and effortfully tried to engage in further online memory processing in relation to their own developing mental models. Effortful processing is, however, not specific to a particular sensory domain. It would therefore be interesting in the future to see if giving subjects verbal cues instead of pictures (e.g. story titles) would affect activations observed during subsequent scans. A recent PET study used visual stimuli which appeared meaningless when initially viewed (because they were degraded) but became recognizable on second viewing after seeing a version of the same image which had not been degraded (Dolan et al., 1997). This design did not permit direct comparison of images with and without prior knowledge on first viewing. Nonetheless, comparison of second viewings with first revealed activation of the same posterior medial parietal (precuneus) region. This suggests that this region responds to the repetition of stimuli in both the auditory (this study) or visual domains (Dolan et al., 1997).
Left middle frontal gyrus (BA 10)
In the present study, activation of the left middle/superior frontal gyrus occurred during repetition of stories, with the degree of activation covarying with number of idea units recalled. In many functional neuroimaging studies of episodic memory that document posterior medial parietal activation, there is typically an associated prefrontal activation. During verbal episodic memory retrieval, this is often found to be a right prefrontal activation, while left prefrontal activations are typically reported during memory encoding. Such results are consistent with the hemispheric encoding retrieval asymmetry (HERA) theory (Nyberg et al., 1996). At first sight, our left prefrontal finding may seem to contradict the HERA model of prefrontal involvement in memory processing, as it might be thought to predict that the retrieval of information during second hearing of stories compared with their first hearing would result in right prefrontal activation. In many previous studies where the asymmetry of memory processing is reported, there is a clear-cut distinction between new information (encoding) and old material (retrieval). In the real world, however, this distinction is often blurred and the dynamic nature of learning is such that encoding of new and retrieval of old often mix. This seems to be the case in the present study where even the second hearing of stories triggered new encoding and thus, consistent with the HERA theory, left prefrontal activation. It should be noted, however, that other studies do report bilateral prefrontal activations with memory processing, while others find left prefrontal activation during retrieval (Kapur et al., 1995; Maguire et al., 1998). More recently, Nolde and colleagues have suggested that the laterality of prefrontal activations in neuroimaging studies has less to do with encoding/retrieval but depends on what they term `the reflective demands of the task' (Nolde et al., 1998). They contend that the left prefrontal cortex is active when retrieval is complex, for example when information is being maintained whilst being evaluated. The Nolde et al. cortical asymmetry of reflective activity (CARA) hypothesis may fit broadly with the present findings, although it is unclear whether they would predict covariation of activity with increased recall of idea units. The location of the present activation is quite ventral in left prefrontal cortex. Similarly located activations in left frontal cortex have been reported in association with the attempt to retrieve information (Kapur et al., 1995) and with retrieval success (Rugg et al., 1996). The latter finding may be most relevant to understanding activation of the left frontal cortex in the present study and certainly fits with the finding of a correlation between activity in this region and the number of idea units that were successfully recalled after scanning.
We did not observe activation in the hippocampal formation. While many early imaging studies of episodic memory revealed only activation in the prefrontal cortex, more recent memory studies have shown hippocampal activation during both memory encoding and retrieval (see Lepage et al., 1998 for review). The present results suggest that hippocampal activation does not vary as a function of the comprehensibility of the material to be remembered. Some studies have found differential parahippocampal (and prefrontal) activation as a function of memory success (Brewer et al., 1998; Wagner et al., 1998), but in our cognitive data and unlike Bransford and Johnson (Bransford and Johnson, 1972), we did not find dramatic changes in the amount remembered across conditions. The failure to see hippocampal activation may therefore reflect an insensitivity in our protocols or, more likely, that the hippocampus was active in all conditions, as memory was required in all cases.
Medial ventral orbitofrontal cortex (BA 11)
This area of the ventromedial orbitofrontal cortex is seldom activated in imaging memory studies; indeed it is often deactivated. Plots of the adjusted blood flow activity in the present study confirm that this region is more active with increasing comprehension. This differs from the response pattern of the anterior medial parietal/posterior cingulate cortex in that activation in the ventromedial frontal area was not observed in the categorical comparison of unusual stories with and without prior knowledge (C4 – C2). This indicates that its response to increasing comprehension reflects something other than linking incoming information to a pre-existing mental framework. A speculative possibility, given that the ratings of comprehension were made subjectively by the subjects, is that they used this information as feedback to themselves in a `reward/punishment' manner. Neurophysiological and lesion evidence has indicated a role for the orbitofrontal cortex in reward (Rolls, 1996), and the change in orbitofrontal activation may reflect an emotional response to the `reward' of increasing comprehension.
Temporal pole (BA 38)
Activation of the left temporal pole was also observed when standard stories were compared with unusual stories, or on second exposure between unusual stories with and without prior knowledge. Activation of the temporal pole seemed to be responsive to extremes of coherence differences (very high versus very low), irrespective of prior knowledge. Other studies have also found this area to be active when stories were compared with unlinked sentences (Mazoyer et al., 1993; Fletcher et al., 1995a). The processing of sentences compared with random word strings also revealed left temporal pole activation (Bottini et al., 1994). Overall, this suggests that this region may be involved when a level of linguistic processing is required beyond that associated with the lexical and semantic analysis of individual words, when sentences are linked to form a narrative.
Conclusion
The key new finding of this experiment is that the anterior medial parietal/posterior cingulate cortex is concerned with linking narrative information with prior knowledge. The ventromedial orbitofrontal cortex is also responsive to increasing comprehension, and we also find that increased activity in the temporal pole occurs during the linking of propositions to build a narrative. Each of these changes in brain activity occurs in the context of a general memory processing/retrieval system that includes the posterior parietal (precuneus) cortex, prefrontal cortex and presumably medial temporal lobe (e.g. hippocampus). Anatomical connections between precuneus and prefrontal cortex have been documented in non-human primates, area 7m projecting to area 46 (Goldman-Rakic, 1988). The anterior medial parietal/posterior cingulate cortex is connected to area 7m, but also has extensive connections with orbitofrontal cortex and the medial temporal region (Vogt and Pandya, 1987). Given that much of cognition involves relating what is new to what is known, these findings throw light on why damage to different brain regions can compromise distinct aspects of the comprehension–memory process. Further work will continue to examine the integrative role of the anterial medial parietal/posterior cingulate cortex (including in domains beyond oral discourse), with the current story stimuli perhaps providing a means for testing effects of lesions in this brain region.
|
Appendix 1 |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Example of a standard story, with global coherence, comprehensible with or without prior knowledge. Adapted from Fancher (1985):
Long ago / a musician was astounded / when his four-year-old son / wrote out a concerto for harpsichord, / ink-spattered / and too difficult for anyone to play / but completely correct musically. / This was but one of several signs that convinced the man / that God had entrusted him / with the care and upbringing of an extra-ordinary genius. / He abandoned his own serious professional ambitions / to devote himself instead to the musical education of his son / and to the promotion of his son's career. / In due course / the boy became not only the most celebrated child prodigy in Europe / but also the greatest composer of his time. / (16 idea units.)
Example of an unusual story, with no global coherence, comprehensible only with prior knowledge (taken from Bransford and Johnson, 1972):
If the balloons popped / the sound would not be able to carry / since everything would be away from the correct floor. / A closed window would also prevent the sound from carrying / since most buildings tend to be well insulated. / Since the whole operation depends on a steady flow of electricity, / a break in the middle of the wire would also cause problems. / Of course, the fellow could shout / but the human voice is not loud enough to carry that far. / An additional problem is that a string could break on the instrument / then there could be no accompaniment to the message. / It is clear that the best situation would involve less distance / then there would be fewer potential problems. / With face to face contact the least number of things could go wrong. / (14 idea units.)
Consider the effect of reading the unusual story again, but having first looked at Fig. A1. This picture provides a mental framework for the unusual story illustrating the powerful effect of prior knowledge on comprehension.
|
|
Acknowledgments |
|---|
We wish to thank Professor Dick Passingham and Dr Jennifer Coull for helpful discussions. E.A.M. and C.D.F. are supported by the Wellcome Trust, and R.G.M.M. by the Medical Research Council.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Aguirre GK, Detre JA, Alsop DC, D'Esposito M. The parahippocampus subserves topographical learning in man. Cereb Cortex 1996; 6: 823–9.
Beeman M. Semantic processing in the right hemisphere may contribute to drawing inferences from discourse. [Review]. Brain Language 1993; 44: 80–120.[Web of Science][Medline]
Bottini G, Corcoran R, Sterzi R, Paulesu E, Schenone P, Scarpa P, et al. The role of the right hemisphere in the interpretation of figurative aspects of language. Brain 1994; 117: 1241–53.
Bransford JD. Human cognition: learning, understanding and remembering. Belmont (CA): Wadsworth; 1979.
Bransford JD, Johnson MK. Contextual prerequisites for understanding: some investigations of comprehension and recall. J Verb Learn Verb Behav 1972; 11: 717–26.
Brewer JB, Zhao Z, Desmond JE, Glover GH, Gabrieli JD. Making memories: brain activity that predicts how well visual experience will be remembered [see comments]. Science 1998; 281: 1185–7. Comment in: Science 1998; 281: 1151–2.
Buckner RL, Raichle ME, Miezin FM, Petersen SE. Functional anatomic studies of memory retrieval for auditory words and visual pictures. J Neurosci 1996; 16: 6219–35.
Dolan RJ, Fink GR, Rolls E, Booth M, Holmes A, Frackowiak RS, et al. How the brain learns to see objects and faces in an impoverished context. Nature 1997; 389: 596–9.[Medline]
Dooling DJ, Lachman R. Effects of comprehension on retention of prose. J Exp Psychol 1971; 88: 216–22.[Web of Science]
Evans AC, Collins DL, Mills SR, Brown ED, Kelly, RL, Peters TM. 3D statistical neuroanatomical models from 305 MRI volumes. In: Klaisner LA, editor. Proceedings of the IEEE-Nuclear Science Symposium and Medical Imaging Conference. Piscataway, NJ: IEEE Service Centre; 1993. p. 1813–7.
Fancher RE. The intelligence men, and makers of the I.Q. controversy. New York: W.W. Norton; 1985.
Fletcher PC, Happe F, Frith U, Baker SC, Dolan RJ, Frackowiak RSJ, et al. Other minds in the brain: a functional imaging study of `theory of mind' in story comprehension. Cognition 1995a; 57: 109–28.[Web of Science][Medline]
Fletcher PC, Frith CD, Grasby PM, Shallice T, Frackowiak RS, Dolan RJ. Brain systems for encoding and retrieval of auditory-verbal memory. Brain 1995b; 118: 401–16.
Fletcher PC, Frith CD, Baker SC, Shallice T, Frackowiak RS, Dolan RJ. The mind's eye – precuneus activation in memory-related imagery. Neuroimage 1995c; 2: 195–200.[Web of Science][Medline]
Fletcher PC, Shallice T, Frith CD, Frackowiak RS, Dolan RJ. Brain activity during memory retrieval: the influence of imagery and semantic cueing. Brain 1996; 119: 1587–96.
Fletcher PC, Shallice T, Frith CD, Frackowiak RS, Dolan RJ. The functional roles of prefrontal cortex in episodic memory. II. Retrieval. Brain 1998; 121: 1249–56.
Friston KJ, Holmes AP, Worsley KJ, Poline JB, Frith CD, Frackowiak RSJ. Statistical parametric maps in functional imaging: a general linear approach. Hum Brain Map 1995; 2: 189–210.
Gardner H, Brownell HH, Wapner W, Michelow D. Missing the point: the role of the right hemisphere in the processing of complex linguistic materials. In: Perecman E, editor. Cognitive processing in the right hemisphere. Orlando: Academic Press; 1983. p. 169–91.
Goldman-Rakic PS. Topography of cognition: parallel distributed networks in primate association cortex. [Review]. Annu Rev Neurosci 1988; 11: 137–56.[Web of Science][Medline]
Graesser AC, Millis KK, Zwaan RA. Discourse comprehension. Annu Rev Psychol 1997; 48: 163–89.[Web of Science][Medline]
Johnson-Laird PN. Mental models: towards a cognitive science of language, inference, and consciousness. Cambridge: Cambridge University Press; 1983.
Kapur S, Craik FI, Jones C, Brown GM, Houle S, Tulving E. Functional role of the prefrontal cortex in retrieval of memories: a PET study. Neuroreport 1995; 6: 1880–4.[Web of Science][Medline]
Krause BJ, Schmidt D, Mottaghy FM, Taylor J, Halsband U, Herzog H, et al. The precuneus is a major player in a network of distributed brain regions in episodic memory retrieval. Neuroimage 1998; 7 (4 Pt 2): S828.
Leonard CL, Baum SR. On-line evidence for context use by right-brain-damaged patients. J Cogn Neurosci 1998; 10: 499–508.[Web of Science][Medline]
Lepage M, Habib R, Tulving E. Hippocampal PET activations of memory encoding and retrieval: the HIPER model. Hippocampus 1998; 8: 313–22.[Web of Science][Medline]
Maguire EA, Burgess N, Donnett JG, Frackowiak RS, Frith CD, O'Keefe J. Knowing where and getting there: a human navigation network. Science 1998; 280: 921–4.
Mazoyer BM, Tzourio N, Frak V, Syrota A, Murayama N, Levrier O, et al. The cortical representation of speech. J Cogn Neurosci 1993; 5: 467–79.
McNamara DS, Kintsch W. Learning from texts: effects of prior knowledge and text coherence. Discourse Processes 1996; 22: 247–88.[Web of Science]
McNamara DS, Kintsch E, Songer NB, Kintsch W. Are good texts always better? Text coherence, background knowledge, and levels of understanding in learning from text. Cognit Instruction 1996; 14: 1–43.
Nolde SF, Johnson KM, Raye CL. The role of prefrontal cortex during tests of episodic memory. Trends Cogn Sci 1998; 2: 399–406.[Web of Science]
Nyberg L, Cabeza R, Tulving E. PET studies of encoding and retrieval: the HERA model. Psychonomic Bull Rev 1996; 3: 135–48.
Rolls ET. The orbitofrontal cortex. [Review]. Philos Trans R Soc Lond B Biol Sci 1996; 351: 1433–44.[Web of Science][Medline]
Rudge P, Warrington EK. Selective impairment of memory and visual perception in splenial tumours. Brain 1991; 114: 349–60.
Rugg MD, Fletcher PC, Frith CD, Frackowiak RS, Dolan RJ. Differential activation of the prefrontal cortex in successful and unsuccessful memory retrieval. Brain 1996; 119: 2073–83.
Shallice T, Fletcher PC, Frith CD, Grasby P, Frackowiak RS, Dolan RJ. Brain regions associated with acquisition and retrieval of verbal episodic memory. Nature 1994; 368: 633–5.[Medline]
Squire LR, Ojemann JG, Miezin FM, Petersen SE, Videen TO, Raichle ME. Activation of the hippocampus in normal humans: a functional anatomical study of memory. Proc Natl Acad Sci USA 1992; 89: 1837–41.
St George M, Mannes S, Hoffman JE. Global semantic expectancy and language comprehension. J Cogn Neurosci 1994; 6: 70–83.
Talairach J, Tournoux P. Co-planar stereotaxic atlas of the human brain. Stuttgart: Thieme; 1988.
Valenstein E, Bowers D, Verfaellie M, Heilman KM, Day A, Watson RT. Retrosplenial amnesia. Brain 1987; 110: 1631–46.
van Dijk TA, Kintsch W. Strategies of discourse comprehension. New York: Academic Press; 1983.
Vargha-Khadem F, Gadian DG, Watkins KE, Connelly A, Van Paesschen W, Mishkin M. Differential effects of early hippocampal pathology on episodic and semantic memory [see comments] [published erratum appears in Science 1997; 277: 1117]. Science 1997; 277: 376–80. Comment in: Science 1997; 277: 330–1.
Vogt BA, Pandya DN. Cingulate cortex of the rhesus monkey: II. Cortical afferents. J Comp Neurol 1987; 262: 271–89.[Web of Science][Medline]
Voss JF, Silifies LN. Learning from history text: the interaction of knowledge and comprehension skill with text structure. Cognit Instruction 1996; 14: 45–68.
Wagner AD, Schacter DL, Rotte M, Koutstaal W, Maril A, Dale AM, et al. Building memories: remembering and forgetting of verbal experiences as predicted by brain activity [see comments]. Science 1998; 281: 1188–91. Comment in: Science 1998; 281: 1151–2.
Received January 25, 1999. Revised March 26, 1999. Accepted April 27, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Whitehead, J. L. Marchant, D. Craik, and C. D. Frith Neural correlates of observing pretend play in which one object is represented as another Soc Cogn Affect Neurosci, June 17, 2009; (2009) nsp021v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Kapogiannis, A. K. Barbey, M. Su, G. Zamboni, F. Krueger, and J. Grafman Cognitive and neural foundations of religious belief PNAS, March 24, 2009; 106(12): 4876 - 4881. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Tavares, A. D. Lawrence, and P. J. Barnard Paying Attention to Social Meaning: An fMRI Study Cereb Cortex, August 1, 2008; 18(8): 1876 - 1885. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Wilson, I. Molnar-Szakacs, and M. Iacoboni Beyond Superior Temporal Cortex: Intersubject Correlations in Narrative Speech Comprehension Cereb Cortex, January 1, 2008; 18(1): 230 - 242. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Tse, R. F. Langston, M. Kakeyama, I. Bethus, P. A. Spooner, E. R. Wood, M. P. Witter, and R. G. M. Morris Schemas and Memory Consolidation Science, April 6, 2007; 316(5821): 76 - 82. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-F. Demonet, G. Thierry, and D. Cardebat Renewal of the Neurophysiology of Language: Functional Neuroimaging Physiol Rev, January 1, 2005; 85(1): 49 - 95. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Pollmann, J. Lepsien, K. Hugdahl, and D. Y. von Cramon Auditory Target Detection in Dichotic Listening Involves the Orbitofrontal and Hippocampal Paralimbic Belts Cereb Cortex, August 1, 2004; 14(8): 903 - 913. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. C. Lou, B. Luber, M. Crupain, J. P. Keenan, M. Nowak, T. W. Kjaer, H. A. Sackeim, and S. H. Lisanby Parietal cortex and representation of the mental Self PNAS, April 27, 2004; 101(17): 6827 - 6832. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. I. Scahill, J. M. Schott, J. M. Stevens, M. N. Rossor, and N. C. Fox Mapping the evolution of regional atrophy in Alzheimer's disease: Unbiased analysis of fluid-registered serial MRI PNAS, April 2, 2002; 99(7): 4703 - 4707. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. A. Maguire, F. Vargha-Khadem, and M. Mishkin The effects of bilateral hippocampal damage on fMRI regional activations and interactions during memory retrieval Brain, June 1, 2001; 124(6): 1156 - 1170. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Elliott, R. J. Dolan, and C. D. Frith Dissociable Functions in the Medial and Lateral Orbitofrontal Cortex: Evidence from Human Neuroimaging Studies Cereb Cortex, March 1, 2000; 10(3): 308 - 317. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Lepage, O. Ghaffar, L. Nyberg, and E. Tulving From the Cover: Prefrontal cortex and episodic memory retrieval mode PNAS, January 4, 2000; 97(1): 506 - 511. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||