Response-related fMRI of veridical and false recognition of words

European Psychiatry 19 (2004) 42–52 www.elsevier.com/locate/eurpsy

Original article

Response-related fMRI of veridical and false recognition of words Reinhard Heun a,*, Frank Jessen a, Uwe Klose b, Michael Erb b, Dirk-Oliver Granath a, Wolfgang Grodd b b

a Department of Psychiatry, University of Bonn, Sigmund-Freud-Street 25, Venusberg, 53105 Bonn, Germany Department of Neuroradiology, Section Experimental NMR of the CNS, University of Tübingen, 72026 Tübingen, Germany

Received 12 April 2002; received in revised form 19 November 2002; accepted 8 September 2003

Abstract Objectives. – Studies on the relation between local cerebral activation and retrieval success usually compared high and low performance conditions, and thus showed performance-related activation of different brain areas. Only a few studies directly compared signal intensities of different response categories during retrieval. During verbal recognition, we recently observed increased parieto-occipital activation related to false alarms. The present study intends to replicate and extend this observation by investigating common and differential activation by veridical and false recognition. Methods. – Fifteen healthy volunteers performed a verbal recognition paradigm using 160 learned target and 160 new distracter words. The subjects had to indicate whether they had learned the word before or not. Echo-planar MRI of blood-oxygen-level-dependent signal changes was performed during this recognition task. Words were classified post hoc according to the subjects’ responses, i.e. hits, false alarms, correct rejections and misses. Response-related fMRI-analysis was used to compare activation associated with the subjects’ recognition success, i.e. signal intensities related to the presentation of words were compared by the above-mentioned four response types. Results. – During recognition, all word categories showed increased bilateral activation of the inferior frontal gyrus, the inferior temporal gyrus, the occipital lobe and the brainstem in comparison with the control condition. Hits and false alarms activated several areas including the left medial and lateral parieto-occipital cortex in comparison with subjectively unknown items, i.e. correct rejections and misses. Hits showed more pronounced activation in the medial, false alarms in the lateral parts of the left parieto-occipital cortex. Conclusions. – Veridical and false recognition show common as well as different areas of cerebral activation in the left parieto-occipital lobe: increased activation of the medial parietal cortex by hits may correspond to true recognition, increased activation of the parieto-occipital cortex by false alarms may correspond to familiarity decisions. Further studies are needed to investigate the reasons for false decisions in healthy subjects and patients with memory problems. © 2003 Elsevier SAS. All rights reserved. Keywords: Activation; Memory; Episodic memory; Veridical recognition; Familiarity; Parietal cortex

1. Introduction Memory problems are common in many psychiatric disorders such as dementia, depression and schizophrenia. However, we know very little about the cerebral correlates of memory disturbances in these different patient groups. Positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) revealed activation of the following brain areas during memory retrieval: i.e. the prefrontal cortex [1,7,9,24,53,58,60,61,66], the anterior cingulate gyrus [20,24,54,71,73], the supplementary motor area * Corresponding author. E-mail address: [email protected] (R. Heun). © 2003 Elsevier SAS. All rights reserved. doi:10.1016/j.eurpsy.2003.09.005

[73], the parahippocampus [58 64], the bilateral precuneus [1,9,16,31,35,39,66], the medial and lateral parietal cortices [1,11,16,24,58,70], the occipital cortex [20,63,71], the primary visual cortex (v1 and v2, [32]), visual integration areas [53] and the cerebellum [1,16,20]. Several brain areas usually interact in different types of memory tasks ([36], for review see [8]). Only few studies investigated the effect of memory performance on the cerebral activation during encoding or retrieval of words. Kopelman et al. [31] observed increased PET activation of the right superior frontal cortex and the precuneus related to improved learning of repeated items. Nyberg et al. [43] observed a correlation of retrieval success and cerebral blood flow in left medio-temporal structures. In

R. Heun et al. / European Psychiatry 19 (2004) 42–52

reanalysis of four PET studies, Tulving et al. [69] detected various brain areas to be related to recognition success, i.e. the frontal lobe, the medial temporal lobe, the cingulate gyrus, the precuneus, the medial occipital lobe, the cerebellum and the brainstem. In an fMRI paradigm, Buckner et al. [6] found increased activation in the right anterior prefrontal cortex during successful retrieval after deep encoding of words. Another recognition paradigm examined by this group revealed, that the activation of the right anterior prefrontal cortex was delayed in comparison with other areas indicating that this activation may be related to retrieval verification [5]. A comprehensive overview on these studies was recently provided by Cabeza and Nyberg [8]. PET and fMRI studies using blocked paradigms may be confounded by differences in task severity and attentional requirements during different conditions. In contrast, eventrelated studies allow to relate the response to individual items with cerebral activation and, thus, may be less prone to such bias. In a response-related fMRI study, Brewer et al. [4] reported increased activation of frontal and parahippocampal areas for successfully encoded pictures when compared with later forgotten and less well remembered pictures. Wagner et al. [72] observed increased prefrontal and temporal activation during encoding for remembered in comparison with later forgotten words. Otten et al. [48] observed increased activation of the left anterior hippocampus and the left ventral inferior frontal gyrus for semantically encoded words which were successfully recognised. Using an event-related verbal recognition paradigm, Schacter et al. [61] tried to find activation differences between hits, false alarms, misses and correct rejections. His failure to find consistent differences may have been the consequence of the small numbers of misses and correct rejections available for comparison in their study. On the other hand, the expected activation differences between items which cannot be discriminated by the study subjects, i.e. hits and false alarms, correct rejections and misses are likely to be extremely small. Comparing hits and correct rejections, Konishi et al. [30] observed activation of the left medial and lateral parietal cortex related to successful retrieval from episodic memory. In a recent response-related fMRI study of verbal retrieval with large numbers of false decisions, we observed that false alarms (falsely identified distracter words) caused increased activations in left and right occipital cortex during retrieval in comparison with hits, misses and correct rejections [23]. The present study re-addresses this issue and investigates the differences of cerebral activation between different item categories defined by the subjects’ retrieval success. Some improvements in comparison with the previous study should increase the validity of results: (1) A non-active control condition was included to assess the validity of the design, (2) the number of fMRI acquisitions was doubled to investigate early and late parts of the hemodynamic response, (3) conjunction analysis was used to investigate common activations of subjectively known decisions, i.e. hits and false alarms, in contrast to their respective counterparts.

43

2. Methods 2.1. Subjects Fifteen volunteers (nine males, six females) were recruited from the student population and faculty of the University of Tübingen, Germany (mean age = 27.4 years; SD = 4.5 range = 21–36). Handiness was determined using the Edinburgh Handiness Inventory [45]. All subjects were right-handed, i.e. a maximum of 3 of 17 common tasks were performed with the left hand (mean = 0.4, SD = 0.9). No subject had ever changed handiness. According to medical and psychiatric history, the subject suffered neither from physical, nor from major psychiatric disorders. The study was conducted in accordance with the declaration of Helsinki and was approved by the local ethics-committee. All subjects gave informed consent for study participation. 2.2. Recognition paradigm The recognition paradigm was designed to identify response-related activation during verbal recognition. First, subjects were instructed to learn 400 words for later recognition. Hundred and sixty of these learned words were used as targets during the following retrieval task and were randomly intermixed with the other words. The 400 words were consecutively presented out-side the scanner in a quiet room on a laptop for 2 s each without any break. This high number of words and their fast presentation was chosen to achieve imperfect retrieval performance. Retrieval was examined 10–15 min later during acquisition of functional MR images when the subjects were lying supine in the MR-scanner. Three hundred and thirty two items were consecutively presented, i.e. 160 target words already presented and learned, 160 distracters, i.e. new nontarget words, and 32 non-response control conditions. Targets and distracters were matched according to imagery, concreteness and meaningfulness [2,46]. The non-response control condition consisted of the presentation of a string of seven o’s. These 352 items were presented every 10 s in a fixed random order (identical for all subjects) for 2 s each followed by a visual control condition consisting of a string of seven x for 8 s. Strings of seven letters were chosen as controls because this was the average number of letters of the target and distracter words. During retrieval, subjects had to discriminate between target and distracter words: targets had to be indicated by pressing a response button twice (defined as positive answer), distracters by pressing once with the right index finger (defined as negative answer). Response type and reaction time were documented for every individual item. Reaction times were measured from the beginning of word presentation to the beginning of the first button press. The nonresponse control condition served as baseline for the activation by different word categories, but was not used for the investigation of the main hypotheses, i.e. the direct comparison of activations by response category.

44

R. Heun et al. / European Psychiatry 19 (2004) 42–52

Four response categories were possible during recognition: (1) hits, i.e. correctly identified target words; (2) misses, i.e. missed target words incorrectly assumed to be new; (3) correct rejections, i.e. novel distracter words which were correctly rejected, and (4) false alarms, i.e. distracter words subjectively, but incorrectly assumed to be known. The distinction by response category during recognition was used for data analyses. (Hits and false alarms were related to positive responses and to subjectively known items; correct rejections and misses to negative responses and subjectively unknown items. Hits and correct rejections corresponded to correct decisions, false alarms and misses to false decisions.) 2.3. fMRI acquisition The recognition paradigm was consecutively performed in the MR-scanner. The subjects were lying supine in the scanner with a secure fixation of the head and the proximal limb to minimise involuntary movements. Words were projected on a transparent screen which could be seen via a mirror placed in front of the subject’s head. All words could be easily seen without eye movement; the angle of view was below 3.5 degree to both sides for all words. In an effort to reduce movement artefacts, subjects were instructed not to move the eyes during any of the functional measurements. Whole brain fMRI examination was performed with a commercial 1.5 Tesla whole body tomograph (Siemens Vision) using a multislice echo planar imaging sequence (EPI) with 29 axial slices (4 mm slice thickness, 1 mm gap, 64 × 64 matrix, field of view (FOV) 192 × 192 mm, echo time (TE) 46 ms, flip angle 90 degrees, measurement time 3 s [29]). The paradigm was run in six blocks of 128 repetitive fMRI measurements, the repetition intervals were (TR) 5 s. Due to technical limitations, i.e. the number of possible total brain acquisition in a run (n = 128), the continuous measurement including the full hemodynamic response was not possible at the time of the study: previous studies revealed that a large number of words was necessary to identify differences of activation by response category. Consequently, we decided to measure only two full brain acquisitions for every word during retrieval. The acquisition period for every full brain measurement started 4 and 9 s after the end of the word presentation (i.e. two scans corresponding to one item). The image acquisition between 4 and 7 s and between 9 and 12 s after the presentation of individual items was chosen to examine early and late parts of the blood oxygen level dependent (BOLD) response. The early acquisition period lay around the usual maximum of the hemodynamic response and thus was expected to be most useful to compare differential activations by response category. The fact that the different slices were measured at different time points in relation to the onset of item presentation may have slightly influenced activation patterns, but could not be excluded due to technical limitations. Extrapolation to one single time point (slice timing) was not chosen, because the slices had to be measured in a fixed sequence and because the precise

delay and course of the BOLD response function was unknown. The acquisition of the slices in a fixed sequence also prevented the analysis of a complete item-related BOLD hemodynamic response. A corresponding set of T1-weighted spin echo images (3-D MPRAGE, FOV 256 × 256 × 128, voxel size 1.5 × 1 × 1 mm) was acquired for the identification of anatomical regions. 2.4. Statistical analysis Reaction times for the four response categories were compared using ANOVA. Analysis of fMRI data was performed using Statistical Parametric Mapping (1999 version, SPM 99, [12]). Individual subjects’ functional images were corrected for motion and realigned using the first scan of each block of items as the reference. Functional scans were normalised to the EPI template of the SPM 99. Finally, the functional images were smoothed with 9 mm full width half maximum (FWHM) Gaussian filter. Contrasts between item types were calculated using the GLM by applying a boxcar function without delay or convolution with a hemodynamic response function (the delay of hrf having already been compensated for in the measurement design). A highpass filter of 200 s was applied to eliminate low-frequency drifts, and scaling to the mean intensity of each whole volume was used for global normalisation. Multiple acquisitions from a single subject were treated as repetitions. The scans were defined according to the categories in the sequence of the appearance of the corresponding stimuli in the trial. Activation associated with all four response categories during recognition were first compared with the control condition and second between different categories. To be able to detect any differences of activations related to different response categories, the study design—including number of subjects used and numbers of items in different categories—was chosen to correspond to the recent study in which we detected significant differences in activation during retrieval of words according to different categories [23,24]. For comparison of signal intensities between different items categories, an uncorrected voxel level of P < 0.01 was chosen. This level was selected because the activation differences of interest were that between items not discriminated and identically classified by the subjects (i.e. hits and false alarms; correct rejections and misses). The differences of activation during retrieval by response category are likely to be relatively small. However, to prevent false positive results, we required a number of at least 50 adjacent activated voxels to represent an activated cluster. For homogeneity in data presentation, we present all results using these criteria (Figs 1 and 2), even though all activations by words in comparison with the control condition easily reached z-values above P = 0.001 (see Table 1) which is the common standard in most fMRI studies. To identify common activations by several item categories, conjunction analysis as implemented in SPM 99 was used [52]. Activations were displayed on surface rendered standard brains (SPM 99, [12]).

R. Heun et al. / European Psychiatry 19 (2004) 42–52

45

Fig. 1. Cortical projections of cerebral activation during verbal retrieval by different item categories in comparison with the control condition: hits (1a), misses (1b), correct rejections (1c) and false alarms (1d). (Threshold P < 0.01, >50 adjacent voxels, columns from left to right: right lateral, left lateral, left medial, right medial view).

3. Results 3.1. Response rates during recognition The following response rates were observed in the 15 subjects during recognition: hits 92.20 ± 26.23 (mean ± SD), misses 53.27 ± 27.22, correct rejections 100.93 ± 26.75, false alarms 41.47 ± 22.61. (The non-response rate was below 10% in all subjects.) Visual inspection of the subjects’ complete response sequences showed that there was no indication for a significant primacy or recency effect, or for a better recognition of specific items. Reaction times were different for the four response categories, i.e. hits 0.49 ± 0.06 s (mean ± SD), misses 0.53 ± 0.06 s, correct rejections 0.52 ± 0.06 s, false alarms 0.51 ± 0.07 s (ANOVA with response category as independent within-subject factor, F = 17.23, df = 3, P = 0.0001), but did not significantly change over time (P > 0.2). Secondary comparisons using Scheffé-tests revealed that hits were answered significantly faster than all other categories (misses, correct rejections, false alarms) (Scheffé-test, all P < 0.01). Words of different

response categories differed according to imagery, concreteness, meaningfulness and relative frequency in the German language (ANOVA F = 6.2, df = 3 P = 0.0001; F = 6.0, df = 3, P = 0.0001; F = 2.8, df = 3 P = 0.040; F = 23.2, df = 3 P = 0.0001). Secondary analysis revealed that false alarms had lower scores on imagery and concreteness, and were reported to be less frequent in the German language than hits and correct rejections (all P < 0.001); meaningfulness scores were lower for false alarms and misses than for hits and correct rejections without reaching significance (all P > 0.1). It should be mentioned that the subjects’ responses were identical for hits and false alarms, and for misses and correct rejections, respectively. Visual inspection of the responses over time in all individual subjects revealed that items of different categories were continuously distributed during the trial period. Consequently, it is unlikely that item sequence effects (i.e. primary or recency effects) should have significantly influenced the fMRI results. In addition, there was no consistent reduction or increase of the reaction times during retrieval in the 15 subjects indicating a relative stability of attention over the trial period.

46

R. Heun et al. / European Psychiatry 19 (2004) 42–52

Fig. 2. Cortical projections of cerebral activation during verbal retrieval by different item categories in comparison with other item categories: activation of subjectively known vs. unknown items (i.e. conjunction analysis of hits vs. misses and false alarms vs. correct rejections (2a), activation by hits vs. misses (both item types learned, 2b), activation by false alarms vs. correct rejections (both item types new, 2c), activation by hits vs. false alarms (both item types subjectively known, 2d), activation by false alarms vs. hits (2e) and activation by subjectively new vs. known items (i.e. conjunction analysis of misses vs. hits and correct rejections vs. false alarms, 2f). (Threshold P < 0.01 and >50 adjacent activated voxels (Fig. 2b–e); for statistical reasons, conjunction analysis (Fig. 2 a,f), as implemented in SPM 99 does not allow to exclude clusters consisting of less than a specified number of voxels, thus the presentation of results of conjunction analyses included clusters 50 voxels even though these are not further interpreted; columns from left to right: right lateral, left lateral, left medial, right medial view).

R. Heun et al. / European Psychiatry 19 (2004) 42–52

47

Table 1 Cerebral activations in 15 healthy subjects comparing response-related activation with the non-response baseline Contrast

Brain region

Hits-baseline (Fig. 1a)

Left inferior frontal gyrus and left precentral gyrus Right inferior frontal gyrus Supplementary motor area

Misses-baseline (Fig. 1b)

Correct rejections-baseline (Fig. 1c)

False alarms-baseline (Fig. 1d)

Bilateral occipital lobe, bilateral inferior temporal gyrus, brainstem Left inferior frontal gyrus Right inferior frontal gyrus Supplementary motor area Bilateral occipital lobe and inferior temporal gyrus Brain stem Left inferior frontal gyrus Right inferior frontal gyrus Supplementary motor area Bilateral occipital lobe and inferior temporal gyrus Brainstem Left inferior frontal gyrus Right inferior frontal gyrus Supplementary motor area Bilateral occipital lobe and left inferior temporal gyrus Brainstem

Co-ordinates of maximum voxel in cluster x, y, z (mm) –36, 24, –9

Maximal activation in cluster z-value

Statistical significance uncorrected P-value (corrected P-value) 0.000 (0.000)

Infinity

Number of voxel in this cluster (n) with P< 0.01 Overlapping with occipital lobe cluster 1503 Overlapping with occipital lobe cluster 8810

33, 24, –12 0, 24, 42

7.40 6.56

3, –72, 9

–33, 24, –6 30, 24, –12 3, 24, 42 –33, –90, –3

5.52 4.97 6.60 6.11

924 229 735 3134

0.000 (0.000) 0.000 (0.003) 0.000 (0.000) 0.000 (0.000)

9, –21, –15 –51, 27, –6 30, 24, –12 –3, 15, 48 –30, –93, –3

3.35 7.50 5.53 6.67 6.84

152 1334 207 694 3045

0.000 (0.675) 0.000 (0.000) 0.000 (0.000) 0.000 (0.000) 0.000 (0.000)

6, –21, –21 –48, 27, –9 30, 24, –12 –3, 15, 48 –33, –90, –3

4.04 4.97 3.52 4.81 6.60

201 936 92 285 1623

0.000 (0.119) 0.000 (0.003) 0.000 (0.496) 0.000 (0.006) 0.000 (0.000)

0, –21, –18

3.46

303

0.000 (0.563)

Infinity

0.000 (0.000) 0.000 (0.000) 0.000 (0.000)

3.2. Activation during retrieval of words

3.3. Activation differences by word categories

All word categories in comparison with the control condition consistently activated several brain areas: the left inferior frontal gyrus including Broca’s area, the right inferior frontal gyrus (the right hemispheric equivalent of Broca’s area), left precentral gyrus, the supplementary motor area (SMA), the bilateral occipital lobe, the inferior temporal gyrus, and the brainstem (see Table 1 and Fig. 1, row a–d). The activation by hits in comparison with the control condition seemed more extended than that by other categories. This difference in the extent of activation in comparison with the control condition might be relevant only in relation to the activation by correct rejections which consists of an equivalent number of items. The increased activations of hits in comparison with the activation by false alarms and misses are likely to be the result of different numbers of items included in the respective comparisons. The reported activations were based on the early full brain acquisitions 4–7 s after word presentations. The activations resulting from the second set of later acquisitions (9–12 s after presentation) were much less intense, but did not reveal any additional significant activation. Consequently, they were not used for further analysis. The observed higher activation at the short delay was in good agreement with recent studies showing a peak of activation around 4 s after stimulation, but no relevant activation after 8–9 s [13,55], even though there is some interindividual variation [13].

A conjunction analysis referring to decisions for subjectively known words included the comparison of activation by false alarms vs. correct rejections (new words in both item categories) and by hits vs. misses (both previously presented and learned words): subjectively known words (vs. subjectively unknown items) activated the left precentral gyrus, the anterior cingulate gyrus, the left lateral parietal cortex, the left medial parieto-occipital cortex and the brainstem (see Table 2 and Fig. 2a). The equivalent, but less powerful individual comparisons showed consistent results: hits in comparison with misses (both learned words) showed increased activation of the left inferior frontal gyrus (including Broca’s area), the left prefrontal gyrus, the medial parietal and parieto-occipital cortex (see Table 1 and Fig. 2b); false alarms in comparison with correct rejections (both new words) showed increased activation in the left lateral parietooccipital cortex (see Table 2 and Fig. 2c). These observations also reflect the differences between activations by hits and false alarms observed in their direct comparison: hits in comparison with false alarms showed a small activation in the medial parietal cortex (see Table 2 and Fig. 2d), false alarms in comparison with hits (identical subjective decisions) revealed an activation in the parieto-occipital cortex (see Table 2 and Fig. 2e). In contrast to these observations, subjectively unknown decisions, i.e. misses and correct rejections both did not show any increased activation in com-

48

R. Heun et al. / European Psychiatry 19 (2004) 42–52

Table 2 Cerebral activation in 15 healthy subjects directly comparing different response-related activation Contrast

Conjunction for subjectively known decisions: hit vs. misses, and false alarms vs. correct rejections (see Fig. 2a) Hits vs. misses (Fig. 2b)

False alarms vs. correct rejections (Fig. 2c) Hits vs. false alarms (Fig. 2d) False alarms vs. hits (Fig. 2e) Conjunction analysis for subjectively unknown decisions: misses vs. hits, and correct rejections vs. false alarms (Fig. 2f)

Brain region

Co-ordinates of max. voxel in cluster X, y, z (mm) Left precentral gyrus and left –39, –69, 39 parietal lobe Anterior cingulate gyrus 0, 36, 3 Medial parieto-occipital –18, –54, 9 cortex Brainstem –15, 3, 3 Left inferior frontal gyrus –45, 21, –12 Left precentral gyrus –42, –18, 57 Bilateral medial 3, –57, 6 parieto-occipital lobe Left medial parietal cortex –6, –33, 36 Brainstem –6, –9, 3 Left lateral parieto-occipital –33, –66, 39 cortex

Maximal activation in cluster z-value 3.85

Number of voxel in this cluster (n) with P< 0.01 664

Statistical significance uncorrected P-value (corrected P-value) 0.000 (0.340)

3.00 2.94

53 84

0.001 (0.990) 0.002 (0.995)

3.28 4.24 3.22 3.70

266 148 88 422

0.001 (0.902) 0.000 (0.059) 0.001 (0.799) 0.000 (0.332)

3.22 2.90 3.05

58 75 497

0.001 (0.804) 0.002 (0.972) 0.001 (0.919)

Bilateral medial parietal cortex Left lateral parieto-occipital cortex Left occipital lobe Right medial frontal gyrus

–3, –51, 36

2.74

60

0.003 (0.993)

–33, –60, 18

2.96

147

0.002 (0.955)

–33, –90, –3 21, 30, 36

3.71 3.08

55 110

0.000 (0.324) 0.001 (0.979)

parison with hits or false alarms (all individual comparisons, P > 0.01, uncorrected voxel level). The conjunction analysis of misses vs. hits (both item types previously presented and learned) and correct rejections vs. false alarms (both item types new) relating to the subjectively unknown (negative) decisions only revealed a small significant activation in the right medial frontal gyrus (see Table 2 and Fig. 2f) which could not be supported by other coherent observations.

4. Discussion 4.1. Activation during verbal retrieval The main focus of the present study is the common and differential activation by different item categories, i.e. hits, false alarms, correct rejection and misses (see next paragraph). However, for ease of understanding, the comparisons with the control condition are described first. The observed activations can be interpreted reasonably on the basis of previous publications, even though all interpretations must be seen as speculative in the present complex paradigm: Verbal retrieval of all item categories in comparison with the control condition consistently activated brain areas possibly related to specific cognitive functions (see Fig. 1 and Table 1). (1) Activation of the left and right inferior frontal gyrus (including Broca’s area and its right hemispheric equivalent) may be related to language processing

[3,25,57,70]; (2) The supplementary motor area (SMA), the anterior cingulate and the left precentral gyrus are likely to be involved in selection, preparation and conduction of the motor response [24,44]; (3) The bilateral occipital lobe and the inferior temporal gyrus may be involved in visual perception, object identification and recognition [18,19,28,37]; (4) the observed brainstem activation might be related to attention control and keeping of awareness [49,65]. In summary, the observed activations are in good agreement with the expected task-related cognitive activities thus supporting the validity of the fMRI measurements. 4.2. Activation during veridical and illusory recognition of words Subjectively known item (i.e. hits and false alarms) revealed a stronger activation than subjectively unknown items (misses and correct rejections) in the left prefrontal gyrus, the left medial and lateral parieto-occipital cortex, and the brainstem (see Fig. 2 and Table 2). This observation in the first conjunction analysis (Fig 2a) was supported by corresponding activations in individual comparisons (between hits and misses, Fig. 2b, and between false alarms and correct rejections (Fig. 2c). However, these comparisons also showed some differences that were supported by direct comparisons: hits showed more pronounced activations of the left Broca’s area of the inferior frontal gyrus, the medial parietal and parietal-occipital cortex (in comparison with misses, i.e. both

R. Heun et al. / European Psychiatry 19 (2004) 42–52

item types learned, Fig 2b, and in comparison with false alarms, identical decisions, Fig. 2d). In contrast, false alarms show a more pronounced activation of the lateral parietooccipital cortex (in comparison with correct rejections and hits). Again, the observed activations make sense: the activation of the left precentral gyrus by subjectively known decisions is likely to be related to the more intense motor response by double vs. single button clicks conducted with the left index finger (observed in conjunction analysis and in the comparison between hits and misses). This activation can be seen as a relevant internal control and supports the validity of the results. Activation of left Broca’s area in the inferior frontal gyrus is most likely related to increased language processing in subjectively known decisions even though there are no previous experiences supporting this speculation. Activation of the left parieto-occipital cortex by hits and false alarms may be related to increased memory recall as indicated by recent publications supporting a relevance of this area for memory performance (see Section 1). In good agreement with our study, others [30,31,69] observed that the activation of the left lateral and medial parietal cortex is related to the retrieval success. Activation of the lateral part of the parietooccipital cortex by subjectively known decisions may also be related to increased subjective associations by hits and false alarms in comparison with misses and correct rejections. In agreement, Tulving et al. [70] reported left parietal activation during auditory sentence recognition, Osherson et al. [47] observed an involvement of parieto-occipital brain areas in deductive reasoning, Grill-Spector et al. [19] showed a correlation of object recognition performance with activation in the temporo-occipital cortex. In the present study, false alarm responses showed an increased activation in the left parieto-occipital lobe in comparisons with correct rejections and hits. A comparable activation peak by false alarms in comparison with hits has already been described [23], even though the observed activations related to false alarms were less focused in the present study. Hits in comparison with false alarms showed an increased activation of the medial parietal cortex, which may be relevant for memory retrieval. Differences between veridical and false recognition of words have also been examined by other authors. Schacter et al. [63] observed activation of the temporoparietal region during veridical vs. false recognition. An event-related fMRI study [61] revealed activation in the anterior cingulate gyrus, premotor cortex, medial and lateral parietal cortex, bilateral anterior prefrontal cortex, bilateral frontal opercular cortex and visual cortex during true and false recognition compared to fixation. However, direct comparisons by response categories revealed no significant activations which might be the consequence of the small numbers of misses and correct rejections used in their study. Separate analyses revealed a delayed onset of anterior prefrontal activity during recognition. In the present study, the time course of activation was not investigated since all available fMRI acquisitions were used to present a maximum

49

of different items. In agreement with our observations, a recent study using event-related potentials reported differences between hits and false alarms in the positivity after 600–900 ms in posterior brain areas [15]. In the present and the recent study [23], false alarms decision were more difficult than hits as indicated by longer reaction times. False alarms had lower scores on imagery and concreteness than other categories, and were less frequent in the German language than hits and correct rejections. Thus, the activation by false alarm responses vs. hits in the lateral parieto-occipital cortex could be the result of the increased processing needs [21]. These activation differences may also indicate that veridical and false recognition are based on different cognitive processes, even though these may not be aware to the subjects. Mather et al. [40] described that characteristics of false memories were different from correct ones, i.e. less precise and vivid. Recognition may be divided in recollection, an intentional use of memory, and the use of familiarity, often a more automatic non-critical use in circumstances such as reduced attention and memory overload [27,33,74]. Norman and Schacter [42] indicated that both veridical and false recognition memory is usually composed of associative information in contrast to sensory and contextual detail. Schacter et al. [59,61,62] proposed that the false recollection of the new items might represent as a source monitoring deficit. Thus, false alarm decisions may be the consequence of falsely assumed familiarity, but less the result of conscious recollection [33]. In agreement, Tulving et al. [71] observed that activation of the striate and prestriate cortex (and in other areas including the anterior cingulate gyrus) is related to familiarity of items; Saykin et al. [58] observed that familiar words activated the right prefrontal cortex, the posterior left parahippocampal gyrus, the left medial parietal cortex and the right superior temporal gyrus. Henson et al. [21] reported activation of the left prefrontal and parietal cortices either by recollection of learned and by subjectively familiarity of learned words. In agreement with our study, recollection decisions for studied words were associated with enhanced activation of the left prefrontal, left parietal and posterior cingulate cortex and precuneus; familiarity decisions were associated with enhanced responses in lateral and medial prefrontal regions. Consequently, veridical decisions related to activation in the medial parieto-occipital cortex might be the consequence of true recognition; false alarm decisions related to activation of the lateral part of the parietal cortex may be the consequence of subjective familiarity of items. However, these results does not clarify whether the observed increased activation is the cause or the consequence of the erroneous decision. Further studies are needed to investigate this issue. 4.3. Limitations and conclusions In contrast to other authors, we could not detect significant activation of several brain areas during recognition: there are several lines of evidence that the prefrontal cortex is involved

50

R. Heun et al. / European Psychiatry 19 (2004) 42–52

in successful encoding and retrieval (e.g. electrophysiology, [38,56]; lesion studies, [50,51]; PET, [10,17,41,70]; fMRI, [22,26]; in a multistudy analysis, Lepage et al. [34] identified five prefrontal areas related to the retrieval mode, i.e. the left and right frontal poles, the frontal operculum (Brodman area 47/45) and the right dorsolateral area (BA/8/9) and the anterior cingulate gyrus; Gabrieli et al. [14] and Stark and Squire [67,68] reported bilateral hippocampal activity during verbal and object recognition. In the present study, prefrontal and hippocampal activation by different item categories might be either equivalent or not sufficiently different to be detected. On the other hand, the fact that some item types, i.e. false alarms and misses, were less frequent than others might lead to an underestimation of activation related to these response categories. It should be mentioned that other authors have also failed to detect any differential activations by response category [61]. These caveats in mind, the present study is one of the first to show different activation during retrieval of differently processed items using a response-related functional MRI design. This was the case even though the subjects’ responses were identical for hits and false alarms and were likely to be unaware to the subjects. Further studies are necessary to investigate the rationale for false retrieval decisions in healthy subjects and psychiatric patients with different types of memory problems. False memories might be relevant in schizophrenia as well as in depressive disorders.

[7]

[8]

[9]

[10] [11]

[12]

[13] [14]

[15]

[16]

[17]

[18]

Acknowledgements We thank Anja Höpfner for skilled technical assistance and Prof. Dr. Lars Nyberg, Umea, Sweden for helpful discussions concerning the data analysis. The authors received the young investigator award of the European Association of Psychiatrists 2002 for a poster presenting the data.

[19]

[20]

[21]

References [22] [1]

[2]

[3]

[4]

[5]

[6]

Andreasen NC, O’Leary DS, Arndt S, Cizadlo T, Hurtig R, Rezai K, et al. Short-term and long-term verbal memory. A positron emission tomography study. Proceedings of the National Academy of Science USA 1995;92:5111–5. Baschek IL, Bredenkamp J, Oehrle B, Wippich W. Bestimmung der Bildhaftigkeit (I), Konkretheit (C) und der Bedeutungshaltigkeit (m’) von 800 Substantiven. Zeitschrift für experimentelle angewandte Psychologie 1977;24:353–96. Bookheimer SY, Zeffiro TA, Blaxton T, Gaillard W, Theodore W. Regional cerebral blood flow during object naming and word reading. Human Brain Mapping 1995;3:93–106. Brewer JB, Zhao Z, Desmond JE, Glover GH, Gabrieli JDE. Making memories: brain activity that predicts how well visual experience will be remembered. Science 1998;281:1185–7. Buckner RL, Koutstaal W, Schacter DL, Dale AM, Rotte M, Rosen BR. Functional-anatomic study of episodic retrieval. II. Selective averaging of event-related fMRI trials to test the retrieval success hypothesis. Neuroimage 1998;7:163–75. Buckner RL, Koutstaal W, Schacter DL, Wagner AD, Rosen BR. Functional-anatomic study of episodic retrieval using fMRI. I. Retrieval effort versus retrieval success. Neuroimage 1998;7:151–62.

[23]

[24]

[25]

[26]

[27]

[28]

Busatto G, Howard RJ, Ha Y, Brammer M, Wright I, Woodruff PWR, et al. A functional magnetic resonance imaging study of episodic memory. NeuroReport 1997;8:2671–5. Cabeza R, Nyberg L. Imaging cognition II: an empirical review of 275 PET and fMRI studies. Journal of Cognitive Neuroscience 2000; 12:1–47. Fletcher PC, Frith CD, Grasby PM, Shallice T, Frackowiak RSJ, Dolan RJ. Brain systems for encoding and retrieval of auditory-verbal memory. An in vivo study in humans. Brain 1995;118:401–16. Fletcher PC, Frith CD, Rugg MD. The functional neuroanatomy of episodic memory. Trends in Neuroscience 1997;20:213–8. Fletcher PC, Shallice T, Frith CD, Frackowiak RSJ, Dolan RJ. The functional roles of prefrontal cortex in episodic memory. II. Retrieval Brain 1998;121:1249–56. Friston KJ. Statistical parametric mapping and other analyses of functional imaging data. In: Toga AW, Mazziotta JC, editors. Brain mapping. The methods. San Diego, New York: Academic Press; 1996. p. 363–86. Friston KJ, Josephs O, Rees G, Turner R. Nonlinear event-related responses in fMRI. Magnetic Resonance in Medicine 1998;39:41–52. Gabrieli JDE, Brewer JB, Desmond JE, Glover GH. Separate neural bases of 2 fundamental memory processes in the human medialtemporal lobe. Science 1997;276:264–6. Gonsalves B, Paller KA. Neural events that underlie remembering something that never happened. Nature Neuroscience 1997:31316– 21. Grasby PM, Frith CD, Friston KJ, Bench C, Frackowiak RSJ, Dolan RJ. Functional mapping of brain areas implicated in auditory—verbal memory function. Brain 1993;116:1–20. Grasby PM, Frith CD, Friston KJ, Simpson J, Fletcher PC, Frackowiak RSJ, et al. A graded task approach to the functional mapping of brain areas implicated in auditory-verbal memory. Brain 1994;117: 1271–82. Grill-Spector K, Kushnir K, Edelman T, Itzchak S, Malach YR. Cue-invariant activation in object-related areas of the human occipital lobe. Neuron 1998;21:191–201. Grill-Spector K, Kushnir T, Hendler T, Malach R. The dynamics of object-selective activation correlate with recognition performance in humans. Nature Neuroscience 2000;3:837–43. Haxby JV, Ungerleider LG, Horwitz B, Maisog JM, Rapoport SI, Grady CL. Face encoding and recognition in the human brain. Proceedings of the National Academy of Sciences USA 1996;93:922–7. Henson RNA, Rugg MD, Josephs O, Dolan RJ. Recollection and familiarity in recognition memory: an event-related functional magnetic resonance imaging study. The Journal of Neuroscience 1999;10: 3962–72. Henson RNA, Rugg MD, Shallice T, Dolan RJ. Confidence in recognition memory for words: dissociating right prefrontal roles in episodic retrieval. Journal of Cognitive Neuroscience 2000;12:913–23. Heun R, Jessen F, Klose U, Erb M, Granath D-O, Grodd W. Responserelated fMRI analysis during encoding and retrieval revealed differences in cerebral activation by retrieval success. Psychiatry Research: Neuroimaging 2000;99:137–50. Heun R, Klose U, Jessen F, Erb M, Papassotiropoulos A, Lotze M, et al. Functional MRI of cerebral activation during encoding and retrieval of words. Human Brain Mapping 1999;8:157–69. Hirano S, Kojima H, Naito Y, Honjo I. Cortical speech processing mechanisms while vocalizing visually-presented languages. Neuroreport 1996;8:363–7. Iidaka T, Sadato N, Yamada H, Yonekura Y. Functional asymmetry of human prefrontal cortex in verbal and non-verbal episodic memory as revealed by fMRI. Cognitive Brain Research 2000;9:73–83. Jacoby LL. A process dissociation framework: separating automatic from intentional uses of memory. Journal of Memory and Language 1991;30:513–41. Kanwisher N, McDermott J, Chun MM. The fusiform face area: a module in human extrastriate cortex specialized for face perception. Journal of Neuroscience 1997;11:4302–11.

R. Heun et al. / European Psychiatry 19 (2004) 42–52 [29] Klose U, Erb M, Wildgruber D, Müller E, Grodd W. Improvement of the acquisition of large amount of mr-images on a conventional whole body system. Magnetic Resonance Imaging 1999;17:471–4. [30] Konishi S, Wheeler ME, Donaldson DI, Buckner RL. neural correlates of episodic retrievalo success. NeuroImage 2000;12:276–86. [31] Kopelman MD, Stevens TG, Foli S, Grasby P. PET activation of the medial temporal lobe in learning. Brain 1998;121:875–87. [32] Le Bihan D, Turner R, Zeffiro TA, Cuénod CA, Jezzard P, Bonnerot V. Activation of human primary visual cortex during visual recall: a magnetic resonance imaging study. Proceedings of the National Academy of Science USA 1993;90:11802–5. [33] LeCompte DC. Recollective experience in the revelation effect: separating the contributions of recollection and familiarity. Memory & Cognition 1995;23:324–34. [34] Lepage M, Ghaffar O, Nyberg L, Tulving E. Prefrontal cortex and episodic memory retrieval mode. Proceedings of the National Academy of Science USA 2000;97:506–11. [35] Maguire EA, Frith CD, Morris RGM. The functional neuroanatomy of comprehension and memory: the importance of prior knowledge. Brain 1999;122:1839–50. [36] Maguire EA, Mummery CJ. Differential modulation of a common memory retrieval network revealed by positron emission tomography. Hippocampus 1999;9:54–61. [37] Malach R, Reppas JB, Benson RR, Kwong KK, Jiang H, Kennedy WA, et al. Object-related activity revealed by functional magnetic resonance imaging in human occipital cortex. Proceedings of the National Academy of Science USA 1995;2:8135–9. [38] Mark RE, Rugg MD. Age effects on brain activity associated with episodic memory retrieval. An electrophysiological study. Brain 1998;121:861–73. [39] Markowitsch HJ, Calabrese P, Fink GR, Durwen HF, Kessler J, Härting C, et al. Impaired episodic memory retrieval in a case of probable psychogenic amnesia. Psychiatry Research: Neuroimaging Section 1997;74:119–26. [40] Mather M, Henkel LA, Johnson MK. Evaluating characteristics of false memories: remember/know judgments and memory characteristics questionnaire compared. Memory & Cognition 1997;25:826–37. [41] McIntosh AR, Nyberg L, Bookstein FL, Tulving E. Differential functional connectivity of prefrontal and medial temporal cortices during episodic memory retrieval. Human Brain Mapping 1997;5:323–7. [42] Norman KA, Schacter DL. False recognition in younger and older adults: exploring the characteristics of illusory memories. Memory & Cognition 1997;25:838–48. [43] Nyberg L, McIntosh AR, Houle S, Nilsson L-G, Tulving E. Activation of medial temporal structures during episodic memory retrieval. Nature 1996;380:715–7. [44] Ojemann JG, Buckner RL, Corbetta M, Raichle ME. Imaging studies of memory and attention. Neurosurgery Clinics of North America 1997;8:307–19. [45] Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 1971;9:97–113. [46] Oldigs-Kerber J, Adamus WS, Kitzinger M. Zur Beeinflussung von verbalen Lern- und Gedächtnisprozessen durch Anticholinergica am Beispiel von Scopolamin. Ein pharmakopsychologischer Beitrag für die neuropsychologische Praxis. Zeitschrift für Neuropsychologie 1991;2:29–40. [47] Osherson D, Perani D, Cappa S, Schnur T, Grassi F, Fazio F. Distinct brain loci in deductive versus probabilistic reasoning. Neuropsychologia 1998;36:369–76. [48] Otten LJ, Henson RNA, Rugg MD. Depth of processing effects on neural correlates of memory encoding. Relationship between findings from across- and within-task comparisons. Brain 2001;124:399–412. [49] Perry E, Walker M, Grace J, Perry R. Acetylcholine in mind: a neurotransmitter correlate of consciousness? Trends in Neuroscience 1999;22:273–80. [50] Petrides M. Visuo-motor conditional associative learning after frontal and temporal lesions in the human brain. Neuropsychologia 1997;35: 989–97.

51

[51] Petrides M. Specialized systems for the processing of mnemonic information within the primate frontal cortex. Philosophical Transactions of the Royal Society of London B 1996;351:1455–62. [52] Price CJ, Friston KJ. Cognitive conjunction: a new approach to brain activation experiments. Neuroimage 1997;5:261–70. [53] Ragland JD, Gur RC, Lazarev MG, Smith RJ, Schroeder L, Raz J, et al. Hemispheric activation of anterior and inferior prefrontal cortex during verbal encoding and recognition: a PET study of healthy volunteers. NeuroImage 2000;11:624–33. [54] Riddle W, O’Carroll RE, Dougall N, Van Beck M, Murray C, Curran SM, et al. A single photon emission computerised tomography study of regional brain function underlying verbal memory in patients with Alzheimer-type dementia. British Journal of Psychiatry 1993; 163:166–72. [55] Rosen BR, Buckner RL, Dale AM. Event-related functional MRI: past, present, and future. Proceedings of the National Academy of Science USA 1998;95:773–80. [56] Rugg MD, Fletcher PC, Frith CD, Frackowiak RSJ, Dolan RJ. Differential activation of the prefrontal cortex in successful and unsuccessful memory retrieval. Brain 1996;119:2073–83. [57] SakuraiY, Momose T, Iwata M, Watanabe T. Semantic process in kana word reading: activation studies with positron emission tomography. Neuroreport 1993;4:327–30. [58] Saykin AJ, Johnson SC, Flashman LA, McAllister TW, Sparling M, Darcey TM, et al. Functional differentiation of medial temporal and frontal regions involved in processing novel and familiar words: an fMRI study. Brain 1999;122:1963–71. [59] Schacter DL. Illusory memories: a cognitive neuroscience analysis. Proceedings of the National Academy of Science USA 1996;93: 13527–33. [60] Schacter DL, Alpert NM, Savage CR, Rauch SL, Alberts MS. Conscious recollection and the human hippocampal formation: evidence from positron emission tomography. Proceedings of the National Academy of Science USA 1996;3:321–5. [61] Schacter DL, Buckner RL, Koutstaal W, Dale AM, Rosen BR. Late onset of anterior prefrontal activity during true and false recognition: an event-related fMRI study. Neuroimage 1997;6:259–69. [62] Schacter DL, Koutstaal W, Johnson MK, Gross MS, Angell KE. False recollection induced by photographs: a comparison of older and younger adults. Psychology and Aging 1997;12:203–15. [63] Schacter DL, Reiman E, Curran T, Yun LS, Bandy D, McDermott KB, et al. Neuroanatomical correlates of veridical and illusory recognition memory: evidence from positron emission tomography. Neuron 1996;17:267–74. [64] Schacter DL, Reiman E, Uecker A, Polster MR, Yun LS, Cooper AL. Brain regions associated with retrieval of structurally coherent visual information. Nature 1995;376:587–90. [65] Sewards TV, Sewards MA. Visual awareness due to neuronal activities in subcortical structures: a proposal. Consciousness and Cognition 2000;9:86–116. [66] Shallice T, Fletcher P, Frith CD, Grasby P, Frackowiak RSJ, Dolan RJ. Brain regions associated with acquisition and retrieval of verbal episodic memory. Nature 1994;368:633–5. [67] Stark CEL, Squire LR. fMRI activity in the medial temporal lobe during recognition memory as a function of study-test interval. Hippocampus 2000;10:329–37. [68] Stark CEL, Squire LR. Functional magnetic resonance imaging (fMRI) activity in the hippocampal region during recognition memory. The Journal of Neuroscience 2000;20:7776–81b. [69] Tulving E, Habib R, Nyberg L, Lepage M, McIntosh AR. Positron emission tomography correlations in and beyond medial temporal lobes. Hippocampus 1999;9:71–82. [70] Tulving E, Kapur S, Markowitsch HJ, Craik FIM, Habib R, Houle S. Neuroanatomical correlates of retrieval in episodic memory: auditory sentence recognition. Proceedings of the National Academy of Science 1994;91:2012–5.

52

R. Heun et al. / European Psychiatry 19 (2004) 42–52

[71] Tulving E, Markowitsch HJ, Craik FIM, Habib R, Houle S. Novelty and familiarity activations in PET studies of memory encoding and retrieval. Cerebral Cortex 1996;6:71–9. [72] 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. Science 1998;281: 1188–91.

[73] Warburton E, Wise RJS, Price CJ, Weiller C, Hadar U, Ramsay S, Frackowiak RSJ. Noun and verb retrieval by normal subjects. Studies with PET. Brain 1996;119:159–79. [74] Yonelinas AP, Jacoby LL. Noncriterial recollection: familiarity as automatic, irrelevant recollection. Consciousness and Cognition 1996;5:131–41.