- Email: [email protected]
Learning face-name associations and the effect of age and performance: a PET activation study
Neuropsychologia 39 (2001) 643– 650 www.elsevier.com/locate/neuropsychologia
Learning face-name associations and the effect of age and performance: a PET activation study Karl Herholz a,*, Patricia Ehlen a, Josef Kessler a, Timm Strotmann a, Elke Kalbe a, Hans-Joachim Markowitsch b a
Max-Planck-Institut fu¨r neurologische Forschung and Neurologische Uni6ersita¨tsklinik, Gleuler Str. 50, D-50931 Cologne, Germany b Physiologische Psychologie, Uni6ersita¨t Bielefeld, Germany Received 27 January 2000; received in revised form 30 August 2000; accepted 16 October 2000
Abstract Learning face-name associations is a complex task to be mastered in every day life that approaches the limits of cognitive capacity in most normal humans. We studied brain activation during face-name learning using positron emission tomography (PET) in 11 normal volunteers. The most intense activation was seen in occipital association cortex (BA 18) bilaterally, also involving lingual and fusiform gyrus (BA 37). In the left hemisphere additional activation were located in inferior temporal gyrus, the inferior part of pre- and postcentral gyrus, and orbitofrontal cortex (BA 11), whereas in the right hemisphere only a region in the precuneus (BA 19) was activated additionally. There was considerable interindividual variation of encoding success, which was significantly related to activation of BA 18 bilaterally. Subject ages covered a range of 26 – 72 years, but — in contrast to the effect of encoding success — there was no significant age effect on activations. Task-independent habituation effects were seen in cerebellum and left middle temporal gyrus. These results indicate that the intensity of information processing in ventral occipital association cortex is most important for success of face-name encoding. Learning is further mediated by a predominantly left-hemispheric network including inferior temporal and orbitofrontal cortex. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: PET; Activation; Face-name association; Learning; Ageing
1. Introduction Memory for faces is an essential cognitive ability and is supported by specialised neuroanatomical structures in ventral temporo-occipital association cortex. Faces can be associated with names. Encoding and retrieval of the names associated with other persons’ faces is a daily task that is mastered with less than perfect success by most normal subjects. Although a huge number of such associations can be stored during life, and often can be retrieved after several decades, the task also regularly seems to reach the limits of human cognitive abilities if one has brief contact with many people. With advancing age, the transfer of new information into long-term memory and retrieval of recently encoded information, such as proper names may become * Corresponding author. Tel.: +49-221-4726229; fax: + 49-2214726298. E-mail address: [email protected] (K. Herholz).
particularly difficult (see recent review by Grady and Craik [17]). The purpose of the present study was to investigate activations during a learning task that was close to daily-life reality in order to provide a baseline for the understanding of memory deficits, as they are experienced by elderly subjects and patients with memory impairment. In addition, face memory association tasks appear to be particularly well suited because of their ecological validity. They have, therefore, been included in behavioural memory tests, such as the Rivermead Behavioural Memory Test [56]. Some previous studies indicated that memory for faces and words involves different neural substrates that are located predominantly in the right and the left hemisphere, respectively [28,48], but may also share a common modality-independent semantic system [16]. The combined associative processing of faces and names has been studied previously mainly in the context of recognition of famous faces [8,16,46]. In addi-
0028-3932/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 8 - 3 9 3 2 ( 0 0 ) 0 0 1 4 4 - 5
644
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
tion to the activation of ventral temporo-occipital cortex by face processing, bilateral fronto-temporal regions have been activated in these studies. Correspondingly, fronto-temporal brain damage impairs memory for past events and famous faces [29]. Yet, the areas activated by recognition of famous faces appear to be different from those activated by faces that have been encoded only recently [31]. Impairment of recognition is not the only possible reason for impairment of learning, which may also be due to impairment of encoding. We, therefore, wanted to study the activation results during encoding of facename associations and compare the results with those of previous functional imaging studies in which encoding of faces or words was performed separately [13,21,26,30,34,43,53]. From these studies, we expected that ventral occipital association cortex, hippocampus, and frontal association cortex would be among the activated areas in a face-name association learning task. We also wanted to analyse the effects of age and performance on activated areas.
2. Subjects and method Eleven normal right-handed volunteers, eight men and three women, were studied. Ages covered a broad range from 26 to 72 years (mean 46.7917.6). Subjects had no history of neurological or psychiatric disease and a normal neurological status. They were tested neuropsychologically with the abbreviated Wechsler-Intelligence-Test (WIP) [7], the Wechsler-Memory-Test revised (WMS-r) [22], and the Recognition-MemoryTest for faces (RMT) [54]. All underwent a PET activation study with a learning task. Magnetic resonance imaging (MRI) of the brain excluded structural abnormalities and was used for anatomical localisation of activated areas. (MRI could not be performed in one subject because of claustrophobia). All subjects gave their informed consent, and the study was approved by the ethics committee of the medical faculty.
The neuropsychological profile of participating subjects is listed in Table 1. Their intelligence was clearly above the population average. The WIP is, however, known to overestimate intelligence compared with the non abbreviated form. They had also high average general and visual memory, and delayed recall was above average. Verbal memory, attention, and recognition memory for faces were not significantly different from population normal scores (see 95% confidence limits). Attention was significantly correlated with age (r= − 0.684, P = 0.02), whereas the other tested parameters were independent from age.
2.1. Cogniti6e task The experimental task was composed of two different conditions, reference and learning, each repeated four times. The learning condition (L) was designed as a face-name-association paradigm. During this task, faces of unfamiliar male and female persons with a first name were shown to the subjects. Faces were presented as black and white halftone images on a neutral background. Common distinguishing features, such as eyeglasses and facial hair, were not removed. The faces did not express specific emotions. The names had been selected from a larger list of first names. Independent control subjects had evaluated the names as not unusual, and the combination of names and faces as consistent. For the activation study, subjects were told to bear the combinations of face and name in mind and speak out the name aloud. During each run, ten new unfamiliar face-name combinations were presented. This set-up was chosen to examine the encoding activity during the association of faces (visual stimuli) and names (verbal stimuli). The reference condition (R) required gender identification. Furthermore, during each run, ten other unfamiliar male and female faces in random order were presented with ‘Frau’ (woman) or ‘Mann’ (man), also in random sequence, as written words underneath. The subjects were instructed to respond verbally, whether the combination presented was
Table 1 Neuropsychological profile of participating subjects Tests
Norm
Subjects
95% confidence limits of the mean
Mean 9 S.D.
Mean
S.D.
WIP72:IQ
1009 15
134.18
10.41
127.2
141.2
WMS-R Verbal memory Visual memory General memory Attention Delayed recall Recognition memory for faces
1009 15 1009 15 1009 15 100 9 15 1009 15 42.79 3.8
101.55 113.82 108.45 111.45 108.27 43
15.45 5.19 11.15 18.59 11.82 3.54
91.2 110.3 101.0 99.0 100.3 40.6
111.9 117.3 115.9 123.9 116.2 45.4
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
Fig. 1. Brain areas significantly (P 50.01, uncorrected) activated across all subjects during face-name association learning, superimposed on a standard MRI template. Activation intensity is displayed on a colour scale as z-values. The transaxial levels of the slices in the left column are given as Talairach z coordinates.
wrong or right. Reference and learning condition were presented in the balanced sequence RRLLLLRR to avoid an order effect in the comparison between learning and reference. With this design of the reference condition a subtraction of irrelevant visual, verbal and speech activity could be accomplished. The goal was to show the activation associated with learning of face-name associations. Each of the eight measurements took a period of one minute and pictures were presented on a computer screen 50 cm in front of the subjects’ eyes. To monitor learning performance, the examiner asked the subjects to retrieve the encoded combinations after each learning condition run. Blood flow was not recorded during these monitoring sessions. Faces from the previous run were presented again, reordered randomly, but now with four different first names, where all the names were presented in the other combination before, but only one was correct. The subject had to determine which of the names belongs to the face.
2.2. Functional imaging Local cerebral blood flow (CBF) was recorded with positron emission tomography (PET) using the O–15water bolus technique [42] on a high resolution scanner (ECAT EXACT HR, Siemens-CTI, Knoxville, TN) [55] recording the entire brain in 3D acquisition mode. Acquisition over 45 s was started automatically when the activity after i.v. bolus injection of 370 MBq (10 mCi) O-15-water reached the brain. After reconstruction by filtered backprojection with a Hanning filter
645
(0.4 cycles per pixel) and correction for random coincidences, attenuation and scatter 47 transaxial slices of 3.125 mm thickness comprising the whole brain were obtained. For data analysis, statistical parametric mapping (SPM 96) was used with standard parameter settings. Reconstructed PET images were realigned, spatially normalised using a 12 parameter affine transformation, followed by non-linear matching with 8 iterations, and smoothed with a 12 mm FWHM isotropic Gaussian filter. Regional activation effects were assessed using the general linear model [15] with age as a covariate, and global CBF as a confounding covariate. Relation to task performance was studied by introduction of the total score of correctly identified faces as a covariate for each individual, with age and global CBF as confounding covariates. Similarly, habituation effects as a possible confounder of learning effects were studied by using the order of scans within each condition as a covariate. The interaction between the covariates of interest and the effects due to study condition were analysed by appropriate contrasts. The effect of age was also studied for resting and activated CBF separately, again with global CBF as a confounding covariate. The uncorrected z-threshold to detect regional effects was set to P= 0.01. Regional effects were accepted as significant if the p value for effect size (z) or cluster size (k) was B0.05 [14] with correction for multiple comparisons. Approximate anatomical locations associated with the coordinates were determined by consulting the structure probability maps [6] as provided by the internet Talairach daemon (http://ric.uthscsa.edu/projects/ talairachdaemon.html), brain atlases [12,51], and Brodmann maps [4].
3. Results
3.1. Effect of learning Across all subjects, the areas with most significant increase of CBF during learning were found bilaterally in occipital association cortex, mainly corresponding to BA 18 (Fig. 1). This large activation was centred in the inferior temporo-occipital gyri, with a local maxima in lingual and fusiform gyri on both sides (BA 18 and 37, Table 2). These activations extended to the parietal lobe on both sides, more extensively reaching BA 39/40 on the left side, and with a significant local maximum in the lateral part of the right precuneus, close to the parieto-occipital sulcus (BA 19). On the left side a local maximum was also seen in left inferior temporal gyrus (BA 37). There was some involvement of the superior border of cerebellar hemispheres, but close inspection of the MRI template and individually spatially nor-
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
646
Table 2 Significant local activation maxima across all subjects during learning Side
Structure
Brodmann area
Z
Cluster sizea
x
y
z
L L L L R R R R R L L
Inferior occipital gyrus Lingual gyrus Fusiform gyrus Inferior temporal gyrus Lingual gyrus Fusiform gyrus Inferior occipital gyrus Fusiform gyrus Precuneus Pre- and postcentral gyrus Middle/superior frontal gyrus (orbitofrontal)
18 18 37 37 18 37 18 18 19 4, 6, 43 11
6.93 6.72 5.11 4.77 6.5 6.16 6.13 5.93 4.73 5.2 4.93
8814 same same same 7408 same same same same 1428 2450
−24 −16 −36 −66 20 40 34 40 28 −54 −26
−88 −92 −36 −44 −90 −58 −84 −76 −68 −8 44
−12 −18 −26 −18 −20 −22 −12 −16 40 22 −14
a Cluster size is number of voxels, each 2×2×2 mm, ‘same’ indicates that these local maxima are part of the same activation cluster as the structure listed above.
malised MRI scans showed that all activation maxima were located supratentorially. Minor activation of the cerebellum proper cannot be excluded, but cannot be separated reliably from supratentorial activity spread by smoothing and residual anatomical variation of the tentorium and occipital lobe. Significant activations outside of visual association areas were seen in the left hemisphere only, including pre- and postcentral (BA 4, 6, and 43), and frontal cortex (BA 11). Relative deactivations during learning were observed extensively in the right hemisphere, in particular in auditory and frontal association cortex, with significant maxima in all three temporal gyri (BA 20, 21, 22) extending also to the supramarginal gyrus (BA 40), in the inferior frontal gyrus (BA 44), and in mesial and polar parts of the superior frontal gyrus (BA 8, 9, 10). Further deactivations were present in the right anterior and posterior cingulate gyrus (BA 10, 31), extending to the mesial part of the right precuneus (BA 7), and in the upper part of the precentral gyrus. In the left hemisphere, deactivations were only seen in the superior temporal gyrus (BA 22).
neuropsychological parameters as listed in Table 1 were in the range of − 0.18 to 0.52 and did not reach statistical significance. The correlation coefficient with age was − 0.56, also not statistically significant. Subjects with better task performance had significantly higher activations in BA 18 (Fig. 3), with local maxima of covariance in left fusiform (BA18, maximum effect at coordinates − 22, − 72, -14, z=4.50, cluster size 2501 voxels) and right lingual gyrus (BA18, coordinates 8, − 80, 18, z= 4.49, within the same cluster).
3.4. Habituation Across the four learning sessions there was a significant training effect, albeit always new faces and names were shown. Correct answers were 5.49 2.4 after the first, 6.79 2.0 after the second, 5.792.3 after the third, and 7.29 2.6 after the fourth session (P= 0.023 for repeats as a linear factor in repeated-measures ANOVA). The order of learning scans was associated with a significant decline of CBF in left middle tempo-
3.2. Effect of age There was no significant effect of age on the increase of CBF during activation. Yet, an analysis of the effect of age on resting and activated CBF separately showed that CBF declined significantly with age in large and similar brain areas bilaterally, in particular in frontal association cortex including cingulate gyrus and in perisylvian cortex (Fig. 2).
3.3. Relation to task performance Task performance differed considerably among participants. On average, the sum of correct answers across the 4 runs was 25.09 7.6, out of 40 possible. Correlation coefficients of task performance with the tested
Fig. 2. Brain areas with significant decline of resting and activated CBF (P5 0.01), shown in transparent orthogonal brain projections. The proximity of a large part of affected regions to the interhemispheric and Sylvian fissures suggests a very similar influence of age-related atrophy on CBF measurements at rest and during activation.
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
Fig. 3. Brain areas in which activation effects were significantly (P 5 0.01) related to encoding success.
Fig. 4. Brain areas in which CBF during both conditions showed significant (P 5 0.01) habituation effects.
ral gyrus (maximum effect at coordinates − 58, − 20, − 8, z= 4.71, cluster size 1287 voxels) and in left cerebellum (coordinates −8, − 76, − 32, z= 4.38, cluster size 1455) (Fig. 4). A similar effect was also observed during reference scans in a large cluster including the same regions bilaterally.
4. Discussion The intense activation of occipital association cortex included BA 18 within and adjacent to the lateral occipital sulcus, which has been associated previously with visual word processing [38]. It also included ventral occipitotemporal association cortex, as observed in former studies of face processing and memory for faces
647
[21,26,30]. The pivotal role of ventral occipitotemporal association cortex in face processing has already been known from studies of patients with prosopagnosia [9,47]. The central role of the fusiform gyrus for face recognition was emphasised in several studies [3,25,30,46]. In contrast to some of these studies [30,50] which emphasised the critical role of the right fusiform gyrus [34], the activation in our study was stronger on the left side, which may be due to the presentation of face-name associations instead of sole faces. Intensity of fusiform activity can be modulated by covert attention [57], which can be directed to different aspects of the stimuli. The associative learning process in our study may involve the recall of faces of familiar persons bearing the name to be learned. Subjects may also have checked the faces to be learned for similarities with familiar people. Probably, such processes that are directly related to associative learning contributed to the intensity and extent of the activation of ventral temporo-occipital association cortex in our study. Our study suggests that the intensity of processing of the stimuli in visual association cortex is critical for memory performance. This is related to findings by Rosier et al. [45], who found that the level of fusiform activity correlated with performance in a shape recognition task. Another recent study [30] describes a correlation of rCBF in a right mid-fusiform area with encoding success in a face memory task. During face encoding, Rajah et al. [43] found a positive influence of occipitotemporal regions on medial temporal regions, and they were identified as part of functional network related to face matching accuracy using a statistical model of regional covariance [1]. In a study involving recognition of degraded faces and objects, a significant activation associated with learning was also localised in the fusiform gyrus, on the left side for objects, and on the right side for faces [10]. In electrophysiological studies, the ventral mesial occipital cortex area displayed early potentials (N100 and P100) by stimulation with faces as well as other visual stimuli, including words [2]. Thus, the fusiform and lingual gyrus which have been identified as structures that are essential for perception and identification of faces [24,46] also appear to play a critical role in associative learning of faces and names. Our associative learning task obviously requires selective attention which may contribute to the observed fusiform activation [36] and to encoding success. Further studies are needed to clarify whether attentional mechanisms or differences in learning strategies are responsible for the link between performance and ventral occipitotemporal activation in our task. Activations of frontal association cortex, mainly on the left side of the brain have been noted during encoding in many previous studies. In our study the activation maximum was located in orbitofrontal cortex
648
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
( − 26, 44, − 14, BA11), similar to a finding in one previous study of face encoding (−33, 27, −12) [21]. Face-sensitive neurons in orbitofrontal cortex have also been observed in monkeys [44]. In humans, lesions in this area may be associated with impairment in the identification of facial emotional expression [23], and it can be activated by covert processing of such expressions [11]. An activation of a very similar area was also seen during processing of famous faces or famous names [16]. Yet, in our study the activated area was not limited to orbitofrontal cortex, but also extended to more lateral and dorsal prefrontal regions (Fig. 1), which have been activated not only in previous studies of face encoding [21,30], but also during encoding of verbal material [20,53]. Thus, this dorsal left prefrontal activation may be related more to encoding-related activity than to the type of material to be encoded. Activation of pre- and postcentral gyrus was not a common observation in studies of human memory. In our study, verbal answers were requested, ‘right’ or ‘wrong’ in the gender identification control task, or the first name associated with the face, as presented. Since the activation was located in the mouth area, we assume that pronouncing the new names was motorically more demanding than pronouncing ‘right’ or ‘wrong’ repeatedly. Thus, this activation is probably not related to memory functions. When designing the study, we anticipated that we would probably see some hippocampal activation, because hippocampus is an essential structure for encoding of new information [49], and the lack of significant activation in the present study was therefore unexpected. In a study of anterograde face memory with a gender classification task as reference [26] a left hippocampal region (at coordinates − 10, − 32, − 12) was activated. The closest match in our study was the local activation maximum in left fusiform gyrus (at coordinates −36, −36, − 26), located close to the parahippocampal gyrus. This may be explained by differences in details of the study design. Kapur et al. used 3-min data acquisition sessions, faces only instead of face-name associations, and a recognition component (the task included raise of index finger if any face was repeated within the session). Our data are in accordance with a study by Kuskowski and Pardo [30], who did not find hippocampal activation during encoding of unfamiliar faces, and with an electrophysiological study by Puce et al. [40], who found no specific neuronal activity in hippocampus during a face-name association learning task. As usual with this technique, our study did not provide absolute measurements of CBF, but potential global effects were adjusted for by ANCOVA in SPM. Thus, we would not detect activations that did not exceed average brain activation. Our data therefore do not prove absence of hippocampal activation. For the
same reason, the finding of relative deactivations (in temporal, frontal and cingulate cortex) may be a technical consequence of adjustment for potential global effects by ANCOVA in SPM. We have seen such global effects on regional cerebral glucose metabolism (measured in absolute units) in an earlier study using a continuous visual recognition task [27]. Furthermore, in the present study there was no major task component added in the control task that would explain widespread higher regional CBF during the control condition as a specific control task-related effect. We did not find an age effect on CBF activation. In a previous study significantly less frontal activation during encoding of faces had been seen in old compared with the young subjects [19]. In subsequent papers [5,18,35] use of multivariate techniques (partial least squares analysis and path analysis) revealed that age-related differences were mainly due to alterations in functional connections. In contrast, Madden et al. [32] found CBF activation during encoding of words in frontal association cortex bilaterally in older normals but not in young normals. The lack of an age-effect in our study may be due to the more complex task, combining faces and names, differences in statistical analysis, and the relatively small sample. Yet, our data indicate that a hypothetical effect of age, which could not be demonstrated, is certainly weaker than the effect of task performance, which has been found significant even in the presence of age as a confounding covariate. The lack of a significant age effect on CBF activation also contrasts with the strong and widespread effect of age on resting, as well as activated blood flow in frontal and perisylvian cortex, which was evident in our study in accordance with the older literature on CBF and ageing [33,37]. Since ageing is also associated with atrophy, partial volume effects may contribute to this decline. Among the two brain areas with habituation of the CBF response during learning, the cerebellum has a well documented record of habituation effects during various tasks, including the acoustic startle response [52] and verbal response selection [41]. The left middle temporal gyrus, which is a multimodal association area, has been previously implicated in face perception [2,39]. Similar habituation was also observed during the reference task and it, therefore, does not seem specific for learning, but rather associated with attentional or face processing demands common to both tasks. Due to their spatial separation, the effects of habituation apparently did not confound the learning-related effects. The presented activation technique may be helpful to understand memory deficits in elderly subjects and patients with memory impairment. Our study indicates that there is no major age-related change in the neural network that supports encoding of new face-name associations. It will be interesting to examine in future
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
studies whether subjects with age-related impairment of proper name retrieval still maintain an intact encoding network. We will also examine whether lesions or focal atrophy in mesial temporal structures in patients with memory deficits will be associated with changes in frontal or even in occipito-temporal activations, or whether other factors, such as attention, may have a more important influence on performance and regional activations in face-name association learning.
Acknowledgements Supported by DFG grant HE2664/2-3.
References [1] Alexander GE, Mentis MJ, Van Horn JD, Grady CL, Berman KF, Furey ML, Pietrini P, Rapoport SI, Schapiro MB, Moeller JR. Individual differences in PET activation of object perception and attention systems predict face matching accuracy. Neuroreport 1999;10(9):1965–71. [2] Allison T, Puce A, Spencer DD, McCarthy G. Electrophysiological studies of human face perception. I: potentials generated in occipitotemporal cortex by face and non-face stimuli. Cereberal Cortex 1999;9(5):415–30. [3] Bentin S, Deouell LY, Soroker N. Selective visual streaming in face recognition: evidence from developmental prosopagnosia. Neuroreport 1999;10(4):823–7. [4] Braak H. Architectonics of The Human Telencephalic Cortex. Berlin: Springer, 1980. [5] Cabeza R, McIntosh AR, Tulving E, Nyberg L, Grady CL. Age-related differences in effective neural connectivity during encoding and recall. Neuroreport 1997;8(16):3479–83. [6] Collins DL, Holmes CJ, Peters TM, Evans AC. Automatic 3-D model-based neuroanatomical segmentation. Human Brain Mapping 1995;3(3):190–208. [7] Dahl G. WIP-Reduzierter Wechsler Intelligenztest. Meisenheim am Glan: Hain, 1972. [8] Damasio H, Grabowski TJ, Tranel D, Hichwa RD, Damasio AR. A neural basis for lexical retrieval [see comments]. Nature 1996;380(6574):499–505. [9] De Renzi E, Perani D, Carlesimo GA, Silveri MC, Fazio F. Prosopagnosia can be associated with damage confined to the right hemisphere — an MRI and PET study and a review of the literature. Neuropsychologia 1994;32(8):893–902. [10] Dolan RJ, Fink GR, Rolls E, Booth M, Holmes A, Frackowiak RSJ, Friston KJ. How the brain learns to see objects and faces in an impoverished context. Nature 1997;389(6651):596–9. [11] Dolan RJ, Fletcher P, Morris J, Kapur N, Deakin JFW, Frith CD. Neural activation during covert processing of positive emotional facial expressions. Neuroimage 1996;4(3 Part 1):194– 200. [12] Duvernoy H. The Human Brain. Wien: Springer, 1991. [13] Fiez JA, Petersen SE. Neuroimaging studies of word reading. Proceedings of the National Academy of Science USA 1998;95(3):914– 21. [14] Friston KJ, Holmes A, Poline JB, Price CJ, Frith CD. Detecting activations in PET and fMRI-Levels of inference and power. Neuroimage 1996;4(3Part 1):223–35. [15] Friston KJ, Holmes AP, Worsley KJ, Poline JB, Frith CD, Frackowiak RSJ. Statistical parametric maps in functional imaging: a general linear approach. Human Brain Mapping 1995;2(4):189– 210.
649
[16] Gorno-Tempini ML, Price CJ, Josephs O, Vandenberghe R, Cappa SF, Kapur N, Frackowiak RS. The neural systems sustaining face and proper-name processing. Brain 1998;121(Pt 11):2103– 18. [17] Grady CL, Craik FI. Changes in memory processing with age. Current Opinion in Neurobiology 2000;10(2):224– 31. [18] Grady CL, McIntosh AR, Bookstein F, Horwitz B, Rapoport SI, Haxby JV. Age-related changes in regional cerebral blood flow during working memory for faces. Neuroimage 1998;8(4):409– 25. [19] Grady CL, McIntosh AR, Horwitz B, Maisog JM, Ungerleider LG, Mentis MJ, Pietrini P, Schapiro MB, Haxby JV. Age-related reductions in human recognition memory due to impaired encoding. Science 1995;269(5221):218– 21. [20] Halsband U, Krause BJ, Schmidt D, Herzog H, Tellmann L, Muller-Gartner HW. Encoding and retrieval in declarative learning: a positron emission tomography study. Behavioral Brain Research 1998;97(1-2):69– 78. [21] 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 Science USA 1996;93(2):922– 7. [22] Ha¨rting C, Markowitsch HJ, Neufeld H, Calabrese P, Deisinger K, Kessler J. Die Wechsler-Memory-Scale- Revised. In: Deutschsprachige Adaption. Bern: Huber, 2000. [23] Hornak J, Rolls ET, Wade D. Face and voice expression identification in patients with emotional and behavioural changes following ventral frontal lobe damage. Neuropsychologia 1996;34(4):247– 61. [24] Kanwisher N, McDermott J, Chun MM. The fusiform face area: a module in human extrastriate cortex specialized for face perception. Journal of Neuroscience 1997;17(11):4302– 11. [25] Kanwisher N, Stanley D, Harris A. The fusiform face area is selective for faces not animals. Neuroreport 1999;10(1):183–7. [26] Kapur N, Friston KJ, Young A, Frith CD, Frackowiak RS. Activation of human hippocampal formation during memory for faces: a PET study. Cortex 1995;31(1):99– 108. [27] Kessler J, Herholz K, Grond M, Heiss WD. Impaired metabolic activation in Alzheimer’s disease: a PET study during continuous visual recognition. Neuropsychologia 1991;29(3):229– 43. [28] Kim JJ, Andreasen NC, O’Leary DS, Wiser AK, Ponto LL, Watkins GL, Hichwa RD. Direct comparison of the neural substrates of recognition memory for words and faces. Brain 1999;122(Pt 6):1069– 83. [29] Kroll NEA, Markowitsch HJ, Knight RT, von Cramon DY. Retrieval of old memories — the temporofrontal hypothesis. Brain 1997;120(Part 8):1377– 99. [30] Kuskowski MA, Pardo JV. The role of the fusiform gyrus in successful encoding of face stimuli. Neuroimage 1999;9(6 Pt 1):599– 610. [31] Leveroni CL, Seidenberg M, Mayer AR, Mead LA, Binder JR, Rao SM. Neural systems underlying the recognition of familiar and newly learned faces. Journal of Neuroscience 2000;20(2):878– 86. [32] Madden DJ, Turkington TG, Provenzale JM, Denny LL, Hawk TC, Gottlob LR, Coleman RE. Adult age differences in the functional neuroanatomy of verbal recognition memory. Human Brain Mapping 1999;7(2):115– 35. [33] Martin AJ, Friston KJ, Colebatch JG, Frackowiak RS. Decreases in regional cerebral blood flow with normal aging. Journal of Cerebral Blood Flow and Metabolism 1991;11(4):684–9. [34] McCarthy G, Puce A, Gore JC, Allison T. Face-specific processing in the human fusiform gyrus. Journal of Cognitive Neuroscience 1997;9(5):605– 10. [35] McIntosh AR, Sekuler AB, Penpeci C, Rajah MN, Grady CL, Sekuler R, Bennett PJ. Recruitment of unique neural systems to support visual memory in normal aging. Current Biology 1999;9(21):1275– 8.
650
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
[36] O’Craven KM, Downing PE, Kanwisher N. MRI evidence for objects as the units of attentional selection. Nature 1999;401(6753):584–7. [37] Pantano P, Baron JC, Lebrun-Grandie P, Duquesnoy N, Bousser MG, Comar D. Regional cerebral blood flow and oxygen consumption in human aging. Stroke 1984;15(4):635– 41. [38] Petersen SE, Fiez JA. The processing of single words studied with positron emission tomography. Annual Review of Neuroscience 1993;16:509–30. [39] Puce A, Allison T, Asgari M, Gore JC, McCarthy G. Differential sensitivity of human visual cortex to faces, letterstrings, and textures: a functional magnetic resonance imaging study. Journal of Neuroscience 1996;16(16):5205–15. [40] Puce A, Allison T, McCarthy G. Electrophysiological studies of human face perception. III: effects of top- down processing on face-specific potentials. Cereberal Cortex 1999;9(5):445–58. [41] Raichle ME, Fiez JA, Videen TO, MacLeod AM, Pardo JV, Fox PT, Petersen SE. Practice-related changes in human brain functional anatomy during nonmotor learning. Cerebral Cortex 1994;4(1):8– 26. [42] Raichle ME, Martin WR, Herscovitch P, Mintun MA, Markham J. Brain blood flow measured with intravenous H2(15)O. II. Implementation and validation. Journal of Nuclear Medicine 1983;24(9):790–8. [43] Rajah MN, McIntosh AR, Grady CL. Frontotemporal interactions in face encoding and recognition. Cognitive Brain Research 1999;8:259– 69. [44] Rolls ET. The functions of the orbitofrontal cortex. Neurocase 1999;5:301– 12. [45] Rosier AM, Cornette L, Dupont P, Bormans G, Mortelmans L, Orban GA. Regional brain activity during shape recognition impaired by a scopolamine challenge to encoding. European Journal of Neuroscience 1999;11(10):3701–14. [46] Sergent J, Ohta S, MacDonald B. Functional neuroanatomy of face and object processing. A positron emission tomography study. Brain 1992;115(Pt 1):15–36.
.
[47] Sergent J, Signoret JL. Functional and anatomical decomposition of face processing: evidence from prosopagnosia and PET study of normal subjects. Philosophical Transactions of the Royal Society of London — Series B Biological Sciences 1992;335(1273):55– 61. [48] Sommer W, Komoss E, Schweinberger SR. Differential localization of brain systems subserving memory for names and faces in normal subjects with event-related potentials. Electroencephalography Clinical Neurophysiology 1997;102(3):192– 9. [49] Squire LR, Zola-Morgan S. The medial temporal lobe memory system. Science 1991;253:1380– 6. [50] Takahashi N, Kawamura M, Hirayama K, Shiota J, Isono O. Prosopagnosia: a clinical and anatomical study of four patients. Cortex 1995;31(2):317– 29. [51] Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of The Human Brain. Stuttgart: Georg Thieme Verlag, 1988. [52] Timmann D, Musso C, Kolb FP, Rijntjes M, Juptner M, Muller SP, Diener HC, Weiller C. Involvement of the human cerebellum during habituation of the acoustic startle response: a PET study. Journal of Neurology Neurosurgery Psychiatry 1998;65(5):771– 3. [53] Tulving E, Kapur S, Craik FI, Moscovitch M, Houle S. Hemispheric encoding/retrieval asymmetry in episodic memory: positron emission tomography findings. Proceedings National Academy of Science USA 1994;91(6):2016– 20. [54] Warrington EK. Recognition Memory Test (RMT). Windsor: NFER-Nelson Publ. Comp. Ltd, 1984. [55] Wienhard K, Dahlbom M, Eriksson L, Michel C, Bruckbauer T, Pietrzyk U, Heiss WD. The ECAT EXACT HR: performance of a new high resolution positron scanner. Journal of Computer Assisted Tomography 1994;18(1):110– 8. [56] Wilson BA, Cockburn J, Baddeley A. The Rivermead Behavioural Memory Test (RBMT). Reading: Thames Valley Test Group, 1985. [57] Wojciulik E, Kanwisher N, Driver J. Covert visual attention modulates face-specific activity in the human fusiform gyrus: fMRI study. Journal of Neurophysiology 1998;79(3):1574–8.
.
Learning face-name associations and the effect of age and performance: a PET activation study Karl Herholz a,*, Patricia Ehlen a, Josef Kessler a, Timm Strotmann a, Elke Kalbe a, Hans-Joachim Markowitsch b a
Max-Planck-Institut fu¨r neurologische Forschung and Neurologische Uni6ersita¨tsklinik, Gleuler Str. 50, D-50931 Cologne, Germany b Physiologische Psychologie, Uni6ersita¨t Bielefeld, Germany Received 27 January 2000; received in revised form 30 August 2000; accepted 16 October 2000
Abstract Learning face-name associations is a complex task to be mastered in every day life that approaches the limits of cognitive capacity in most normal humans. We studied brain activation during face-name learning using positron emission tomography (PET) in 11 normal volunteers. The most intense activation was seen in occipital association cortex (BA 18) bilaterally, also involving lingual and fusiform gyrus (BA 37). In the left hemisphere additional activation were located in inferior temporal gyrus, the inferior part of pre- and postcentral gyrus, and orbitofrontal cortex (BA 11), whereas in the right hemisphere only a region in the precuneus (BA 19) was activated additionally. There was considerable interindividual variation of encoding success, which was significantly related to activation of BA 18 bilaterally. Subject ages covered a range of 26 – 72 years, but — in contrast to the effect of encoding success — there was no significant age effect on activations. Task-independent habituation effects were seen in cerebellum and left middle temporal gyrus. These results indicate that the intensity of information processing in ventral occipital association cortex is most important for success of face-name encoding. Learning is further mediated by a predominantly left-hemispheric network including inferior temporal and orbitofrontal cortex. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: PET; Activation; Face-name association; Learning; Ageing
1. Introduction Memory for faces is an essential cognitive ability and is supported by specialised neuroanatomical structures in ventral temporo-occipital association cortex. Faces can be associated with names. Encoding and retrieval of the names associated with other persons’ faces is a daily task that is mastered with less than perfect success by most normal subjects. Although a huge number of such associations can be stored during life, and often can be retrieved after several decades, the task also regularly seems to reach the limits of human cognitive abilities if one has brief contact with many people. With advancing age, the transfer of new information into long-term memory and retrieval of recently encoded information, such as proper names may become * Corresponding author. Tel.: +49-221-4726229; fax: + 49-2214726298. E-mail address: [email protected] (K. Herholz).
particularly difficult (see recent review by Grady and Craik [17]). The purpose of the present study was to investigate activations during a learning task that was close to daily-life reality in order to provide a baseline for the understanding of memory deficits, as they are experienced by elderly subjects and patients with memory impairment. In addition, face memory association tasks appear to be particularly well suited because of their ecological validity. They have, therefore, been included in behavioural memory tests, such as the Rivermead Behavioural Memory Test [56]. Some previous studies indicated that memory for faces and words involves different neural substrates that are located predominantly in the right and the left hemisphere, respectively [28,48], but may also share a common modality-independent semantic system [16]. The combined associative processing of faces and names has been studied previously mainly in the context of recognition of famous faces [8,16,46]. In addi-
0028-3932/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 8 - 3 9 3 2 ( 0 0 ) 0 0 1 4 4 - 5
644
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
tion to the activation of ventral temporo-occipital cortex by face processing, bilateral fronto-temporal regions have been activated in these studies. Correspondingly, fronto-temporal brain damage impairs memory for past events and famous faces [29]. Yet, the areas activated by recognition of famous faces appear to be different from those activated by faces that have been encoded only recently [31]. Impairment of recognition is not the only possible reason for impairment of learning, which may also be due to impairment of encoding. We, therefore, wanted to study the activation results during encoding of facename associations and compare the results with those of previous functional imaging studies in which encoding of faces or words was performed separately [13,21,26,30,34,43,53]. From these studies, we expected that ventral occipital association cortex, hippocampus, and frontal association cortex would be among the activated areas in a face-name association learning task. We also wanted to analyse the effects of age and performance on activated areas.
2. Subjects and method Eleven normal right-handed volunteers, eight men and three women, were studied. Ages covered a broad range from 26 to 72 years (mean 46.7917.6). Subjects had no history of neurological or psychiatric disease and a normal neurological status. They were tested neuropsychologically with the abbreviated Wechsler-Intelligence-Test (WIP) [7], the Wechsler-Memory-Test revised (WMS-r) [22], and the Recognition-MemoryTest for faces (RMT) [54]. All underwent a PET activation study with a learning task. Magnetic resonance imaging (MRI) of the brain excluded structural abnormalities and was used for anatomical localisation of activated areas. (MRI could not be performed in one subject because of claustrophobia). All subjects gave their informed consent, and the study was approved by the ethics committee of the medical faculty.
The neuropsychological profile of participating subjects is listed in Table 1. Their intelligence was clearly above the population average. The WIP is, however, known to overestimate intelligence compared with the non abbreviated form. They had also high average general and visual memory, and delayed recall was above average. Verbal memory, attention, and recognition memory for faces were not significantly different from population normal scores (see 95% confidence limits). Attention was significantly correlated with age (r= − 0.684, P = 0.02), whereas the other tested parameters were independent from age.
2.1. Cogniti6e task The experimental task was composed of two different conditions, reference and learning, each repeated four times. The learning condition (L) was designed as a face-name-association paradigm. During this task, faces of unfamiliar male and female persons with a first name were shown to the subjects. Faces were presented as black and white halftone images on a neutral background. Common distinguishing features, such as eyeglasses and facial hair, were not removed. The faces did not express specific emotions. The names had been selected from a larger list of first names. Independent control subjects had evaluated the names as not unusual, and the combination of names and faces as consistent. For the activation study, subjects were told to bear the combinations of face and name in mind and speak out the name aloud. During each run, ten new unfamiliar face-name combinations were presented. This set-up was chosen to examine the encoding activity during the association of faces (visual stimuli) and names (verbal stimuli). The reference condition (R) required gender identification. Furthermore, during each run, ten other unfamiliar male and female faces in random order were presented with ‘Frau’ (woman) or ‘Mann’ (man), also in random sequence, as written words underneath. The subjects were instructed to respond verbally, whether the combination presented was
Table 1 Neuropsychological profile of participating subjects Tests
Norm
Subjects
95% confidence limits of the mean
Mean 9 S.D.
Mean
S.D.
WIP72:IQ
1009 15
134.18
10.41
127.2
141.2
WMS-R Verbal memory Visual memory General memory Attention Delayed recall Recognition memory for faces
1009 15 1009 15 1009 15 100 9 15 1009 15 42.79 3.8
101.55 113.82 108.45 111.45 108.27 43
15.45 5.19 11.15 18.59 11.82 3.54
91.2 110.3 101.0 99.0 100.3 40.6
111.9 117.3 115.9 123.9 116.2 45.4
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
Fig. 1. Brain areas significantly (P 50.01, uncorrected) activated across all subjects during face-name association learning, superimposed on a standard MRI template. Activation intensity is displayed on a colour scale as z-values. The transaxial levels of the slices in the left column are given as Talairach z coordinates.
wrong or right. Reference and learning condition were presented in the balanced sequence RRLLLLRR to avoid an order effect in the comparison between learning and reference. With this design of the reference condition a subtraction of irrelevant visual, verbal and speech activity could be accomplished. The goal was to show the activation associated with learning of face-name associations. Each of the eight measurements took a period of one minute and pictures were presented on a computer screen 50 cm in front of the subjects’ eyes. To monitor learning performance, the examiner asked the subjects to retrieve the encoded combinations after each learning condition run. Blood flow was not recorded during these monitoring sessions. Faces from the previous run were presented again, reordered randomly, but now with four different first names, where all the names were presented in the other combination before, but only one was correct. The subject had to determine which of the names belongs to the face.
2.2. Functional imaging Local cerebral blood flow (CBF) was recorded with positron emission tomography (PET) using the O–15water bolus technique [42] on a high resolution scanner (ECAT EXACT HR, Siemens-CTI, Knoxville, TN) [55] recording the entire brain in 3D acquisition mode. Acquisition over 45 s was started automatically when the activity after i.v. bolus injection of 370 MBq (10 mCi) O-15-water reached the brain. After reconstruction by filtered backprojection with a Hanning filter
645
(0.4 cycles per pixel) and correction for random coincidences, attenuation and scatter 47 transaxial slices of 3.125 mm thickness comprising the whole brain were obtained. For data analysis, statistical parametric mapping (SPM 96) was used with standard parameter settings. Reconstructed PET images were realigned, spatially normalised using a 12 parameter affine transformation, followed by non-linear matching with 8 iterations, and smoothed with a 12 mm FWHM isotropic Gaussian filter. Regional activation effects were assessed using the general linear model [15] with age as a covariate, and global CBF as a confounding covariate. Relation to task performance was studied by introduction of the total score of correctly identified faces as a covariate for each individual, with age and global CBF as confounding covariates. Similarly, habituation effects as a possible confounder of learning effects were studied by using the order of scans within each condition as a covariate. The interaction between the covariates of interest and the effects due to study condition were analysed by appropriate contrasts. The effect of age was also studied for resting and activated CBF separately, again with global CBF as a confounding covariate. The uncorrected z-threshold to detect regional effects was set to P= 0.01. Regional effects were accepted as significant if the p value for effect size (z) or cluster size (k) was B0.05 [14] with correction for multiple comparisons. Approximate anatomical locations associated with the coordinates were determined by consulting the structure probability maps [6] as provided by the internet Talairach daemon (http://ric.uthscsa.edu/projects/ talairachdaemon.html), brain atlases [12,51], and Brodmann maps [4].
3. Results
3.1. Effect of learning Across all subjects, the areas with most significant increase of CBF during learning were found bilaterally in occipital association cortex, mainly corresponding to BA 18 (Fig. 1). This large activation was centred in the inferior temporo-occipital gyri, with a local maxima in lingual and fusiform gyri on both sides (BA 18 and 37, Table 2). These activations extended to the parietal lobe on both sides, more extensively reaching BA 39/40 on the left side, and with a significant local maximum in the lateral part of the right precuneus, close to the parieto-occipital sulcus (BA 19). On the left side a local maximum was also seen in left inferior temporal gyrus (BA 37). There was some involvement of the superior border of cerebellar hemispheres, but close inspection of the MRI template and individually spatially nor-
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
646
Table 2 Significant local activation maxima across all subjects during learning Side
Structure
Brodmann area
Z
Cluster sizea
x
y
z
L L L L R R R R R L L
Inferior occipital gyrus Lingual gyrus Fusiform gyrus Inferior temporal gyrus Lingual gyrus Fusiform gyrus Inferior occipital gyrus Fusiform gyrus Precuneus Pre- and postcentral gyrus Middle/superior frontal gyrus (orbitofrontal)
18 18 37 37 18 37 18 18 19 4, 6, 43 11
6.93 6.72 5.11 4.77 6.5 6.16 6.13 5.93 4.73 5.2 4.93
8814 same same same 7408 same same same same 1428 2450
−24 −16 −36 −66 20 40 34 40 28 −54 −26
−88 −92 −36 −44 −90 −58 −84 −76 −68 −8 44
−12 −18 −26 −18 −20 −22 −12 −16 40 22 −14
a Cluster size is number of voxels, each 2×2×2 mm, ‘same’ indicates that these local maxima are part of the same activation cluster as the structure listed above.
malised MRI scans showed that all activation maxima were located supratentorially. Minor activation of the cerebellum proper cannot be excluded, but cannot be separated reliably from supratentorial activity spread by smoothing and residual anatomical variation of the tentorium and occipital lobe. Significant activations outside of visual association areas were seen in the left hemisphere only, including pre- and postcentral (BA 4, 6, and 43), and frontal cortex (BA 11). Relative deactivations during learning were observed extensively in the right hemisphere, in particular in auditory and frontal association cortex, with significant maxima in all three temporal gyri (BA 20, 21, 22) extending also to the supramarginal gyrus (BA 40), in the inferior frontal gyrus (BA 44), and in mesial and polar parts of the superior frontal gyrus (BA 8, 9, 10). Further deactivations were present in the right anterior and posterior cingulate gyrus (BA 10, 31), extending to the mesial part of the right precuneus (BA 7), and in the upper part of the precentral gyrus. In the left hemisphere, deactivations were only seen in the superior temporal gyrus (BA 22).
neuropsychological parameters as listed in Table 1 were in the range of − 0.18 to 0.52 and did not reach statistical significance. The correlation coefficient with age was − 0.56, also not statistically significant. Subjects with better task performance had significantly higher activations in BA 18 (Fig. 3), with local maxima of covariance in left fusiform (BA18, maximum effect at coordinates − 22, − 72, -14, z=4.50, cluster size 2501 voxels) and right lingual gyrus (BA18, coordinates 8, − 80, 18, z= 4.49, within the same cluster).
3.4. Habituation Across the four learning sessions there was a significant training effect, albeit always new faces and names were shown. Correct answers were 5.49 2.4 after the first, 6.79 2.0 after the second, 5.792.3 after the third, and 7.29 2.6 after the fourth session (P= 0.023 for repeats as a linear factor in repeated-measures ANOVA). The order of learning scans was associated with a significant decline of CBF in left middle tempo-
3.2. Effect of age There was no significant effect of age on the increase of CBF during activation. Yet, an analysis of the effect of age on resting and activated CBF separately showed that CBF declined significantly with age in large and similar brain areas bilaterally, in particular in frontal association cortex including cingulate gyrus and in perisylvian cortex (Fig. 2).
3.3. Relation to task performance Task performance differed considerably among participants. On average, the sum of correct answers across the 4 runs was 25.09 7.6, out of 40 possible. Correlation coefficients of task performance with the tested
Fig. 2. Brain areas with significant decline of resting and activated CBF (P5 0.01), shown in transparent orthogonal brain projections. The proximity of a large part of affected regions to the interhemispheric and Sylvian fissures suggests a very similar influence of age-related atrophy on CBF measurements at rest and during activation.
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
Fig. 3. Brain areas in which activation effects were significantly (P 5 0.01) related to encoding success.
Fig. 4. Brain areas in which CBF during both conditions showed significant (P 5 0.01) habituation effects.
ral gyrus (maximum effect at coordinates − 58, − 20, − 8, z= 4.71, cluster size 1287 voxels) and in left cerebellum (coordinates −8, − 76, − 32, z= 4.38, cluster size 1455) (Fig. 4). A similar effect was also observed during reference scans in a large cluster including the same regions bilaterally.
4. Discussion The intense activation of occipital association cortex included BA 18 within and adjacent to the lateral occipital sulcus, which has been associated previously with visual word processing [38]. It also included ventral occipitotemporal association cortex, as observed in former studies of face processing and memory for faces
647
[21,26,30]. The pivotal role of ventral occipitotemporal association cortex in face processing has already been known from studies of patients with prosopagnosia [9,47]. The central role of the fusiform gyrus for face recognition was emphasised in several studies [3,25,30,46]. In contrast to some of these studies [30,50] which emphasised the critical role of the right fusiform gyrus [34], the activation in our study was stronger on the left side, which may be due to the presentation of face-name associations instead of sole faces. Intensity of fusiform activity can be modulated by covert attention [57], which can be directed to different aspects of the stimuli. The associative learning process in our study may involve the recall of faces of familiar persons bearing the name to be learned. Subjects may also have checked the faces to be learned for similarities with familiar people. Probably, such processes that are directly related to associative learning contributed to the intensity and extent of the activation of ventral temporo-occipital association cortex in our study. Our study suggests that the intensity of processing of the stimuli in visual association cortex is critical for memory performance. This is related to findings by Rosier et al. [45], who found that the level of fusiform activity correlated with performance in a shape recognition task. Another recent study [30] describes a correlation of rCBF in a right mid-fusiform area with encoding success in a face memory task. During face encoding, Rajah et al. [43] found a positive influence of occipitotemporal regions on medial temporal regions, and they were identified as part of functional network related to face matching accuracy using a statistical model of regional covariance [1]. In a study involving recognition of degraded faces and objects, a significant activation associated with learning was also localised in the fusiform gyrus, on the left side for objects, and on the right side for faces [10]. In electrophysiological studies, the ventral mesial occipital cortex area displayed early potentials (N100 and P100) by stimulation with faces as well as other visual stimuli, including words [2]. Thus, the fusiform and lingual gyrus which have been identified as structures that are essential for perception and identification of faces [24,46] also appear to play a critical role in associative learning of faces and names. Our associative learning task obviously requires selective attention which may contribute to the observed fusiform activation [36] and to encoding success. Further studies are needed to clarify whether attentional mechanisms or differences in learning strategies are responsible for the link between performance and ventral occipitotemporal activation in our task. Activations of frontal association cortex, mainly on the left side of the brain have been noted during encoding in many previous studies. In our study the activation maximum was located in orbitofrontal cortex
648
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
( − 26, 44, − 14, BA11), similar to a finding in one previous study of face encoding (−33, 27, −12) [21]. Face-sensitive neurons in orbitofrontal cortex have also been observed in monkeys [44]. In humans, lesions in this area may be associated with impairment in the identification of facial emotional expression [23], and it can be activated by covert processing of such expressions [11]. An activation of a very similar area was also seen during processing of famous faces or famous names [16]. Yet, in our study the activated area was not limited to orbitofrontal cortex, but also extended to more lateral and dorsal prefrontal regions (Fig. 1), which have been activated not only in previous studies of face encoding [21,30], but also during encoding of verbal material [20,53]. Thus, this dorsal left prefrontal activation may be related more to encoding-related activity than to the type of material to be encoded. Activation of pre- and postcentral gyrus was not a common observation in studies of human memory. In our study, verbal answers were requested, ‘right’ or ‘wrong’ in the gender identification control task, or the first name associated with the face, as presented. Since the activation was located in the mouth area, we assume that pronouncing the new names was motorically more demanding than pronouncing ‘right’ or ‘wrong’ repeatedly. Thus, this activation is probably not related to memory functions. When designing the study, we anticipated that we would probably see some hippocampal activation, because hippocampus is an essential structure for encoding of new information [49], and the lack of significant activation in the present study was therefore unexpected. In a study of anterograde face memory with a gender classification task as reference [26] a left hippocampal region (at coordinates − 10, − 32, − 12) was activated. The closest match in our study was the local activation maximum in left fusiform gyrus (at coordinates −36, −36, − 26), located close to the parahippocampal gyrus. This may be explained by differences in details of the study design. Kapur et al. used 3-min data acquisition sessions, faces only instead of face-name associations, and a recognition component (the task included raise of index finger if any face was repeated within the session). Our data are in accordance with a study by Kuskowski and Pardo [30], who did not find hippocampal activation during encoding of unfamiliar faces, and with an electrophysiological study by Puce et al. [40], who found no specific neuronal activity in hippocampus during a face-name association learning task. As usual with this technique, our study did not provide absolute measurements of CBF, but potential global effects were adjusted for by ANCOVA in SPM. Thus, we would not detect activations that did not exceed average brain activation. Our data therefore do not prove absence of hippocampal activation. For the
same reason, the finding of relative deactivations (in temporal, frontal and cingulate cortex) may be a technical consequence of adjustment for potential global effects by ANCOVA in SPM. We have seen such global effects on regional cerebral glucose metabolism (measured in absolute units) in an earlier study using a continuous visual recognition task [27]. Furthermore, in the present study there was no major task component added in the control task that would explain widespread higher regional CBF during the control condition as a specific control task-related effect. We did not find an age effect on CBF activation. In a previous study significantly less frontal activation during encoding of faces had been seen in old compared with the young subjects [19]. In subsequent papers [5,18,35] use of multivariate techniques (partial least squares analysis and path analysis) revealed that age-related differences were mainly due to alterations in functional connections. In contrast, Madden et al. [32] found CBF activation during encoding of words in frontal association cortex bilaterally in older normals but not in young normals. The lack of an age-effect in our study may be due to the more complex task, combining faces and names, differences in statistical analysis, and the relatively small sample. Yet, our data indicate that a hypothetical effect of age, which could not be demonstrated, is certainly weaker than the effect of task performance, which has been found significant even in the presence of age as a confounding covariate. The lack of a significant age effect on CBF activation also contrasts with the strong and widespread effect of age on resting, as well as activated blood flow in frontal and perisylvian cortex, which was evident in our study in accordance with the older literature on CBF and ageing [33,37]. Since ageing is also associated with atrophy, partial volume effects may contribute to this decline. Among the two brain areas with habituation of the CBF response during learning, the cerebellum has a well documented record of habituation effects during various tasks, including the acoustic startle response [52] and verbal response selection [41]. The left middle temporal gyrus, which is a multimodal association area, has been previously implicated in face perception [2,39]. Similar habituation was also observed during the reference task and it, therefore, does not seem specific for learning, but rather associated with attentional or face processing demands common to both tasks. Due to their spatial separation, the effects of habituation apparently did not confound the learning-related effects. The presented activation technique may be helpful to understand memory deficits in elderly subjects and patients with memory impairment. Our study indicates that there is no major age-related change in the neural network that supports encoding of new face-name associations. It will be interesting to examine in future
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
studies whether subjects with age-related impairment of proper name retrieval still maintain an intact encoding network. We will also examine whether lesions or focal atrophy in mesial temporal structures in patients with memory deficits will be associated with changes in frontal or even in occipito-temporal activations, or whether other factors, such as attention, may have a more important influence on performance and regional activations in face-name association learning.
Acknowledgements Supported by DFG grant HE2664/2-3.
References [1] Alexander GE, Mentis MJ, Van Horn JD, Grady CL, Berman KF, Furey ML, Pietrini P, Rapoport SI, Schapiro MB, Moeller JR. Individual differences in PET activation of object perception and attention systems predict face matching accuracy. Neuroreport 1999;10(9):1965–71. [2] Allison T, Puce A, Spencer DD, McCarthy G. Electrophysiological studies of human face perception. I: potentials generated in occipitotemporal cortex by face and non-face stimuli. Cereberal Cortex 1999;9(5):415–30. [3] Bentin S, Deouell LY, Soroker N. Selective visual streaming in face recognition: evidence from developmental prosopagnosia. Neuroreport 1999;10(4):823–7. [4] Braak H. Architectonics of The Human Telencephalic Cortex. Berlin: Springer, 1980. [5] Cabeza R, McIntosh AR, Tulving E, Nyberg L, Grady CL. Age-related differences in effective neural connectivity during encoding and recall. Neuroreport 1997;8(16):3479–83. [6] Collins DL, Holmes CJ, Peters TM, Evans AC. Automatic 3-D model-based neuroanatomical segmentation. Human Brain Mapping 1995;3(3):190–208. [7] Dahl G. WIP-Reduzierter Wechsler Intelligenztest. Meisenheim am Glan: Hain, 1972. [8] Damasio H, Grabowski TJ, Tranel D, Hichwa RD, Damasio AR. A neural basis for lexical retrieval [see comments]. Nature 1996;380(6574):499–505. [9] De Renzi E, Perani D, Carlesimo GA, Silveri MC, Fazio F. Prosopagnosia can be associated with damage confined to the right hemisphere — an MRI and PET study and a review of the literature. Neuropsychologia 1994;32(8):893–902. [10] Dolan RJ, Fink GR, Rolls E, Booth M, Holmes A, Frackowiak RSJ, Friston KJ. How the brain learns to see objects and faces in an impoverished context. Nature 1997;389(6651):596–9. [11] Dolan RJ, Fletcher P, Morris J, Kapur N, Deakin JFW, Frith CD. Neural activation during covert processing of positive emotional facial expressions. Neuroimage 1996;4(3 Part 1):194– 200. [12] Duvernoy H. The Human Brain. Wien: Springer, 1991. [13] Fiez JA, Petersen SE. Neuroimaging studies of word reading. Proceedings of the National Academy of Science USA 1998;95(3):914– 21. [14] Friston KJ, Holmes A, Poline JB, Price CJ, Frith CD. Detecting activations in PET and fMRI-Levels of inference and power. Neuroimage 1996;4(3Part 1):223–35. [15] Friston KJ, Holmes AP, Worsley KJ, Poline JB, Frith CD, Frackowiak RSJ. Statistical parametric maps in functional imaging: a general linear approach. Human Brain Mapping 1995;2(4):189– 210.
649
[16] Gorno-Tempini ML, Price CJ, Josephs O, Vandenberghe R, Cappa SF, Kapur N, Frackowiak RS. The neural systems sustaining face and proper-name processing. Brain 1998;121(Pt 11):2103– 18. [17] Grady CL, Craik FI. Changes in memory processing with age. Current Opinion in Neurobiology 2000;10(2):224– 31. [18] Grady CL, McIntosh AR, Bookstein F, Horwitz B, Rapoport SI, Haxby JV. Age-related changes in regional cerebral blood flow during working memory for faces. Neuroimage 1998;8(4):409– 25. [19] Grady CL, McIntosh AR, Horwitz B, Maisog JM, Ungerleider LG, Mentis MJ, Pietrini P, Schapiro MB, Haxby JV. Age-related reductions in human recognition memory due to impaired encoding. Science 1995;269(5221):218– 21. [20] Halsband U, Krause BJ, Schmidt D, Herzog H, Tellmann L, Muller-Gartner HW. Encoding and retrieval in declarative learning: a positron emission tomography study. Behavioral Brain Research 1998;97(1-2):69– 78. [21] 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 Science USA 1996;93(2):922– 7. [22] Ha¨rting C, Markowitsch HJ, Neufeld H, Calabrese P, Deisinger K, Kessler J. Die Wechsler-Memory-Scale- Revised. In: Deutschsprachige Adaption. Bern: Huber, 2000. [23] Hornak J, Rolls ET, Wade D. Face and voice expression identification in patients with emotional and behavioural changes following ventral frontal lobe damage. Neuropsychologia 1996;34(4):247– 61. [24] Kanwisher N, McDermott J, Chun MM. The fusiform face area: a module in human extrastriate cortex specialized for face perception. Journal of Neuroscience 1997;17(11):4302– 11. [25] Kanwisher N, Stanley D, Harris A. The fusiform face area is selective for faces not animals. Neuroreport 1999;10(1):183–7. [26] Kapur N, Friston KJ, Young A, Frith CD, Frackowiak RS. Activation of human hippocampal formation during memory for faces: a PET study. Cortex 1995;31(1):99– 108. [27] Kessler J, Herholz K, Grond M, Heiss WD. Impaired metabolic activation in Alzheimer’s disease: a PET study during continuous visual recognition. Neuropsychologia 1991;29(3):229– 43. [28] Kim JJ, Andreasen NC, O’Leary DS, Wiser AK, Ponto LL, Watkins GL, Hichwa RD. Direct comparison of the neural substrates of recognition memory for words and faces. Brain 1999;122(Pt 6):1069– 83. [29] Kroll NEA, Markowitsch HJ, Knight RT, von Cramon DY. Retrieval of old memories — the temporofrontal hypothesis. Brain 1997;120(Part 8):1377– 99. [30] Kuskowski MA, Pardo JV. The role of the fusiform gyrus in successful encoding of face stimuli. Neuroimage 1999;9(6 Pt 1):599– 610. [31] Leveroni CL, Seidenberg M, Mayer AR, Mead LA, Binder JR, Rao SM. Neural systems underlying the recognition of familiar and newly learned faces. Journal of Neuroscience 2000;20(2):878– 86. [32] Madden DJ, Turkington TG, Provenzale JM, Denny LL, Hawk TC, Gottlob LR, Coleman RE. Adult age differences in the functional neuroanatomy of verbal recognition memory. Human Brain Mapping 1999;7(2):115– 35. [33] Martin AJ, Friston KJ, Colebatch JG, Frackowiak RS. Decreases in regional cerebral blood flow with normal aging. Journal of Cerebral Blood Flow and Metabolism 1991;11(4):684–9. [34] McCarthy G, Puce A, Gore JC, Allison T. Face-specific processing in the human fusiform gyrus. Journal of Cognitive Neuroscience 1997;9(5):605– 10. [35] McIntosh AR, Sekuler AB, Penpeci C, Rajah MN, Grady CL, Sekuler R, Bennett PJ. Recruitment of unique neural systems to support visual memory in normal aging. Current Biology 1999;9(21):1275– 8.
650
K. Herholz et al. / Neuropsychologia 39 (2001) 643–650
[36] O’Craven KM, Downing PE, Kanwisher N. MRI evidence for objects as the units of attentional selection. Nature 1999;401(6753):584–7. [37] Pantano P, Baron JC, Lebrun-Grandie P, Duquesnoy N, Bousser MG, Comar D. Regional cerebral blood flow and oxygen consumption in human aging. Stroke 1984;15(4):635– 41. [38] Petersen SE, Fiez JA. The processing of single words studied with positron emission tomography. Annual Review of Neuroscience 1993;16:509–30. [39] Puce A, Allison T, Asgari M, Gore JC, McCarthy G. Differential sensitivity of human visual cortex to faces, letterstrings, and textures: a functional magnetic resonance imaging study. Journal of Neuroscience 1996;16(16):5205–15. [40] Puce A, Allison T, McCarthy G. Electrophysiological studies of human face perception. III: effects of top- down processing on face-specific potentials. Cereberal Cortex 1999;9(5):445–58. [41] Raichle ME, Fiez JA, Videen TO, MacLeod AM, Pardo JV, Fox PT, Petersen SE. Practice-related changes in human brain functional anatomy during nonmotor learning. Cerebral Cortex 1994;4(1):8– 26. [42] Raichle ME, Martin WR, Herscovitch P, Mintun MA, Markham J. Brain blood flow measured with intravenous H2(15)O. II. Implementation and validation. Journal of Nuclear Medicine 1983;24(9):790–8. [43] Rajah MN, McIntosh AR, Grady CL. Frontotemporal interactions in face encoding and recognition. Cognitive Brain Research 1999;8:259– 69. [44] Rolls ET. The functions of the orbitofrontal cortex. Neurocase 1999;5:301– 12. [45] Rosier AM, Cornette L, Dupont P, Bormans G, Mortelmans L, Orban GA. Regional brain activity during shape recognition impaired by a scopolamine challenge to encoding. European Journal of Neuroscience 1999;11(10):3701–14. [46] Sergent J, Ohta S, MacDonald B. Functional neuroanatomy of face and object processing. A positron emission tomography study. Brain 1992;115(Pt 1):15–36.
.
[47] Sergent J, Signoret JL. Functional and anatomical decomposition of face processing: evidence from prosopagnosia and PET study of normal subjects. Philosophical Transactions of the Royal Society of London — Series B Biological Sciences 1992;335(1273):55– 61. [48] Sommer W, Komoss E, Schweinberger SR. Differential localization of brain systems subserving memory for names and faces in normal subjects with event-related potentials. Electroencephalography Clinical Neurophysiology 1997;102(3):192– 9. [49] Squire LR, Zola-Morgan S. The medial temporal lobe memory system. Science 1991;253:1380– 6. [50] Takahashi N, Kawamura M, Hirayama K, Shiota J, Isono O. Prosopagnosia: a clinical and anatomical study of four patients. Cortex 1995;31(2):317– 29. [51] Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of The Human Brain. Stuttgart: Georg Thieme Verlag, 1988. [52] Timmann D, Musso C, Kolb FP, Rijntjes M, Juptner M, Muller SP, Diener HC, Weiller C. Involvement of the human cerebellum during habituation of the acoustic startle response: a PET study. Journal of Neurology Neurosurgery Psychiatry 1998;65(5):771– 3. [53] Tulving E, Kapur S, Craik FI, Moscovitch M, Houle S. Hemispheric encoding/retrieval asymmetry in episodic memory: positron emission tomography findings. Proceedings National Academy of Science USA 1994;91(6):2016– 20. [54] Warrington EK. Recognition Memory Test (RMT). Windsor: NFER-Nelson Publ. Comp. Ltd, 1984. [55] Wienhard K, Dahlbom M, Eriksson L, Michel C, Bruckbauer T, Pietrzyk U, Heiss WD. The ECAT EXACT HR: performance of a new high resolution positron scanner. Journal of Computer Assisted Tomography 1994;18(1):110– 8. [56] Wilson BA, Cockburn J, Baddeley A. The Rivermead Behavioural Memory Test (RBMT). Reading: Thames Valley Test Group, 1985. [57] Wojciulik E, Kanwisher N, Driver J. Covert visual attention modulates face-specific activity in the human fusiform gyrus: fMRI study. Journal of Neurophysiology 1998;79(3):1574–8.
.