functional anatomy of high-resolution visual mental imagery...t/ourio, denis & mazoyer, 1998a) no...

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/12547900

Functional Anatomy of High-Resolution Visual Mental Imagery

Article in Journal of Cognitive Neuroscience · February 2000

DOI: 10.1162/08989290051137620 · Source: PubMed

CITATIONS

152READS

90

6 authors, including:

Some of the authors of this publication are also working on these related projects:

Archeoneuro View project

Hypertension and neuropsychiatric diseases View project

Emmanuel Mellet

Institut des Maladies Neurodegeneratives

105 PUBLICATIONS 6,277 CITATIONS

SEE PROFILE

Bernard Mazoyer

Université of Bordeaux Hospital and Medical school


SEE PROFILE

Stephen M Kosslyn

Harvard University


SEE PROFILE

Michel Denis

Computer Sciences Laboratory for Mechanics and Engineering Sciences


SEE PROFILE

All content following this page was uploaded by Bernard Mazoyer on 05 June 2014.

The user has requested enhancement of the downloaded file.

https://www.researchgate.net/publication/12547900_Functional_Anatomy_of_High-Resolution_Visual_Mental_Imagery?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_2&_esc=publicationCoverPdfhttps://www.researchgate.net/publication/12547900_Functional_Anatomy_of_High-Resolution_Visual_Mental_Imagery?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_3&_esc=publicationCoverPdfhttps://www.researchgate.net/project/Archeoneuro?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_9&_esc=publicationCoverPdfhttps://www.researchgate.net/project/Hypertension-and-neuropsychiatric-diseases?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_9&_esc=publicationCoverPdfhttps://www.researchgate.net/?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_1&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Emmanuel-Mellet?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_4&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Emmanuel-Mellet?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_5&_esc=publicationCoverPdfhttps://www.researchgate.net/institution/Institut-des-Maladies-Neurodegeneratives?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_6&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Emmanuel-Mellet?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_7&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Bernard-Mazoyer-3?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_4&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Bernard-Mazoyer-3?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_5&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Bernard-Mazoyer-3?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_7&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Stephen-Kosslyn?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_4&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Stephen-Kosslyn?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_5&_esc=publicationCoverPdfhttps://www.researchgate.net/institution/Harvard_University?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_6&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Stephen-Kosslyn?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_7&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Michel-Denis-2?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_4&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Michel-Denis-2?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_5&_esc=publicationCoverPdfhttps://www.researchgate.net/institution/Computer-Sciences-Laboratory-for-Mechanics-and-Engineering-Sciences?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_6&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Michel-Denis-2?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_7&_esc=publicationCoverPdfhttps://www.researchgate.net/profile/Bernard-Mazoyer-3?enrichId=rgreq-6996fb11952b7c5473a67db8421f4d45-XXX&enrichSource=Y292ZXJQYWdlOzEyNTQ3OTAwO0FTOjEwNDU5MDIzNDA5NTYyOEAxNDAxOTQ3NjYzOTUx&el=1_x_10&_esc=publicationCoverPdf

Functional Anatomy of High-Resolution VisualMental Imagery

E. Mellet, N. Tzourio-Mazoyer, S. Bricogne, and B. MazoyerUPRES 2127 Universite cle Caen & CEA-LRC 13V, Caen. France

S. M. KosslynHarvard University

M. DenisLIMSI CNRS. Orsay, France

Abstract

• This study had two purposes. First, in order to address thecontroversy regarding activation of the primary visual area(PVA) during visual mental imagery, regional cerebral bloodflow (rCBF) was recorded while subjects performed a task tha trequired high-resolution visual mental imagery. Second, inorder to discover whether verbal descriptions can engagevisual mechanisms during imagery in the same way as v isualstimuli, subjects memori/ed 3D scenes that were visual lypresented or were based on a verbal description. Comparisonof the results from the imagery conditions to a non-imagery

baseline condition revealed no activation in PVA for imagerybased on a verbal description and a significant decrease ofrCBF in this region for imagery based on visual learning. Thepattern of activation in other regions was very similar in thetwo conditions, including parietal, midbrain. cerebellar. pre-frontal. left insular, and right inferior temporal regions. Theseresults provide strong evidence that imagery based on verbaldescriptions can recruit regions known to be engaged in high-order visual processing. H

INTRODUCTION

The functional anatomy of visual mental imagery hasbeen a topic of intense debate, particularly as regardsthe role of early visual areas in the generation anil useof mental images (for example. Roland & Gulyas. 1994).At the center of the debate is the hypothesis thatduring imagery, primary visual cortex anil nearby cor-tical areas are used to reconstruct the local geometry ofthe surface of the visuali/ed object or scene (Kosslyn.1994a). This hypothesis is based on the fact that .thanks to the small receptive fields of its neurons, thisarea specifies shape with high-resolution (Fox et al.,1986). Indeed a number of researchers have reportedthat this region is activated during tasks requiring visualmental imagery of objects (Bookheimer et al.. 1998;Chen et al., 1998; Damasio et al.. 1993: Kosslyn et al.,1993, 1999; Kosslyn. Thompson. Kim & Alpert. 1995:Kosslyn. Thompson. Kim, Ranch & Alpert, 1996: LeBihan et al., 1993; Menon et al.. 1993; Sabbah et al.,1995). Moreover, it has been reported that , as occursduring actual perception, mental images rely on atopographical representation in primary visual cortex:mental imagery of small objects activates the mostposterior part of calcarine sulcus. whereas mental

imagery of increasingly larger objects activates increas-ingly more anterior parts of this sulcus (Kosslyn et al.1993, 1995). However, other laboratories have noifound evidence that early visual cortex is activateiduring visual mental imagery. During various visuamental imagery tasks, such as mental imagery of 2Fcolored patterns (Roland & Gulyas, 1995), mental exploration of a 2D scene (Mellet. Txourio, Denis, £Ma/over, 1995), mental construction of 3D pattern(Mellet et al., 1996), mental navigation (Ghaem et al.1997), or mental imagery of objects based on theidefinitions (Mellet. T/ourio, Denis & Mazoyer, 1998a)no activation of the primary or adjacent early visuacortex was observed. In fact, in some cases the regionacerebral blood flow (rCBF) values were actually lowcthan those found in the baseline conditions.

The disparity in results could arise from a number isources. For example, none of the studies that fountactivation in early visual cortex during visual ment;'imagery used spatial tasks, whereas most of the stuiliethat failed to find such activation relied upon sucitasks. In addition, most of the studies that failed t-find activation in these areas relied on stimuli stored ilong-term memory, whereas most of those thai di

r J(MK) .\l

find such activation relied on stimuli learned in theminutes prior the experiment. Moreover, many of thestudies that failed to find such activation tested sub-jects who scored high on spatial abilities tests, whereasnone of those that found such activation selectedsubjects in this way. And the studies differed in manyother ways, such as the baseline conditions used.Unfortunately, the corpus of available results doesnot reveal any clearcut pattern that allows one to inferwhich factors are crucial and which incidental forproducing activation in early visual areas during visualmental imagery.

One recent attempt to reconcile these divergentresults posits that the recruitment of early visual areasduring visual mental imagery depends on the spatialresolution of the mental images (Kosslyn, Thompson &\lpert, 1997; Thompson & Kosslyn, 1998). As mentalmages become increasingly vivid and accurate, they

come to resemble more closely percepts—and hence itis plausible that such images might recruit early visualareas that may underlie such properties in perception.The present study was designed to determine whethertasks that require high-resolution visual mental imageswill always engender activation of early visual cortex. Inthe present task, the subjects must make relativelydifficult discriminations, which can be performed wellonly if the image specifies shape with high-resolution.The discriminations we use are of comparable diff icul tyto those used in the Kosslyn et al. (1995) study, whichdid find activation of the primary visual area duringimagery. Thus, if we failed to obtain such activation inthe present study, this failure cannot be ascribed to thisfactor.

Another aspect of the task could have an effect onactivity in the primary visual cortex during mental'magery: the learning modality. Visual mental imagesan be created either on the basis of previously stored

visual memories of objects or scenes, or on the basisof verbal descriptions. However, previous behavioralstudies have shown that visual mental images con-structed from verbal descriptions are functionally verymuch like visual mental images constructed fromvisual memories (Denis £ Cocude, 1992). In a. pre-vious neuroimaging study, Mellet et al. found that theconstruction of mental images from verbal instructionsrelied on associative visual areas, but did not give riseto activation in PVA. It is possible that early visualareas tend to be recruited only when previously storedvisual memories are recalled. In order to clarify thispoint, our experimental design included the two typesof learning.

The goal of the present study was to elucidate theeffect of both, requiring the subjects to use high-resolu-tion images and the learning modality on the functionalanatomy of visual mental imagery. In order to addressthese issues, we studied performance and functionalanatomy during the same mental imagery task when the

images arise from memories that initially were encodedvisually or verbally. Using ISO-labeled water, nine se-quential PET measurements of regional rCBF wereobtained for each of seven subjects. We conductedthree replications of each of three experimental condi-tions: a high-resolution mental imagery task based onvisually learned material, the same task based on verb-ally learned material, and a baseline task. During thelearning phase in the visual learning condition, thesubjects were asked to study and memorix.e scenes priorto scanning. During the learning phase in the verballearning condition, the subjects were instructed to listento verbal descriptions of how shapes were to be ar-ranged, and to form and memorix.e a visual image foreach of the described scenes. This was also done priorto scanning. In both imagery conditions, each scene wascomposed of four simple geometric shapes, arranged ona base; the scenes differed only in the ordering of theelements on the base (see Figure 1A). During theimagery conditions, the subjects visualized one of thescenes, then they heard the labels of two positions onthe base and a statement of relative height; they were toconsult their image and decide whether the statementcorrectly characteri/.ed the relative heights of the sceneabove the two named positions on the base. In all cases,the judgments required high-resolution in the images.Finally, in addition to the two imagery conditions,subjects simply heard the stimuli and responded with-

Figure 1. (A) The base on which four geometric shapes were placed.Before the PET experiment, each subject learned the locationsspecified by the letters. (B) An example of a typical 3D scene that thesubject had to learn from a visual or a verbal presentation. During thePET measurements, the subjects, eyes closed, had to form a visualmental image of the scene and evaluate statements about heightcomparisons, such as "D lower than B", heard through earphones. Inthe present example, the answer would be "wrong".

Mellet et al. 99

out visualizing or making judgments; this was the base-line condition.

RESULTS

Behavioral Results

During the PET measurements, we recorded the num-ber of correct responses and the response times. Thepercentage of correct responses in the visual learningcondition was 81.5±3.2 %, and in the verbal learningcondition it was 80.6±4.8 % (mean±standard error ofthe mean). In both conditions, the subjects clearlyperformed better than chance (p=0.005 and p=0.01,respectively, Mest), and there was no difference inaccuracy between the two conditions (p=0.8 paired t-test). A slight improvement in the number of correctresponses in the third replication of the task was evidentin the visual learning condition (p=0.05, ANOVA forrepeated measures), but no such practice effect wasevident in the verbal learning condition (p=0.7). Themean response time over subjects was 8.852±0.295 sec(mean±standard error of the mean). The mean re-sponse time in the visual learning condition was8.691±0.412 sec, which did not differ from that in the

verbal construction condition (9.014±0.424 sec, />=0.7paired f-test). No replication effect was observed for themean response time in either condition (p>0.3 in bothcases).

PET Results

We report below the results of a series of contrasts, eachof which reveals different aspects of the neuroimagingresults.

Imagery Based on Visual Learning Compared toBaseline

As presented in Table 1 and Figure 2 (left) during themental imagery task based on visually learned material,the activation exhibiting the largest Z score extendedfrom the precuneus symmetrically in the two hemi-spheres toward the intraparietal sulcus and in the super-ior occipital gyrus. This activation was not symmetrical:It had a larger anterior extent in the left intraparietalsulcus, up to the postcentral sulcus.

A large cluster of activation, extending from z=10mm above the AC-PC line to z=-36 mm below this

Foci of significant normalized regional cerebral blood flow (NrCBF) increases when imagery was based on visual learning compared with thebaseline condition. The data, based on seven subjects, are local maxima detected with SPM software. Activated region volumes are given in voxels.Within these regions, the anatomical localization of the maximum Z scores of the voxel is given on the basis of the Talairach and Tournoux atlas(1988), using their stereotactic coordinates in mm. (R., right; L., left).

"Significant at p

Figure 2. Statistical parametric map revealing the significant activations when subjects formed images based on visual learning compared to thebaseline condition (Left) anil uheri they formed images haseil on verbal learning compared to the baseline condition (Right) . The volumes arethrcsholdcd 10 project in thivc orthogonal direilions, sagittal, coronal, anil transverse and reach threshold at '/.—3-1 (/>

conditions ( tha t is. a decrease in rCBF during the visuallearning imagery condition).

DISCUSSION

Vie now return to the two major issues that led us toconduct this study, the question of whether the require-ment to use high-resolution imagery will invoke activa-

n in earlv visual areas, and the relation between

imagery based on visually learned s t imul i vs. imagerybased on verbally learned stimuli. VC'e interpret theresults as they bear on these issues.

Medial Occipital Cortex and High-ResolutionMental Images

One goal of this study was to test the hypothesis that themedial occipital cortex, including the primary visual

Foci of significant NrCBF increases when imagery' was based on visual learning compared with when imagery was based on verbal learning and viceversa (for details, see Table 1 legend).

Mellet et al. 103

Figure 3. (a) Statistical para-metric map revealing the sig-nificant activation clusters whensubjects formed imagery basedon verbal learning minus ima-gery based on visual learning (inyellow) and vice versa (in red).The volumes are thresholded toproject in three orthogonaldirections, sagittal, coronal, andtransverse, and reach thresholdat Z=3.1 (p

way in the same subjects, our results provide an anato-nio-functional substrate to these behavioral reports. Thesimilarity in the results from the two imagery conditionsindicates that the cerebral structures that underlie thegeneration and the maintenance of the mental image, aswell as the operations performed on the image, did notdepend on the modality in which the information wasoriginally acquired. This finding is all the more impress-ive because the imagery task, performed following visualor verbal learning, included several different steps. First,when they heard the number identifying the scene,subjects had to visually recall the scene and then retaina high-resolution visual image un t i l the comparisonstatement was delivered. Then, they had to identifythe locations corresponding to the named letters, andassess the height difference at the two points. Followingthis, they had to select and produce the appropriate

•spon.se. In the following sections we wil l discuss howmemory retrieval, visual working memory, attentionalshift ing, and high-resolution shape processing couldaccount for the activation that we observed.

Spalial Components and Occipito-Parielal Act/ration

The bilateral act ivat ion of superior occipito-parietalareas could reflect the spatial processing required bythe task: The subject hail to form a mental image inwhich each geometric shape was accurately placed onthe base and positioned over the corresponding let-ters. The superior occipital gyms anil the posteriorpart of the intraparieta! sulcus are portions of thedorsal system, which is speciali/ed for the processingof spatial information (Haxby et al., 1994). This path-wax' is also involved in the spatial processing of visualmental images, as indicated in numerous neuroima-ging studies of mental imagery that included a spatial

mponent (Cohen et al., 1996; Kavvashima, Roland &u'Sullivan. 1995; Kosslyn et al., 1993, 1998; Mellet etal., 1995; 1996; Tagaris et al., 1997). In this contextour results are consistent with the role played by thedorsal system in the spatial processing of visual mentalimages and show, in addition, that these highly spe-cialized visual areas are recruited independently of themodality in which the information was acquired, vi-suallv or verbally.

Object Visual Imagery and the Ventral System

In order to perform our task, the subjects had to formimages of objects—and, thus, we would expect activationin the ventral system. The involvement of the ventralsystem during object imagery has been consistentlyreported in previous studies (D'Esposito et al., 1997;Kosslyn et al., 1993; Mellet et al., 1996; 1998a; Roland &Gulyas, 1995). The right inferior temporal activationdetected in both imagery tasks is thus consistent withthe role of this cortical area in the processing of shapes in

both visual perception anil visual imagery. However, theright laterali/ation of this activation challenges the left-ward functional speciali/ation of the inferior temporalgyrus for mental image generation proposed by others(D'Esposito et al., 1997). It has previously been sug-gested that the right inferior temporal gyrus may beactivated by the generation of complex mental images(Mellet et al., 1998a; Mellet, Petit, Ma/oyer, Denis &T/ourio. 1998b), anil the images required in the presentstudy appear to qualify as "complex." Indeed, most ofthe studies based on complex and or detailed mentalimagery reported an activation of the right inferiortemporal gyrus (Kosslyn et al., 1993; Mellet et al., 1996;1998a). whereas those that used simple and/or poorlydetailed imagery failed to activate this cortical area(Bookheimer et al., 1998; D'Esposito et al., 1997). Ourresults are thus consistent with a specific role of the rightinferior temporal gyrus in this type of imagery.

Memory Components and Frontal Activations

During the imagery tasks, the frontal lobe was activatedin three distinct locations: (1) a right dorsolateral pre-frontal area that included an activation in the superiorfrontal gyrus, middle frontal, anil inferior frontal gyri; (2)the intersection of the left middle frontal sulcus and theprecentral gyrus; and, (3) the border of the left inferiorfrontal gyrus and of the anterior insula. These regionsmay constitute a network that underlies the memoryact iv i ty involved in visual mental imagery. The rightdorsolateral prefrontal activations that we founil couldbe related to retrieval from episodic memory, a crucialprerequisite of our mental imagery task (Buckner. Rai-chelle. Miezin. &. Petersen, 1996; Moscovitch, Kapur,Kohler, & Houle, 1995; Roland & Gulyas. 1995: Tulving,Kapur, Craik, & Moscovitch, 1994). VC'e also foundactivation in the left middle frontal sulcus. Note thatalthough this activation just failed to reach thep

The activation of the left anterior insuJa remainsdifficult to interpret. Most of our subjects reported that,during the mental inspection of the image, they silentlyrehearsed the comparison term they had just heard. Theactivation of the left anterior insula could be related tosuch verbal rehearsal, which is consistent with thefinding that this cortical region was activated duringverbal working memory studies (Paulesu, Frith, & Frack-owiak, 1993; Smith, Jorrides, & Koeppe, 1996).

Mesencephalo-Thalamic and Cerebellar Activations

The imagery task required the subjects to attend to thevisualized shapes, and such attention may in turn havedrawn on systems that keep the brain aroused. Themesencephalon encompasses a part of the ascendingactivating reticular system in humans, and it has beendemonstrated that it plays a key role, together with themedial thalamus, in maintaining a state of high vigilanceand attention (both in vision, for example, Kinomura,Larsson, Gulyas, & Roland, 1996, and audition, forexample, Paus et al., 1997). However, in our study, mostof the attentional resources were focused on a purelymental scene. Our results, thus, emphasize that themidbrain activating system is not only engaged byattention to sensory input, but also can be involved inabsence of perceptual input.

We also observed an important activation of thevermis that extended bilaterally to an output nucleus,the dentate nucleus. A cerebellar contribution to cogni-tive operations is 1 of the major findings provided byrecent imaging studies (see Schmahmann. 1996). Medialactivation of the cerebellum has been repeatedlyreported during visual episodic memory tasks (Finket al., 1996; Moscovitch et al., 1995) and during short-term verbal maintenance (Fiez et al., 1996). As notedabove, these cognitive processes were required to per-form our task. We found cerebellar activations thatencompassed bilaterally the dentate nucleus. Amongthe cerebellar deep nuclei, the dentate seems to beparticularly involved in higher cognitive functions (Mid-dleton £ Strick, 1994). Although it is not yet possible toassign a specific role to this output nucleus, the presentresults, together with those from previous studies, sup-port the inference that this region deals with internalrepresentation; indeed, bilateral activations of this nu-cleus have been described during a pegboard puzzlesolving task (Kim, Ugurbil, & Strick, 1994), a sensorydiscrimination task (Gao et al., 1998), and a conceptualreasoning task derived from the Wisconsin Sorting CardTest (Rao et al., 1997). It is of interest that in nonhumanprimates this region is connected via the medial thala-mus to the prefrontal cortex (Middleton & Strick, 1994).Consistent with the results of the present study, Andrea-sen et al. propose that such a network plays a role inhumans in the coordination and sequencing of thoughts(Andreasen, Paradiso, & O'Leary, 1998).

EXPERIMENTAL PROCEDURES

Subjects

Seven right-handed healthy male French students (age:20-25 years) volunteered to participate in this studyAll were free from nervous disease or injury and had n < ,abnormality on their Tl-weighted magnetic resonanceimages (MRI). Written, informed consent was obtainedfrom each subject after the procedures had been fullyexplained. This study was approved by the local In-stitutional Review Board. In order to ensure optimalhomogeneity of the sample of the subjects with respectto their imagery abilities, subjects were selected as"high imagers" (or, more specifically, as having highspatial ability) on the basis of their scores on theMinnesota Paper Form Board (MPFB) and on theMental Rotations Test (MRT); all subjects scored be-yond the 75th percentile of a population of 208 malesubjects. The mean MPFB score for the subjects was23.9±1.3 (mean±SD, whole population 19.7±4.4) andtheir mean MRT score was 15.6±1.8 (whole population11.8±4.4).

Materials

The subjects were first trained to form a vivid mentalimage of a base, as illustrated in Figure 1A. This baseincluded letters that demarcated specific positions alongit. Scenes were constructed by placing four shapes onthe base, a red triangle, a green rectangle, a yellowrectangle, and a blue cone. Figure IB provides anillustration of one of the scenes. Twenty-four differentscenes were thus available, but 16 only were used in theexperimental paradigm: four in the training phase priorthe PET session (two for training in the visual learningimagery condition and two for training in the verballearning imagery condition), two in each PET conditionrepeated three times for each type of learning (12scenes). In addition to the 16 scenes, 16 sets of nineauditorily delivered comparison statements were pre-pared. These statements were delivered by computer toheadphones. The notion of high-resolution mentalimages has been operationalized elsewhere as follows:A task requires high-resolution if one needs to attend todistinct portions of shapes that have subtle variations orif the to-be-visualized patterns have parts that subtendless than 0.5 of visual angle and these parts need to bekept distinct from other part (Thompson & Kosslyn.1998). The high-resolution mental imagery task wedesigned in our study fulfilled both criteria.

Procedure

Bebai ioral Procedure

We began with a training phase, prior to scanning. In thei n i t i a l t raining, the subjects learned two scenes. Asoccurred in the subsequent experimental conditions.

Volume /_'. A'

the two scenes were identified as "I" and "2." Prior tolearning the scenes proper, subjects first memori/ed thebase, including the letters and their precise positions.They then were shown the geometric forms, one at atime, and asked to memorixe their appearance.

The subjects were taught two methods of learning thescenes: from a visual presentation and from a verbaldescription. In the visual learning condition, they saw acomplete scene, formed by placing four shapes on thebase. The subjects were requested to memorixe thescene as accurately as possible. After studying a scene,the subjects closed their eyes and v isuali/ed it, and thenopened their eyes and compared the image against theactual stimulus, correcting their image as necessary. Thisprocedure was repeated until the subjects felt that theirmental image corresponded to the display. The st imuliwere left in free view for as long as the subjects required

> memorixe the display. The same procedure was usedwith scene "2".

In the verbal learning condition, the subjects neveractually saw the scenes. Vi'ith eyes closed, the subjectswere first asked to recall the mental image of the base,including the letters and their precise positions alongthe base. Then they heard a description of the order inwhich geometric forms should be arranged on the basefor s t imulus number 1; for example, the construction ofFigure 115 would be described as "red triangle, . . . greenrectangle. . . . blue cone, ... yellow rectangle". Thesubjects were asked to build and memorixe the visualmental image of the corresponding scene. They thenheard the description of scene "2". and again buil t andmemori/ed the corresponding scene. This procedurewas repeated unti l the subjects reported that they hadmemorix.ed the two scenes (3 times on average).

After the subjects had learned the initial two scenes,whichever the type of learning, they were trained to

jrform the mental imagery task proper: with eyesclosed, they first heard the identifying number of oneof the two scenes through earphones, which promptedthem to generate a mental image of the whole scene, asaccurately as possible, with sharp corners and edges.Four seconds later, they heard a question correspondingto a height comparison term, such as "D lower than B,"and had to mentally compare the heights of the twoshapes being visualized immediately above the namedpoints (see Figure IB, dotted lines). If the statement wascorrect, the subjects were to respond by saying "right";it it was incorrect, they were to say "wrong." Thesubjects were asked to respond as quickly as possible,while still remaining as accurate as possible. Once thesubjects had responded, the identification number ofthe other scene was delivered after a lag of 750 msec, thesubjects were then to generate the corresponding men-tal image, and four seconds later they heard a newcomparison term.

The subjects were interviewed to ensure that theyunderstood the nature of the learning procedures and

task, and any confusions were corrected anil questionsanswered.

Pl-T Procedure

PliT conditions. There were three PET conditions:Imagery after visual learning, imagery after verballearning, and a baseline. For the two imagery condi-tions, the subjects memori/ed two scenes during the15 min prior to scanning. They used the samemethod to memorixe the scenes as was taught duringthe training phase. During the PET measurements,the subjects performed the imagery task, with thecues and comparison statements being deliveredthrough earphones. The subjects had to respond bysaying "right" or "wrong" into a microphone con-nected to a computer. After each response, thecomputer recorded the response time, then 750 mseclater delivered the identification number of the otherscene, and 4 sec later a new comparison statement.The investigator recorded whether each response wascorrect or incorrect. Each condition consisted of ninecomparison statements, alternating from one scene tothe other. A different set of nine comparison termswas used in each of the conditions in each replica-tion, so that the subjects could not know in advancewhich judgments they would need to perform. Thus,the subjects could not simply perform the judgmentsand memorixe them during the learning phase. Inaddition to the two imagery conditions, there was abaseline task. During the baseline task, the subjectsclosed their eyes, listened to randomly chosen com-parison statements delivered even seven seconds,and alternatively said "right" and "wrong" after eachterm. The subjects were instructed not to producemental images. Although the subjects had. at thispoint, already gone through the training phase, and,thus, knew the nature of the task, they all reportedthat they could refrain from forming images orperforming the task during the baseline measure-ments. All conditions were conducted with eyesclosed in total darkness.

Image data acquisition. For each rCBF measurement,63 2.425 mm thick contiguous brain slices were ac-quired simultaneously on an ECAT HR+PET camera. Ablack tent-like chamber was set up all around the PETtomograph, to ensure that the subjects were in totaldarkness. Emission data were acquired in 3D mode.Tasks were started 20 sec before the intravenous bolusinjection of 8 mCi of 15O-labeled water. A single 90-secscan was acquired and reconstructed (including acorrection for head attenuation using a measuredtransmission scan) with a Manning filter of 0.5 mm.~cut off frequency and a pixel size of 2x2 mm". Thebetween-scan time interval was 15 min, and the condi-tion order was the same in all subjects, specifically:

Mellet et al. 107

baseline, imagery after visual learning, and imageryafter verbal learning, and this series was replicatedthree times.

Data analysis. After automatic realignment (AIR,Woods, Grafton, Holmes, Cherry, & Mazziotta, 1997),the original brain images were transformed into thestandard stereotactic Talairach space using the MNItemplate. The images were smoothed using a Gaussianfilter of 12 mm. FWHM leading to a final smoothness of15 mm. FWHM. The rCBF was normalized within andbetween subjects using a proportional model (scaling).The comparisons across conditions were made by way of/-statistics. Statistical parametric maps corresponding tocomparisons between the three conditions were gener-ated with the 1996 version of SPM (Friston et al., 1995).As indicated earlier, the experimental protocol wasdesigned to use the baseline condition as a controlcondition for both imagery tasks. The two imageryconditions were also compared directly. For each com-parison, the voxel amplitude t-map was transformed in aZ volume that was threshokled at Zu = 3.09, whichcorresponds to a 0.001 confidence level (without correc-tion for multiple comparisons).

The conjunction analysis was performed using thesame SPM package. It revealed the common sites ofactivation in the two imagery conditions compared tothe baseline condition (Price et al., 199"). This type ofanalysis required at least two pairs of tasks. In our study,the first pair corresponded to the two imagery tasks,whereas the second pair consisted of two differentreplications of the baseline condition. As in the subtrac-tion analysis, the resulting Z volume was threshokled atZ0=3.09.'

Acknowledgments

The authors arc deeply indebted to their colleagues V.Beaudouin. P. Lochon. O. Third, and G Perchey for theirinva luable help in tracer production anil data acquisition andto. L Petit. M Pesenti and O. Hondo for their thoughtfulcomments. S. M. Kosslyn was supported by a grant from theJames S. McDonnell Foundation.

Reprint requests should be sent to E. Mellet, Crouped'Imagerie Neurofonctionnelle, G1P Cyceron, BP 5229. l40"4CAEN Cedex.

REFERENCESAndreasen, N. C.. Paradiso. S., & O'Leary, D. S. (1998). "Cog-

nitive Dysmetria" as an integrative theory of schi/ophrenia: Adysfunction in cortical-sulx'ortical-cerebellar circuitry?Schizophrenia Bulk-tin, 2~t, 203-218.

Bookheimer. S. Y.. X.cffiro. T. A., Blaxton, T. A., Gaillard, W.D., Malow, B., it Theodore. W. I I . (1998). Regional cere-bral blood flow during auditory responsive naming: evi-dence for cross-modality neural activation. NeuroKeporl,9. 2i()9-24 13.

Buckner, K. 1... Kaichle. M. K.. Miexin, F. M.. it Peterson. S. K.(199(i). Functional anaiomk .studies of memon retrieval for

auditory words and visual pictures. The Journal ofNeu-roscience, 16, 6219-6235-

Chen, W., Kato, T., Zhu, X. -H., Ogawa, S., Tank, D. W. &Ugurbil, K. (1998). Human primary visual cortex and lateralgeniculate nucleus activation during visual imagery. Neii-roReport, 9, 3669-3674.

Cohen, J. D., Peristein, W. M., Braver, T. S., Nystrom, L. E., Noll.D. C., Jonides, J., & Smith, E. E. (1997). Temporal dynamicof brain activation during a working-memory task. Nature.386, 604-607.

Cohen, M. S., Kosslyn, S. M., Breiter, H. C., DiGirolamo, G. J..Thompson, W. L, Anderson, A. K., Bookheimer, S. Y., Rosen.B. R., & Belliveau, J. W. (1996). Changes in cortical activityduring mental rotation: A mapping study using functionalmagnetic resonance imaging. Brain, 119, 89-100.

Courtney, S. M., Ungerleider. L. G., Keil, K.. & Haxby, J. V.(1997). Transient and sustained activity in a distributedneural system for human-working memory. Nature, 386.608-611'.

D'Esposito. M., Detre, J. A., Aguirre, G. K.. Stallcup, M., Alsop.D. C., Tipett, L. J., & Farah, M. J. (199"). A functional MR1study of mental-image generation. .\'ettrops\-cbolo}>ia, 35.725-~30.

Damasio. H., Grabowski, T. J., Damasio. A. R.. Tranel, D., Boles-Ponto. L. L.. & VTatkins, G. L. (1993). Visual recall with eyesclosed and covered activates early visual cortices. Nen-mxcience Abstracts, 19, 1-603 (Abstract).

Denis. M. £ Cocude. M. (1992). Structural properties of visualimages constructed from poorly, or well-structured, verbaldescriptions. Memoiy and Cognition. 20. 49~-506.

Denis. M.. Goncalves, M. R., & Memmi. D. (1995). Mentalscanning of visual images generated from verbal descrip-tions: Toward a model of image accuracy. \eiiropSYcbolo-Hia. 33. 1511-1530.

Fiex, |. A.. Raife. F... Balota. D. A.. Schwarx. .1. P.. Raichle. M. E..it Peterson. S. E. (1996). A positron emission tomographystudy of the short-term maintenance of verbal information.The Journal oj ~5-4282.

Friston. K. J.. Holmes, A.. VCorsloy, K., Poline. J. -B., Frith. C. D..& Frackowiak, R. S. J. (1995). Statistical parametric maps infunctional imaging: A general approach. Human BrainMapping, J. 189-201.

Gao.J. -11., Parson. I.. M.. Bower. J. M., Xiong.J.. U.J., & Fox. P.T. (1998). Cerebellum implicated in sensory acquisition anildiscrimination rather than motor control. Science, 272,

Ghaem, O.. Mellet. E., Crivello, F.. Txourio. N., Max.oyer. B.,Borthox. A.. & Denis. M. (199T). Mental navigation alongmemorixed routes activates the hippocampus, precuneus,and insula. Neuroreport, 8, ~39-744.

Haxby, J. V., Horwitx, B., Ungerleider, L. G., Maisog, J. M.,Pietrini. P., & Grady, C. L (1994). The functional organixa-tion of human extrastriate cortex: A PET-rCBF study of se-lective attention to faces and locations. The Journal ofNeuroscience, 1-4, 6336-6353-

Kawashima, R.. Roland, P. E.. & O'Sullivan, B. T. (1995).Functional anatomy of reaching anil visuo-motor learning:A positron emission tomography. Cerebral Cortex, J,111-122.

Kim, S. G., I 'gurbil, K.. it Strick. P. I.. (1994). Activation of acerebellar output nucleus during cognitive processing.Science. J65, 9)9-951.

Kinomura. S.. Lirsson. |.. Gulvas, B.. it Roland. P. K. ( 1996).

KM Journal oj'('o^niiire \enroscience Volume 12. \nniln'r /

Activation by attention of the human-reticular formation andthalamic intralaminar nuclei. Science, 271. 512-515.

Kosslyn, S. M. (199-»a). Image, awl brain. Cambridge: MITPress.

Kosslyn. S. M., Al|X.-rt, N. M., Thompson, W. [.., Chabris, C. F.,Ranch, S. I... it Anderson. A. K. (1994b). Identifying objectsseen from different viewpoints: A PET investigation. Brain,117, 1055-10"!.

Kosslyn, S. M., Alpert. N. M., Thompson, W. L, Maljkovic,V.. Weise, S. B., Chabris. C. F., Hamilton, S. E., Rauch,S. L, & Buonanno, F. S. (1993). Visual-mental imageryactivates topographically-organized visual cortex: PF.T in-vestigations. Journal of Cognitive Neuroscience, 5,263-28".

Kosslyn, S. M., DiGirolamo, G. J., Thompson, W. I.., & Alpert, N.M. (1998). Mental rotation of objects versus hands: Neuralmechanisms revealed by positron emission tomography.Psychopbysio/ogy. .i, 151-161.

Kosslyn. S. M.. Pascual-Leone, A., Felician, O., Camposano, S.,Keenan. J. P., Thompson, W. L, Ganis, G., Sukel, K. E., &Alpert, N. M. (1999). The role of area I7 in visual imagery:

*"" Convergent evidence from PET and rTMS. Science, 28-i,ur-ro.

Kosslyn. S. M.. Thompson. VX'. I... &. Alpert. N. M. (199"). Neuralsystems shared by visual imagery and visual perception: Apositron emission tomography study. Neumimage. 6, 320-33-».

Kosslyn, S. M.. Thompson, W. I... Kim, I. |.. & Alpert, N. M.(1995). Topographical representations of mental images inprimary visual cortex. Nature. i7H. 496—198.

Kosslyn. S. M.. Thompson. VO. I... Kim. I. J.. Ranch, S. I.. £Alpert. N. M. (1996). Individual differences in cerebral bloodflow in area 1" predict the time to evaluate visualized letters.Journal of Cognitive Nenmscience, K. "8-82.

I.e Bihan. D., R. Turner. T. A. Xeffim, C. A. Cunod, P. Jex/ard &V. Bonnerot. (1993). Activation of human primary visualcortex during visual recall: A magnetic resonance imagingstudy. Proceedings of the National Academy of Sciences ofthe United States of America. 5. 11802-11805.

Mellet. E., Txourio. N., Denis, M., & Mazoyer, B. (1995). Apositron emission tomography study of visual, and mental,spatial exploration. Journal of Cognitive Neuroscience, 7,433-445.llet, E., Tzourio, N., Crivello, F., Joliot, M., Denis, M., &

Mazoyer, B. (1996). Functional anatomy of spatial mentalimagery generated from verbal instruction. The Journal ofNeuroscience, 16, 6504-6512.

Mellet, E., Tzourio, N., Denis, M., & Mazoyer, B. (1998a). Cor-tical anatomy of mental imagery of concrete nouns based ontheir dictionary definition. NeuroReport, 9, 803-809.

Mellet. E., Petit, L, Mazoyer, B., Denis, M., & Tzourio, N.(1998b). Reopening the imagery debate: Lessons fromfunctional anatomy. Neitroimage, 8, 129-139.

Menon, R. S., Ogawa, S., Tank, D. W., Ellermann, J. M., Merkele,H., & Ugurbil, K. (1993). Visual-mental imagery by functionalbrain MRI. In D. Le Bihan, R. Turner, M. Mos'ley, & J. Hyde(Eds.), Functional MRI of the brain: A workshop presented!>y the society of magnetic resonance in medicine and thesociety for magnetic resonance imaging. Arlington, VA:Society of Magnetic Resonance in Medicine.

x'idclleton, F. A. & Strick, P. L. (1994). Anatomical evidence forcerebellar and basal ganglia involvement in higher cognitivefuntion. Science, 266, 458-461.

Moscovitch. M.. Kapur, S., Kohlcr, S.. & Houle, S. (1995).Distinct neural correlates of visual long-term memory forspatial location antl object identity: A positron emission to-mography study in humans. Proceedings of the NationalAcademy of Sciences of the I 'nited States of America, 92,3^21-3725.

Paulesu. E.. Frith, C. D.. & Frackowiak, R. S. J. (1993). TheNeural Correlates of the Verbal Component of WorkingMemory. Nature, 362, 342-3-15.

Pans. T., Xatorre, R. J., Hofle, N., Carmanos, '/... Gotman, J.,Petride.s, M.. &. Evans, A. C. (199"). Time-related changes inneural systems underlying attention and arrousal during theperformance of an auditory-vigilance task. Journal of Cog-nitive Neuroscience. 9, 392—i08.

Price. C. J., Moore, C. J.. it Friston. K. J. (199~). Subtractions,conjunctions and interactions in experimental design of ac-tivations studies. Unman Krain Mapping. 5, 264-272.

Rao, S. M.. Bobholz, J. A., Hammeke, T. A., Rosen, A. C.,Woodley. S. J., Cunningham, J. M., Cox, R. VC'., Stein, E. A., &Binder, J. R. (199"). Functional MRI for subcortical partici-pation in conceptual reason n ing skills. NenroReport, 8,198--1993.

Roland, P. E. it Gulps, B. (1994). Visual imagery and visualrepresentation. Trends in Nenrosciences, 17, 281-286.

Roland. P. E. cS: Gulyas. B. (1995). Visual memory, visual ima-gery, and visual recognition of large field patterns by humanbrain: Functional anatomy by positron emission tomogra-phy. Cerehral Cortex. 1. 79-93.

Sabbah. P.. Simond. G.. Levrier. O.. Habib, M., Trabaud, V.,Murayama. N., Mazoyer. B. M., Briant, J. F., Raybaud, C., &Salamon. G. (1995). Functional magnetic resonance imagingat 1.5 T during sensorimotor and cognitive task. EuropeanNeurology. J5. 131-136.

Schmahmann. J. D. (1996). From movement to thought: Ana-tomic substrates of the cerebellar contribution to cognitiveprocessing Human Brain Mapping, -i, l~-4-198.

Smith. E. E.. Jonides. J.. & Koeppe. R. A. (1996). Dissociatingverbal and spatial working memory using PET. CerebralCortex. (•>. 11-20.

Smith, E. E.. Jonides. J.. Koeppe, R. A., Awh, E., Shumacher, E.H.. & Minoshima, S. (1995). Spatial vs. object workingmemory: PET investigations. Journal of Cognitive Neit-roscience, 7, 337-356.

Tagaris, G. A., Kim, S. G., Strupp, J. P., Andersen, P., Ugurbil, K.,&. Georgopoulos, A. P. (1997). Mental rotation studied byfunctional magnetic resonance imaging at high field (4 Te-sla): Performance and cortical activation. Journal of Cogni-tive Neuroscience, 9, 419-432.

Thompson, W. L. & Kosslyn, S. M. (1998). Neural systemsactivated during visual mental imagery: A review andmeta-analyses. In A. W. Toga &. J. C. Mazziotta, (Eds.),Brain mapping. The applications. New York: AcademicPress.

Tulving, E., Kapur, S., Craik, F. I. M., Moscovitch, M., &Houle, S. (1994). Hemispheric encoding/retrieval asym-metry in episodic memory: Positron emission tomogra-phy findings. Proceedings of the National Academy ofSciences of the United States of America, 91, 2016-2020.

Woods, R. P., Grafton, S. T., Holmes, C. J., Cherry, S. R., &Mazziotta, J. C. (1997). Automated image registration: I.General methods and intrasubject validation. Journal ofComputer-Assisted Tomography, 22, 139-152.

Mellet et al. 109

View publication statsView publication stats
https://www.researchgate.net/publication/12547900

functional anatomy of high-resolution visual mental imagery...t/ourio, denis & mazoyer, 1998a) no...

Documents