A Positron Emission Tomography Study of Visual and Mental Spatial Exploration E. Mellet and N. Tzourio zyxwvu CEA-DRM,Orsay, France M. Denis LIMSI-(:NKS, Orsay, France B. Mazoyer CEA-DKM, Orsay and Ilniversitk, Paris 7, France Abstract vc'e me;isured normalized regional cerebral blood flow (NrCBF) using positron emission tomography (PET) and zyxwvut ox)'- gem1 5-l;iheled water in eight young right-handed healthy vol- unteers \elected zyxwvutsrqp as high-imagers. during 2 runs of zyxwvuts 3 different conditicins: I, rest in total darkness 2; visual exploration of a map 3; mental exploration of the Same map in total darkness. NrCBF images were aligned with individual magnetic reso- nance images (MKI), and NrCBF variations between pairs of measurvnients (N zyxwvutsrqp = 15) were computed in regions of interest having ;inatoniical boundaries that were defined using a three- dimensional (3-D) reconstruction of each subject MKI. During INTRODUCTION Despite tletiri Bergson's famous statement in Matikre et M6moiw (Hergson, 1896) that the nervous system is not a machine to create representations, cognitive psychol- zyxwvut ogy has found in people's abilities to generate and ma- nipulatc internal representations a wide field of investigations. The concept of mental representation has become increasingly important in theories advanced to account for mechanisms of brain activity (Rumelhart zyxwvu Sr Norman. 1988; Kosslyn Sr Koenig, 1992). Representations are now viewed as psychological entities resulting from subject's interactions with the external world, and stand- ing for absent or distant objects. Some representations are highly symbolic by nature, as is the case for linguistic or conceptual representations, whereas others preserve features zyxwvutsr ;is well as the internal structure of perceptual events in an highly analog fashion. This is typically the casc for visual mental images for which recent theories assume the existence of stored representations common to visual perception, and a similarit) of processes that access and nianipulate these representations (Kosslyn, 1980, 1987). Indeed. most visual exploration, we found bilateral activations of primary visual areas, superior and inferior occipital gyri? fusiforrn and lingual gyri, cuneus and prrcuneus, bilateral superior parietal, and angular gyri. The right lateral premotor area was also activated during this task while superior temporal gyri and Uroca's ;ma were deactivated. Hy contrast. mental exploration activated the right superior occipital cortex, the supplemen- tary motor area, and the cerebellar vermis. No activation was observed in the primary visual area. These results argue for a specific participation of the superior occipital cortex in the generation and maintenance of visual mental images. W investigations about visual mental imagery postulate strong functional relationships between mechanisms serving imagery and those serving perception (Finktt, 1989; Denis, 1991; Kosslyn Sr Koenig, 1992). Neuropsy- chological studies have shown that bilateral lesions in the occipital cortex give rise to similar impairments in visual perception and image generation parah, 1984). Strong arguments are also provided by studies showing similar deficits in perceptual and imaginal tasks in pa- tients suffering from hemineglect (Basso, Bisiach, & Luz- zati, 1980; Bisiach Sr Berti, 1988). Finally, the use of imaging techniques in healthy volunteers, such as meas- urements of regional cerebral blood flow (rCBF) vari- ations by single photon emission tomography (SPECT), has provided data supporting the involvement of tem- porooccipital areas in mental imagery tasks (Goldenberg, Podreka, Steiner, Sr Willms, 1987; Charlot, Tzourio, Zil- bovicius, Mazoyer, Sr Denis, 1992; Goldenberg, Steiner, I'odreka, Sr Deecke, 1992). Moreover, using positron emission tomography (PET), the current most sensitive technique for investigating human functional neuro- anatomy (Posner, 1993), Kosslyn recently reported that an area located near the primary visual area was more