Technical Comments

The Hippocampus and Human Navigation

See allHide authors and affiliations

Science  18 Dec 1998:
Vol. 282, Issue 5397, pp. 2151
DOI: 10.1126/science.282.5397.2151-a

Eleanor A. Maguire et al. (1) used positron emission tomography (PET) to scan the brains of humans while they navigated in a familiar virtual reality town. Maguireet al. observed activation in several brain regions of the volunteers performing the task, and they concluded that accurate navigation is associated with activation of the right hippocampus. These findings are in line with mammalian studies of hippocampal function, particularly the discovery (shared by one of the authors of the report, J. O'Keefe) of place cells in the rodent hippocampus (2). It appears, however, that the activation ascribed to the hippocampus in the report (1) as shown in figures 1B and 2B of (1), is primarily outside the hippocampus. Although the edge of the activated region may include the subiculum, the activation shown on the average magnetic resonance image (MRI) of the actual subjects is in the parahippocampal cortex and fusiform gyrus. The Talairach (3) stereotactic coordinates supplied by Maguire et al. (1) for the presumed peak activation of the right hippocampus are outside the hippocampus as well. Recent functional MRI studies have reportedparahippocampal activity associated with navigational tasks in humans (4, 5); another study demonstrated selective activation of an area of the parahippocampal gyrus by visual spatial scenes that depict places (6). The distinction between the hippocampus and its neighboring structures—perirhinal cortex, entorhinal cortex, and parahippocampal gyrus—is important, because these areas are functionally different, as shown by lesion experiments in primates (7).

Also, the region identified in figure 1B in the report as the left tail of the caudate is in fact in the corona radiata as it converges to form the internal capsule. This region lies between the caudate and the insula, according to the MRI depiction and the Talairach coordinates given in the report. It could be argued that the proximity to the caudate is within the spatial resolution limits of the method, but if so, then consideration should be given to interpretation of the data as activation of the insula.


Response: Fried raises two issues: the loca- tion of the hippocampal activations reported in our report (1) in relation to the Talairach and Tournoux (2) brain, and the location of the hippocampal activations on the figures as shown in our report. While the Talairach and Tournoux x, y, and z coordinate system was used to locate the voxels of peak activation in our study, the Talairach and Tournoux brain was not. Instead, the scans of volunteers' brains were normalized with reference to the Montreal Neurological Institute's (MNI) template brain (3) (based on 305 normal brains), as we describe in note 7 of our report (1). These brains are different, with the Talairach and Tournoux brain being smaller and less precisely characterized as compared with the MNI template. Thus, although our locations of peak activation are within the hippocampal formation when correctly referenced to the MNI template (CA1 or subiculum, Fig. 1), if inappropriately plotted on the smaller Talairach and Tournoux brain, they appear to be extra-hippocampal.

Figure 1

Cross-hairs show the location of the peaks of activity for the two relevant activations on the MNI template brain. Right medial temporal region is featured with cross-hairs located at the closest structural slice to the activation peaks. Peak activation falls within the CA1 or subicular regions of the hippocampal formation. (A and B) Coronal and sagittal views, respectively. Successful navigation by following arrows. x = 30, y = −16, andz = −22; shown at x = 30, y = −15, and z = −21. (C and D) Coronal and sagittal views, respectively. Correlation: accuracy/rCBF. x = 36,y = −12, and z = −20; Shown at x = 36, y = −12, and z = −21.

We appreciate Fried's concern about the lack of clarity afforded by displaying activations on a structural image averaged across the ten subjects, where medial temporal structures are difficult to visualize. Nevertheless, the peak activations as shown on the MNI template brain can be observed in the right hippocampal formation (Fig. 1). In the case of both activations, our discussion was focused on the location of the peaks of activity. In the case of the categorical comparison (successful navigation-arrows), the spread of the activation undoubtedly includes parahippocampal cortex as Fried suggests; however, the peak is in the hippocampal formation. In the case of the correlation between accuracy of navigation and rCBF, the activation remains within the hippocampal formation. Finally, the activations we reported (1) were definitely in the hippocampal formation as compared with the activation observed by Epstein and Kanwisher (4) that Fried mentions. The activation they observed in their study (4)—which called for volunteers to perform a passive, 2-dimensional scenes task—was at least 3 cm posterior in the brain to the activation that we observed (1) while our volunteers performed an active, 3-dimensional navigation task.

  • * Neil Burgess, James G. Donnett, Department of Anatomy and Developmental Biology, and Institute of Cognitive Neuroscience, University College London, London WC1E 6BT, United Kingdom; Richard S. J. Frackowiak, Christopher D. Frith, Wellcome Department of Cognitive Neurology, Institute of Neurology, University College London; John O'Keefe, Department of Anatomy and Developmental Biology, and Institute of Cognitive Neuroscience, University College London.


Navigate This Article