Info

% 6C% 8C% 1CC% 12

Fig. 2. Comparison between model simulation and experimental results. (Left) Bar graphs of the percent signal change in simulations using the visual model (blue bars) in each brain region and the corresponding results (red bars) from the comparable PET study of (Haxby et al. 1995). The percent signal changes were evaluated as the difference between the DMS task of object shape vs. a passive viewing task of degraded shapes (see Tagamets & Horwitz (1998) for details). (Right) A similar comparison for the auditory model and the corresponding fMRI study. The percent signal changes were computed as the normalized difference of the percent signal changes of both the DMS and control tasks relative to baseline (see Husain et al. (2004) for details)

Fig. 2. Comparison between model simulation and experimental results. (Left) Bar graphs of the percent signal change in simulations using the visual model (blue bars) in each brain region and the corresponding results (red bars) from the comparable PET study of (Haxby et al. 1995). The percent signal changes were evaluated as the difference between the DMS task of object shape vs. a passive viewing task of degraded shapes (see Tagamets & Horwitz (1998) for details). (Right) A similar comparison for the auditory model and the corresponding fMRI study. The percent signal changes were computed as the normalized difference of the percent signal changes of both the DMS and control tasks relative to baseline (see Husain et al. (2004) for details)

summary, the visual model was able to generate simulated electrophysiological data and simulated PET data that generally were in close agreement with experimental data. Moreover, the model could perform the DMS task.

We have dwelt at length on the visual model because many of the assumptions used in its construction have significant experimental support, and consequently, require little justification on our part. The really new thing is the way in which these multiple mechanisms are put together. What the modeling shows is (1) these quantitative agreements between simulation and experiment strongly argue for the specific hypotheses that were made concerning the neural mechanisms by which multiple interacting brain regions implement this visual DMS task; (2) the assumptions used to related the neural activity to the functional neuroimaging data are sound; and (3) it is possible to account for neuroscientific data at a macroscopic (brain area) level in terms of activity at a mesoscopic (cellular/columnar) level.

0 0

Post a comment