| Visual motion perception is one of the critical functions of the visual system.Direction and speed are the most important features of visual motion stimuli,and the processing of these features mainly occurs in the dorsal visual pathway of the visual cortex.The middle temporal complex(MT+)has sensitive direction-selectivity and plays a key role in visual motion perception.The primary visual cortex(V1),as a key visual brain area,has been widely studied and exhibits different visual motion perception abilities from MT+.Previous studies generally found that V1 responds greater to incoherent motion than to coherent one,while there is no unified conclusion on how MT+ responds to these two motion patterns.Most studies suggest that MT+ is more sensitive to coherent motion,but some studies have found that human MT+ has a stronger response to incoherent motion or has no difference in response to them.In addition,V1 is thought to be more sensitive to slow motion,while MT+ shows higher responses as speed increases.These studies have used univariate analysis methods and have not clearly explained how visual motion information is represented in the dorsal visual pathway.Studies that benefit from multivariate pattern analysis have revealed that object information in the ventral visual pathway is represented based on activity patterns across voxels,but it is still unknown how motion information is represented in the dorsal visual pathway.Random dot kinematograms(RDKs)have been widely used in studies of visual motion perception as they do not have specific information of location.Therefore,this study used RDKs as stimuli and designed two visual motion perception experiments based on3 T and 7T functional magnetic resonance imaging(fMRI)to collect brain imaging data of healthy human subjects with different spatial resolutions.The study employed both univariate analysis and multivariate pattern analysis based on support vector machine(SVM)classification to investigate how MT+ and V1 encode visual motion coherence and motion velocity.First,SVM classification results at both 3T and 7T showed that MT+ and V1 have the ability to discriminate between coherent and incoherent random dot motion,and the univariate analysis results supported the view that MT+ has a stronger response to coherent motion while V1 has a stronger response to incoherent motion.Second,velocity showed a slight effect on the encoding of visual motion coherence in MT+ and V1.MT+ showed the highest accuracy in classifying motion coherence at 10.0 deg/s,while V1 was most likely to show a difference in response to the two motion patterns at 2.5 deg/s.The results from 3T and 7T mutually supported each other in the performance of MT+ and V1 in encoding visual motion coherence,while the decoding ability of them for motion velocity was only observed in the multivariate analysis at 7T,which demonstrated the sensitivity of the multivariate activity pattern analysis and suggested that 7T fMRI could improve decoding performance when studying visual motion perception based on SVM classification.Finally,the results from both univariate and multivariate analyses showed that V1 did not exhibit hemispheric bias in encoding visual motion coherence and velocity,while MT+ showed left-sided dominance in most univariate results and right-sided dominance in some SVM classification results,suggesting that MT+ may have some hemispheric bias when encoding visual motion coherence and velocity.In summary,this study not only investigated the encoding of visual motion coherence and velocity in MT+ and V1 using traditional univariate analysis but more importantly,it was the first study to investigate this issue from a multivariate perspective.This provided evidence for the representation of visual motion information in the dorsal visual pathway based on multi-voxel activity patterns and demonstrated the advantages of fMRI with higher resolution in studying visual motion information encoding based on SVM classification by comparing the results from 3T and 7T fMRI. |
PDF Full Text Request