122.06 Chang from Watanabe lab of Boston U. compared the thickness and size of the early visual areas between young (20-30) and old participants (>55) . They found the surface size of V1-v3 reduced with age even though thickness remains the same. They also tried to associate these differences in anatomy with perceptual learning and attention performance. They found a strong negative correlation between both attention and perceptual learning performance with V1 surface area.
[CC: I remembered a study which shows the thickness is positive correlated with IQ. However, that studies mentioned no surface area. Here, the surface area is strongly associate with certain psychological ability. I am wondering whether we can apply the same paradigm to other types of population, such as color blind and normal, high and low creativity etc.]
125.03 It is an interesting concept. In multivoxel pattern classification, it is necessary to run the same condition more than two times such that one can have both training and test sessions. However, repeating the same condition several times means fMRI adaptation. The authors here analyze the same data set both with MVPC and fMRI adaptation and found different conclusions.
125.09 Tootell measured the cortical magnification factor with fMRI by mapping activated to object images at different image size. Due o the cortical magnification factor, one would expect the bold activation increase fast with size when the size is small, and the activation slow down when the size is large. However, if there is cortical magnification factor, BOLD activation show increases with area linearly. They found that V2 and LOC activation showed strong cortical magnification factor while PPA does not. Hence, some inverse transform must occur between these stages.
[CC: I think this is more due to spatial summation than cortical magification. Roger should hollow out his stimuli.]
271.03, 06 & 07 Simocelli developed a nice model for V2 neurons. What a V2 neuron does is compute the correlation among V1 neurons. Hence, it is a correlation computer of V1 responses. They tested this model by measuring single cell electrophysiology responses and psychophysical performance to reconstructed image based on the properties of the model neurons.
[CC: It may be a good idea to test this model with texture contrast]
324.10 This study from Gallant's lab is more suitable to a psychology conference than a neuroscience one. They had participants watching a 30-min movie and naming the objects they saw. They then construct a semantic network of the terms. Then, they asked participants to attend to one category of objects and showed that the semantic network changed with attention.
428.04 Aguirre et al. applied Schira's template of retinotopic mapping (with some modification, of course) and found that one can define V1 from anatomy as accurate as using retinitopic mapping fMRI.
[CC: If this scheme works, this will have a greatest impact of all studies I saw this year so far. Most people using retinotopic mapping just want to get V1. It took about 3-5 min to get a good anatomy scan but about 40min to 1 hour to get stable a retinotopic mapping (and literarily hours of data analysis) . If they yield the same information, why spend the whole time doing the experiment over and over again. Not to say many patients are just not available for a long fMRI experiment. ]
487.13 This study from Wilson's lab suggested that FFA responses can be predicted by some principle components of faces.
[CC: This idea itself is interesting. However, Since their PCA was done by Wilson's model of distance between face features, I am not sure how this result differed from all those 2nd order configuration effects on face recognition.]
[There were nearly 190 posters on Monday afternoon. After reading so many posters in two hours, I found myself cannot remember anyone in detail.]
536.01 electrophysiologyI for contour integration.
536.02 The authors adapted their participant with a disk of a fixed size then place the projection screen at different distance to get the size illusion. They then measured BOLD activation related to the size illusion at V1 ROIs corresponding to different eccentricity. They showed that the V1 activation changed with perceived, rather than real sizes.
[CC: this is actually quite similar to Ansgar's exp a few years ago. But we did not get a as clean data as they did.
694.15 Kim & Freeman studied the single neuron responses to a target presented either with collinear flankers or with a surround at different conext-target distance. They found that the surrounds were mostly suppressive while the collinear flankers were mostly facilitative. For time course, surrounds with different distance has different amount of suppression between 100-300 ms while the collinear flankers only show distance effect at the late (200-300) component.
[CC: I like the experiment. But their model of feedback connection was just weird]
[the population receptive field almost dominated the last two days of talk sessions]
