TopWe can also read out image scale and position
Classification performance for scale, location, classification and identification. To read out scale and position the classifier was trained on 70% of the repetitions of the 77 pictures at different scales and positions (using the standard set and the two additional scales or the standard set and the two additional positions). Read-out of scale was evaluated at a fixed position and read-out of position was evaluated at a fixed scale. The example labels indicated only the scale or the position of the image (regardless of the image identity). Then the performance was evaluated by asking what the scale or position was for the remaining 30% of the repetitions. The dashed line indicates chance performance (which is different for each task). Chance was 1/3 for read-out of scale and position because there were 3 possible scales and positions. The blue rectangles indicate the range of performances for the interval [-100;0) ms with respect to stimulus onset (control). Classifier parameters: MUA, n=64 sites, [100;300) ms interval, bin size = 50 ms.
