Reversing a Performance Asymmetry through Practice


Zenger B. & Fahle M. (1997). Investigative Ophthalmology & Visual Science 38 , p. S963.

Purpose: We studied practice effects in contrast-based figure-ground segmentation. Methods: Stimuli were textures consisting of 10 x 10 vertical Gabor patches (GP). In one quadrant, we presented a target texture (2x2 GPs with contrast c_tg) and in the three other quadrants distractor textures (2x2 GPs with contrast c_dt). The remaining GPs had a contrast of c_bg. Stimuli were presented for 83 msec and observers had to indicate the target quadrant (4AFC). For different contrasts c_bg and fixed contrasts c1 and c2 (c1 < c2) we compared performance for c_dt = c1, c_tg = c2 (condition A) and c_dt = c2, c_tg = c1 (condition B). Results: For low c_bg it is easier to find the high contrast target (condition A), whereas for high c_bg it is easier to find the low contrast target (condition B). The asymmetry reverses at a c_bg between c1 and c2 for naive observers, but the reversal point can shift to higher contrasts (> c2) after training. For c_bg-values around c2, the performance asymmetry (between condition A and B) thus changes its sign due to practice. Conclusions: Practice in contrast-based figure-ground segmentation can lead to a highly specific performance improvement.





back to my homepage