Perceptual learning of orientation discrimination by more than one attribute

Using a perceptual learning paradigm, we evaluated whether information from the attributes: color, luminance and motion is combined to provide orientation coding. Four observers were trained to discriminate the orientation between color-defined bars, four between luminance-defined bars, and four between motion-defined bars. Before and after training, they were tested with each of the three attributes separately and all superimposed, at the same and at a different location as the one seen during training. A similar improvement was found whether the bars seen after training were defined by the same, or by a different attribute as the one seen during training, or by the three attributes superimposed. This improvement was significantly more substantial at the location where the bars were presented during training. Moreover, orientation discrimination was always better when the bars were defined by three attributes than by any one alone. Because the improvement was retinotopic and not restricted to the attribute seen during training, we suggest that training changed the sensitivity of orientation-selective cells responsive to color, luminance and motion. Moreover, the overall better performance with additional attributes supported an integration of information from color, luminance, and motion at a common site for orientation coding.