New research : Inception of color : Learning to Associate Orientation with Color in Early Visual Areas by Associative Decoded fMRI Neurofeedback
Associative learning is an essential brain process where the contingency of different items increases after training. Although associative learning has been found to occur in many brain regions, there has been no clear evidence that associative learning of visual features occurs in early visual areas. In a new paper published in Current Biology, Amano et al. have shown that associative learning between orientation and color occurs in the early visual areas.
Study participants took part in four different experimental stages: retinotopic mapping, color classifier (decoder) construction, A-DecNef training, and post-test stages. In the color classifier construction stage, we measured blood-oxygen-level dependent (BOLD) signal multi-voxel patterns in V1/V2 evoked by the presentation of red-black, green-black, and gray-black gratings of both vertical and horizontal orientations, and constructed a color classifier. The outputs of the classifier represented the calculated likelihood of red color being presented to the participants. The classifier’s mean percentage of correct color classification was approximately 70%, which was significantly above chance.
The color classifier construction stage was followed by three days of the A-DecNef training stage. Participants were unknowingly trained to create an internal association between a physically-presented achromatic vertical grating and the neural activation for a specific target color (red), although no chromatic stimulus was presented in the display. Participants were asked to “Maintain your gaze at the fixation point at the center of the display. While the achromatic grating is being presented, try to somehow regulate your brain activity to make a to-be-presented solid gray disk as large as possible”. The participants were not informed that the size of the disk was proportional to the red likelihood. In short, participants were trained to induce the activation patterns for red in V1/V2 without having any real red stimulus presented. By pairing such activation patterns with a physically-presented vertical grating, we tested whether associative learning of red and vertical orientation occurred.
To test whether associative learning between the color and orientation was indeed created as a result of A-DecNef, we conducted a psychophysical measurement during the post-test stage where a two-alternative-forced-choice task was performed to construct chromatic psychometric functions for each participant. Vertical, horizontal, and oblique gratings were used (Panel A). The oblique stimuli were used as control orientations for the vertical orientation, since the A-DecNef training should have no effect on oblique stimuli. The color of the inner grating was tinted from green to red in 8 steps, passing through a neutral gray. Participants were instructed to judge whether the inner grating was red or green. Panels B and C show the mean red response (in percent) in the A-DecNef group, and in the control group who did not participate in the A-DecNef training, respectively. These results suggested that A-DecNef training, which targeted V1/V2, created perception of a color associated with an orientation.
We found that the association between orientation and color is long-lasting, as has been reported for other types of associative learning. Further, an additional analysis confirmed that there was little chromatic information in higher visual areas during the A-DecNef training, suggesting that the A-DecNef predominantly modified BOLD-signal patterns and the red likelihood in early visual areas.
In summary, using A-DecNef, we created long-lasting associative learning of color and orientation in early visual areas. These results suggest that early visual areas are most likely to be the main locus of the associative learning, although we still cannot completely reject the possibility of the contribution of other areas to the associative learning. Further investigation will be necessary to clarify the exact roles of early visual areas, and perhaps other areas, if any, in the processes underlying associative learning of visual features.
“Learning to Associate Orientation with Color in Early Visual Areas by Associative Decoded fMRI Neurofeedback”
Amano, K., Shibata, K., Kawato M., Sasaki, Y., Watanabe, T. (2016): Current Biology 26 (14), 1861–1866.
URL : http://www.cell.com/current-biology/abstract/S0960-9822(16)30472-9