Department of Psychology and Center for Neural Science, New York University, New York, New York 10003, and Department of Psychology, and Neuroscience Graduate Program, University of California, San Diego 92093.
J Neurosci. 2014 Mar 5;34(10):3586-96. doi: 10.1523/JNEUROSCI.3484-13.2014.
Selective attention modulates activity within individual visual areas; however, the role of attention in mediating the transfer of information between areas is not well understood. Here, we used fMRI to assess attention-related changes in coupled BOLD activation in two key areas of human visual cortex that are involved in motion processing: V1 and MT. To examine attention-related changes in cross-area coupling, multivoxel patterns in each visual area were decomposed to estimate the trial-by-trial response amplitude in a set of direction-selective "channels." In both V1 and MT, BOLD responses increase in direction-selective channels tuned to the attended direction of motion and decrease in channels tuned away from the attended direction. Furthermore, the modulation of cross-area correlations between similarly tuned populations is inversely related to the modulation of their mean responses, an observation that can be explained via a feedforward motion computation in MT and a modulation of local noise correlations in V1. More importantly, these modulations accompany an increase in the cross-area mutual information between direction-selective response patterns in V1 and MT, suggesting that attention improves the transfer of sensory information between cortical areas that cooperate to support perception. Finally, our model suggests that divisive normalization of neural activity in V1 before its integration by MT is critical to cross-area information coupling, both in terms of cross-area correlation as well as cross-area mutual information.
选择性注意调节个体视觉区域内的活动；然而，注意在介导区域间信息传递中的作用还不是很清楚。在这里，我们使用 fMRI 评估了在人类视觉皮层中两个关键的运动处理区域（V1 和 MT）中与注意力相关的 BOLD 激活的耦合变化。为了研究跨区域耦合的注意力相关变化，我们在每个视觉区域的多体素模式中进行了分解，以估计一组方向选择性“通道”中的逐试反应幅度。在 V1 和 MT 中，对注意到的运动方向进行调谐的方向选择性通道中的 BOLD 反应增加，而对远离注意方向进行调谐的通道中的 BOLD 反应减少。此外，类似调谐群体之间的跨区域相关性的调制与它们的平均反应的调制呈反比关系，这种观察可以通过 MT 中的前馈运动计算和 V1 中的局部噪声相关性的调制来解释。更重要的是，这些调制伴随着 V1 和 MT 中的方向选择性反应模式之间的跨区域互信息的增加，这表明注意力可以改善合作支持感知的皮质区域之间的感觉信息传递。最后，我们的模型表明，在 MT 整合之前，V1 中神经活动的可分离归一化对于跨区域信息耦合至关重要，无论是在跨区域相关性还是跨区域互信息方面都是如此。
