摘要
To form a unified percept, the brain needs to integrate information from different sensory modalities. Although several brain areas have been found to encode multisensory information, little is known about how it is decoded and accumulated in downstream decision-making areas. We recorded from single neurons in lateral intraparietal area(LIP) in macaque monkeys performing a visual-vestibular multisensory decision-making task. We found that:(i) the rising phase of the vestibular ramping activity was substantially earlier than the visual one, indicating that LIP accumulates vestibular acceleration signals but visual velocity signals;(ii) LIP ramping activity in the visual, but not vestibular condition, significantly depended on the task-difficulty, a hallmark of gradual evidence accumulation in decision-making;(iii) task parameters(choice and modality) were randomly mixed at the single neuron level but largely separated at the population level in LIP; and(iv) when presented simultaneously, the two distinct sources of evidence(visual and vestibular) were integrated,resulting in an enhanced divergence of LIP population trajectories. However, the improvement is far from that predicted from optimal cue integration theory, as observed in the behavior. Our results may place important constraints on theoretical considerations for neural basis of multisensory integration and multisensory decision-making.
To form a unified percept, the brain needs to integrate information from different sensory modalities. Although several brain areas have been found to encode multisensory information, little is known about how it is decoded and accumulated in downstream decision-making areas. We recorded from single neurons in lateral intraparietal area(LIP) in macaque monkeys performing a visual-vestibular multisensory decision-making task. We found that:(i) the rising phase of the vestibular ramping activity was substantially earlier than the visual one, indicating that LIP accumulates vestibular acceleration signals but visual velocity signals;(ii) LIP ramping activity in the visual, but not vestibular condition, significantly depended on the task-difficulty, a hallmark of gradual evidence accumulation in decision-making;(iii) task parameters(choice and modality) were randomly mixed at the single neuron level but largely separated at the population level in LIP; and(iv) when presented simultaneously, the two distinct sources of evidence(visual and vestibular) were integrated,resulting in an enhanced divergence of LIP population trajectories. However, the improvement is far from that predicted from optimal cue integration theory, as observed in the behavior. Our results may place important constraints on theoretical considerations for neural basis of multisensory integration and multisensory decision-making.
引文
[1]Gu Y,Angelaki DE,De Angelis GC(2008)Neural correlates of multi-sensory cue integration in macaque area MSTd.Nature Neuroscience,11(10):1201-10.