Conference on Cognitive Computational Neuroscience

Brain activity during mental imagery is often characterized as a reactivation of sensory activity that is especially robust in high-level visual areas and less so in low-level visual areas. However, brain areas vary considerably in their response to visual stimuli, and so it is possible that each area's engagement by imagery could depend upon the content of imagery itself. To investigate this possibility, we tested if the activity profile across the visual cortex remains stable when subjects imagine two different kinds of stimuli: simple oriented bars and crosses, and complex natural scenes. Specifically, we measured and mapped signal-to-noise ratios (SNR) during imagery and vision across the human visual system using 7T fMRI. We find that SNR profiles are highly stable across stimulus types: imagery SNR is identical to visual SNR in high-level visual areas, but dramatically lower than visual SNR in lower visual areas. These findings indicate that engagement of cortical visual areas during voluntary mental imagery is highly stereotyped, and varies much more across visual areas than it does across imagery content.