Sixty-eight percent of cells (17/25) responded to the contralateral cage, more than for any
other Selleck EGFR inhibitor scene part (α = 0.05, ANOVA; p < 10−15, binomial test). However, significant numbers of units also responded to the contralateral wall (44%, 11/25), ipsilateral wall (36%, 9/25), and ipsilateral cage (32%, 8/25) (α = 0.05, ANOVA). In total, 81% of cells modulated by the cage scene (17/21) were sensitive to ipsilaterally presented stimuli or interactions involving ipsilaterally presented stimuli (α = 0.05, ANOVA). Intriguingly, despite the large spatial separation between the two cages, the populations modulated by each showed significant overlap: six of the eight cells responding to the ipsilateral cage responded to the contralateral cage as well, and 44% of cells (11/25) were modulated by the interaction between the cages. In this Article, we used a combination of Selleck Birinapant fMRI, targeted electrical microstimulation, and single-unit electrophysiology to identify and functionally characterize two nodes within the network for processing visual scenes in the macaque brain. First, using fMRI, we identified the most robust activation to scene versus nonscene images within area LPP, a bilateral region in the fundus of the occipitotemporal sulcus anterior to area V4V. Next, microstimulation of LPP
combined with simultaneous fMRI revealed that LPP is strongly connected to areas DP and V4V posteriorly, and to MPP, a discrete, more medial region within parahippocampal cortex located at the same anterior-posterior location as LPP. Finally, single-unit recordings targeted to LPP and MPP allowed us
to characterize the selectivity of Phosphatidylinositol diacylglycerol-lyase single cells within these two scene-selective regions to scene versus nonscene stimuli, as well as to a large number of different scene stimuli, revealing three major insights. First, the single-unit recordings showed that both regions contain a high concentration of scene-selective cells. Second, they showed that cells in both LPP and MPP exhibit a preference for stimuli containing long, straight contours, and responses of LPP neurons to photographs and line drawings of scenes are significantly correlated. Third, experiments presenting two sets of combinatorially generated scene stimuli revealed a rich population code for scene content in LPP. Synthetic room stimuli multiplexing spatial factors (depth, viewpoint) with nonspatial factors (texture, objects) revealed that LPP cells are modulated not only by pure spatial factors but also by texture and objects, and decomposed scene stimuli revealed that individual LPP cells are selective for the presence of subsets of scene parts and part combinations. In LPP and MPP, the average response across cells does not strongly depend upon the presence of objects but instead depends upon the presence of spatial cues (Figures 1C, S1, 2, and 4).