Two parallel streams for face- and body processing
Elsie Premereur, Jessica Taubert, Peter Janssen, Wim Vanduffel, Rufin Vogels · Lirias · 2015
Face patches in macaque inferotemporal (IT) cortex (e.g. Tsao et al., 2003, 2008) form an interconnected network as shown in combined electrical microstimulation (EM) - fMRI experiments (Moeller et al., 2008). More recent research showed the existence of body patches in macaque IT, adjacent to and even partially overlapping with regions more activated by faces (e.g. Pinsk et al., 2005; Popivanov et al., 2012, 2014). We aimed to directly compare the effective connectivity of face patches ML and AL with that of the mid STS body patch using fMRI-EM (Ekstrom et al., 2008). We first identified face and body patches based on fMRI and single-unit recordings in two macaque monkeys (G and D). During the subsequent anesthetized fMRI-EM experiments, EM blocks were interleaved with no-EM blocks. In EM blocks, the individual patches were stimulated at 1 mA using a platinum-iridium electrode (impedance: 40-150 kΩ) inserted in a recording grid. EM lasted for 250 ms (pulsewidth: 0.48 ms, frequency: 200 Hz) and was repeated on average every 2 seconds. Both monkeys were injected with a contrast agent and scanned on a 3T Siemens MR scanner with an 8-channel phased-array coil. Data were acquired in two sessions per animal for every EM-site. Corroborating previous research (Moeller et al., 2008), EM of AL and ML caused increased fMRI activations in both animals throughout the face patches in IT. EM of the mid STS body patch elicited increased activation in a network of IT areas, which was largely segregated from the network activated by AL-EM (p < 0.05, FWE corrected). A conjunction analysis only showed a few voxels co-activated by AL-EM and body-EM. We also obtained a negative correlation between the voxel’s t-score values for the contrasts AL-EM vs no -EM and body-EM vs no-EM (Spearman rank correlation, monkey D: -0.35, monkey G: -0.35), indicating that voxels significantly driven by one contrast were not activated by the other. Furthermore, ML-EM and body-EM also increased activation in two separate networks in anterior IT (p < 0.05, FWE corrected), although co-activated voxels were found in posterior IT (conjunction analysis, p < 0.05, FWE corrected). The latter finding may be due to the very close proximity of both patches and spill-over of the EM effect from the target to the neighbouring patch. Finally, similar to EM of the face patches, body-EM increased activation throughout the body-selective regions. These results suggest that face and body patches form two interconnected hierarchical networks that are largely separated within monkey IT.