Tolerance of macaque middle superior temporal sulcus body patch neurons to shape-preserving stimulus transformations.

• ¹ Katolieke Universiteit Leuven, Biomedische Wetenschapen, Belgium
• ² KU Leuven, Res. Group Psychiatry, Belgium
• ³ MGH Martinos Ctr., United States
• ⁴ Harvard Med. Sch., United States
• ⁵ Univ. of Glasgow, Inst. of Neurosci. and Psychology, United Kingdom

Previously we demonstrated that although neurons in the middle Superior Temporal Sulcus (midSTS) body patch responded greater to images of bodies than to faces and objects, the body patch neurons show profound within-class stimulus selectivity (Popivanov et al. JNS 2014). Applying Bubbles showed that those neurons respond to relatively small stimulus fragments (Popivanov et al. SFN 2013). Here we assess the tolerance of the midSTS body patch neurons to stimulus transformations that preserve shape: changes in retinal location, in size and silhouette and outline rendering of the images (maintaining only shape information, but lacking textural or shading features). Also, we measured the effect of in-plane image rotation. We recorded spiking activity of single neurons from the left midSTS body patch of 2 awake and fixating macaque monkeys. After a search test with 100 stimuli (monkey and human bodies, faces, man-made objects and fruits/vegetables), stimuli were selected for further tests. All stimuli were presented for 200 ms with an inter-stimulus interval of ~ 400 ms. Analyses were done on baseline subtracted firing rates. RF mapping (4º size; 35 locations; 3º steps) showed that the RF of most of the neurons included the fovea. The RFs varied in size (mean > 8º, SD > 3º) and peak location with a bias towards the lower contralateral visual field quadrant. Stimulus preference (5 stimuli; 4º size) was preserved across 2 tested positions inside the RF. Presentations of two stimuli at 2º, 4º or 8º sizes showed strong tolerance of stimulus preference to size changes. Moreover, applications of Bubbles (Popivanov et al. SFN 2013) using two stimulus sizes revealed image fragments at similar relative locations that were correlated across the two sizes, demonstrating size tolerance. To assess whether these neurons still respond selectively when only shape features are present, we showed 10 stimuli in 3 versions: original (shaded and textured), their silhouettes and outlines. The large majority of the neurons responded as well to the silhouette as to the original image and maintained their selectivity for the silhouette versions (median correlation between the responses to the original and the silhouette r = 0.63). Furthermore, the degree of tolerance to the silhouette transformation correlated positively with the body category selectivity (r = 0.35). These data demonstrate that shape features are sufficient to drive body selective neurons of the midSTS body patch. Responses to outlines were reduced and showed less preserved selectivity. In-plane rotation (step: 45º) of the effective stimulus showed that responses of midSTS body patch cells strongly depended on orientation.