| abstract |
Neurons of cortical area V4 are selective for shape – the conformation of contrast boundaries
that outline objects in a visual scene. However, it is unknown how this selectivity arises in the
ventral visual pathway, and whether shape tuning depends on visual experience in early life. To
answer these questions, the Visual Neuroscience Laboratory (VNL) at New York University im-
planted 96-channel "Utah" electrode arrays in areas V2 and V4 of two fixation-trained juvenile
Macaca nemestrina. VNL recorded neural responses to shape stimuli with a single parametrically-
varied curvature component (El-Shamayleh and Pasupathy, 2016), at both 7 and 12 months of age
in macaques which translates to approximately an age of 2.5 and 4.3 human years (Kiorpes, 2016).
To assess curvature selectivity we fit neural responses with the Angular Position and Curvature
(APC) model (Pasupathy and Connor, 2001). We found that it explained the response variance
(r2 > 0.1) of many sites in both V2 (74% of n = 64) and V4 (75% of n = 95). Interestingly,
we found that a simple image-computable model based on reverse correlation of local stimulus
luminance could explain most sites in V2 (97%) and V4 (93%). 1% and 5% of sites (V2 and V4,
respectively) were unexplainable by either model.
To account for experimental differences between our work and prior studies, we performed
control analyses using a previously-published dataset of well-isolated shape-selective V4 neurons
(Pasupathy and Connor, 2001). We quantified our ability to characterize the selectivity of simulated
multi-unit responses with an expanded shape set, and of single-unit responses with a range of
shape stimuli chosen to match our experimental conditions. Collectively, these results support our
observation of shape selectivity in ventral visual areas of the juvenile macaque. A comparison of
model fits across developmental age points suggest that V2 and V4 sites are well-tuned for shape
features from the earliest ages tested.
|
