Neural models of biological motion perception so far have focused on the processing of features that characterize the 2D structure and motion of the human body. We extend such models for the processing of shading cues in order to analyze the three-dimensional structure of walkers. As extension of a learning-based neural model, we add a ‘shading pathway’ that computes the internal contrast gradients that vary with the 3D view of the walker, even if the silhouette information remains identical. These model neurons are embedded within a recurrent neural field that jointly represents the sequential temporal structure of the stimulus and the view of the walker. The neural field dynamics reproduce the perceptual multi-stability and the spontaneous perceptual switching between stimulus views, observed for silhouette stimuli in psychophysical experiments. Also the model reproduces the processing of shaded stimuli, and a new perceptual illusion, which illustrates a lighting-from-above prior in the processing of biological motion stimuli.

Figure 1: Architecture of shading pathway. After initial filter stage modelling V1 simple cells by gabor filters the strong gradients of the outer contour are suppressed. Internal gradients are analyzed with appropriate filters and pooled, producing input tot Gaussian radial basis funcion usints that have been trained with example patterns of shaded walkers. 

Figure 2: Responses of internal gradient detectors in the  model to walker silhouettes with different internal shading gradients. (Arrow lenbgth corresponds to the response amplitude of internal gradient detectors.)