Significance and implications of visual shape processing at intermediate cortical levels

It is our fortune that misperceptions of the world are very rare. Only in highly contrived laboratory settings do they occur regularly; or in very rare human cases do they show up consistently. Either way, valuable information can be garnered from these misperceptions about the hierarchical levels of processing that occur between the presentation of a stimulus and its visual registration in consciousness. Vannuscorps et al. (2021) present an intriguing case study of Davida’s misperceptions of 2D stimuli that shed light on what happens in the visual cortex between the unconscious registration of basic visual features at the striate level and the perception, the conscious registration, of the stimulus that relies on the integration of neural activity in multiple and distributed areas of post-striate processing. Vannuscorps et al. propose that (a) from an initial retinotopic construction of separate 2D bounded regions of visual space, the post-striate, intermediate cortical levels (b) additionally construct correspondingly intermediate shape-centered representations (ISCRs) of visual stimuli, and all this (c) at unconscious levels of processing. The proposal is very reasonable and offers fruitful avenues of research of visual scene perception. Moreover, it is consistent with several lines of theoretical and empirical research. As to theory, Stephen Grossberg has presented increasingly sophisticated theoretical models of visual perception (cited in Grossberg, 2016) which shed light on the intermediate cortical levels of processing – up to V4 – that are of primary concern to Vannuscorps et al. The registration of primitive features (brightness contrast, colour, orientation, size, etc.) in early cortical processing is followed at the ISCRs by the segregation of bounded 2-D regions of the visual field. In the Grossberg model, 2-D bounded regions result from the interaction of the boundary-contour system (BCS) delimiting the contours of a bounded region, e.g., a rectangle, and the feature-contour system (FCS) that fills in its surface properties, say a particular grey level. Exploring the cortical hierarchy in rhesus monkey, Felleman and coworkers (Felleman et al., 1997; Wang et al., 2007; Xiao et al., 1999) have investigated how the processing of contour (e.g., orientation) and surface properties (e.g., colour, brightness) proceeds in segregated cortical modules from early V1 levels to intermediate (V2) and higher (V4) levels of the hierarchy. Since similar processing most likely occurs in humans, Felleman et al.’s work sheds important light on the nature of Vannuscorps et al.’s proposed ISCR processing. In his chapter “Filling-in the forms: Surface and boundary interactions in visual cortex” Grossberg (2003) suggested that surfaces are for seeing. In other words, seeing or perception, the conscious registration of the segregated two-dimensional shapes – and the later three-dimensional ones – can occur only as and after the surface filling-in process initiated. How real-time filling-in processes of a stimulus’s surface features can occur has been investigated by Breitmeyer and Jacob (2012). The form or shape itself, can be processed entirely at the unconscious ISCR levels of processing, as Vannuscorps et al. propose. Empirical evidence for this comes from a variety of masked priming studies in which processing of the prime shape is confined to unconscious levels. For instance, Breitmeyer et al. (2005) compared the priming effects of entire square and rhombic stimuli, their vertices/corners only, and their sides only. The strongest priming effects were obtained when entire primes were masked, somewhat weaker