Unstable orientation perception as a failure of perceptual binding.

In the target article, Vannuscorps, Galaburda and Caramazza describe a fascinating case, Davida, who most frequently misperceives the orientation of 2D shapes. Davida’s deficit is remarkable by its specificity: it only applies to shapes defined by high-contrast bounded regions of space, but not to 3D shapes, low spatial frequency stimuli, or shapes defined by unconnected colinear dots. On the other hand, it generalizes across perception and action tasks, with very similar patterns of errors. This prompted Vannuscorps and colleagues to propose that Davida has a deficit in mapping a shape-centred representation processed specifically in the parvocellular pathway onto higher-level spatial reference frames centred on the viewer and the environment that are responsible for delivering our conscious perception and guide our actions. They suggest that these representations, which they call intermediate shape-centred representations (ISCRs) occur at an early – and pre-conscious – stage of visual processing that is common to the ventral and dorsal visual streams. In this commentary I highlight some other evidence that would support the existence of such early shape-centred representations and speculate that Davida’s faulty perception of orientation is due to a difficulty in maintaining a stable binding between ISCRs and a spatial reference frame. Davida clearly has no problem recognizing shapes and identifying objects (with the exception of orientation-defined letters, such as b, d, p, q), even though her perception of their orientation is mostly incorrect. Thus, her performance strongly suggests that the shape and identity of an object is extracted before an orientation vector that describes how the object is positioned in space is assigned to the object. In this, she is similar to previously-described cases of orientation agnosia, who can recognize objects presented in a variety of orientations, but cannot interpret their orientations (Cooper & Humphreys, 2000; Fujinaga et al., 2005; Harris et al., 2001; Karnath et al., 2000; Turnbull et al., 1997; Turnbull et al., 1995). A similar dissociation was described in a number of studies conducted with neurotypical participants, which found that object identity is determined faster than object orientation (De Caro, 1998; De Caro & Reeves, 2000; Harris et al., 2020), as well as other studies that show implicit object recognition from brief displays, as indexed by priming and repetition blindness, is orientation-invariant and that any costs incurred by presenting an object in an unfamiliar orientation occur at a later stage of conscious report (Dux & Harris, 2007; Harris & Dux, 2005; Harris et al., 2008; Hayward et al., 2010). Collectively, these findings provide empirical support for an early theoretical position advocated by Corballis (1988) who argued that one must know an object’s identity before one can determine its orientation, because orientation can only be judged relative to an objectcentred reference frame (i.e., that object’s intrinsic principal axis and axis polarity). He also argued that viewpoint-dependent costs in object recognition are due to a time-consuming process of mapping an object-centred axis onto a spatial reference frame – that is, the process of determining the object’s orientation. Davida’s perceptual problems provide additional and very compelling evidence for the notion that objects are initially represented in a shape-centred reference frame before being mapped to viewer-centred or environment-centred reference frames. A striking and rather unique aspect of Davida’s perception is that she perceives shapes as rapidly alternating between different orientations, and these orientations obey an orderly relationship with the