Parietal encoding of action in depth

Abstract

The posterior parietal cortex is a crucial node in the process of coordinates transformation for the visual control of eye and hand movements. This conviction stems from both neurophysiological studies in the behaving monkey and from the analysis of the consequences of parietal lobe lesions in humans. Despite an extensive literature concerning varying aspects of the composition and control of eye and hand movements, there is little information about the physiological processes responsible for encoding target distance and hand movement in depth or about their control and impairment in parietal patients. This review is an attempt to provide a comprehensive picture from the fragmentary material existing on this issue in the literature. This should serve as a basis for discussion of what we consider to be a prototypical function of the dorsal visuomotor stream in the primate brain, that of encoding eye and hand movement in depth.

Introduction

The idea that the visual system is divided into two main streams of information is an old one, and can be tracked back to the work of Max Schultze (1866) and to the Duplicity Theory proposed in 1894 by von Kries (1894).

More recently, Ungerleider and Miskin (1982) suggested that visual processing in the cerebral cortex uses a dorsal visual stream, devoted to the process of spatial localization of objects, and a ventral stream more concerned with the identification of object features, such as form, color, etc., in line with earlier work by Schneider (1969) showing that a similar dichotomy could be found when comparing the consequences of lesions affecting cortical and subcortical visual centers in the hamster. Ungerleider and Miskin's hypothesis was based on behavioral observations, but also benefited from a large amount of data available on cortico-cortical connectivity. In the early nineties, an interesting extension and specification of these ideas on the functional organization of the visual system was proposed by Goodale and Milner (1992), who suggested that the neural substrates of visual perception may be distinct from those underlying visual mechanisms used to guide motor behaviour. According to their view, the dorsal stream projecting from the primary visual cortex (V1) to the parietal lobe through extrastriate areas would be more involved in visual processing for the control of action, while the ventral stream projecting to the inferotemporal cortex would be more concerned with visual perception. These ideas were quasi-contemporary to the description of the anatomical and functional organization of the parieto-frontal network underlying arm reaching (Caminiti, Johnson & Ferraina, 1996; Johnson, Ferraina, & Caminiti 1993; Johnson, Ferraina, Bianchi, & Caminiti, 1996), and with the demonstration offered by these studies that the main source of visual information to the monkeys frontal areas controlling limb movement was the Superior Parietal Lobule (SPL), a cortical region directly projecting to premotor and motor cortices. Therefore, these anatomical and physiological studies in monkeys were along the same line of reasoning proposed by Goodale and Milner on the role of the dorsal stream.

The neuropsychological evidence supporting the existence of a dorsal visuomotor channel, as opposed to a ventral one, derives first from the observation of a patient (DF) suffering from carbon monoxide poisoning affecting the lateral and ventral prestriate areas (Milner et al., 1991) and, probably disconnecting V1 from the inferotemporal cortex. Patient DF showed normal ability to use visual attributes to perform movements, while the same attributes could not be reported by her in naming, matching or forced choice tasks (Goodale, Milner, Jakobson, & Carey, 1991; Milner et al., 1991).

Directly related to the subject of our review is the observation that in patient DF normal reaching performance was based on intact sensitivity to distance information and to binocular cues, such as the eyes vergence angle (Carey, Dijkerman, & Milner, 1998; Carey, Jacobson, & Goodale, 1991; Mon-Williams, Tresilian, McIntosh, & Milner, 2001) and retinal disparity (Mon-Williams et al., 2001). Similar observations have been obtained in a different patient with visual agnosia (patient DM; Turnbull, Driver, & McCarthy, 2004), displaying an important deficit in depth processing for pictorial perception, but normal depth perception for action. Thus, the mechanisms of encoding depth features for the composition of motor commands seem to survive lesions of the ventral stream.

Conversely, optic ataxia's patients, characterized by lesions centered around the intraparietal sulcus (IPS) and SPL (for reviews see Battaglia-Mayer & Caminiti, 2002; Caminiti, Ferraina, Battaglia-Mayer, Mascaro, & Burnod, 2005), exhibit a specific deficit in localizing visual targets with respect to their body (Rondot, de Recondo, & Dumas, 1977). These patients are able to identify objects properly although they cannot accurately perform a goal directed action. More in general, patients with Posterior Parietal Cortex (PPC) lesions are able to recognize objects, but not their spatial relationship (Andersen, 1987, Critchley, 1953), especially in depth (Brain, 1941). In patient WF (Holmes & Horrax, 1919) the parietal cortex lesion caused a severe disturb of object's perception in depth. Patient WF was unable to state which of two objects was near his body or to an external landmark, as well as incapable to avoid obstacle during reaching (Holmes & Horrax, 1919). Thus in humans normal three-dimensional (3D) visual perception seems to depend on parietal cortex. Furthermore, patients with PPC lesions fail to converge the eyes on near objects (reported by Lynch, Mountcastle, Talbot, & Yin, 1977). In conclusion, neuropsychological data obtained from optic ataxia and visual agnosia patients clearly make the case for a double dissociation between perceptual recognition of objects and object-oriented actions. Furthermore, the dissociation seems particularly evident when considering the 3D signals used for visual perception. Binocular signals, such as angular vergence and retinal disparity, seems to rely on the normal functioning of the dorsal stream, while pictorial cues like texture, illumination gradients, and perspective seems mainly processed along the ventral visual stream.

In monkeys, the inactivation of parietal area CIP (Tsutsui, Taira, & Sakata, 2005), in the caudal part of the IPS produces a selective impairment in the recognition of objects surface orientation. Unfortunately, the available lesion studies in monkeys did not analyze in a systematic and quantitative fashion the effect of cortical lesions on encoding of target distance. Together with the difficulty in distinguishing between encoding of visual target distance and reach amplitude, these studies do not provide a solid basis for a critical evaluation of the consequences of cortical lesion on action in depth in this species. In spite of its importance, and probably due to the inherent difficulty of its study, the neural bases of action in depth remain a neglected topic of modern neuroscience. This stands in contrast to the extensive psychophysical and neurophysiological literature (see also Carey et al., 1998) available on depth perception.

In conclusion, little is known about distance processing for action in patients with brain damage in general, and after parietal lesions in particular, and the same holds true for neurophysiological studies concerning the relevance of depth information for the composition of eye and hand movement.

In this review we will attempt a critical discussion of the available literature in this field. First, we will discuss the importance of binocular signals for deriving an egocentric representation of target distance in parietal cortex, as well as the operations underlying the spatial remapping necessary to provide the constancy of the visual word in front of the continuous change of the fixation point in depth. Then, we will illustrate data showing how in parietal cortex binocular information about target distance can be combined with proprioceptive information about hand position to guide reaching in depth, and the relevance of this combination for the coordinates transformation underlying visually guided reaching. Finally, we will highlight the command and control function for eye and hand movement of the parieto-frontal system, as portion of the dorsal visuomotor stream, and its relevance for understanding the consequence of parietal lesion in humans within the conceptual scheme offered by the two visual systems hypothesis.