The objects of action and perception
Introduction
It is a common assertion that the fundamental task of vision is to construct a representation of the three-dimensional layout of the world and the objects and events within it. But such an assertion begs at least two fundamental and interrelated questions. First, what is vision? Second, what is the nature of the representation that vision delivers? These questions, which are central to the entire research enterprise in understanding human vision, form the framework for the present paper. In attempting to answer these questions, we will contrast what we believe are two major functions of the visual system. One function of vision is the creation of an internal model or percept of the external world – a model that can be used in the recognition of objects and understanding their interrelations. Most research in object vision has concentrated on this function (witness the current volume). There is another function of vision, however, which is concerned not with object recognition, but with object-directed action. In this case, vision guides our actions with respect to the world by transforming visual inputs into appropriate motor outputs. We will suggest that separate, but interacting, visual systems have evolved for the perception of objects on the one hand and the control of actions directed at those objects on the other. This `duplex' approach to high-level vision suggests that Marrian or `reconstructive' approaches and Gibsonian or `purposive-animate-behaviorist' approaches need not be mutually exclusive and may be actually complementary.
Section snippets
What is vision?
Vision gives us sight. In other words, vision gives us an experience of the world beyond our immediate body surface, a world full of objects and events that are imbued with meaning and significance. Research in human psychophysics and perception has concentrated almost entirely on the way in which the visual system delivers this visual experience (for related discussions of this issue see Georgeson, 1997; Watt, 1991Watt, 1992). Although a good deal of this research has concentrated on
Action and perception systems in the primate brain: dorsal and ventral streams
The evolution of separate systems for visual perception and for the visual control of action is reflected in the organization of the visual pathways in the primate cerebral cortex. Over fifteen years ago, Ungerleider and Mishkin (1982)identified two distinct `streams of processing' in the macaque monkey brain: a so-called ventral stream projecting from primary visual cortex to inferotemporal cortex and a so-called dorsal stream projecting from primary visual cortex to posterior parietal cortex (
Electrophysiological and behavioural studies in the monkey
The functional division of labour between the two streams proposed by Goodale and Milner is also supported by a large number of studies in the macaque monkey. Thus, monkeys which show profound deficits in object recognition following inferotemporal lesions are nevertheless as capable as normal animals at picking up small food objects (Klüver and Bucy, 1939), at catching flying insects (Pribram, 1967), and at orienting their fingers in a precision grip to grasp morsels of food embedded in small
Neuro-imaging studies in humans
Ten years ago little was known about the organization of the cerebral visual pathways beyond V1 in humans. With the advent of functional neuroimaging, however, a wealth of data has suddenly become available. The careful work of Tootell et al. (1996)has revealed an organization of visual areas in the human brain that is remarkably similar to that seen in the macaque. Although clear differences in the topography of these areas emerges as one moves from monkey to human, the functional separation
Differences in the visual transformations mediating action and perception
The division of labour within the organization of the cerebral visual pathways in primates reflects the two important trends in the evolution of vision in higher vertebrates that were identified earlier. First, the emergence of a dorsal `action' stream reflects the need for more flexible programming and on-line control of visually guided motor outputs. It is interesting to note that this stream is intimately connected not only with the primate forebrain but also with those brainstem structures
Dissociations between action and perception in normal subjects
Although the visual fields of the two eyes together span about 200°, most of our perceptual experience is confined to the few degrees subtended by the foveal and parafoveal region. In short, we see what we are looking at. Yet as we move through the world, stepping over curbs, negotiating doorways, and grasping door handles, we often utilize visual information from the far periphery of vision. This differential use of the fovea and peripheral visual fields by perception and action systems may
The action/perception distinction in computational vision
We would suggest that the distinction between vision for perception and vision for action is relevant to some aspects of a current debate in the computational vision literature. The debate could be characterized as one between `behaviorist or purposive' approaches and `reconstructive' approaches to vision. Here we will make some general remarks that capture only some aspects of the various positions in the debate as there are many theoretical divergences within both `behaviorist' and
Getting it together: interactions between action and perception
Throughout this paper, we have been advancing the idea that the ventral perception system and the dorsal action system are two independent and decidedly different visual systems within the primate brain. We realize that in doing this we have overstated our position to some extent. This was a deliberate attempt to counter the tendency in object vision research to focus on issues such as recognition and other cognitive operations, without taking into account the actions that are performed on
Unlinked references
Goodale and Milner, 1982, Milner and Goodale, 1993
Acknowledgements
The preparation of this manuscript was helped in part by grants from the Medical Research Council of Canada to M.A.G. and the Natural Sciences and Engineering Research Council to G.K.H.
References (101)
- et al.
Size-contrast illusions deceive the eye but not the hand
Current Biology
(1995) - et al.
Principles of animate vision
CVGIP: Image Understanding
(1992) - et al.
Orientation dependence in the recognition of familiar and novel views of three-dimensional objects
Vision Research
(1992) - et al.
Levels of categorization in visual recognition studies using functional magnetic resonance imaging
Current Biology
(1997) - et al.
Visual illusion and action
Neuropsychologia
(1996) - et al.
Sensorimotor transformation during eye movements to remembered visual targets
Vision Research
(1991) Visual pathways supporting perception and action in the primate cerebral cortex
Current Opinion in Neurobiology
(1993)- et al.
Separate visual pathways for perception and action
Trends in Neurosciences
(1992) Some effects of pretectum lesions in the frog's detection of stationary objects
Behavioural Brain Research
(1980)The representation of visual scenes
Trends in Cognitive Sciences
(1997)