David Milner & Melvyn Goodale, University of Durham and University of Western Ontario
In:- Perception, Action and Consciousness – Eds. Gangopadhyay, N., Madary, M. & Spicer, F.
Keywords: ventral stream, dorsal stream, Milner & Goodale, Aglioti, Biegstraaten
The authors had previously proposed a model in which there are distinct paths for vision for perception and vision for action. This involves to pathways the ventral and the dorsal streams. The authors understand the distinct streams in terms of their output rather than their input. It is claimed that initial visual information is transformed in different ways for different purposes. The ventral stream is argued to turn inputs into representations of the characteristics of objects including their spatial position. Both conscious and unconscious perceptions are here thought to be processed in the ventral stream. The dorsal stream is seen as guiding bodily actions such as reaching and grasping objects. The ventral stream is seen as providing our subjective experience of the world, but not the basis for physical action. It is accepted that perception can influence action, in fact this is one of the things it evolved to do, but the connection is argued to be indirect and flexible, and to involve memory and planning.
The authors distinguish between movement and action. For them movement refers to the simple physical movement of an arm or a leg etc., while action involves planning and possibly more than one movement. It is interesting in this context that the Libet and similar experiments that are taken as an absolute refutation of the existence of freewill refer exclusively to single movements, and not to the planning of actions involving multiple movements. The authors argue that the ventral stream is only involved in the planning of actions to deal with goal-objects that have been perceived. The actual implementation of the action depends on the dorsal stream.
The authors refer to evidence for the importance of the dorsal stream in control of action such as the acts of reaching or grasping. Such actions are impaired in patients with damage to the dorsal stream. Furthermore patients who have damage to the ventral stream, so that they cannot identify object, but who do have damage to the dorsal stream, can reach and grasp the objects that they cannot identify. This is similar to the phenomenon of blindsight, which is also thought to result from the distinction between the ventral and the dorsal stream.
The ways in which subjects are prone to visual illusions is also argued to support this distinction between the ventral and the dorsal functions. In particular tests subjects misjudge the length of objects, but when they reach to grasp these objects they correctly estimate their lengths. The author’s findings have proved controversial within neuroscience. One criticism has been that the distinction in D.F.’s behaviour has not been between ventral and dorsal processing but between tasks where viewing is performed from the point-of-view of the subject (egocentric) and where it is based on a separate reference point (allocentric). However, the authors argue that the test quoted in fact showed the subject D.F. using her dorsal functioning to report on her perceptual judgement.
The authors argue that perception and action evolved to perform separate tasks in different ways. Action or grasping requires an accurate estimate of the size of objects, whereas for perception it may be more useful to have an estimate of the relative size of more than one remote object, preserved in time to allow the recognition of objects and their relationships. A study by Aglioti et al (1. 1995) showed that a grip aperture in a disc was less subject to a standard illusion when it had to be grasped than when the object was merely viewed. A study by Biegstraaten et al (2. 2007) showed no reason to think that perceived size guided how we grasped an object.
The authors think that their model suggests that the dorsal-based vision for action works only in real time. A study on the patient D.F. by Goodale et al (3. 1994a) showed that while she could grasp correctly when the object was present, she could not remember the correct grasp soon after the object was removed, indicating that working memory was not functioning in this respect. The authors accept that there will not be complete immunity to visual illusion and some element of illusion could come through from the visual cortex prior to the division of the dorsal and the ventral stream. Particular types of illusion arise in the early cortex rather than the ventral stream.
It is also admitted that more unpractised actions may require cognitive involvement and therefore be less dependent on the dorsal stream, and require more perceptual involvement. Control appears to pass to the dorsal stream once the movement becomes more practised. A study by Gonzalez et al (4. 2006) showed that unfamiliar actions where subject to the effect of visual illusions in the same manner as visual perception, but in contrast to familiar actions. The authors argue that some studies that have suggested that grasping type are subject to illusions only reflect the fact that equipment involved in the experiment has rendered the action required unfamiliar, and therefore subject to illusion. The Gonzalez study also showed that the right hand was less prone to illusions, presumably because it is more practised in most subjects. In studies involving patients pointing at objects, the performance of those with ventral stream damage deteriorated when they had to remember a position. This is ascribed to reliance on the dorsal stream. In contrast a patient with damage to the dorsal stream showed improved performance when having to remember a position, because as time passed they relied more on the perceptual/ventral stream.