Commentary on Norman
Abstract: 60 words
Main Text: 989 words
References: 251 words
Total Text: 1386 words
Our commentary focuses on the functional link between the ventral and dorsal systems implied by Norman as they relate to overt movement. While issues relating to space perception and size constancy are the primary justification for this dual-process theory, the philosophical extensions of this approach are less consistent with examination of motor control and in particular motor learning.
Overall, Norman presents a compelling argument for associating the visual streams with function specific philosophies: Helmhotzian and Gibsonian. While this position is inherently attractive, due to the demonstrated successes of each philosophy in isolation and congruence with the neurophysiological evidence of Goodale and colleagues (e.g., Milner & Goodale 1992,1995), the weakness lies in its ability to unify the viewpoints in a cohesive manner.
At the surface there are a number of aspects of the proposed system / theory harmonization that are consistent with current approaches to the study of motor skill. Most descriptive (Fitts, 1964; Gentile, 1972) and theoretical (Adams, 1971) accounts of skill acquisition separate the learning process into distinct stages where early stages are characterized by verbal/cognitive process and subsequent to practice, tasks become more motor. It is also commonly demonstrated that skilled performers are able to perform exceptional motor skill while being unable to provide a verbal report of what exactly they did or how they did it. These examples are at least superficially consistent with a dual process approach. Both examples suggest that initial unskilled attempts at performing motor actions are verbal/cognitive dominated supposedly using the ventral stream and constructivist processes. With a large amount of practice, control appears to switch to a 'motor frame', consistent with dorsal stream information, and begins to arise in a more automatic fashion -- non-conscious or ecological.
"When the visuomotor behaviour in question is complex and yet not well learned ... many functions that are later performed solely by the dorsal system are supported by the ventral system".(Section 5.1, line 26)
The important feature here is how the dorsal system is able to benefit from the learning apparently accomplished within the ventral stream. According to the basic assumptions of Gibsonian psychologists (section 2.2) all required information must be attainable from the visual array by a process2 of resonating toward the environmental invariants. Consider simple 'skills' of riding a bike and running. Imagine the scenario where an object 'appears' immediately before the person. For a runner, this array affords jumping or stopping -- presumably pre-set action-perception coupling. However, given the identical visual array, for the cyclist steering or breaking are required. How does the ecological stream acquire this coupling without memory? The obvious answer, and the one we think would be acceptable to many movement theorists is that ecological processes (e.g., t, dt/dt) amount to a motor pre-filter; perhaps transforming the complex visual array into symbolic form (e.g., Self organizing maps). Moreover, if one permits the adjustment of these pre-filters in even a limited sense, either by resonance (Adaptive Resonance Theory, Grossberg 1976) or any unsupervised approach, the role of the ecological structure becomes more reasonable. But, without structural (or virtual) memory, whether by self-organization or supervised (executive) control, the dorsal stream is incomplete as a component of a learning system.
Similar problems arise in the expression of an ecological/ventral combined system for simple, well learned, actions if "the dorsal stream is faced with difficulties"(Section 5.1, line 28). Although it would appear contrary to the intent of Norman, this example also appears to necessitate some agent1 (homunculus) role -- deciding the relative efficacy of the visual array. Such an executive would require real-time access to the "ecological" stream in order to make such evaluations. While it could again be argued that such an evaluation may be made without executive control (a dis-order invariant perhaps or resonance), this seems inconsistent with several experimental findings. In particular, Hu et al. (1999) suggest that the dorsal streams remains in control of movements for a, short, interval following occlusion. Moreover, according to the Norman account, the influence of spatial perception should influence motor output if, and only if, the visual array is degraded or removed during movement execution. While both of these findings have been demonstrated (e.g., Binsted and Elliott 1999; Westwood, McEachern, and Roy, 2000), the required level of degradation/interference is equivocal.
While neither example seems overly supportive of Norman's view, the general goal and much of the implementation of Norman is valid and consistent with an earlier account of perception and action. Specifically, Kugler and Turvey (1987) suggests a "minimally intelligent agent" required to "act minimally" may perform exactly the combining role necessary. In figure 1, we've summarized Norman's view (a) and an alternate view where the dorsal and ventral pathways merely conduct information. This view is certainly contrary to Norman but perhaps more consistent with the original conceptions of Goodale and other proponents of separated visual streams. Our contention is, that instead of two fundamentally different systems interacting (across some measure of space) and requiring the induction of a variety of new constructs, the information converges on a 'Kugleresk' agent -- here evaluating the efficacy of each stream, perhaps updating the perceptual and action prefilter to account for context or memory (Figure 1b). What about other forms of information: proprioceptive, tactile, auditory? How are these controlled/accessed? There are strong indications that there are similar ecological information sources from within haptics but that this system is independent of contextual (perceptual) manipulation (e.g., Cooper, Carello & Turvey, 1999). While this is consistent with the a dualistic approach to information, there begins to be a question of the benefit of a unique process for any and all streams. The problem is further confounded if all 'processes' are required to interconnect in the rapid and low-level manner Norman suggests. In summary, the control of movement provides a number of challenges to Norman's view of separate theoretical approaches to thinking and doing. In the end a single theoretical standpoint will be the most coherent explanation of the extant literature, not a concatenation of historical viewpoints.
1. Here the term agent is intended to infer an 'intelligent' (computational or philosophical) entity. No presumption of consciousness is to be inferred implicitly or explicitly.
2. The term 'process' is attributed to both the constructivist and ecological events, while this term may be somewhat unpalatable to ecological psychologists it is intended to reflect both the resonance/transformations of their viewpoint and the 'boxes'/computations of constructivists.
Adams,J.A.(1971). A closed-loop theory of motor learning. Journal of Motor Behavior,3,111-150.
Carpenter, G.A. & Grossberg, S.(1987). The art of adaptive pattern recognition by a self-organizing neural network. IEEE Computer, 21(3), 129-152.
Cooper, M., Carello, C., & Turvey, M.T. (1999)Further evidence of perceptual independence (specificity) in dynamic touch.Ecological Psychology,11(4),269-281.
Fitts, P.M.(1964). Perceptual-motor skills learning. In A.W.Melton (Ed.),Categories of human learning(pp. 243-285). New York: Academic Press.
Gentile, A.M.(1972)A working model of skill acquisition with application to teaching.Quest,17,3-23.
Grossberg, S.(1976a). Adaptive pattern classification and universal recoding. I. Parallel development and coding of neural feature detectors. Biological Cybernetics, 23,121-134.
Grossberg, S.(1976b). Adaptive pattern recognition and universal recoding. II. Feedback, expectation, olfaction, and illusions. Biological Cybernetics, 23, 187-202.
Hu, Y., Eagleson, R. & Goodale, M.A.(1999). The effects of delay on the kinematics of grasping.Experimental Brain Research,126,109-116.
Kugler, P.N. & Turvey, M.T.(1987)Information, natural law, and the self-assembly of rhythmic movement.Hillsdale, NJ: Erlbaum
Milner & Goodale(1992). Seperate visual pathways for perception and action. Trends in Neuroscience, 15,20-25.
Milner & Goodale(1995) The visual brain in actionNew York: Oxford University Press.
Westwood, D.A., McEachern, T., & Roy, E.A.(2000). Delayed grasping of a Müller-Lyer figure.Experimental Brain Research,141,166-173.