The concepts of invariants and cues are useful, as are those of dorsal and ventral streams, but Norman overgeneralizes when interweaving them. Cues are not confined to identification tasks, invariants not to action, and both can be learned.

Norman tries to link the ecological and constructivist approaches to the dorsal and ventral pathways of the visual system. Such a link implies that the distinction is not only one of approach, but that different issues are studied. Norman identifies these issues as perception and action. The influence of contextual illusions is critical for Norman's arguments. We point out that fast (dorsal) actions can be fooled by contextual illusions while (ventral) perceptual judgements can be insensitive to them. We conclude that both approaches can, in principle, be used to study visual information processing in both pathways.

The main function of vision in many animals is to control movement. In rodents, some visuomotor acts require the construction of models of the external world while others rely on Gibsonian invariants. Such findings support Norman's dual processing approach but it is not clear that the two types of processing rely on homologs of visual processing streams described in primates.

Several lines of evidence indicate that rapid target-aiming movements, involving both the eyes and hand, can be biased by the visual context in which the movements are performed. Some of these contextual influences carry-over from trial to trial. This research indicates that dissociation between the dorsal and ventral systems based on speed, conscious awareness, and frame of reference is far from clear.

Constructivist and ecological approaches are also observed in tactile perception studies. The question is whether identification and localization are dissociated in the tactile modality as well, and whether Norman's conception may be generalized to the field of touch. An analogue to blindsight was evidenced in passive touch, but no such dissociation was observed in active touch. A study is in progress in this domain.

The ability to recognise the actions of conspecifics from displays of biological motion is an essential perceptual capacity. Physiological and psychological evidence suggest that the visual processing of biological motion involves close interaction between the dorsal and ventral systems. Norman's strong emphasis on the functional differences between these systems may impede understanding of their interactions.

Norman's assigning of the constructivist percept-percept coupling approach and the ecological affordances approach to the ventral and dorsal visual systems, respectively, makes a more workable metatheory than each taken separately, but brings both under closer inspection.

The two modes of visual processing “localizing” versus “identifying” as expressed by four authors in 1967 are more encompassing than the “two visual systems” dichotomies posed by later theorists. Norman's view of parietal cortex functions of vision seems much too narrow.

I argue that Norman's thesis works in sofaras one is prepared to be lax about the alleged congruency between “ecological” and “dorsal,” and unconcerned by the results of recent studies showing that grasping movements are subject to the same geometric distortion illusions as perceptual judgements.

Whether or not the correspondence of dorsal stream functions to Gibsonian ecological psychology and the ventral stream functions to “constructivism” hold up, the overall goal of capturing a pragmatic realism should not be forgotten.

Norman presents intriguing arguments in support of a mapping between ecological and constructivist visual cognition, on the one hand, onto the dorsal ventral dual route processing hypothesis, on the other hand. Unfortunately, his account is incompatible with developmental data on the functional emergence of the dorsal and ventral routes. We argue that it is essential for theories of adult visual cognition to take constraints from development seriously.

Currently there is considerable interest in the notion that dorsal and ventral visual systems might differ in their specializations for thought and action. Behavior invariably involves multiple processes such as perception, judgment, and response execution. It is not clear that characteristics of the dorsal and ventral processing streams, as described by Norman, are entirely of a perceptual nature.

Norman's aim to reconcile two longstanding and seemingly opposed philosophies of perception, the constructivist and the ecological, by casting them as approaches to complementary subsystems within the visual brain is laudable. Unfortunately, Norman overreaches in attempting to equate direct perception with dorsal/unconscious visual processing and indirect perception with ventral/conscious visual processing. Even a cursory review suggests that the functional and neural segregation of direct and indirect perception is not as clear as the target article would suggest.

Perception, as Gibson described it – picking up information that specifies the real local situation – includes not only perceiving affordances and controlling small movements, but also seeing the large-scale environmental layout and the position/movement of the “ecological self.” If the dorsal cortical system is also responsible for that very significant achievement, its activity must be at least partly conscious.

Comparisons are drawn between two theories of visual perception and two modes of information processing. Characteristics delineating dorsal and ventral visual systems lack internal consistency, probably because they are not completely separable. Mechanism is inherent when distinguishing these systems, and becomes more apparent with different processing domains. What is lacking is a more explicit means of linking these theories.

Opposed to Norman's proposal, processing of affordance is likely to occur not solely in the dorsal stream but also in the ventral stream. Moreover, the dorsal stream might do more than just serve an important role in motor actions. It supports egocentric location coding as well. As such, it would possess a form of representational memory, contrary to Norman's proposal.

Berkeley and Helmholtz proposed different indirect mechanisms for size perception: Berkeley, that size was conditioned to various cues, independently of perceived distance; Helmholtz, that it was unconsciously calculated from angular size and perceived distance. The geometrical approach cannot explain size-distance paradoxes (e.g., moon illusion). The dorsal/ventral solution is dubious for close displays and untestable for far displays.

Norman relates two theoretical approaches, the constructivist and ecological, to two cortical visual streams, the ventral and dorsal systems, respectively. This commentary reviews a similar approach in order to increase our understanding of complex skill development and to advance Norman's goal of stimulating and guiding research on the two theoretical approaches and the two visual systems.

Perception of (and during) events is a necessary feature of any perceptual theory. Norman's dual-process approach cannot account for the perception of events without substantial interactions between the dorsal and ventral systems. These interactions, as outlined by Norman, are highly problematical. The necessity for interactions between the two systems makes the distinction useless.

Norman's reconciliation of the two theories of perception is challenged because it directly leads to the nature-nurture dichotomy in the development of the two visual systems. In contrast, the proposition of a separate development of the two visual systems may be better understood as involving different types of information that follow a distinct temporal sequence.