40 Informational environments
architecture to computer interfaces.
2
This divergence of influences did not come by
coincidence, as Marr had while Gibson had not much to say about processes on the
neuronal level of perception, whereas Gibson had while Marr had not much to say
about the guidance of action by perception in an environment. The illustrations to
each author’s main work will provide a hint of the fundamental difference at issue
here: the figures in Marr (2010) are, for the most part, patterns that are supposed to
represent retinal images and the stages of their processing, circuit diagrams of the
computational architecture of perception and function graphs of the algorithms that
govern visual processing, whereas Gibson (1979) provides the reader with a number of semi-abstract, schematic visualisations of how ambient energies impinge
on organisms placed in, and perceptually directed towards, an environment with
certain properties.
Perception, according to Gibson’s view, does not amount to an image-like representation of an outer physical world. One should not take experimental settings
as the paradigm of perception, where the subject’s visual apparatus is exposed to
momentary stimuli detached from environmental settings (“snapshot” or “aperture vision”, Gibson 1979, 1). Nor do pictures or other mediated representations
of world affairs provide a suitable paradigm of perception, as the information
they provide is confined to a few aspects of their subject matter, allowing the
viewer to capture only a limited subset of the information available in the environment (Gibson 1979, Chapter 15). When moving in relation to a picture, one
will discover the difference to a real scene with ease. It is the dynamics of spatial
and somatic relations between perceiving organism and object that has to be systematically accounted for. This is a distinction that Marr could not have made in
a principled way. An identical retinal image caused by a natural scene and a pictorial representation would be treated in identical fashion by the perceptual system.
Only the context of that representation could account for the difference, but that
context is not part of Marr’s inquiry.
The idea that visual perception begins with the projection of an image onto
the retina, Gibson holds, will be misguiding to begin with. Animals with compound eyes, that is animals whose eyes neither have a lens nor a retina but are
composed of an array of closely packed light-sensitive tubes, can produce reasonably accurate visually guided behaviour without even the possibility of deriving
their visual perceptions from retinal images, or equivalents thereof (Gibson 1979,
61f ). The concept of a retinal image misguides us into believing, first, that we
not only process such images but also actually see them, or that some instance
in our brain could see them, and make inferences – which Gibson derides as
the “ ‘little man in the brain’ theory of the retinal image” (Gibson 1979, 60).
We might stop and reflect upon our perceptions by wondering what that red blot
over here may be, or what shade of red it displays. Arguably, most philosophers
think of perception in this fashion, but part of Gibson’s mission is to demonstrate
that this is not how perception works. Second, and more subtly, the notion of the
retinal image suggests that we perceive stimuli, and that perception is a response
to those stimuli in which we derive information from them. Information would
come into play, or even would be generated, only here, in the stages of perceptual
architecture to computer interfaces.
2
This divergence of influences did not come by
coincidence, as Marr had while Gibson had not much to say about processes on the
neuronal level of perception, whereas Gibson had while Marr had not much to say
about the guidance of action by perception in an environment. The illustrations to
each author’s main work will provide a hint of the fundamental difference at issue
here: the figures in Marr (2010) are, for the most part, patterns that are supposed to
represent retinal images and the stages of their processing, circuit diagrams of the
computational architecture of perception and function graphs of the algorithms that
govern visual processing, whereas Gibson (1979) provides the reader with a number of semi-abstract, schematic visualisations of how ambient energies impinge
on organisms placed in, and perceptually directed towards, an environment with
certain properties.
Perception, according to Gibson’s view, does not amount to an image-like representation of an outer physical world. One should not take experimental settings
as the paradigm of perception, where the subject’s visual apparatus is exposed to
momentary stimuli detached from environmental settings (“snapshot” or “aperture vision”, Gibson 1979, 1). Nor do pictures or other mediated representations
of world affairs provide a suitable paradigm of perception, as the information
they provide is confined to a few aspects of their subject matter, allowing the
viewer to capture only a limited subset of the information available in the environment (Gibson 1979, Chapter 15). When moving in relation to a picture, one
will discover the difference to a real scene with ease. It is the dynamics of spatial
and somatic relations between perceiving organism and object that has to be systematically accounted for. This is a distinction that Marr could not have made in
a principled way. An identical retinal image caused by a natural scene and a pictorial representation would be treated in identical fashion by the perceptual system.
Only the context of that representation could account for the difference, but that
context is not part of Marr’s inquiry.
The idea that visual perception begins with the projection of an image onto
the retina, Gibson holds, will be misguiding to begin with. Animals with compound eyes, that is animals whose eyes neither have a lens nor a retina but are
composed of an array of closely packed light-sensitive tubes, can produce reasonably accurate visually guided behaviour without even the possibility of deriving
their visual perceptions from retinal images, or equivalents thereof (Gibson 1979,
61f ). The concept of a retinal image misguides us into believing, first, that we
not only process such images but also actually see them, or that some instance
in our brain could see them, and make inferences – which Gibson derides as
the “ ‘little man in the brain’ theory of the retinal image” (Gibson 1979, 60).
We might stop and reflect upon our perceptions by wondering what that red blot
over here may be, or what shade of red it displays. Arguably, most philosophers
think of perception in this fashion, but part of Gibson’s mission is to demonstrate
that this is not how perception works. Second, and more subtly, the notion of the
retinal image suggests that we perceive stimuli, and that perception is a response
to those stimuli in which we derive information from them. Information would
come into play, or even would be generated, only here, in the stages of perceptual