Varieties of perception 51
against databases of frequencies of occurrence of retinal images corresponding to
commonly occurring natural scenes containing that kind of object.
In the present example, equally luminous objects or patterns are perceived as
darker when placed in a brightly illuminated context and lighter when placed
in a darker context because the differential rates of occurrence of the retinal
projections caused by the same objects under lighter vs. darker conditions are
matched by adaptive behaviours towards those same objects under the respective conditions (Purves et al. 2011, 15589). At first, the probability distributions
of luminance values of objects under lighter vs. darker conditions and the perceived brightness will seem to be skewed towards greater perceived brightness
than measurement of luminance would suggest for those retinal patterns which
occur most often, namely under poor lighting conditions. However, perception is
not skewed in terms of the behaviours that respond to the world affairs as they are
encountered under these conditions. There will be some use to perceiving objects
as exceedingly bright under poor lighting conditions. The use of this seeming illusion should be expected to lie in more reliable or efficient recognition of the kind
of object in question when lighting conditions are poor, and hence in more reliable
responses to the presence or behaviour of that object. Similar conditions apply to
other seeming illusions, such as illusions of length or colour.
Alternatively (a possibility not discussed by the Empirical Strategists), there
might be cases where a seeming illusion is a side-effect without adaptive functions of its own (or, in the terms of Gould and Lewontin 1979, a “spandrel”)
of perceptual processes whose mechanisms of production serve some adaptive
purpose. Such side effects would then either be adaptively neutral or only mildly
detrimental – so mildly in fact that they are more than outbalanced by the benefits
gained from the successful performance of the adaptive functions of the mechanisms that happen to generate them. Under these latter conditions, getting the
luminance values wrong would be the artefact of the operations of mechanisms
of perception whose success is more relevant to the organism in positive terms
than the bias in perception of luminance is in negative terms. But even then, there
would have to be some kind of correlation between the frequency distributions,
across the perceptual situations the organism encounters, of the physical variable
under observation ( f 1) and the presence of some other factor that is relevant to the
rates of success ( f 2) of those behaviours guided by that other mechanism. Otherwise, no statistical relation between f 1 and f 2 could be identified to begin with.
Either way, the normative implications of getting the luminance values or other
physical variables wrong are either inverted or at least much relativised. The
length, shape, colour and so forth of some object may be invariably misrendered
in perception, hence misperceived, while being appropriately treated on the level
of behaviours and their adaptive functions. If the effects of the behaviours towards
the object in question are conducive to the perceiving organism’s welfare, in the
light of his overall constitution and abilities, and if this conduciveness is rooted in
a history of phylogenetic or ontogenetic selection of whatever ways of perceptual
rendering of the respective properties may come to pass, that perceptual rendering
will be vindicated. Systematically misperceiving the physical properties of some
against databases of frequencies of occurrence of retinal images corresponding to
commonly occurring natural scenes containing that kind of object.
In the present example, equally luminous objects or patterns are perceived as
darker when placed in a brightly illuminated context and lighter when placed
in a darker context because the differential rates of occurrence of the retinal
projections caused by the same objects under lighter vs. darker conditions are
matched by adaptive behaviours towards those same objects under the respective conditions (Purves et al. 2011, 15589). At first, the probability distributions
of luminance values of objects under lighter vs. darker conditions and the perceived brightness will seem to be skewed towards greater perceived brightness
than measurement of luminance would suggest for those retinal patterns which
occur most often, namely under poor lighting conditions. However, perception is
not skewed in terms of the behaviours that respond to the world affairs as they are
encountered under these conditions. There will be some use to perceiving objects
as exceedingly bright under poor lighting conditions. The use of this seeming illusion should be expected to lie in more reliable or efficient recognition of the kind
of object in question when lighting conditions are poor, and hence in more reliable
responses to the presence or behaviour of that object. Similar conditions apply to
other seeming illusions, such as illusions of length or colour.
Alternatively (a possibility not discussed by the Empirical Strategists), there
might be cases where a seeming illusion is a side-effect without adaptive functions of its own (or, in the terms of Gould and Lewontin 1979, a “spandrel”)
of perceptual processes whose mechanisms of production serve some adaptive
purpose. Such side effects would then either be adaptively neutral or only mildly
detrimental – so mildly in fact that they are more than outbalanced by the benefits
gained from the successful performance of the adaptive functions of the mechanisms that happen to generate them. Under these latter conditions, getting the
luminance values wrong would be the artefact of the operations of mechanisms
of perception whose success is more relevant to the organism in positive terms
than the bias in perception of luminance is in negative terms. But even then, there
would have to be some kind of correlation between the frequency distributions,
across the perceptual situations the organism encounters, of the physical variable
under observation ( f 1) and the presence of some other factor that is relevant to the
rates of success ( f 2) of those behaviours guided by that other mechanism. Otherwise, no statistical relation between f 1 and f 2 could be identified to begin with.
Either way, the normative implications of getting the luminance values or other
physical variables wrong are either inverted or at least much relativised. The
length, shape, colour and so forth of some object may be invariably misrendered
in perception, hence misperceived, while being appropriately treated on the level
of behaviours and their adaptive functions. If the effects of the behaviours towards
the object in question are conducive to the perceiving organism’s welfare, in the
light of his overall constitution and abilities, and if this conduciveness is rooted in
a history of phylogenetic or ontogenetic selection of whatever ways of perceptual
rendering of the respective properties may come to pass, that perceptual rendering
will be vindicated. Systematically misperceiving the physical properties of some