Blidsight
Blidsight
Blidsight
1. Introduction
Norton Nelkin, who taught at the University of New Orleans and had a
penchant for the city’s endemic jazz, was a philosopher who tried to under-
stand how things are. This distinguished him from those of his colleagues
who contemplated how they could be, or how they should be. Homing in
on perception, cognition, and consciousness, he would turn to neuropsycho-
logy, and test his ideas against its data and findings in patients with neuro-
logical disorders. In the context of perception, blindsight probably figures
largest in his arguments on the dissociability of conscious states.
The evidence indicates that blindsight is blind in the sense that the blindsight
patient does not see a phenomenal image of the stimulus although he can
‘guess’ its presence and, again in a guessing mode, even discriminate its
wavelength. Nelkin repeatedly comes back to this ‘hue’ discrimination,
because it has been argued (for instance by Hardin, 1988) that colours are
essentially phenomenal. If blindsight is non-phenomenal, the discrimin-
ability of red and green stimuli must be based on their difference in wave-
length or on an ensuing difference in physiological processing, but not on
colour which is the phenomenal counterpart of wavelength in conscious
vision. The colour words (for instance ‘red’ and ‘green’) that we asked the
patients to use during the two-alternative forced-choice guessing procedures
could easily be replaced with ‘1’ and ‘2’ or with ‘630’ and ‘550 nm’ (Stoerig,
1987; Stoerig and Cowey, 1992, Brent et al., 1994), and should thus not be
seen as indicating an underlying phenomenality of hue. Nelkin acknowl-
edges this possibility (p. 178) but nevertheless holds on to the view that it
is colours and not wavelengths (or physiological indices of chromaticity) that
are being discriminated.
To stress this point, he accumulates several additional hypotheses
(pp. 178–9):
(1) Monkeys with striate cortex ablations have been reported to discrimi-
nate stimuli of different colour (Keating, 1979), and indeed extrastriate
cortical areas may well be involved in their discriminability. How-
ever, the existing evidence shows that while the occipitoparietal vis-
ual areas retain visual responsivity when primary visual cortex is
inactivated, the occipitotemporal areas do not, or to a minimal extent
(Bullier et al., 1993). V4 is part of this ventral processing stream and
therefore less likely a candidate for wavelength discrimination.
(2) Similarly, there is no evidence available that would show an area with
colour-constancy coded cells to be involved. More to the point, neither
is there evidence for colour-constancy in monkey blindsight.
(3) Several areas in human extrastriate cortex respond selectively to col-
our (Gulyàs and Roland, 1991). Whether any, some, or all of these are
involved in blindsight wavelength discrimination is an open issue.
As in the monkeys, there is presently only evidence for visual respon-
sivity in the dorsal (occipitoparietal) cortical processing stream of
blindsight patients.
(4) Hemidecorticated patients who do not retain any extrastriate cortex
in the hemisphere contralateral to their hemianopic field, do not show
blindsight as demonstrated with direct (guessing) methods (King et
al., 1996; Stoerig et al., 1996), let alone wavelength discrimination. The
descriptions ‘blue as the sky’ or ‘like a sunset’ that the patients uttered
appropriately in response to light blue and orange stimuli, and that
I mentioned to Nelkin (see p. 179) were indeed evidence of a phenom-
enal perception, only we could in the end not find any evidence that
this was mediated by the hemianopic field. Instead, statistically sig-
nificant detection and colour identification only occurred when the
stimuli were intense enough to be detected by the unaffected hemi-
retina on the basis of intraocular light reflection and diffusion (Stoerig
et al., 1996).
It is clear that the latter finding does not preclude the possibility that
patients and monkeys with damage confined to the visual cortex demon-
Blackwell Publishers Ltd. 1996
Phenomenal Vision and Apperception 229
strate wavelength discriminability indicative of colour constancy. If there
was colour constancy in blindsight—and it would not matter in this context
which part of the remaining visual system was responsible for it—then its
presence would provide a (still very controversial) argument in favour of a
phenomenal representation of chromatic information in blindsight (see Stoe-
rig, 1997). However, so far there is no answer to this empirical question.
Indeed, the related evidence from a dyschromatopsic patient (a patient with
a central colour vision deficit) demonstrated that colour vision and colour
constancy were affected in tandem (Kennard et al., 1995). Although colour
constancy may well turn out to be a composite of several mechanisms,
encompassing retinal as well as extrastriate cortical stages, what little empiri-
cal evidence there is indicates that constancy is lost along with colour vision.
It follows that at least for now there is no empirical reason to think that
there are phenomenal images involved in blindsight. The patients claim they
do not have them, the monkeys appear to claim something similar, and the
chromatic discriminations that blindsight patients can perform do not sup-
port a divergent view: blindsight, like purely reflexive visual processing,
appears to be an instance of non-phenomenal vision.
Although Nelkin concedes that this straightforward reading of the avail-
able evidence—that blindsight patients lack the phenomenal representation
of visual stimuli—is at least ‘empirically and theoretically possible’, he does
not share this view: ‘Actually, I don’t think that the right interpretation of
these cases is that no phenomena are experienced. Rather, it is only that the
clinical subjects are not apperceptively conscious of their phenomenal states’
(p. 57). This brings us to apperception, another kind of consciousness.
(1) The patients have lost the direct conscious knowledge of their seeing.
The patients certainly know that they are performing in a test, but
Blackwell Publishers Ltd. 1996
230 Mind & Language
their commentaries indicate that they do not know that they see and
cannot judge the quality of their performance. ‘When shown his
results he [patient DB] expressed great surprise, and reiterated that he
was only guessing’ (Weiskrantz et al., 1974, p. 4); the patients ‘actually
believed their performance was completely random’ (Pöppel et al.,
1973, p. 296); the patient ‘had no awareness of her above-chance per-
formance’ (Magnussen and Mathiesen, 1989, p. 727); the patient
‘claimed he was only guessing, and could hardly believe that his per-
formance was above chance’ (Stoerig et al., 1985, p. 596). Blindsight
patients know not that they see; they may know that indirectly, for
instance because the experimenter tells them, but they have no direct
knowledge of the fact. This is comparable to the normally sighted
subjects’ ignorance of their pupil responses; you can acquire that
knowledge indirectly, for instance by watching your eyes in a mirror,
but you have no direct conscious access to the response.
Furthermore, the patients cannot even distinguish between the
highly significant performance which they show in response to visual
stimuli in their blind field and the chance level performance which
they show in response to stimuli confined to their blind spot, a recep-
tor-free zone of the retina used for control experiments (own
observation). As a consequence, without feedback, blindsight patients
do not know how they fare in their responses, just as the normally
sighted observer does not know of his pupil responses and their
variability.
(3) The patients cannot explicitly access the ‘blindseen’ visual information.
The patients cannot use the visual information they have processed
for direct control of reasoning or reporting. Control of action, another
Blackwell Publishers Ltd. 1996
Phenomenal Vision and Apperception 231
use visual information can be put to, is slightly more complicated an
issue because visually guided movements such as pointing and grasp-
ing are less affected than verbal responses and matching motor
responses in both blindsight (Perenin and Rossetti, 1996) and agnosia
(Goodale et al., 1991). It seems that motor control can to some extent
be executed in response to unconsciously processed visual infor-
mation while this same information cannot be used for conscious
planning of actions, reporting, reasoning, and remembering.
It has been known for a century that even patients with complete cortical
blindness caused by bilateral occipital damage can for a limited period of
time experience visual hallucinations (Seguin, 1886). The hallucinated images
are assumed to be caused by pathological hyper-excitation in temporal lobe
structures (Gloning et al., 1967; Kölmel, 1985). The patients consciously
apperceive these images, and they report them, at least if they don’t worry
that the asking party will judge them to be out of their mind. It follows
that if a phenomenal image is generated in a cortically blind field, it can be
apperceived, indicating that the disturbance is not at the level of appercep-
tion.
Stored visual information about the shape, the colour, and the particular
movement of objects can be accessed by patients who have suffered a lesion
that caused a cortical blindness. Phenomenal imagery appears to be lost
along with phenomenal vision; at least that is the conclusion rendered prob-
able by Farah et al.’s (1992) finding that following an occipital lobe resection
the patient’s ‘internal screen’ shrank in tandem with her visual field. In con-
trast, such cortically blind patients are still able to describe the shape of an
Blackwell Publishers Ltd. 1996
Phenomenal Vision and Apperception 233
animal’s eyes and ears (Goldenberg and Artner, 1991) and respond to a var-
iety of questions requiring recall of detailed visual information (Chatterjee
and Southwood, 1995)—always provided they have accumulated the infor-
mation before the lesion. This indicates again that it is not the conscious
access to visual information that is lost in blindsight. If the visual information
has passed through the phenomenal stage before the lesion occurred, it
remains consciously accessible; if it has not passed through that stage
because it was only presented after the lesion, it cannot be apperceived.
For Nelkin to be right, for blindsight to be a loss of apperception in the
presence of visual phenomena, the results I listed above have to be explained
in a different manner. This may be possible in some instances. One could
argue that while the cortically blind patient loses apperception of the
phenomenal representation of the external visual world, previously stored
visual information remains apperceivable, as do endogenously generated
images (hallucinations). Blindsight would then be an extraordinary discon-
nexion syndrome with the patient losing the many different ways in which
we consciously access phenomenal visual information. The patient loses the
connection between vision and language areas because he cannot report that
he sees, let alone what he sees, although it is phenomenally represented; the
patient loses the connections between vision and motor areas because he
cannot deliberately sign that or what he sees; the patient loses the connec-
tions between vision and memory areas because he cannot consciously recall
anything he has phenomenally seen after his lesion; and the patient loses
the connections between vision and planning areas because he cannot use
the phenomenally represented information for any explicit future plan or
action. For this, his visual cortex would have to be disconnected effectively
from all the rest of the cortex. Such a situation could result from a lesion
that leaves striate and early extrastriate cortical areas intact while destroying
the surrounding tissue in a way that disrupts all connections to parietal,
temporal and frontal lobe areas (see Nakamura and Mishkin, 1986, for an
experimental study). At the same time though, not only visual memories
but also in addition a particular kind of phenomenal image, namely the
hallucinated one, could be apperceived.
Although this scenario is not impossible, I do not find it a plausible scen-
ario for cortical blindness and blindsight. I do however agree that phenom-
enal representations can exist without apperception. I also agree that visual
information can be consciously accessed when it is not phenomenally rep-
resented provided it has at some time passed through a phenomenal state.
References
Albert, M.L., Soffer, D., Silverberg, R. and Reches, A. 1979: The Anatomic Basis
of Visual Agnosia. Neurology, 29, 876–9.