Categories of residual visual capacities

As in other areas of neuropsychology, single cases or small-group studies can be especially illuminating because they can reveal what dissociations may be possible. In addition, for reasons alluded to above, work has tended to focus on patients with the appropriately restricted and defined lesions. Much of the intensive research has been carried out with subjects D.B., G.Y., and F.S. [D.B. had surgical removal of tumor from V1 (see Weiskrantz et al., 1974); a magnetic resonance imaging (MRI) scan is not possible because of metal aneurysm clips and CT scans are badly distorted for the same reason, but a high resolution CT scan is possible for upper bank of calcarine cortex, and shows complete loss of V1 in that bank (Weiskrantz, Cowey, and Hodinott-Hill, in preparation), corresponding to the lower visual field where testing has been concentrated. G.Y. suffered a head injury; he has been scanned several times (see Barbur et al., 1993; Baseler et al., 1999) showing complete left V1 removal except for occipital pole, corresponding to 3 degrees of macular sparing. F.S. suffered a severe craniocerebral trauma, almost completely destroying his left primary visual cortex; the damage is well documented by MRI (see Goebel et al., 2001)]. Taking the evidence collectively, not all necessarily found in all subjects, the findings are as follows:

Localizing and Detection

Subjects have been asked to attempt to localize the position of a brief, fixated stimulus by making an eye movement to the supposed locus of the stimulus, typically a briefly flashed spot (Pöppel et al.,1973), or by pointing or touching the locus of a target on a perimeter screen (Weiskrantz et al., 1974), even though they deny seeing the stimulus. In yet another study, G.Y. was trained to give a numerical score on a ruled scale to successfully describe the locus (Barbur et al., 1980). Localizing accuracy is typically not as good as in the intact visual field but nevertheless can be very impressive. Successful detection is obviously a necessary prerequisite for localization or any other discriminative capacity, but detection has also been studied independently of localizing (Azzopardi and Cowey, 1997; Barbur et al., 1980; Stoerig, 1987; Stoerig et al., 1985; Stoerig and Pöppel, 1986; Weiskrantz, 1986).

Acuity

D.B.'s acuity was measured with large Moirè fringe interference gratings (approximating sine-wave gratings) based on forced-choice guessing for the presence of a grating versus a luminance-matched homogeneous patch. The reduction in acuity for the region 16–20 degrees eccentric was about 2 octaves compared to the mirror-symmetric region of the intact field (2.5 c/deg compared with 10 c/deg) (Weiskrantz, 1986). Other acuity measurements have been derived from contrast-sensitivity functions in the scotoma (e.g., with G.Y.), using forced-choice detection of a grating versus a homogeneous patch. Acuity falls to zero at approximately 7 c/deg, which again is approximately a drop of about 2 octaves compared to his intact field (Barbur et al., 1994a; Weiskrantz et al., 1998). The reduction is in agreement with animal evidence (Miller et al., 1980).

Orientation

D.B. has repeatedly demonstrated good discriminative capacity for orientation of lines and a grating in the frontal plane in the blind hemifield, although not as sharp as in his good field. He could nevertheless discriminate a difference in orientation of 10 degrees between two gratings presented successively and briefly, even at an eccentricity of 45 degrees in the impaired field, with no acknowledged experience with the gratings or even with the brief flash. Other subjects who have been tested have shown more variable residual capacity for orientation (e.g., Barbur et al., 1980). For example, G.Y. does not show orientation discrimination using gratings, but positive evidence has been obtained using single lines (Morland et al., 1996).

Color

Spectral sensitivity functions of the blind hemifields of several hemianopes have been measured by Stoerig and Cowey (1989, 1991, 1992; Cowey and Stoerig, 1999) and been found to be qualitatively similar to those of their normal hemifields, although with quantitatively reduced sensitivity. The profiles include the humps and troughs thought to reflect color opponency; they also show the characteristic loss of long-wavelength sensitivity following dark adaptation, the Purkinje shift. Strikingly, wavelength discrimination, such as red versus green, but even of more closely spaced wavelengths, is possible in some subjects (Stoerig and Cowey, 1992) using forced-choice guessing. Barbur et al. (1994b) demonstrated good discrimination between “colored” patches and achromatic patches matched in luminance using a two-alternative forced-choice (2AFC) paradigm in G.Y. Ruddock and his colleagues (Brent et al., 1994) have also found good discrimination between red and achromatic stimuli in G.Y. It appears, in this connection, that the discriminative sensitivity of the blind field is biased toward the red end of the spectrum. In addition to wavelength discrimination, a strong inference can be drawn from pupillometry that successive color contrast is intact in G.Y., another hemianopic blindsight subject (Barbur et al., 1999), and recently also confirmed with D.B.

While residual color mechanisms can be found, it should be stressed that in all of these studies, the subjects never reported any experience of color per se. It is worthwhile to note Stoerig and Cowey's comment about the subjects' commentaries:

the patients were often asked whether they could perceive anything when the blind field was tested. Throughout the experiments, which involved from 2 to 4 three-hour sessions per month for approximately six months, they consistently claimed that this was not the case and that they never saw or felt anything that was related to stimulus presentation. (1991, p. 1496)

Movement and Transient Stimuli

This subject has long antecedents given the classical evidence of Riddoch (1917) and Poppelreuter (1917/1990), who described wartime gunshot-wound patients who could see moving but not stationary stimuli. Reports of detection and tracking of moving stimuli in their impaired fields are among the earliest observations of human subjects (Barbur et al., 1980; Brindley et al., 1969; Denny-Brown and Chambers, 1955) and of the monkey (Humphrey, 1974). Using threshold determinations, D.B. was shown to have good ability to detect moving bars and spots, although with reduced sensitivity, depending on the location in the blind field (Weiskrantz, 1986). A number of parametric psychophysical studies of directional discrimination of moving spots or bars have been carried out by Barbur and his colleagues on G.Y. (Barbur et al., 1993; Sahraie et al., 1997, 1998; Weiskrantz et al., 1995; cf. also Perenin, 1991), who measured the limits of velocity and contrast for successful directional discrimination in the blind field. Some of these studies led to the experimental exploration of the distinction between awareness and unawareness modes and also to brain imaging (see below).

While there is no question that blindsight subjects can detect moving bars or spots and discriminate their direction of movement, other modes of movement discrimination, namely, the discrimination of direction of movement of random dot kineograms and plaids, has been found to be lacking (Azzopardi and Cowey, 2001; Cowey and Azzopardi, 2001).

Temporal onset/offset transients also are of particular significance in the blind field. Weiskrantz et al. (1991) carried out a systematic study in G.Y. by varying the temporal slope of the Gaussian envelope of the onset and offset of stimuli in a 2AFC paradigm for gratings versus equiluminant homogeneous patches. Performance improved as the temporal slope increased. In a related study, the manipulation of spatial as well as temporal transients also allowed a specification to be made of both the spatial and temporal parameters required for good detection. From such determinations, it was possible to conclude that G.Y.'s blind field possesses a narrowly tuned spatiotemporal visual channel, with a peak of about 1 c/deg and a cutoff (acuity) of about 7 c/deg (Barbur et al., 1994a). This can now be linked to a closely similar channel in monkeys revealed by pupillometry (see below). It appears to be characteristic of a number of hemianopic subjects, although with a small variation in the peak spatial frequency.

A somewhat different approach had been used in an early study of D.B. in which the temporal rate of onset was varied systematically over a wide range in a forced-choice discrimination paradigm between a circular, homogeneous luminous disc versus no stimulus. Although the sharper the rate of onset the better the performance, D.B. still performed reliably well above chance, and without acknowledged awareness, even with extremely slow rates of onset (Weiskrantz, 1986, Chapter 9). Therefore, a rapid temporal transient is not a necessary feature for good detection. It also indicates that this mode of blindsight is quite different from that in the early reports by Riddoch (1917) that described vigorous movement in the blind field.

“Form”

Early reports of evidence for shape discrimination were negative or weak (cf. Weiskrantz, 1986). With reaching responses, however, when subjects are asked to grasp solid objects in their blind fields, the situation may be different. Both Marcel (1998) and Perenin and Rossetti (1996) have reported that the subject's hand adopts the appropriate configuration in advance of grasping an object in the blind field. This result is nicely in accord with Milner and Goodale's (1995) thesis that the shape and orientation of objects can be involved in directed visual actions via the dorsoventral stream toward the objects even when subjects are unable to perceive the objects correctly. Less easy to accommodate on such a basis, on the other hand, is the evidence for color processing and the striking claim by Marcel (1998) that words flashed into the blind field can influence the interpretation of meanings of words subsequently shown in the intact field. This intriguing report is isolated and deserves to be followed up. It would require considerable expansion of the known capacity for residual processing of stationary shapes in the blind field.

Emotional Content

There are reports that conditioned aversive properties of stimuli in normal human subjects can give rise to autonomic responses even when the subjects have no awareness of them (Esteves et al., 1994; Øhman and Soares, 1998). Functional imaging studies have also demonstrated that conditioned fear stimuli rendered invisible by backward masking can nevertheless activate the amygdala via a colliculo-pulvinar pathway (Breiter et al., 1996; Morris et al., 1999). Behavioral evidence from monkeys with V1 removal suggests that the blind field is inert to the emotional content. For example, when a fear-evoking stimulus, such as a strange doll, is presented to the affected hemifield, the animal appears to ignore it completely, although it emits loud shrieks of fear and outrage when it is confronted in the normal visual field. Nor does it react, for example, to a highly prized banana in the blind field (Cowey, 1967; Cowey and Weiskrantz, 1963). On the other hand, no concurrent measures of autonomic activity were made, and it remains possible that emotion-provoking stimuli produce responses in the absence of overt behavioral responses, just as, for example, galvanic skin responses can be recorded to familiar faces in prosopagnosic patients who cannot distinguish familiar from unfamiliar faces perceptually (Tranel and Damasio, 1985).

A recent study of faces with emotional expression projected to G.Y.'s blind field has some direct relevance (de Gelder et al., 1999). G.Y. has been shown to be able to discriminate between different facial expressions in moving (but not stationary) video images projected into his blind field, such as happy versus sad, angry versus fearful. Strikingly, he could also identify which of four different emotional expressions were presented on any single exposure. He failed to “see” the faces as such, and his responses were at chance with inverted faces. Using cross-field interactions between blind and intact hemifields (see below), evidence was found for face processing with stationary exposures. It remains unknown whether his autonomic system would also be sensitive to emotional stimuli that he can discriminate in the absence of seeing. A recent functional MRI (fMRI) study using a conditioned fear paradigm in G.Y. demonstrated that the amygdala is activated by the conditioned stimulus presented to the blind field, and that the amygdala's activity correlates with activity levels in the superior colliculus and pulvinar. Thus, a route that bypasses V1 evidently remains intact for activation by emotional events (Morris et al., 2001).

Attention in the Blind Field

It has been demonstrated with G.Y. that attention can confer an advantage in processing stimuli in the blind field either within an attended time interval or at an attended spatial location. Using a Posner cueing paradigm, G.Y.'s discrimination of the location of a stimulus was improved by a visual cue that provides temporal information about the time window within which the targets will appear (Kentridge et al., 1999a). Cues in the blind field that provide information about the likely spatial location of a target also confer an advantage, strikingly in the complete absence of reported awareness of either the cue or the target. Cues in G.Y.'s blind field were effective even in directing his attention to a second location remote from that at which the cue was presented, again under conditions in which there was no reported awareness of the cue or the target (Kentridge et al., 1999b). Such evidence suggests that spatial selection by attention, on the one hand, and conscious awareness of attentional control, on the other, cannot be the same process.