Residual vision in multiple retinal locations within a scotoma: implications for blindsight.

INTRODUCTION

"Blindsight" is the ability of subjects with damage to the postgeniculate visual system to make correct judgments about stimuli within their scotoma while reporting that they see nothing in that region. It has been suggested that blindsight is mediated by retinal projections to the midbrain, or to extrastriate cortical areas that bypass the damaged primary visual cortex (e.g., Cowey & Stoerig, 1991). Alternatively, blindsight may be mediated by surviving vestiges of geniculostriate function (Campion, Latto, & Smith, 1983; Poppel, Held, & Frost, 1973). Fendrich et al. (1992) proposed that islands of spared striate cortex within a larger damaged region may represent one form of such vestigial geniculostriate function underlying blindsight. Thus, they found that a subject with a dense left homonymous hemianopia (with some macular sparing), as a result of a stroke at age 54, performed a two-alternate forced-choice (2AFC) discrimination at levels significantly above chance within a small region of his hemianopia, approximately 1 [degree] in diameter. Although his 2AFC performance was above chance in the upper left, blind portion of his visual field, the subject reported that he "never saw anything." The combination of discrimination with lack of awareness is consistent with the definition of blindsight, albeit only within a very small region of the visual field. Fendrich, Wessinger, and Gazzaniga (1992) also present magnetic resonance images that show a small isolated region of apparently preserved tissue within the patient's otherwise damaged striate cortex. They argued that since their subject's good discrimination performance was limited to a small region of his blind visual field, it is more parsimonious to attribute his ability to the island of apparently spared cortex rather than an intact retino-tectal pathway which, if it were the substrate for blindsight, should allow discriminations to be made over the entire visual field.

They extended the proposal to explain other cases of blindsight in humans. If this were the case then one would expect blindsight in other subjects to be restricted to patchy isolated regions of the visual field. Moreover, these regions are presumably so small that associated islands of spared cortex are not readily detected by neuroimaging. They suggest that methodological differences in studies of blindsight may account for the difference between the patchy spared discrimination performance they predict and the apparently continuous regions of blindsight found in other cases of blindsight. For their study they used a dual-Purkinje image eye-tracker together with an image stabilization attachment to ensure that computer-generated stimuli were maintained at fixed retinal, and hence cortical, locations regardless of the subject's eye-movements. They suggest that similar stabilized perimetry would reveal patchy discrimination performance in other cases. They further suggest that results showing blindsight distributed continuously over large regions of blind fields may be explained by eye-movements fortuitously passing stimuli into positions corresponding to islands of spared cortex, even though there are studies in which no detectable eye-movements of such magnitude could be detected (Weiskrantz, Harlow, & Barbur, 1991). The corollary of this hypothesis is that normal conscious awareness of visual stimuli depends on the integrity of large connected areas of cortex, while small disconnected islands allow some aspects of visual information to be accessed, even though these islands are insufficient to support conscious awareness. Methods that eliminate shifts in the retinal positions of stimuli should therefore reveal patchy discrimination performance in other cases of blindsight. The present study set out to measure 2AFC discrimination performance for targets in a number of retinal positions in a subject GY, in whom blindsight has previously been demonstrated (e.g., Barbur, Ruddock, & Waterfield, 1980). We employed identical stimuli to those used by Fendrich et al. (1992), in conjunction with methods that reduced movement in the retinal location of these stimuli to insignificant levels. If Fendrich et al. (1992) are correct in their hypothesis that blindsight is mediated through small islands of spared cortex, then it should be very unlikely to find good 2AFC discrimination performance without awareness in all but a small proportion of the locations examined.

Experiment 1

This experiment used procedures and stimulus conditions similar to those reported in Fendrich et al. (1992). GY was required to detect a black 1 [degree] diameter circle, flashed three times over a 600 msec interval at the target location, with an onset of 96 msec.

Results

The subject maintained fixation remarkably well, as had been expected from earlier published studies. Examination of his eye-movement records revealed only one trial during which he made a saccade of more than 1 [degree]. The amplitude of this saccade was only 1.1 [degrees] and it was not directed toward the target. This trial was, nevertheless, removed from the analysis. Apart from this one saccade the eye-movement records show steady fixation with superimposed square-wave jerks typically between 0.2 [degrees] and 0.4 [degrees] amplitude (typical of involuntary eye-movements during fixation; see for example Yarbus (1967), p. 108, [ILLUSTRATION FOR FIGURE 55 OMITTED]) and a very gentle nystagmus of about 0.2 [degrees] per second drift with corrective microsaccades showing some overshoot (probably inertial overshoot and return of the eyeball). The low-level nystagmus may possibly have been induced by the image stabilization method used here as they have not been reported for GY elsewhere. A representative eye-movement record is shown in Figure 1.

The subject's discrimination performance and responses on the "commentary key" reporting awareness of stimuli in his blind field are shown in Figure 2. Before Bonferroni correction, discrimination performance is significantly above chance in 10 of the 15 locations tested. After Bonferroni correction for 15 comparisons, discrimination performance remained above chance in 8 of the 15 locations. After most blocks the subject reported that he had not had any experience of events occurring in his blind field and judged his responses simply to have been random. After two of the blocks he reported that he felt he "did not do very well apart from a few" and "not very well, but on a few towards the end I was aware of something; the rest were chance." These verbal reports are reflected in his responses on the commentary keys, which show little experience of stimuli in any location apart from the one …