Speculations on the neural basis of islands of blindsight.

Blindsight, residual visual function in the absence of conscious awareness, can sometimes be found within the scotomas of patients with lesions to primary visual cortex. However, cases in which blindsight is distributed across large regions of a scotoma are extremely rare. In contrast, blindsight is relatively frequent within small islands of residual visual function. We review the evidence for the existence of these islands. We argue that blindsight is likely to depend on vestiges of geniculostriate function, and that in humans the secondary retinotectal pathway has little functional utility in the absence of geniculostriate support. To account for the frequency of blindsight within residual islands of function, we speculate that patients may be unaware of such islands precisely because they are islands, which are isolated from the integrated network of neural activity that represents visual space. The relationship of blindsight to the hemispatial neglect is considered in this context.

The temporal cross-capture of audition and vision.

We report that when a flash and audible click occur in temporal proximity to each other, the perceived time of occurrence of both events is shifted in such a way as to draw them toward temporal convergence. In one experiment, observers judged when a flash occurred by reporting the clock position of a rotating marker. The flash was seen significantly earlier when it was preceded by an audible click and significantly later when it was followed by an audible click, relative to a condition in which the flash and click occurred simultaneously. In a second experiment, observers judged where the marker was when the click was heard. When a flash preceded or followed the click, similar but smaller capture effects were observed. These capture effects may reveal how temporal discrepancies in the input from different sensory modalities are reconciled and could provide a probe for examining the neural stages at which evoked responses correspond to the contents of conscious perception.

Ocular pursuit of predicted motion trajectories.

The initiation and maintenance of slow eye movements (SEMs) usually depend on the perception of a moving stimulus. However, the endogenous representation of predictable target motion can be sufficient to initiate and maintain brief episodes of SEM even when the stimulus is not present. In this note, we show that expectancies generated by predictable stimulus motion trajectories can also produce smooth deceleration, reversal of direction, and subsequent acceleration in these movements, and explore the limits of the predictive component of the SEM control system quantitatively.

Illusory contour perception and amodal boundary completion: evidence of a dissociation following callosotomy.

A fundamental problem in form perception is how the visual system can link together spatially separated contour fragments to form the percept of a unitary shape. Illusory contours and amodal completion are two phenomena that demonstrate this linking process. In the present study we investigate these phenomena in the divided hemispheres of two callosotomy ("split-brain") patients. The data suggest that dissociable neural mechanisms are responsible for the generation of illusory contours and amodal completion. Although both cerebral hemispheres appear to be equally capable of perceiving illusory contours, amodal completion is more readily utilized by the right hemisphere. These results suggest that illusory contours may be attributable to low-level visual processes common to both hemispheres, whereas amodal completion reflects a higher-level, lateralized process.

The oculomotor gap effect without a foveal fixation point.

Turning off a fixation point prior to or coincident with the appearance of a visual target reduces the latency of saccades to that target. We investigated this 'gap effect' when subjects fixated a central point or the center of a square formed by four points that were 4, 2 or 1 degree eccentric from the square's center. The fixation anchor vanished 200 ms prior to the appearance of a saccadic target in a Gap condition, coincident with the target's appearance in a 0-Gap condition, or remained on in an Overlap condition. Saccadic reaction time was reduced in the Gap relative to 0-Gap condition irrespective of the type of fixation anchor. However, saccadic reaction time was not reduced in the 0-Gap relative to Overlap condition when the points forming the square had eccentricities of 2 or 4 degrees. Results are interpreted in terms of a partial mediation of the gap effect by fixation cells in the rostral pole of the superior colliculus.

Variability of residual vision in hemianopic subjects.

Using stabilized visual field mapping techniques, seven hemianopic subjects were extensively investigated for residual visual abilities. Isolated islands of detection abilities were demonstrated by four of these subjects. Additional abilities demonstrated within these islands included saccadic and verbal localization, wavelength discrimination, form discrimination, and motion detection. These abilities were also accompanied by low-confidence ratings, and thus have the character of blindsight. It is noteworthy that different subjects demonstrated different abilities at different visual field locations, underscoring the between and within subject variability often observed with blindsight. Furthermore, magnetic resonance images obtained for each subject demonstrated variable sparing of occipital cortex. Such cortical sparing, in conjunction with the behavioral variability, supports the notion that some instances of blindsight are mediated by remnants of the primary visual pathway.

Inhibitory Tagging of Locations in the Blind Field of Hemianopic Patients

This study evaluated the potential contribution of extrageniculate visual pathways to oculomotor orienting reflexes in hemianopic patients. It tested whether extrageniculate pathways mediate inhibition of return (IOR)-a phenomenon characterized by slowed target detections at recently stimulated locations (Posner & Cohen, 1984). Because hemianopic subjects cannot overtly respond to stimuli presented within their hemianopic field, we utilized a spatial cueing paradigm that capitalized on the fact that IOR operates in spatiotopic coordinates. Subjects moved their eyes so that a cue and a target presented at the same spatial location were imaged successively onto blind and seeing portions of their retinas. One hemianopic patient showed a similar IOR effect from cues presented within both the seeing and the hemianopic fields. With a second hemianopic patient, only presentations of the cue to the subject's seeing field produced IOR. The explanation for this discrepancy is not evident. These observations highlight both the potential value and the pitfalls inherent in using "blindsight" as a window into human consciousness.

Inhibitory tagging of locations in the blind field of hemianopic patients.

This study evaluated the potential contribution of extrageniculate visual pathways to oculomotor orienting reflexes in hemianopic patients. It tested whether extrageniculate pathways mediate inhibition of return (IOR)-a phenomenon characterized by slowed target detections at recently stimulated locations (Posner & Cohen, 1984). Because hemianopic subjects cannot overtly respond to stimuli presented within their hemianopic field, we utilized a spatial cueing paradigm that capitalized on the fact that IOR operates in spatiotopic coordinates. Subjects moved their eyes so that a cue and a target presented at the same spatial location were imaged successively onto blind and seeing portions of their retinas. One hemianopic patient showed a similar IOR effect from cues presented within both the seeing and the hemianopic fields. With a second hemianopic patient, only presentations of the cue to the subject's seeing field produced IOR. The explanation for this discrepancy is not evident. These observations highlight both the potential value and the pitfalls inherent in using "blindsight" as a window into human consciousness.

Islands of residual vision in hemianopic patients.

Following damage to primary visual cortex, some patients our initial findings. The data reveal a patchy distribution of redemonstrate a limited ability to respond to stimuli they do not sidual visual abilities in the absence of acknowledged awareness. acknowledge seeing. This residual vision, which has been referred to as "blindsight," has been attributed to secondary visual pathways. We previously reported an isolated island of blindsight in one patient and argued it was a likely consequence of cortical sparing in V1. We now report an extension of our initial findings. The data reveal a patchy distribution of residual visual abilities in the absence of acknowledged awareness. Variable patterns of cortical sparing appear to be the most parsimonious way to account for this outcome, suggesting that blindsight is generally mediated by the primary visual pathway.

Residual vision with awareness in the field contralateral to a partial or complete functional hemispherectomy.

Two patients with unilateral disconnection or removal of the entire occipital lobe were tested for residual vision in their blind field. Using image stabilization to eliminate eye motion artifacts, the central portion of each subject's visual field was tested, beginning 1 degree from fixation and extending outward to 13.5 degrees. A narrow zone of residual vision was identified along the retinal vertical meridian of each patient. The lateral edge of this zone was generally within 3.5 degrees of the vertical meridian, though extended farther outward (but not beyond 6 degrees) at one field location for each subject. In one patient, it was present in both superior and inferior quadrants; in the other, it was confined to the superior quadrant. Within their zones of residual vision, both patients could detect stimuli and perform simple shape discriminations, but could not name complex line drawings. The patients were aware of their vision within this zone. No residual vision, with or without awareness, was found in areas tested outside these zones. Given the complete absence of visual cortex contralateral to the observed residual vision, alternate structures must be mediating these abilities.

Collaboration between the hemispheres of a callosotomy patient. Emerging right hemisphere speech and the left hemisphere interpreter.

Split brain patients who are initially unable to produce speech in their right hemispheres sometimes develop the ability to do so. Patient J.W., the subject of this report, is such a patient. At the time of his callosotomy, J.W. had a language dominant left hemisphere; his right hemisphere could understand both spoken and written language, but he was unable to speak. Fourteen years after his surgery, we found that J.W. was capable of naming approximately 25% of the stimuli presented to his left visual field (LVF). Now, 1 year later, we find that he can name about 60% of such stimuli. This late-developing speech ability appears to be consequence of long-term neural plasticity. However, the subject's extended verbal responses to LVF stimuli seem to result from a collaboration between the hemispheres and to involve the left hemisphere interpreter.

Nasotemporal overlap at the retinal vertical meridian: investigations with a callosotomy patient.

To evaluate nasotemporal overlap at the retinal vertical midline, we asked a callosotomy patient to compare the orientation of square wave gratings that were presented in his opposing visual fields. The gratings were horizontally or vertically oriented and had spatial frequencies ranging from 1 to 8 cycles per degree (cpd). Retinal stabilization ensured the gratings remained properly lateralized during sustained presentations. In accord with previous investigations, when the gratings were presented for only 200 msec, or their medial edges were 2 from the vertical meridian, performance was generally at chance. However, when presentations lasted 2 sec and the medial edges of the gratings were 1 from the vertical meridian, above chance performance was obtained. Accuracy rates were highest with 2 and 4 cpd gratings, and dropped at 1 and 8 cpd. Unexpectedly, the subject performed significantly better when the gratings were offset vertically from each other than when both were displayed on the horizontal meridian. Since this improvement did not occur when gratings were presented horizontally aligned above or below the horizontal meridian, it must be distributed to the relative offset between the gratings. The data suggest a narrow zone of nasotemporal overlap at the retinal vertical midline where very limited visual information is encoded by crossed temporal and uncrossed nasal retinal ganglion cells. An experiment in which only one grating in a pair was close to the vertical meridian indicates that this zone may be more pronounced in the nasal hemiretina.

Transcranial magnetic stimulation: delays in visual suppression due to luminance changes.

Transcranial magnetic stimulation (TMS) was used to investigate the delays in visual processing time produced by changes in luminance. By administering a magnetic pulse over the occipital pole 80-140 ms after the onset of visual stimuli, we could suppress perception of a four-digit number within a 30 ms time window. Commensurate with previous studies of visual processing latencies, a drop in luminance from 3.52 logTd to 2.61 logTd delayed the peak of the suppression window by 8.9 ms, while a further drop from 2.61 logTD to 1.75 logTd delayed the peak by an additional 15.4 ms. This study validates the use of TMS as a psychophysical tool.

Extrageniculostriate vision in humans: investigations with hemispherectomy patients.

Residual vision was assessed in the blind hemifield of one hemispherectomized and one partially hemispherectomized patient, using an interval two alternative forced choice detection task. Fixation instabilities were controlled by retinal stabilization. In both patients, residual vision was found in the hemianopic field close to the vertical meridian. This residual vision is largely confined to a band not wider the 3 degrees, but there is a local region in each patient where it extends more than 3 degrees from the meridian. However, more than 6 degrees from the vertical meridian we found no indication of residual function in either patient. Within the band of spared vision, subjects are aware of stimuli and can perform simple shape discriminations. Visual acuity profiles argue against an explanation based on eccentric fixation. Explanations based on the retino-tectal pathway or on retinal naso-temporal overlap are possible.

The emergence of the capacity to name left visual field stimuli in a callosotomy patient: implications for functional plasticity.

Callosotomized patient J.W. has a well-documented history of right hemisphere language abilities, including an auditory and visual lexical-semantic system with limited phonology and syntax. However, J.W. has not previously exhibited the ability to name stimuli presented to the left visual field (LVF). We report the emergence of this ability. Experiments were conducted in which pictures and text were presented to the subject's LVF using retinal stabilization techniques to ensure lateralization. J.W. was able to correctly name approximately one-quarter of these stimuli under a variety of presentation conditions. The newly developed ability to respond verbally to complex LVF stimuli can be the result of (1) enhanced inter-hemispheric transfer of information via sub-cortical pathways, (2) sophisticated cross cueing strategies, or (3) control of motor speech in the right hemisphere. Although it appears that the first two mechanisms make a contribution to J.W.'s LVF naming performance, accuracy for unpredictable stimulus sets and the error patterns require acknowledgement that control of motor speech is now available to the right hemisphere.

Are microsaccades responsible for the gap effect?

Extinguishing a fixation point shortly before, or concurrently with, the onset of a peripheral visual target reduces the latency of saccades to that target. Saslow (1967) hypothesized that this gap effect might occur because fixation point offsets reduce the incidence of corrective microsaccades with an associated saccadic refractory period. In the present study, a robust gap effect was obtained. However, using a Purkinje image eyetracker with 1 arcmin of resolution, we found that fixation point offsets had no effect on the occurrence of microsaccades and that the occurrence of microsaccades had no impact on the magnitude of gap effect. Microsaccades therefore do not appear to play any part in the production of the gap effect.

Visual-auditory interactions in sensorimotor processing: saccades versus manual responses.

Reaction times (RTs) to bimodal (visual and auditory) stimuli were examined using 3 different response systems: saccades, directed manual responses, and simple manual responses. The observed levels of intersensory facilitation exceeded race model predictions and therefore support summation (coactivation) models of bimodal processing. However, response-dependent differences suggest that the processing of bimodal targets also depends on the relevant sensorimotor pathways and requirements of the task. Coactivation of response mechanisms might account for the effects found using simple RTs. The results for saccades are consistent with known patterns of auditory-visual convergence in the oculomotor system.

Residual vision in a scotoma: implications for blindsight.

Blindsight, the ability of some blind patients to describe attributes of stimuli they have no conscious awareness of seeing, has been attributed to a secondary (retinotectal) visual pathway. However, it has also been proposed that blindsight could be due to residual function within the primary (geniculostriate) visual pathway. Data have now been obtained that support the second alternative. With an image stabilizer ensuring the accurate retinal placement of stimuli, dense visual field mapping was carried out with a hemianopic patient. This perimetry revealed, embedded in the patient's scotoma, an isolated 1-degree island of residual vision that was not disclosed by conventional perimetric methods. Stimuli presented to this island could be detected and discriminated, although the subject reported he did not see them. The existence of this island of vision implies a corresponding island of functioning cortex within the patient's lesion. Other instances of blindsight may be mediated by similar islands of functioning cortex.

Oculomotor readiness and covert orienting: differences between central and peripheral precues.

The costs produced by invalid precues can depend on the spatial relationship between the cued location and the target location. If oculomotor programs mediate attention shifts, then the effect of varying the spatial relation between the cue and target should be the same for covert orienting (indexed by manual responses) and saccadic responses. We found this to be true only for central precues. With central precues, both manual and saccadic costs were greater when cue and target occurred on opposite sides of the vertical meridian than when they occurred on the same side. With peripheral precues, there were no meridian effects in either response condition, but there was a significant dissociation in the pattern of saccadic and manual costs. For manual responses costs were greater when the target was eccentric relative to the cue, whereas for saccades costs were greater when the cue was eccentric to the target. These results provide additional support for the idea that different orienting mechanisms are engaged by central and peripheral precues. They further suggest that the relationship between oculomotor and attentional orienting may depend on the nature of the precue, with the potential for interdependence being greater with central precues.