Impaired attentional selection following lesions to human pulvinar: Evidence for homology between human and monkey

Abstract

We examined the contributions of the human pulvinar to goal directed selection of visual targets in 3 patients with chronic, unilateral lesions involving topographic maps in the ventral pulvinar. Observers completed 2 psychophysical tasks in which they discriminated the orientation of a lateralized target grating in the presence of vertically-aligned distracters. In experiment 1, where distracter contrast was varied while target contrast remained constant, the patients' contralesional contrast thresholds for discriminating the orientation of grating stimuli were elevated only when the task required selection of a visual target in the face of competition from a salient distracter. Attentional selectivity was restored in the patients in experiment 2 where target contrast was varied while distracter contrast remained constant. These observations provide the first evidence that the human pulvinar plays a necessary role in modulating physical saliency in attentional selection, and supports a homology in global pulvinar structure between humans and monkey.

Multiple items within a visual scene compete for our focal attention. This competition is resolved on the basis of both the perceptual salience of the stimulus and its behavioral salience in relation to the goals of ongoing behavior (1). Visual items can compete for representation in ventral occipito-temporal brain areas, with this competition varying according to the physical distinctiveness of the items and according to whether they demand processing through the same receptive fields (2, 3). The competition can also be biased in favor of less conspicuous objects if they are nonetheless more relevant for behavior (2, 4, 5). These “goal-driven” attentional control signals arise within dorsal fronto-parietal networks (6–8) and they lead to behavioral improvements in discriminating the features of the attended object (9, 10). What remains unclear is how such “dorsal” attentional signals are communicated to ventral occipital and temporal areas to bias visual analysis. Here, we report the first direct behavioral evidence in humans for the role of the pulvinar in coordinating these goal-driven and stimulus-driven interactions. We used a sensitive psychophysical task to examine target selection in a special group of patients with well documented chronic, unilateral lesions involving topographic maps in the ventral pulvinar. Our findings demonstrate that the pulvinar plays an important role in filtering irrelevant but salient visual distracters.

The pulvinar nucleus of the thalamus has been hypothesized to play a central role in coordinating attentional effects on visual processing (11, 12). Most of our current knowledge on patterns of connectivity of the pulvinar stems from anatomical studies in non-human primates. The primate pulvinar has extensive connectivity with the cortex. Based on this connectivity, several general organising principles within the pulvinar have been suggested: a global dorsal/ventral division, and an anterior/posterior organization (13, 14).

Neurons in dorsal pulvinar [PU(d)] predominantly project to areas within the fronto-parietal network and superior anterior temporal cortex (15, 16). Ventral pulvinar [PU(v)], conversely, which includes cell populations within inferior, lateral and medial subnuclei, exhibits reciprocal connections with successive occipito-temporal cortical areas along the ventral processing stream, and contains several topographic maps of the contralateral visual field (13, 17–20). Although there are few direct connections between PU(d) and PU(v), attentional control areas such as the frontal eye fields (FEF) and lateral intraparietal area (LIP) have strong connections with the superior colliculus (SC), which subsequently projects to PU(v) (13, 14). Fronto-parietal control signals may therefore influence occipito-temporal cortex via colliculo–ventral pulvinar–cortical feedback connections.

With respect to anterior/posterior pulvinar organization, neurochemical tracer studies in macaques have revealed that neurons within visual areas in the ventral processing pathway (i.e., from V1 to anterior temporal cortex) show an orderly pattern of connectivity within ventral pulvinar. Specifically, early visual areas are connected with the rostrolateral pole of the pulvinar nucleus, whereas neurons in progressively more anterior temporal areas show overlapping patterns of connectivity with more caudomedial regions (16, 19–21).

Putative homologies between the monkey and the pulvinar's internal structure and connectivity in humans, however, have yet to be defined. Despite theoretical claims and strong anatomical plausibility for pulvinar involvement in attentional selection and filtering of distracters, empirical evidence on the pulvinar's function in human visual selection has not been forthcoming. Heretofore, neuropsychological studies in patients with pulvinar damage have found no evidence that distracters interfere with performance (22–24), and one study reports reduced flanker interference from contralesional distracters (25). Pulvinar activation has nevertheless been reported using fMRI or positron emission tomography in healthy human observers during filtering tasks or other attentional paradigms (8, 26–30). The lack of distracter interference effects in human neuropsychological studies may reflect the use of behavioral tasks that measured the effects of response conflict, rather than perceptual selection of stimuli based on their behavioral salience.

If patterns of pulvinar organization and connectivity evident in non-human primates can be extended to humans, then isolated damage to the pulvinar should lead to specific impairments in selecting relevant targets in the context of physically salient distracters—conditions in which task-based control of selection is necessary. Specifically, deficits in selectivity should be evident after damage to ventral pulvinar, which should disrupt the coordination of attentional feedback signals within visual cortex and result in contralesional impairment in filtering salient distracters. To assess these predictions, we measured contrast thresholds for discriminating simple grating stimuli in patients with chronic, unilateral lesions that involved ventral pulvinar areas that represent the contralateral visual field.

In experiment 1, the performance of 3 patients on an attentional filtering task was compared with 3 control participants matched for sex and age. The patients (2 with left hemisphere lesions and 1 with a right hemisphere lesion) had sustained hemorrhages in the posterior thalamus between 2 and 8 years before testing (see Patients in the Methods section.) For all 3 patients damage to topographic maps of the ventral pulvinar has been extensively documented based on both behavioral and radiological criteria. Previous studies have reported contralesional impairments of stimulus localization (31, 32), oculomotor performance (33), response channel activation (25) and visual feature binding (31, 32). The radiological extent of their lesions has also been documented (31, 32, 34), including lesion reconstruction on the Morel et al. Multiarchitectonic and Stereotactic Atlas of the Human Thalamus (35) based on high resolution (3T) multispectral MR imaging (33). All observers had normal or corrected-to-normal vision. For all patients, visual fields were full, and there were no clinical signs of extinction or hemispatial neglect. A second experiment, completed by 2 patients, showed that deficits in filtering distracters were reversed in the pulvinar lesioned patients by altering the physical strength of the target stimulus.

Results

Experiment 1.

In experiment 1 (Fig. 1A), the participant's task was to decide whether a lateralized target sine wave grating was tilted to the left or right of vertical. On each trial, the target appeared alone, or in the context of vertically aligned physically salient distracters. The distracters were Gaussian contrast discs that were brighter (+50%, +80% contrast) or darker (−50% contrast) than the background. A no-distracter (0% contrast) condition served as a baseline with which to compare the relative interference from distracters. Orientation thresholds (i.e., the smallest detectable difference from vertical) within each hemifield were established using 2 simultaneous interleaved staircases. Fixation was monitored throughout all trials using a remote eye-tracker (see Methods).

Fig. 1.

Effects of distracters on grating orientation discrimination. (A) Example display sequence used in the orientation discrimination task. Observers fixated a central point throughout all trials. Lateralized target gratings appeared with/without distractor discs (shown here with distracters). Stimuli appeared briefly within the left or right visual field. observers were required to discriminate whether the grating was tilted to the left or right of vertical. In experiment 1, target orientation thresholds for each observer were calculated at 4 levels of distractor disk contrast (−50%, 0%, +50%, 80%). Distractor contrast was manipulated by changing disk luminance relative to the background (see Methods). (B) Orientation discrimination performance as a function of distracter contrast for pulvinar-lesioned patients and controls. Data represent mean (SE) change in orientation threshold from baseline. Data points in each graph are averages based on 4–6 thresholds. Performance in the “zero-contrast” distracter condition (baseline) is indicated by the dotted line. Examples of the 4 distractor conditions are illustrated above the graph. Open squares, patients; filled squares, controls. (C) Mean (SE) change in orientation threshold from baseline plotted separately for patients D.G. (Top), C.R. (Middle), and T.N. (Bottom). Data are plotted separately for targets appearing within the contralesional (filled squares) and ipsilesional (open squares) fields.

To ensure that measurement of orientation discrimination performance (without distracters) was comparable across conditions, we first established each observer's contrast threshold for detecting gratings in each visual field using a present-absent Contrast Sensitivity task (see Methods). For experiment 1, targets in the orientation sensitivity task were presented at 3x the mean contrast detection threshold obtained for each observer and visual field.

Contrast thresholds for detecting single gratings in the present-absent task did not reliably differ between patients and controls [F(1,4) = 2.68, P > 0.1] and did not differ between fields for either group [F(1,4) = 1.11, P > 0.1] (Table 1). Further, baseline orientation discrimination thresholds (i.e., with no distracters) did not differ between patients and controls (Table 1), [F(1,4) = 0.04, P > 0.1], nor did orientation thresholds differ between fields for either group [F(1,4) = 0.29, P > 0.1].

Table 1.

Grating contrast thresholds and ″baseline″ (no distractor) grating orientation thresholds in experiment 1

Subject	Target Contrast		Baseline Orientation Threshold
	Contralesional	Ipsilesional	Contralesional	Ipsilesional
Patients
D.G.	13.0	8.3	5.5	6.0
C.R.	6.0	5.0	3.4	2.7
T.N.	8.3	8.3	3.0	4.5
Mean	9.1	7.2	4.0	4.4
SD	3.6	1.9	1.4	1.7
Controls
J.H.	6.0	4.3	4.6	4.4
F.O.	3.6	6.1	3.8	3.9
J.W.	7.3	6.0	3.8	3.6
Mean	5.6	5.5	4.1	4.0
SD	1.9	1.0	0.5	0.4

Each threshold measurement for each control observer in each condition was submitted to a mixed repeated-measures ANOVA with observer as a between subject factor and distracter contrast (−50, 0, +50, +80 condition) and visual field (left, right) as within subject factors. There was a small but reliable effect of distracter contrast [F(3,6) = 3.295, P = 0.032] indicating that the control observers were not fully able to filter the effects of distracters on orientation sensitivity. The effects of distracter contrast did not differ between left and right fields.

To isolate the relative influence of distracter contrast on orientation sensitivity and to compare patients and controls, baseline orientation thresholds measured without distracters (i.e., in the zero contrast distracter condition) were subtracted from those measured with distracters in each visual field. Mean threshold differences from baseline are plotted in Fig. 1B for patients and controls as a function of distracter contrast, collapsed across left and right visual fields.

A mixed repeated-measures ANOVA, including goup (patient, control) and observer as between subject factors and distracter contrast as a within subject factor revealed an interaction between goup and distracter contrast [F(3,66) = 9.8, P < 0.001] indicating that patients were impaired in filtering distracters compared with control observers. A 3-way interaction between goup × observer × distracter contrast [F(6,66) = 4.2, P = 0.001] reflected the fact that, whereas control observers did not differ from one another in their ability to filter distracters, patients did differ in their degree of impairment.

Mean threshold differences from baseline for each patient are plotted in Fig. 1 C–E as a function of distracter contrast and hemifield. Each contrast threshold for each patient in each condition was submitted to a mixed repeated-measures ANOVA with patient as a between-subject factor and distracter contrast and visual field (ipsilesional and contralesional) as within-subject factors. Effects of field [F(1,11) = 8.0, P = 0.017] and distracter contrast [F(3,33) = 44.8, P < 0.001] were qualified by a 2-way interaction of field × distracter contrast [F(3,33) = 3.3, P < 0.05]. An interaction between patient × distracter contrast [F(6,33) = 6.2, P < 0.001] reflected that the patients differed from one another in the degree to which distracters impaired perceptual sensitivity (with T.N. being least severely affected and impaired only in the contralesional field). However, the difference between ipsilesional and contralesional distracter interference was similar for the 3 patients [patient × field × distracter contrast (F(6,33) = 0.5, P > 0.1)]. The data demonstrate that lesions involving pulvinar result in distracter filtering deficits. In these patients in whom the lesion extends into the ventral pulvinar, the deficit is greater in the contralesional field. In patient T.N., who has the most restricted pulvinar lesion, the deficit is restricted to the contralesional field.

Experiment 2.

If pulvinar damage disrupts attentional signals such that competition in ventral stream visual areas is biased by stimuli that are most physically salient, increasing the target's contrast above that of the distracters should restore performance in the patients [as it does following extrastriate cortical lesions (36)]. In experiment 2, patients C.R. and D.G. again completed an orientation sensitivity task, only this time target contrast was systematically increased while holding distracter contrast constant. (Patient T.N. was unable to complete the experiment). In one condition, a low-contrast target appeared in the context of 2 vertically aligned low-contrast disk distracters (“target and distracter”). In this condition, target and distracter contrast were set to 2× grating detection threshold, as established separately for each patient and visual field in the grating detection task (Table 2). This served to maximize target grating orientation thresholds, while ensuring that patients were able to detect targets on all trials. Grating orientation thresholds were then compared with conditions in which target contrast was systematically increased by 5% (target 5%>), and 40% (target 40%>). A “target-only” condition (a low-contrast target with no distracters) was also incorporated to compare patients' thresholds with experiment 1 (see Methods).

Table 2.

Percentage contrasts for target and distractor stimuli in each visual field for patients C.R. and D.G. in experiment 2

Visual Field	Target only		Target and Distracter		Target 5%>		Target 40%>
	Contra	Ipsi	Contra	Ipsi	Contra	Ipsi	Contra	Ipsi
Patient D.G.
Target	26%	16%	26%	16%	31%	21%	66%	56%
Distractor	0%	0%	26%	16%	26%	16%	26%	16%
Patient C.R.
Target	12%	10%	12%	10%	17%	15%	52%	50%
Distractor	0%	0%	12%	10%	12%	10%	12%	10%

Percent target contrast in the target-only and target and distractor conditions were initialized at 2× grating contrast threshold, as established separately for each patient and visual field in the contrast sensitivity task. Target contrast was systematically increased by 5% and 40% from the baseline established for each patient and visual field. Contra = contralesional visual field, Ipsi = ipsilesional visual field.

The mean (absolute) target orientation thresholds in each visual field for patients D.G. and C.R. are plotted in Fig. 2 as a function of target contrast. Absolute orientation thresholds for each visual field (contralesional, ipsilesional) and target contrast condition (target-only, target and distracter, target 5%>, target 40%>) for each patient, were compared using a mixed repeated measures ANOVA. Across both patients, the orientation threshold decreased (improved) significantly as the target's contrast increased relative to that of the distracters (main effect of target contrast condition: F(3,30) = 17.54, P < 0.001; no other significant main effects or interactions). Follow-up planned comparisons revealed that compared with the target and distracter condition in which a low-salience target appeared in the context of low-salience distracters, orientation sensitivity to the target grating significantly improved when the target's contrast was increased by 5% (contralesional field: [t(11) = 3.86, P = 0.003]; ipsilesional field [t(11) = 2.66, P = 0.02], and 40% (contralesional field: [t(11) = 5.58, P < 0.0001]; ipsilesional field [t(11) = 4.41, P = 0.001]. Indeed, target orientation thresholds for C.R. and D.G. in both visual fields were restored a level equal to or better than those observed in the baseline condition of experiment 1.

Fig. 2.

Mean (SE) target grating orientation thresholds for each patient as a function of target contrast in experiment 2. (Upper) Patient D.G. (Lower) Patient C.R. Target orientation thresholds for each observer were measured with a low-contrast target grating that appeared alone (“T-only” condition), or in the presence of low-contrast distracters (“T and D” condition). In the remaining conditions distracter contrast was held constant while target grating contrast increased by 5% (T 5%) and 40% (T 40%) relative to the no-distracter (T-only) condition. Grating contrast was manipulated by changing the contrast between dark and light stripes (see Methods). Examples of the 4 target/distractor contrast conditions in experiment 2 are illustrated above the graph. Data points are averages based on 4–8 thresholds. Data are plotted separately for targets appearing within the contralesional (filled squares) and ipsilesional (open squares) fields.

Finally, comparing distracter interference effects across experiments in Figs. 1 and 2, the magnitude of distracter interference effects are larger in experiment 1 than in the target and distracter condition in experiment 2. Based on the findings from experiment 1, improvements in target selectivity in pulvinar-lesioned patients should be observed when either target salience is increased, or distractor salience is decreased. Given that distracters were higher contrast in experiment 1 (50–80%) than in experiment 2 (10–26%; see Table 2), these data again confirm that reducing distracter salience improves target selectivity.

For both patients, therefore, the results of experiment 2 confirm the findings of experiment 1, and further demonstrate that increasing the target's salience above that of distracters restored target discrimination performance. Taken together, the results indicate that pulvinar lesions produce impairments in attentional selectivity; visual sensitivity is driven by stimuli of highest physical contrast rather than those that are most behaviorally relevant.

Discussion

These psychophysical data from human observers with discrete pulvinar lesions lend strong support to speculations that the pulvinar plays a role in integrating fronto-parietal attentional control signals within visual processing areas (13, 14, 20). The patients were impaired in discriminating target features, but only when targets were accompanied by salient distracters. Moreover, in these patients with damage to spatiotopic maps in the ventral pulvinar, the filtering deficit was greater in the contralesional than in the ipsilesional visual field.

We may think of this as equivalent to there being increased visual crowding, perhaps because luminance is pooled from distracters and targets without mediation from the pulvinar. These data are in agreement with both patterns of selective impairment in pulvinar-lesioned monkeys (37), and with patterns of pulvinar glucose uptake in healthy human observers when they are required to identify an object within a cluttered visual array (26). The severity of the patients' impairment depended on the relative contrasts of target and distracters; as distracter contrast increased relative to the target, sensitivity to target features declined (experiment 1). Conversely, when target contrast was increased relative to the distracters (experiment 2), selectivity was restored. Importantly, these impairments cannot be explained by reduced visual acuity, an inability to detect the targets or a general difference in cognitive function, given the equivalent performance in the single target condition.

Emerging diffusion tensor imaging tractography evidence indicates that the human pulvinar nuclei connect with a large number of cortical areas including V1, V2, V4, inferior temporal cortex, and the parietal cortex, frontal eye fields and other prefrontal areas (38). The evidence for a specific deficit in selecting low-saliency targets after damage to the pulvinar provides further evidence for a homology in global pulvinar structure between humans and monkeys, and for functional connectivity between ventral pulvinar and ventral-stream visual areas. The decline in attentional selectivity in these patients closely resembles the pattern of deficits observed in non-human primates with extrastriate cortical lesions restricted to the ventral visual pathway (36).

Two of the patients (D.G. and C.R.), who have left hemisphere lesions showed distracter-related impairments in orientation sensitivity for targets in both hemifields, although, for both, the deficit was greater in the contralesional field. In patient T.N. the deficit was restricted to the contralesional hemifield. Chemical tracer studies in the monkey brain indicate that although ventral pulvinar neurons predominantly represent the contralateral hemifield (13, 19, 39) neurons with large, bilateral receptive fields have been recorded from cells within macaque lateral (ventral) pulvinar (40). Similarly, fMRI evidence in healthy humans (27, 30) suggests that, as in the monkey, human pulvinar may have neurons with large receptive fields that can influence attentional competition across the entire visual field. However, a more recent fMRI study reported activation within ventral pulvinar only for contralateral events (41). It is possible, however, that the bilateral effects observed in 2 of our patients (D.G. and C.R.) are attributable to involvement of dorsal parts of pulvinar, which is known in primates to have neurons with bilateral receptive fields (15) and reciprocal connections with posterior parietal cortex (16). Indeed, a patient with bilateral parietal lobe damage has also been shown to perform poorly on the same filtering task used here (42).

We cannot say whether this difference among the patients reflects a smaller and more anatomically restricted lesion in patient T.N., or a possible difference between left and right pulvinar functions. Previous PET and fMRI (26, 30) studies of visual filtering of distracters while identifying a foveal target have revealed only left pulvinar activation. The current observations provide the first direct evidence for a contribution of the right pulvinar in modulating perceptual salience.

Previous lesion studies that have used similar psychophysical paradigms to those reported here have shown that, after damage to “sources” of attentional bias [e.g., parietal cortex (42)], or “target” areas of attentional influence [e.g., V4 or TEO (36, 43)], high-contrast distracters severely impair selectivity. In contrast, our pulvinar patients were still able to make relatively fine target discriminations in the presence of distracters; the outcome of competition in occipito-temporal areas, although impaired, was not solely driven by the relative physical conspicuity of target and distracters. This is consistent with the notion that rather than operating as a primary source of attentional influence, or a critical subcortical relay between dorsal and ventral cortices, the pulvinar likely plays a more regulatory role in coordinating attentional effects (14, 44, 45). For example, intriguing possibilities include that the pulvinar may facilitate temporal synchronicity in topographically specific areas within occipito-temporal cortex, thereby increasing signal-to-noise ratios (13), or similarly, that it may influence neural coherence on a much broader scale between fronto-parietal and occipitotemporal systems (46). We anticipate that examining pulvinar structure and function in more detail using newly emerging diffusion magnetic resonance tractography (38) or high resolution fMRI connectivity techniques should provide powerful new insights into the neuronal mechanisms by which the pulvinar influences attentional filtering in visual cortex.

Materials and Methods

Patients.

D.G. is a 70-year-old right-handed man who suffered an hypertensive hemorrhage in the left posterior thalamus 3 years before testing. He has weakness in his right arm and leg, but he can walk with a cane. The hemorrhage destroyed most of the pulvinar, sparing only the most posterior ventro-lateral part.

C.R. is a 19-year-old right-handed man who suffered a closed head injury in a fall 2 years before testing, resulting in a hemorrhagic contusion and avulsion restricted to the posterior pole of the left pulvinar, with no other lesions to the brain. He has made a full clinical recovery and is now attending university.

T.N., a 60-year-old right-handed woman, suffered hypertensive right thalamic hemorrhage 8 years previously and has residual left arm and leg weakness and sensory loss but can walk with a cane. The lesion is restricted to the anterior pulvinar affecting the most rostral and dorsal part of the topographic maps of the ventral pulvinar, and causes deficits restricted to the inferior left quadrant. The region damaged in the lateral, ventral, and anterior “corner” of the pulvinar corresponds to the locus of activation for contralateral pulvinar maps observed in a recent fMRI investigation (35).

General Methods.

In the main task participants judged whether a sinewave grating target was tilted to the left or right. The target was either presented alone or with distracter discs above and below. In experiment 1 we varied distracter contrast across conditions (i.e., 0%, no distracters, 50%, or 80%) while holding target contrast constant. In experiment 2, distracter contrast was held constant whereas target contrast increased across conditions (i.e., 5%, 40%).

Apparatus, Stimuli, Design, and Procedure.

Stimuli were presented on a 16-inch View-Sonic CRT monitor (100-Hz refresh rate), controlled by a Pentium(R)-IV 3.4 GHz computer. Before the experiment, luminance values at the screen were measured using a photometer. These were used to create a look-up table to voltages that corrected for the nonlinearities of the screen such that an equal voltage increment led to an equal luminance increment at the screen. A remote SMI I-View eye-tracker was used to monitor fixation. Matlab and the Psychophysics Toolbox was used to control stimulus presentation and record responses (47, 48). In all experiments, target stimuli were phase-randomized sinusoidal gratings (spatial frequency 1.5 cycles per degree) presented within a Gaussian envelope (Gabor) with standard deviation of 0.5°. The Gaussian disk distracters had a standard deviation of 0.5° (circular aperture). Stimuli appeared on a gray background of mean luminance. Target contrast was calculated by subtracting dark-stripe luminance from white-stripe luminance, and dividing by the sum of the 2 luminances. Average luminance of the target and background were therefore equal. Distracter contrast was calculated by subtracting maximum disk luminance from the background luminance and dividing by the summed luminances. Target gratings were presented such that their centres were 4° from the central fixation point. For patient D.G. and C.R., stimuli were presented within the upper left and right quadrants of the visual field. For T.N., whose impairments are reportedly limited to the inferior visual field (32), targets were presented in the lower left and right quadrants. Viewing distance was 80 cm. The distracters were vertically aligned with the target and centered 4.79° (11 mm) above and below the grating. With a stimulus aperture size of 0.5°, the spacing between target and distracters was 0.29° (4 mm).

On each trial, a fixation point appeared for 1,000 ms, followed by the grating and distracters for 500 ms. A 500-ms auditory tone signaled target onset on each trial. The fixation point remained on-screen, until a response was made. “Left” and “right” responses were made by using the left and right arrow keys, respectively. Subsequent trials were initiated by the observer via another arrow key-press. To ensure that the difficulty of grating discrimination was comparable at each target position, we first established each observer's contrast detection threshold for gratings at each target location (see Psychophysical Methods).

Participants' gaze was monitored on all trials to ensure constant fixation using a remote I-View X eye-tracker. During all trials the participant's head was stabilized using a chin rest fixed at 80 cm from the computer monitor. Trials in which the eyes deviated outside a 1.5° circular radius around the point of fixation were discarded, and the following trial selected randomly from one of the staircases and left or right tilt direction.

We initially conducted a Contrast Sensitivity task to determine detection thresholds for the target gratings in each visual field. In subsequent experiments target contrast was scaled according to participants' detection performance, to equate task difficulty within contralesional and ipsilesional visual fields and between observers. Similarly, distracter contrast in experiment 2 was scaled using individually-defined target contrast thresholds obtained in the contrast sensitivity task as a guide (see Psychophysical Methods).

Experiments 1 and 2 each had 4 stimulus contrast conditions × 2 target sides, yielding a total of 8 conditions per experiment. Testing sessions took ≈2 h each. Within a testing session participants typically completed 1 block of trials per condition, with each block comprising 2 simultaneous interleaved staircases. Side of target presentation (left/right) was blocked across the 4 contrast conditions in each experiment. This served to minimize performance variability across contrast manipulations due to uncertainty about the target's location. The order of blocks was counterbalanced across sessions, both within and between observers, using an (A–B)(B–A) design. In each testing session, therefore, participants completed 8 blocks of trials, half with left-sided targets and half with right-sided targets. The order of contrast conditions (distracter contrast in experiment 1, and target contrast in experiment 2) was randomized within target side blocks, for each session and participant.

Psychophysical Methods.

Orientation discrimination thresholds were established for each target side and distracter contrast condition in the orientation sensitivity task. An interleaved staircase procedure was used using 2 opposing staircases (starting orientation Staircase 1 = 0.01°, Staircase 2 = 40.0°), with a maximum orientation limited to 80° from vertical. Target orientation was divided by 1.25 after 3 consecutive correct responses, and multiplied by 1.25 after 1 incorrect response. Each staircase terminated after 14 reversal points. Thresholds for each staircase were calculated as the geometric mean of the last 10 reversal points.

In experiment 1, participants T.N. and D.G. completed 4 thresholds per condition, and C.R. completed 6 thresholds per condition (and an identical number of thresholds per condition were established for the respective controls). In experiment 2, patient D.G. completed 4 thresholds per condition and C.R. completed 8 thresholds per condition.

The contrast Sensitivity task was designed to determine the contrast of target stimuli to be used in the orientation sensitivity task in experiment 1 and 2, for each observer. All aspects of the contrast sensitivity task were conducted in the same manner as the Orientation experiment, with the following exceptions. The target grating always appeared alone, without flanking distracters. On half the trials, a target Gabor was present (“target present” trials), whereas on the remaining trials, no target appeared (“target absent trials”). Participants were instructed to make a present/absent judgement about the target. As in the orientation experiment, 2 interleaved staircases were run simultaneously, but this time the contrast of the target Gabor was altered after the required number of consecutive correct or incorrect responses. Present and absent responses were made by using the left and right arrow keys, respectively. The geometric mean of each staircase was averaged to produce a mean contrast threshold. Independent measurements of threshold contrast were made for targets in the left and right hemifield for each observer. This procedure ensured that targets were always above-threshold (and could therefore always be discriminated) and that the task remained sufficiently difficult to enable sampling of performance within the dynamic range of each observers' contrast response function.

In experiment 1, mean contrast thresholds were multiplied by 3 to produce the contrast at which each grating appeared in the orientation sensitivity task, thereby ensuring that observers could reliably detect a target stimulus on each trial. In experiment 2 target contrast thresholds for each patient and visual field were multiplied by 2 to produce the contrast at which the target grating appeared in the target-only and target and distracter condition. Target contrast was increased by 5% and 40% respectively in the remaining conditions, whereas distracter percentage contrast was held constant at a value of 2x grating contrast threshold for each observer and visual field (see Table 2). Distracter contrast was reduced in experiment 2 (i.e., range 10–26% across observers) to permit an exploration of the effect of increased target contrast on orientation selectivity. For all ANOVAs the Hyunh–Feldt correction was used where appropriate.