Visual hemispatial neglect, re-assessed

Abstract

Increased computer use in clinical settings offers an opportunity to develop new neuropsychological tests that exploit the control computers have over stimulus dimensions and timing. However, before adopting new tools, empirical validation is necessary. In the current study, our aims were twofold: to describe a computerized adaptive procedure with broad potential for neuropsychological investigations, and to demonstrate its implementation in testing for visual hemispatial neglect. Visual search results from adaptive psychophysical procedures are reported from 12 healthy individuals and 23 individuals with unilateral brain injury. Healthy individuals reveal spatially symmetric performance on adaptive search measures. In patients, psychophysical outcomes (as well as those from standard paper-and-pencil search tasks) reveal visual hemispatial neglect. Consistent with previous empirical studies of hemispatial neglect, lateralized impairments in adaptive conjunction search are greater than in adaptive feature search tasks. Furthermore, those with right hemisphere damage show greater lateralized deficits in conjunction search than do those with left hemisphere damage. We argue that adaptive tests, which automatically adjust to each individual's performance level, are efficient methods for both clinical evaluations and neuropsychological investigations and have the potential to detect subtle deficits even in chronic stages, when flagrant clinical signs have frequently resolved.

Figure 1

Stimulus displays for the adaptive Feature Search task, illustrating target-absent (top) and target-present (bottom) displays. Displays used colored stimuli, but for illustrative purposes, white replaces red and black replaces blue. Targets were blue (shown as black) “O”s.

Figure 2

Stimulus displays for the adaptive Scattered Feature Search task, illustrating target-absent (top) and target-present (bottom) displays. Displays used colored stimuli, but for illustrative purposes, white replaces blue and black replaces red. Targets were red (shown as black) squares.

Figure 3

Stimulus displays for the adaptive Conjunction Search task, illustrating target-absent (top) and target-present (bottom) displays. Displays used colored stimuli, but for illustrative purposes, white replaces blue and black replaces red. Targets were red (shown as black) squares.

Figure 4

An example of a psychophysical staircase progression for the adaptive Conjunction Search task. Presentation time began at 2000 ms for both left (open white diamonds) and right (solid black circles) staircases. Points of inflection in the plots, i.e., reversals, are indicated in gray. Each staircase ended after ten reversals, and the TPTs for target detection on the left and right sides were calculated from the average of the last eight reversal points.

Figure 5

Scatter plots of healthy performance on the adaptive tasks: (A) Feature versus Conjunction Search, (B) Scattered Feature versus Conjunction Search and (C) Feature versus Scattered Feature. Regression lines and R²-values are shown for each pair of measures. Negative scores indicate that left target detection required longer presentation times than right target detection; conversely, positive scores indicate that right target detection required longer presentation times than left target detection. Scores of 0 indicate no field differences in target detection. Note that the scales vary: for CS it spans +/- 250, and for FS and SFS it spans +/- 25. In plot C, 7 points overlap (with scores of 0 on both measures).

Figure 6

Three standard fixed-measure pencil and paper search tasks (adapted from the SCAN, McGlinchey-Berroth et al., 2000). Each was presented on letter-sized paper, aligned with participant’s midline such that items were evenly distributed on the left and right sides. The experimenter marked the central demonstration item before participants searched for the remaining lines in (A) the line cancellation task, the remaining target letter “A”s in (B) the letter search task, or the remaining target symbols ( s) in (C) the symbol search task. As is shown, lines were “cancelled,” i.e., marked with a pen stroke, whereas target letters and symbols were circled. Three different patients’ performance is shown, and marked with the side of their lesion (RHD=right hemisphere damage, LHD=left hemisphere damage). Asymptomatic performance is illustrated in (A) the line cancellation task, i.e., no lines were missed. Right-sided neglect (i.e., more right- than left-sided misses) is illustrated in (B) the letter search task. Note that the patient omitted one target on the right side, and three on the left side. This patient’s score for the letter search task would be 2 (3 contralesional misses minus 1 ipsilesional miss). Left-sided neglect (i.e., more than left- than right-sided misses) is illustrated in (C) the symbol search task. Note that the patient omitted six targets on the left side and two on the right side. This patient’s score for the symbol search would be 4 (6 contralesional misses minus 2 ipsilesional misses).

Figure 7

Histograms of patient performance on (A) the fixed-measure Standard Search task, (B) the adaptive Feature Search task and (C) the adaptive Conjunction Search task. In all plots, positive values indicate contralesional scores (i.e., hemispatial neglect), whereas negative values indicate ipsilesional scores. No differences between contralesional and ipsilesional target detection result in 0 scores. In each plot, data were sorted into 11 bins spanning: 1 item (A), 50 ms (B) and 500 ms (C). Data are shaded according to the patient’s lesion side, as are mean scores (indicated by downward-pointing arrows)

Figure 8

Scatter plots of patient performance on the Standard Search task versus performance on the adaptive (A) Feature and (B) Conjunction Search tasks. Positive values indicate contralesional scores (i.e., hemispatial neglect), whereas negative values indicate ipsilesional scores. No differences between contralesional and ipsilesional target detection result in 0 scores. Regression lines and R²-values are shown for each pair of measures. Patient lesion side is indicated by filled triangles for LHD and open circles for RHD. Note that the y-axis scales in (A) and (B) are different: spanning +/- 250 and +/-2500, respectively. In plot A, 2 LHD points overlap (with scores of 0 on both measures).

Figure 9

Scatter plots of patient performance on the Conjunction Search task at two different delays post-stroke (on average, with ~1 year delay). Positive values indicate contralesional scores (i.e., hemispatial neglect), whereas negative values indicate ipsilesional scores. No differences between contralesional and ipsilesional target detection result in 0 scores. The regression line and R²-value is shown. Patient lesion side is indicated by filled triangles for LHD and open circles for RHD.