Visual neglect as a "disengage failure"

(Submitted to the 22^nd Midyear Meeting of the International Neuropsychological Society, Durban, South Africa, June 23-26, 1999)

Visual neglect as a "disengage failure"

A.J. Lundervold, K. Hugdahl, A. Lundervold, & K. Brønnick.

Correspondence:
Astri J. Lundervold, Dept. of Neurology, Haukeland University Hospital, N-5021 Bergen, Norway

Abstract
We applied a computer-based cue-target attentional test to a heterogeneous sample of 41 neurological patients with brain lesions verified on CT or MRI scans. Fuzzy c-means clustering of three derived parameters of measured reaction times was used to generate 4 "natural" subgroups. All patients with right hemisphere lesions were allocated to cluster 1 and 2. Both were characterised by severe attentional deficit in the contralateral visual field and cluster 2 also disengage failure. Patients with left hemisphere or bilateral lesions were found in cluster 3 and 4 with mild unilateral deficits without disengage failure. We conclude that this experimental model together with proper feature selection and multivariate analysis can lead to refined understanding of visual neglect with implications for patient selection to rehabilitation programs.

Symptoms of visual neglect are seen in patients with brain lesions affecting cortico-subcortical pathways (1) and are far more frequent after damage to the right than the left hemisphere (2). Visual neglect is a heterogeneous clinical phenomenon, characterised by a failure to respond or orient to new or meaningful stimuli presented in the attentional field contralateral to the damaged hemisphere. According to Posner et al. (3), attention can be described as three interacting networks: the orienting, the executive and the vigilance network. In this model visual neglect is explained as a dysfunction of the disengage operation within the orienting network. In the present study we applied an experimental implementation of this model to a sample of neurological patients. Using cluster analysis techniques we assume that by proper selection of feature variables a subgroup of neglect patients with disengage failure could be revealed.

SUBJECTS An experimental task was administered to 41 subjects: Twenty-seven patients with cerebral infarction, five with head trauma, three with multiple sclerosis and two with encephalopathy of unknown reason. Four patients with peripheral motor dysfunction after polio-myelites were included as a comparison group. Brain lesions and their lateralization were documented by CT or MRI scans. Sixteen patients had right hemisphere lesions, ten patients had left hemisphere lesions and 12 had bilateral lesions. Test-retest data were obtained for some of the patients, giving a total of 49 observations.

MEASURES We designed an attentional task similar to the cue-target paradigm described by Posner and Raichle (3). The cue was presented either on the same side as the target (valid cuing), on the opposite side of the target (invalid cuing), or the target was presented without a cue (noCue) in two test-conditions: one with an external cue (exogenous paradigm) and one where the cue was a central arrow that must be interpreted to help direct attention (endogenous paradigm). Reaction-time (RT) in [ms] was recorded as dependent measure. Psychomotor tempo (f1) was defined as mean RT for targets on invalid and noCue trials Unilateral attentional deficit (f2) was defined by the difference in RT between target presentations to the left and right side of the fixation point. A difference larger than the maximum difference in the comparison group was defined as a deficit. Disengage failure (f3) was defined when RT to targets presented after an invalid cue was slower than targets presented with no cue.

DATA ANALYSIS The patients were categorised according to lesion site and grouped as follows: Right hemisphere=1; left hemisphere = 2; bilateral lesions = 3 and the comparison group = 0. Size and direction of the difference between left and right RT as well as the disengage failure in observations with unilateral attentional deficit were explored by computing scatterplots. Variables f1, f2, and f3 were selected as features to investigate natural subgroups in the data set {(f_i1, f_i2, f_i3)} using fuzzy c-means clustering (6). Here, f_ij denotes the value of feature fj (j=1,…,3) in observation i (i=1,…,n). Four clusters (c=4) were employed.

RESULTS

Figure 1 reveals that 19 of 21 observations with right hemisphere lesions showed an attentional deficit to the left in the endogenous test condition. Nine of 11 observations with left hemisphere lesion demonstrated slower RT to right compared with left targets. The difference was largest in the right hemisphere group and on invalid trials. The 13 observations with bilateral lesions were mixed and the difference in comparison patients around zero.

Figure 1 Figure 2

Unilateral attentional deficit was revealed in 36 observations in the exogenous and 40 in the endogenous condition. In the endogenous condition, the majority of patients with right hemisphere lesions showed disengage failure to the left targets, a subgroup showed disengage failure both to right and left targets and a few patients demonstrated slower reaction time to the noCue trials compared with invalid trials to the left (Figure 2). Disengage failure to the right was shown in less than half of the observations with lesion in the left hemisphere, with an overall less severe impairment than observations with right lesions. Mean RT was about 100 milliseconds slower to targets presented to the right than left (Figure 3).

Figure 3

Table 1

Fuzzy clustering centroids (endogenous left)

................... f1 ...... f2 ...... f3

Cluster #1 771.62 213.37 121.78

Cluster # 2 615.65 121.49 11.59

Cluster #3 387.01 -41.33 -9.12

Cluster #4 534.28 2.43 -21.49

The fuzzy c-means clustering was conducted using f1 to f3 in the 40 patients with unilateral attentional deficit in the endogenous condition. Cluster centroids for left-sided presentations are presented in Table 1. Only patients with right hemisphere lesions were found in Cluster #1 and #2. Cluster #1 is characterised by a disengage failure to the left, a marked left>right difference and slowed reaction-time to targets presented to the left. The patients within this cluster thus have a left sided neglect that can be explained as a disengage failure. Cluster #2 is characterised by slow reaction time and visual neglect to the left, but no clear disengage failure. Both patients with a moderate unilateral attentional deficit and a more general attentional deficit seem to belong to this cluster. Three patients with right hemisphere lesions were categorised in Cluster #4, characterised by mild impairment. Two of these patients improved clinically over time, and the third patient had an infarction in the basal ganglia. All patients with lesions in the left or in both hemispheres were found in Clusters #3 and #4.

DISCUSSION

The most severe unilateral attentional deficit was found in patients with right hemispheric lesions, confirming earlier studies (2) and theories of hemispheric specialisation, suggesting that the left hemisphere is specialised to process details in the right visual field, while the right hemisphere seems attentive to global aspects in both the right and left visual half-fields (3, 4).

Attention is a distributed function in the brain and different attentional disorders are associated with different lesion sites. In our study, some of the patients with no disengage failure showed a deficit within the executive network of attention (e.g. patients with lesions restricted to the basal ganglia). Even symptoms of visual neglect were heterogeneous. Disengage failure was only revealed in a subgroup of patients with unilateral attentional deficit, all characterised by left hemispheric lesions. A more detailed study of lesion site and demarcation, stage of illness and overall neuropsychological function is part of an ongoing study of these patients.

Disengage failure was more prominent in the test condition depending on internal than external cues, making sense of using external cues in training procedures to patients with symptoms of visual neglect. The success in using internal cues depends on intensity and time of training as well as cognitive functioning in more general. By generating subgroups of patients with different attentional deficits, rehabilitation programs might be more adapted to a given patient, his or her cognitive, psychological and social function in activities of daily living. This is still a major challenge in neuropsychological rehabilitation of patients with symptoms of visual neglect.

REFERENCES

(1) Vallar, G. (1993). The anatomical basis of spatial hemineglect in humans. In J.H. Robertson & J.C. Marshall (eds.). Unilateral neglect: Clinical and experimental studies. UK: Lawrence Erlbaum Associates Ltd. Publishers, 27-59

(2) Halligan, P.W., & Marshall, J.C. (1994). Toward a principled explanation of unilateral neglect. Cognitive Neuropsychology, 11, 167-206.

(3) Posner, M.I. & Raichle, M.E. (1994). Images of Mind. New York: Scientific American Library.

(4) Heilman, K., & Van Den Abell, T. (1980). Right hemisphere dominance for attention: The mechanisms underlying hemispheric asymmetries of inattention (neglect). Neurology, 30, 327-330.

(5) Robertson, L.C., & Lamb, M.R. (1991). Neuropsychological contributions to part-whole organization. Cognitive Psychology, 23, 299-332.

(6) Bezdek, J.C. (1981) Pattern Recognition with Fuzzy Objective Function Algorithms. New York: Plenum Press