Validity of the visual object and space perception battery (VOSP) in stroke patients with posterior brain lesions

(1)

Validity of the Visual Object and Space Perception

battery (VOSP) in stroke patients with posterior brain

lesions.

Masterthesis Clinical Neuropsychology Juni 2017

Naam: F.J. Annegarn Studentnummer: 605978

Opleiding: Master Gezondheidszorgpsychologie; Klinische neuropsychologie Onderwijsinstelling: Universiteit van Amsterdam

Begeleider: S. van der Werf Tweede lezer: I. Slighte

(2)

MASTERTHESIS: VALIDITY OF THE VOSP IN STROKE PATIENTS WITH

POSTERIOR BRAIN LESIONS J. ANNEGARN 2

ABSTRACT

The current study sought to answer whether Warrington’s two-factor model, that comprises object and space perception in the VOSP battery, can be confirmed in a population of stroke patients with posterior brain lesions. Furthermore, this study looked at the differences in outcome on the VOSP battery and single subtests between LHL and RHL patients. Previous studies, spread over several decades, showed contradicting results concerning the validity of the VOSP battery and theories about possible differences between LHL and RHL patients. This study involved stroke patients with posterior brain lesions (n=130). Analyses involved confirmatory factor analyses, partially adapted from Rapport et al. (1998). Furthermore, differences between LHL and RHL patients are analysed through nonparametric frequency analysis, T-tests and ANOVA. Results show a generally good fit for a tripartite (three-factor) model. These findings show that object perception and space perception share an underlying common visual perception factor. Our findings support Warrington’s assumption that RHL patients have more difficulty with the VOSP battery. Several subtests of the VOSP battery seem to be independently useful in clinical practice. Further research should be concentrating on a different theory for visual processing, were there are not only two major visual

(4)

4

1) INTRODUCTION

An important component of a neuropsychological assessment is the evaluation of the visual perception. It stems from the possibility that visual perceptual disorders could be an

explanation for poor performance on neuropsychological tests measuring cognitive abilities. Numerous neuropsychological tests combine several aspects of visual processing, as well as motor functions, making it difficult to single out visual perceptual disorders.

Warrington and McCarty (1991) designed a series of tests that were intended to measure the domains of visual processing separately. Nowadays, the Visual Object and Space Perception battery (VOSP) is commonly used by neuropsychologists to evaluate visual perception.

The VOSP includes a number of tests designed to assess specific, dissociable aspects of object and space perception. The test battery intended to evaluate visual perception

problems in brain-damaged individuals, relatively independently of other cognitive and motor processes. This is attractive for clinical use, where testing on visual perception deficits is required before drawing conclusions about cognitive functions such as attention and executive functioning. Thereby, the VOSP operates under the assumption that object perception (OP) and space perception (SP) are two independent constructs. Moreover, several clinical studies have studied left and right differentiation concerning object and space perception. Results have shown conflicting results (Bonello et al., 1997; Merten, 2006; Quental, Brucki & Bueno, 2013; Schintu et al, 2014, Warrington & James, 1991). The two independent constructs and the left and right differentiation are still matter of debate and will be a focus of this study.

(5)

5 1.1 Purpose

Although the VOSP is a widely used instrument with good psychometric properties in terms of test-retest reliability and sensitivity to detect change, its use may be hampered by

questionable validity. Furthermore, single subtests of the VOSP are often used in

neuropsychological evaluations and there is contradicting evidence regarding the validity of single subtests. The manual contains no guidelines for the interpretation of individual score profiles or interpretation on subtest level. Additionally, nothing is mentioned about a relation between performance on specific subtests and perception disorders or agnosia’s. It seems that studies regarding the validity of the VOSP battery mostly used patients without posterior brain lesions, which is surprising because this patient group would supposedly experience more visual perception problems (Bonello et al., 1997; Herrera-Guzmán et al., 2004; Rapport et al., 1998).

1.2 Object perception

It has been described in several articles how the model for object agnosia from Warrington and colleagues (Taylor & Warrington, 1973; Warrington & Taylor, 1973, 1978; Warrington 1982; McCarthy & Warrington, 1990, Warrington et al., 2009) is the theoretical underpinning of the development of VOSP. This model distinguishes three stages supporting object

recognition: (1) Visual analysis, the sensory input processed from the eyes to the brain. (2) Perceptual categorization, the perceived object is categorized. Despite the diversity of an object, it normally still is placed in the right category. (3) Semantic categorization, where the perceptual representation is combined with the knowledge about the characteristics of the object (Heutink & Bouwman, 2012).

Considering this model, three subtypes of impaired object recognition can be distinguished: (1) disorders of visual sensory discrimination, (2) perceptual categorization

(6)

6

disorders and (3) semantic categorization disorders. These disorders can result from

disruptions in different brain areas. For example, visual sensory discrimination disorders can be seen when the occipital areas of both hemispheres are damaged. In addition, perceptual categorization disorders can occur because of lesions in the right hemisphere (RHL) and semantic categorization disorders can be caused by (predominantly) left hemisphere lesions (LHL) (McCarthy & Warrington, 1990).

According to the model, RHL patients have problems with object recognition under difficult conditions. RHL patients view objects incorrect due to overemphasis on details or overlooking important characteristics, for instance, when incomplete object images are

presented, or when an object is shown from an unusual viewpoint. In contrast to LHL patients who have problems with recognizing objects when the meaning of the images requires subtle semantic discrimination (semantic categorization disorders).

1.3 Space perception

In addition of Warrington's model for object perception, the VOSP is also based on the theoretical assumption that object and space perception are two functionally independent domains. McCarthy and Warrington (1990) assume that visual disorientation refers to a disorder in the localization of the position of objects in the space. This disorder is commonly seen after bilateral damages in the occipital-parietal border area. Visuospatial agnosia is a disorder in spatial analysis, which also contains disorders like unilateral neglect (attention disorder resulting in ignoring one side of the visual field), spatial discrimination and visual scanning. In general, these skills are related to the right hemisphere, where posterior lesions lead to the most severe and enduring deficits (McCarthy & Warrington, 1990).

(7)

7 1.4 Earlier research

Over the last three decades several studies have been published, werethe VOSP battery is used in different clinical groups, as well studies especially investigating the psychometric properties of the battery and its intended task sensitivity for right hemisphere patients over left hemisphere patients (Bonello et al., 1997; Merten, 2006; Quental, Brucki & Bueno, 2013; Schintu et al, 2014). Studies on the reliability on (parts) of the VOSP showed that the test-retest reliability is high (Merten, 2006). The study of Herrera-Guzmán et al. (2004) showed a significant effect for age and gender in the same five subtests (SI, OD, PS, PD and CA), education showed a marginally significant effect on OD and SI. The majority of their results are in line with earlier research in an American and English sample (Bonello et al., 1997; Warrington and James, 1991). With regard to the validity of the VOSP battery, results are more variable (Bonello et al., 1997; Merten, 2005).

The research of Kaplan & Hier (1982) showed a significant result for the influence of visuospatial deficits on visuospatial tests. In addition, visuospatial disorders were found to be more severe in patients with posterior right hemisphere damage in contrast with anterior damage. Warrington & James (1991) observed a low percentage of LHL patients with a deviant score, supporting the disadvantage of the right hemisphere concerning object and space perception. It was, however, not clear to what extent lesion location and visual field defects influenced this observation. Contrary, Merten et al. (2006) did not find confirmation for the exclusive involvement of the right hemisphere to low performance on the Silhouette identification subtest.

Furthermore, mixed results were found with regard to the sensitivity of the test in different clinical populations. For instance, the use of cut-off scores from the manual resulted in the study of Bonello et al. (1997) in an unacceptable high percentage of subjects whose performance on specific subtests was wrongly classified as impaired.

(8)

8

Rapport et al. (1998)conducted confirmatory factor analytic research to evaluate the congruence of the dimensional structure in comparison with the theoretical structure, which the development of the VOSP is based on, namely Warrington’s two-factor model of object, and space perception. They indeed found that a two-factor model (object and space perception is measured separately by the VOSP) was a better fit than a one-factor model (VOSP

measures only visual perception, with two dimensions object and space perception). Likewise, support was demonstrated for right hemisphere damage detection properties. The authors argue that the difference between the models could have been greater, if there would be a greater psychometric reliability of the VOSP.

1.5 VOSP

In the VOSP, a form-discrimination task serves as an initial screening test (Shape detection), to assess whether the visual and sensorial capacity of patients are sufficient to complete the other subtests. Administration of the remaining VOSP battery is not advised at a score of 15 or less. Furthermore, there are four subtests evaluating object perception. Starting with Incomplete letters (IL), which evaluates whether a patient has a selective disorder in reading degraded letters. Next is Silhouettes (SI), were patients have to recognize regular animals and objects positioned from an irregular angle. This can be especially difficult for patients with lesions in the occipital lobe of the right hemisphere. In Object decision (OD), patients have to recognize objects; however, they do not have to name the object, which excludes a verbal component. The last one is Progressive silhouettes (PS) were the object recognition threshold is evaluated. Patients with lesions in the right hemisphere have a higher threshold with three-dimensional images rotated from 90° to 0°. There are four tests evaluating space perception. The first one is Dot counting (DC), where patients have to count dots, evaluating object location and space perception. The second test, Position discrimination (PD), evaluates

(9)

9

whether a patient is able to correctly interpret the position of two objects in comparison of each other. Number location (NL) evaluates more complicated space perception. The last subtest, Cube analysis (CA), patients interpret two-dimensional pictures of a three dimensional object. This evaluates the ability to perceive spatial relations of objects.

1.6 This study

The current study sought to answer whether Warrington’s two-factor model, that comprises object and space perception in the VOSP battery, can be confirmed in a population of stroke patients with left and right posterior brain lesions. The first hypothesis is that RHL patients show more problems/difficulty with the total VOSP battery (OP and SP combined) than LHL patients. We assume that there will be a significant difference between the scores of RHL and LHL patients, were RHL patients will score significantly lower on the VOSP battery than LHL patients.

Our second hypothesis constitutes that LHL patients have more difficulty with object perception subtests and RHL patients have more difficulty with space perception subtests (double dissociation). The assumption is that statistical analysis will reveal performance on OP subtests is significantly influenced due to the left hemispheric damage and performance on SP subtest is significantly influenced due to right hemispheric damage.

The last hypothesis constitutes the theoretical structure on which the VOSP battery is based (a two-factor model of object and space perception) corresponds with the dimensional structure of VOSP battery.We propose that a confirmatory factor analysis will confirm that a two-factor model is a better fit than a one-factor model in a group of patients with posterior lesions. This is in line with earlier findings in a group of healthy older adults (Rapport et al., 1998). Furthermore, Warrington’s proposed two-factor model will be compared to a tripartite model. A tripartite model constitutes a common factor besides two independent factors. This

(10)

10

would mean that besides the two independent factors OP and SP, they would share a common visual factor. The goodness of fit indices will show the two-factor model to be a good fit and a better fit for the dimensional structure in comparison with a one-factor model or tripartite model.

2) METHODS

2.1 Participants and procedure

For this study, we used data collected during clinical assessments at the Royal Dutch Visio in Haren, Haarlem, Heerhugowaard and Amsterdam the Netherlands, a rehabilitation institute for blind and partially sighted people, between November 2004 and January 2016. In this period, 229 stroke patients were assessed. All patients were seen for a visual assessment and diagnostic neuropsychological assessment. The neuropsychological assessment consisted of a variation of different neuropsychological tests and questionnaires, specially fitted to the problems and questions of the patient’s situation. The VOSP or parts of the VOSP were a standard element of the assessment.

During the visual assessment, visual acuity was assessed using an eye chart or optical instruments. The acuity was measured separately for the left and right eye (monocular), in addition, the acuity both eyes (binocular) was measured. These measurements were done with and without the patient’s own correction (e.g. glasses, contacts). Furthermore, the ability to see contrast was evaluated using functional acuity contrast test (FACT). Possible visual field impairments were evaluated using a Goldmann and/or Humphrey method, sometimes

alternativemethods were necessary due to the condition of the patient. Whether the visual field loss affected the macula was evaluated in about half the patients, partly to determine the ability to see details (see table 1). Patients were excluded from this study with visual acuity lower than 0.3, a score of 15 or less on the screening test of the VOSP battery and an

(11)

11

incomplete assessment of the VOSP battery. This study has been approved in keeping with ethical and institutional guidelines, by the department of Psychology of the University of Amsterdam.

Table 1 – Descriptive and percentages of perimetry assessment for stroke patients included in this

study.

All (n=130) LHL (n=31) RHL (n=76)

Group n % cum.% n % cum.% n % cum.%

Visual acuity ODS

(Snellen) ≥ 0.4 122 93.8 93.8 29 93.5 93.5 70 92.1 92.1

0.3-0.4 8 6.2 100 2 6.5 100 6 7.9 100

Contrast normal 116 89.2 92.9 26 83.9 83.9 68 89.5 89.5 deviant 8 6.2 100 2 6.5 90.4 6 7.9 97.4 Test visual field Goldmann 48 36.9 38.7 13 41.9 46.4 28 36.8 36.8

Humphrey 6 4.6 43.5 1 3.2 50 5 6.6 43.4

Both 65 50 96 14 45.2 100 38 50 93.4

Peri-Ball 5 3.8 100 5 6.6 100

Visual field deficits Left 73 59.2 59.2 4 12.9 12.9 64 84.2 84.2 Right 33 25.4 84.6 22 71.0 83.9 3 3.9 88.1

Both 6 4.6 89.2 4 5.3 93.4

No visual field loss 18 13.8 100 5 16.1 100 5 6.6 100 Macular sparing yes 63 48.5 79.7 18 58.1 81.8 40 52.6 80

no 16 12.3 100 4 12.9 100 10 13.2 100

2.2 Materials

Visual Object and Space Perception battery.

The VOSP consists of nine subtests: a screening test, four object perception tests and four space perception tests. The screening test measures form discrimination to establish visual sensory deficits and basic shape perception. If the patient scores below 15, administration further testing with the VOSP is not recommended.

(12)

12

1) Incomplete Letters: the subtest contains two trial and twenty stimulus cards. Patients have to identify letters that have been faded out, and only 30% of the original shape of the letters remains.

2) Silhouettes: Patients are shown thirty silhouette drawings of common objects (fifteen animals and fifteen objects) from atypical perspectives. The silhouettes are rotated on the lateral axis and the items increase in difficulty.

3) Object Decision: a multiple-choice subtest with twenty items, with four choices. Each item consists of four silhouette drawings, and the patient has to select a real object among three nonsense shapes.

4) Progressive Silhouettes: consists of two series of ten silhouette drawings, a handgun and a trumpet. These objects are rotated back from 90° to 0° around their longitudinal axis. The score is made up out of the number of images the subject needs to see to recognize the object.

Figure 1. – Examples of the different object perception subtests in the VOSP Battery.

Left to right: Shape detection, Incomplete letters, Silhouettes, Object decision, Progressive Silhouettes. Shape detection and incomplete letters are administrated from a booklet a little smaller than a5. Silhouettes, Object decision and Progressive silhouettes are administrated with an a4-sized booklet (Warrington, E. K. & James, A. M., 1991).

(13)

13

Four tests for space perception (shown in figure 2).

5) Dot Counting: The patient is asked to count randomly arranged black dots on ten cards. The patient has to localize points and use simple visual scanning.

6) Position Discrimination: Twenty items consists of two squares side by side, both with a black dot. In one square, the dot is positioned exactly in the centre; in the other, it is slightly off-centre. The subject has to determine in which of the squares the dot is in the off-centre.

7) Number Location: There are ten items, presenting two squares placed above each other. The upper square contains randomly placed numbers (1-9). The lower square presents a black dot. In this spatial localisation test, the patient has to identify the number in the upper square that corresponds with the dot in the lower square.

8) Cube Analysis:The patient has to determine how many blocks are represented in the line drawing. The blocks that possibly are hidden from the view and cannot be seen have to be considered as well. Ten two-dimensional drawings of three-dimensional configurations of block are presented.

Figure 2. – Examples of the different space perception subtests in the VOSP Battery.

Left to right: Dot counting, Position discrimination, Number location, Cubes. These subtests are administrated from a booklet a little smaller than a5 (Warrington, E. K. & James, A. M., 1991).

The total number of correct answers for each subtest composes the overall results. Except for Progressive Silhouettes, here the total pictures required to recognize the object is counted. For all subtests cut-off scores have been determined for two age categories, below 50 years and 50 years and above (Spreen, 2006).

(14)

14 2.3 Statistical analysis

The significance level was set at p < 0.05 with two-sided testing. Assumptions for the

statistical analyses were checked and all participants fulfilled the referred inclusion/exclusion criteria. A power analysis was performed to check the power of our sample size, using the program GPower 3.1©_{. Analysis of the distribution (Kolmogorov-Smirnov coefficient,}

skewness, and kurtosis), homogeneity (Levene’s coefficient) and descriptive statistic were obtained using IBM SPSS statistics 23©_{. Depending on the distribution of the sample (i.e.}

nonnormality), different methods of statistical analysis were used.

First, a T-test was performed to compare the results of LHL patients versus RHL patients using SPSS. Second, a MANOVA was performed to examine whether LHL patients have more difficulty with object perception subtests and RHL patients with space perception. Third, a confirmatory factor analysis (CFA) was performed to assess whether the theoretical two-factor model proposed by Warrington is a better fit than a one-factor model and tripartite model, using R version 3.3.2©_{. Estimation techniques used in the CFA were adapted from the}

study of Rapport et al. (1998). Primary the SCALED χ2_{technique with robust standard errors}

(Bentler, 1993) was conducted and the maximum likelihood (ML) technique was used as a secondary method. Rapport et al., 1998 does not specify why the secondary method was used. Several studies were consulted to obtain the fit indices and acceptable threshold levels that should be used in this study (Hu & Bentler, 1999; Rapport et al., 1998; Schreiber et al., 2006).

(15)

15

3) RESULTS

3.1 Sample characteristics

The mean age is 61.0 (SD13.0) years and the sample consist of 82 males and 48 females (see table 1). Patients had to complete the whole VOSP to be included in this study. Patients with an acuity (binocular with own correction) lower than 0.3, a score below the cut-off on the VOSP screenings test (n=25), and patients with lesions in both hemispheres (n=23) were excluded from this study. This left a remaining sample of 130 stroke patients in the current study.

The decision was made to proceed with non-parametrical tests. A test for multivariate normality revealed that Incomplete letters D (130)=0.3, p<0.01; Silhouettes D(130)=0.09, p<0.01; Object decision D(130)=0.170, p<0.01; Progressive silhouettes D(130)=0.08, p<0.05; Dot counting D(130)=0.35, p<0.01; Position discrimination D(130)=0.20, p<0.01; Number location D(130)=0.3, p<0.01; Cube analysis D(130)=0.3, p<0.01; Total VOSP score D(130), p<0.01; Total Object perception score D(130)=0.09, p<0.05 and Total space perception score D(130)=0.17, p<0.01 were all significantly non-normal. The variables were divided in LHL, and RHL damage groups. Still most of the variables showed a significantly non-normal distribution. Patients made low numbers of errors on the subtests, translating in mostly negatively skewed subtests, showing a ceiling effect.

Levene’s test for homogeneity showed equal variance for most of the data. The variances were significantly different for Dot counting F(2, 127)=10.73, p<0.01; Position discrimination F(2,127)=5.13, p<0.01 and Number location F(2, 127)=3.4, p<0.05. A power analysis revealed a power of 0.76 for a CFA, 0.64 for a T-test and 0.97 for a MANOVA with a dataset of 107 participants (LHL and RHL groups).

(16)

16

Table 2 – Descriptive (mean ± standard deviations) and percentages for the stroke patients included in

this study.

All (n=130) LHL (n=31) RHL (n=76)

Group N Mean (SD) % N Mean (SD) % N Mean (SD) % Age 130 61.0 (13.0) 31 58.2 (14.1) 76 61.1 (13.0) Sex male 82 63.1 22 71.0 45 59.2 female 48 36.9 9 29.0 31 40.8 Education 105 5.01 (1.2) 23 4.7 (1.2) 62 5.2 (1.1) Time 0-6 months 34 26.2 10 32.3 20 26.3 6-12 months 55 42.3 15 48.4 27 35.5 >12 months 4 30.8 6 19.4 29 38.2 IS 102 78.5 25 19.2 57 43.8 HS 24 18.5 5 3.8 16 12.3 SAH 4 3 1 0.8 3 2.3

Note: LHL: patients with left hemispheric lesion, RHL: patients with right hemispheric lesion. Age: years at the

time of the VOSP administration; Education: Verhage education code (Verhage, 1964). The highest achieved education corresponds with a number between 1 and 7 on the Verhage scale, were 1 represents less than 6 years of lower education and 7 represents a university degree; Time: time passed since brain damage at the time of VOSP administration; IS: Ischemic stroke; HS: haemorrhagic stroke, SAH: subarachnoid haemorrhage

3.2 Right side vs. left side

Warrington proposed that patients with RHL experience more difficulty with the VOSP battery than LHL patients. In this sample, a Mann-Whitney test showed the total VOSP score in RHL patients did not differ significantly from LHL patients (see table 3). A Mann-Whitney test showed that the total score on space perception differs significantly between RHL and LHL patients. Whereas the object perception did not seem affected by the side of brain

damage. In clinical practice, it is commonly presumed that LHL patients show more difficulty with the object perception subtests, and RHL patients with the space perception subtests. From the eight subtests, a Kruskal-Wallis test showed that the side of brain damage significantly affects DC, PD and CA. On subtest level, this indicates RHL patients score significantly lower on these subtests in comparison to LHL patients. Furthermore, IL shows a trend (see table 3).

(17)

17

Table 3 – Comparison between LHL patients and RHL patients on VOSP scores.

LHL (n=31) RHL (n=76) Mann-Whitney test

Scoring on Mean SD Mean SD U z p r

Total VOSP 106,10 14,963 100,26 17,978 960 -1,498 0,134 0,145 Total OP 60,55 11,584 58,46 12,532 1062 -0,797 0,425 -0,07 Total SP 45,55 5,227 41,80 7,219 793 -2,652 0,008** -0,23 Kruskal-Wallis test H(1) p IL 17,13 5,136 16,80 3,889 2,987 0,084 SI 17,74 5,348 16,92 5,489 0,03 0,863 OD 16,61 2,539 16,57 2,665 0 0,986 PS 9,06 2,816 8,17 3,62 1,655 0,198 DC 9,71 0,739 8,68 1,827 9,68 0,002** PD 17,87 2,553 16,30 3,254 5,54 0,019* NL 8,87 1,522 8,70 1,987 0,18 0,671 CA 9,10 1,904 8,24 2,326 5,399 0,020*

Note: OP = Object perception; SP = Space perception; IL = Incomplete Letters; SI = Silhouettes; OD = Object

Decision; PS = Progressive Silhouettes; DC = Dot counting; PD = Position Discrimination; NL= Number Location; CA = Cube Analysis. * p<0.05, **p<0.01.

3.3 Factor structure of the VOSP

The correlations show a correlation between almost all subtests, except for NL (see table 4). More ideal would be correlations between only the OP subtests and only the SP subtests. A closer look at the correlations reveals PD correlates with every subtest and CA with 7 out of 8. In this study visual perception was described in a one-factor model, a two-factor model and a tripartite model (figure 3). The one factor model represented the hypothesis that the eight VOSP subtests measured only one latent construct of visual perception. The two factor model, as proposed by Warrington, assumes that object and space perception are two

independent factors. Additionally, the tripartite model considers object and space perception as two independent factors, however it assumes a common overall visual perception factor as well. Table 5 summarizes the fit measures for all models.

(18)

18

Figure 3 – Representation of the one-factor model, two-factor model and tripartite (or three-factor) model of the VOSP theory.

Table 4 – Correlations, means and standard deviations of the VOSP battery.

IL SI OD PS DC PD NL CA IL 1.00 SI .554** 1.00 OD .297** 0.498** 1.00 PS .415** .458** .458** 1.00 DC .198* .194* .238** .221* 1.00 PD .469** .338* .340** .358** .458** 1.00 NL .253** .245** .166 .183* .182 .133 1.00 CA .572** .355** .268** .352** .497** .402** .402** 1.00 *p<0.05; **p<0.01 Tripartite model One-factor model Two-factor model

(19)

19

The absolute fit indices (chi-square, RMSEA, GFI, AGFI, SRMR) look at the ability of the model to reproduce the observed covariance matrices; it determines how well an a-priori model fits the data. The relative fit indices (TLI/NNFI, CFI, and IFI) involve comparing the theoretical model specified to a baseline model.

3.3.1 One-factor model and two-factor model

For the one-factor model the robust comparative fit index (CFI) = 0.82, the robust Tucker-Lewis fit index (TLI) = 0.75, the root mean square error of approximation (RMSEA) = 0.16, SCALED chi-square = 70.5, degrees of freedom (df) = 20, and p-value = 0.00. The absolute and relative fit indices show a poor fit except for GFI and AGFI. For the two-factor model the robust CFI = 0.90, the robust TLI = 0.85, RMSEA = 0.12, SCALED chi-square = 49.5, degrees of freedom = 19, and p-value = 0.00. The same as for the one-factor model the absolute and relative fit indices show a poor fit except for GFI and AGFI. The fit indices for both the one-factor model and two-factor model are an overall indication of a poor fit between the theorized models and the observed data. However, the two-factor model seems less of a poor fit than the one-factor model. The robust CFI, robust TLI slightly indicate a moderate fit in contradiction to the one-factor model.

3.3.2 Three factor model

A tripartite model was analysed using the maximum likelihood estimation. The absolute fit indices chi-square = 16.96 with df = 12 and p = 0.151, GFI = 0.97, SRMR = 0.042, and RMSEA = 0.06 show a good fit. AGFI = 0.91 shows a moderate fit. The comparative fit indices CFI = 0.99, TLI = 0.97 and IFI = 0.99 show a good fit as well. In contrast to the one- and two-factor models, this tripartite model shows an overall indication of a good fit between the proposed model and the observed data.

(20)

20

Table 5 – Summary of fit for measurement models.

Absolute fit indices Relative fit indices

Model SCALED χ2 P SRMR Robust CFI Robust TLI One-factor 70.54 (20) 0.00 0.077 0.824 0.753 Two-factor 49.54 (19) 0.00 0.069 0.897 0.848

Absolute fit indices Relative fit indices

Model χ2 p GFI AGFI SRMR RMSEA

p-close CFI NNFI (TLI) IFI One-factor 87.34 (20) 0.000 0.99 0.98 0.077 0.16 0.000 0.82 0.75 0.82 Two-factor 59.20 (19) 0.000 0.99 0.99 0.069 0.12 0.003 0.89 0.84 0.89 Tripartite 16.96 (12) 0.151 0.97 0.91 0.042 0.06 0.383 0.99 0.97 0.99 Note: Above: Satorra-Bentler estimation, below: Maximum likelihood estimation. χ2 = chi-square (degrees of freedom), CFI = Comparative fit index, TLI/NNFI =Tucker-Lewis Index/non-normed fit index, GFI = Goodness of fit index, AGFI = Adjusted GFI, IFI = incremental fit index, RMSEA = root mean square error of

approximation, SRMR = standardized root mean squared residual.

3.4 Exploring on subtest level

The frequencies and percentages of patients scoring below the cut-off on the VOSP subtests are shown in table 6. Exploring the sample shows for the total sample the highest a-priori chance to score below cut-off for PD and SI. Because the a-priori chances for the separate groups (LHL and RHL) show the same result, these subtests seem to be the most difficult in this population. On all subtests besides OD, a higher percentage of RHL patients scored below cut-off in comparison to LHL patients, suggesting RHL patients have more difficulty with the VOSP subtests. Moreover, the difference in percentage below cut-off in DC and PD between LHL and RHL were found to be significant. These findings support the theory of Warrington claiming the VOSP is more difficult for RHL patients.

(21)

21

Table 6 – Results on the subtests of the VOSP below cut-off.

All (N=130) LHL (N=31) RHL (N=76)

< cut-off % A priori < cut-off % A priori < cut-off % A priori

IL 26 20 0.2 5 16.1 0.16 18 23.7 0.24 SI 35 26.9 0.27 6 19.4 0.19 23 30.3 0.30 OD 20 15.4 0.15 5 16.1 0.16 9 11.8 0.12 PS 17 13.1 0.13 2 6.5 0.07 12 15.8 0.16 DC 20 15.4 0.15 1* 3.2 0.03 16* 21.1 0.21 PD 58 44.6 0.45 9* 29 0.29 41* 53.9 0.54 NL 13 10 0.1 1 3.2 0.03 9 11.8 0.12 CA 16 12.3 0.12 2 6.5 0.07 12 15.8 0.16

Note: Patients with left hemispheric lesion (LHL), patients with right hemispheric lesion (RHL), patients with brain lesions both left and right hemisphere (BL). IL = Incomplete Letters; SI = Silhouettes; OD = Object Decision; PS = Progressive Silhouettes; DC = Dot counting; PD = Position Discrimination; NL= Number Location; CA = Cube Analysis. *p<0.05; **p<0.01

Table 7 shows the number of subtests where patients scored below cut-off. Of the LHL patients, the majority (almost 84%) scored none or one subtest below cut-off. The number of subtests below cut-off in RHL patients is more varied. Comparable percentages of LHL patients (35.5%) and RHL patients (34.2%) did not fail at any of the subtests. Significantly, there is no difference between these two groups. However, only 16.1% of LHL patients scored two or more subtests below cut-off, while in comparison 44.7% of the RHL patients failed two or more subtests. Analysis shows a significant difference between LHL patients and RHL patients that score two or more subtests below cut-off on the subtests (see table 7). This implies that RHL more often fail multiple subtests in comparison to LHL patients. This finding supports Warrington’s theory that RHL patients have more difficulty with the VOSP.

(22)

22

Table 7 – Number of subtests below the cut-off score.

All (N=130) LHL (N=31) RHL (N=76) Subtests < cut-off N % Cumulative % N % Cumulative % N % Cumulative % 0 48 36.9 36.9 11 35.5 35.5 26 34.2 34.2 1 36 27.7 64.6 15** 48.4 83.9 16** 21.1 55.3 2 17 13.1 77.7 3 9.7 93.6 11 14.5 69.8 3 9 6.9 84.6 1 3.2 96.8 8 10.5 80.3 4 5 3.8 88.4 5 6.6 86.9 5 7 5.4 93.8 5 6.6 93.5 6 3 2.4 96.1 2 2.6 96.1 7 5 3.8 100 1 3.2 100 3 3.9 100

Note: LHL: patients with left hemispheric lesion, RHL: patients with right hemispheric lesion, BL: patients with brain lesions both left and right hemisphere, IL: Incomplete Letters; SI: Silhouettes; OD: Object Decision; PS: Progressive Silhouettes; DC: Dot counting; PD: Position Discrimination; NL: Number Location; CA: Cube Analysis. *p<0.05; **p<0.01

4) DISCUSSION

This study sought to provide support for Warrington’s two-factor model, that comprises object and space perception in the VOSP battery, in a population of patients with right and left hemispheric damage. The first hypothesis constituted that RHL patients show more problems with the total VOSP battery than LHL patients do. RHL patients showed more frequent a deviant score on VOSP subtests. Higher percentages scored below cut-off per subtest and more patients had a higher amount of subtests below cut-off. These findings support Warrington’s theory that RHL patients have more difficulty with the VOSP battery.

The second hypothesis stated that LHL patients have more difficulty with object perception subtests and RHL patients have more difficulty with space perception subtests. In this study both space perception overall scores and number of failed subtests indicated that there were differences between RHL and LHL patients. This suggests that RHL patients indeed seem to have more difficulties with space perception tasks compared to LHL patients. This finding supports the sensitivity of the space perception tasks for RHL patients.No

(23)

23

evidence was found to support the theory that LHL patients have more difficulty with OP subtests. On the contrary, RHL patients seem to fail the OP subtests more often than LHL patients, supporting Warrington’s theory that the VOSP is most sensitive for RHL patients.

The study of Schintu et al. (2014) is one of several studies that looked at the neural underpinnings of object and space perception using the VOSP. Their results showed different hemispheric dominance for object and space perception tasks. Their conclusion incorporated the ventral ‘what’ and dorsal ‘where’ route and stated: “Our findings identified distinct and lateralized brain regions critical for object and space perception within the left ventral stream and the right dorsal stream, respectively”.Although Warrington & McCarty (1991) did not mention the involvement of the ventral and dorsal stream in the manual of the VOSP, it has been for several decades, a reasonable assumption that object perception relies on the ‘left’ ventral route and space perception on the ‘right’ dorsal route.This assumption can be

endorsed by several studies from the ’80 till not that long ago (Batelli et al, 2008; Goodale & Milner, 1992; Haxby et al., 1991; Humphreys & Riddoch, 2006; Kravitz et al., 2011; Kravitz et al., 2013; Milner & Goodale, 2008; Rizzolatti & Matelli, 2003; Schintu et al., 2014;

Ungerleider & Haxby, 1994). However, increasingly more studies propose different theories considering (the neural underpinnings of) visual perception and deviate from the two separate pathways and the left and right hemispheric dominance.

The last hypothesis stated that the theoretical structure on which the VOSP battery is based corresponds with the dimensional structure. Results show that neither the one-factor model nor the two-factor model had a good fit in a population of patients with posterior lesions. The tripartite model showed a generally good fit, which suggests that object and space perception share an underlying common visual perception factor. Contradicting to our hypothesis, this result states that the model underlying the VOSP battery in this patient group might be more hierarchical than initially proposed by Warrington and McCarty (1991).

(24)

24 Comparing our results with the results of Rapport et al. (1998) resulted in a very

different outcome. Because our sample included patients with posterior lesions, an even better fit of the two-factor model was expected. Rapport et al. (1998) found a reasonable good fit for the two-factor model, however in our sample the one-factor nor the two-factor model showed a good fit. Their study only focussed on the one- and two-factor model and did not consider the possibility of a tripartite model. The study of Haan & Cowey (2011) challenged the theory of visual processing organised along only two separate visual pathways. Their review does not dismiss hierarchical organization of the visual system, but argue that the hierarchies may be found within each visual area. An example was described for the perception and

recognition of colour. Several were studies taken into account and demonstrated that the colour processing system includes the supposedly lower order processing of the information from the retina, and shares a number of characteristic with higher-order processing systems such as face recognition. The statement about a more hierarchical organization of the visual system is more in line with our finding that the tripartite model is a better fit in for this population in comparison to the one- and two-factor model.

Several constraints have to be taken into account interpreting of the results of the factor analyses. First, Rapport et al. (1998) mentioned: “An important assumption underlying the ML estimator is that the data is multivariate normal. Violation of this assumption can lead to an inadequate evaluation of the model. The χ2_{goodness-of-fit tests tend to reject too many}

true models and parameter estimates may be biased, yielding too many significant results (Browne, 1984; West, Finch, & Curran, 1995)”. In this study, the assumption of multivariate normality is also violated. Reason to still use the ML estimator is, the estimator is considered to be a relatively robust to violations of normality assumptions (Bollen, 1989). Furthermore, Reinartz et al., 2009 reported Monte-Carlo experiments with samples of different sizes and different Kurtosis and Skewness in terms of the usage of the ML estimator. This study did not

(25)

25

find major difference in SEM analysis results. Another consideration is that relative fit indices presumably, are less affected by nonnormality. An explanation for the nonnormality could be the high ceiling effects seen in the performance on the VOSP battery. To adequately compare the results of our study with the study of Rapport et al. (1998) it seems important to know if their sample suffered from an equally of even higher ceiling effects.

Secondly, considering the experiment design a possible bias in the assessment of performance of patients on different subtests has to be considered. Neuropsychologists could differ in which answers they consider right or wrong. However, neuropsychologists are assumed to follow protocol as much as possible. Moreover, some of the OP subtests are susceptible for aphasia. In this sample, there are no patients with severe aphasia, but patients with a mild form or residues of aphasia could have been present, which makes naming objects more complicated. These patients could do worse on subtests not because of a perception problem, rather a verbal problem. Additionally several if not all subtests are susceptible for visual inattention disorders (neglect). In this sample, it is highly possible patients suffered from a form of an inattention disorder, which could negatively have effected their results. Furthermore, lower order visual problems can be of influence on the performance the subtest. In this study, low acuity has been taken into account. To our surprise most of the participating patients in this suffered from visual field loss (almost 90%), which makes is hard to

distinguish whether this effected the performance on the VOSP battery.

Our clinical experience suggests that not all of the subtests in the VOSP are equally informative in people with posterior brain lesions. Moreover, during a neuropsychological assessment regularly not the whole VOSP is administrated due to limited time, limited energy and expectation of possible cognitive disorders. In our population, SI and PD are the most difficult (and thereby most sensitive) subtests, with a minimum or no ceiling effect. These subtests seem to be a good starting point when administration of the whole VOSP is not

(26)

26

possible. If patients score below cut-off on these subtest, further testing seems necessary. The manual states that the VOSP battery is especially suitable for RHL patients, however in clinical practise it is used for both RHL and LHL patients. This study shows that the VOSP is more sensitive for RHL patients in comparison to LHL patients. However, this does not mean the VOSP is insensitive for LHL patients. Even though RHL patients made more mistakes than LHL patients did, it does not mean LHL patient do not make mistakes on subtests. Furthermore, only a few significant differences were found between RHL and LHL patients on the different subtests.

For future research, it seems to be inevitable to further investigate the “patchwork model” suggested by Haan & Cowey (2011) and deviated from the theories of only two independent visual perception routes. Haan & Cowey (2011) stated, “It is likely that a number of separate maps are involved in the perception and recognition of one visual characteristic, and that these subordinate maps process information that increases in complexity”. The study mentioned the research of Op de Beeck and Baker (2010), in which the system for object recognition is argued in a similar way. This study seemed to support a more complex hierarchy in the visual system withthe tripartite model that showed a far better fit in comparison to the two-factor model.

Finally, the VOSP battery proves to be useful for the assessment of higher visual perception in patients with posterior brain lesions, where it is most sensitive for RHL patients. The subtests Silhouettes and Position discrimination are the best starting point when whole administration of the VOSP is not possible. The system underlying visual perception is more hierarchically organized than previously assumed. Further research of the neuropsychological assessment and (neural) underpinnings of the visual system, should deviate from the two commonly known visual pathways and focus on a more complex and hierarchical visual

(27)

27

system. Moreover, this could give more insight and leads for possibly a new neuropsychological visual perception battery in the future.

(28)

28

5) REFERENCES

Acres, K., Taylor, K. I., Moss, H. E., Stamatakis, E. A., & Tyler, L. K. (2009).

Complementary hemispheric asymmetries in object naming and recognition: A voxel-based correlational study. Neuropsychologia, 47(8-9), 1836-1843.

Arbuckle, J. L. (2014). IBM SPSS Amos 23 Grad Pack [Computer software]. Chicago: Smallwaters.

Battelli, L., Walsh, V., Pascual-Leone, A. & Cavanagh, P. (2008). The ‘when’ parietal

pathway explored by lesion studies. Current opinion in Neurobiology, 18, 120-126. DOI: 10.1016/j.conb.2008.06.004.

Bollen, K.A. (1989). A new incremental fit index for general equation models. Sociological methods & research, 17(3), 303-316.

Bonello, P.J., Rapport, L.J. & Millis, S.R. (1997). Psychometric properties of the visual object and space perception battery in normal older adults. The clinical neuropsychologist 11(4), 436-442.

Byrne, B.M. (2010). Structural equation modelling with AMOS: basic concepts, applications, and programming (2nd_{Ed.). London. : Taylor and Francis Group, LLC}

Coslett, B.H. (1997). Neglect in vision and visual imagery: a double dissociation. Brain, 120, 1163- 1171.

Curran, P.J., West, S.G. & Finch, J.F. (1996). The robustness of test statistics to nonnormality and specification error in confirmatory factor analysis. Psychological methods, 16-29. Goodale, M.A. & Milner, A.D. (1992). Separate visual pathways for perception and action.

Trends in neurosciences, 15(1), 20-25.

Haan, B. & Cowey, A. (2011). On the usefulness of 'what' and 'where' pathways in vision. Trends in cognitive neurosciences, 15(10), 460-466.

(29)

29

Herscovitch, P., Schapiro, M. B. & Rapoport, S. I. (1991). Dissociation of object and spatial visual processing pathways in human extrastriate cortex. Proceedings of the national academy of sciences, 88, 1621-1625.

Herrera-Guzmán, I., Peña-Casanova, J., Lara, J. P., Gudayol-Ferré, E. & Böhm, P. (2004). Influence of age, sex, and education on the visual object and space perception battery (VOSP) in a healthy normal elderly population. The clinical neuropsychologist, 18, 385-394.

Heutink, J. & Bouwman, I. C. M. (2012). Visuele perceptie van object en ruimte (VPOR). In A. Bouma, J. Mulder, J. Lindeboom & B. Schmand (Red.), Handboek

neuropsychologische diagnostiek (2e herziene druk, pp. 341-359). Pearson Assessment and Information B.V.

Hu, L. & Bentler, P.M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural equation modelling 6(1), 1-55. Humphreys, G.W. & Riddoch, M. J. (2006) Features, objects, action: The cognitive

neuropsychology of visual object processing, 1984–2004. Cognitive Neuropsychology, 23:1, 156-183, DOI: 10.1080/02643290542000030.

Kaplan, J. & Hier, D.B. (1982). Visuospatial deficits after right hemisphere stroke. The American journal of occupational therapy, 36(5), 314-321.

Konen, C.S., Behrmann, M., Nishimura, M., & Kastner, S. (2011). The functional neuroanatomy of object agnosia: a case study. Neuron, 71(1), 49-60.

Kravitz, D.J., Kadharbatcha, S.S., Baker, C.I. & Mischkin, M. (2011). A new neural framework for visuospatial processing. Nature news neuroscience, 12, 217-230.

Kravitz, D.J., Kadharbatcha, S.S., Baker, C.I., Ungerleider, L.G. & Mishkin, M. (2013). The ventral visual pathway: an expanded neural framework for the processing of object quality. Trends in cognitive sciences, 17(1), 26-49.

(30)

30

Lissauer, H. (1890). Ein fal von seelenblindheit nebst einem bietrag zur theorie derselben. Archiv für psychiatrie und nervenkrankheiten, 216, 22-270 (English translation in Cognitive Neuropsychologia, 1988, 5, 153-192)

McCarthy, J.C. & Warrington, E.K. (1990). Cognitive neuropsychology. A clinical introduction. San Diego: Academic Press.

Merten, T. (2005). Factor structure of the Hooper visual organization test: a cross-cultural replication and extension. Archieves of clinical neuropsychology, 20, 123-128

Merten, T. (2006). An analysis of the VOSP Silhouettes Test with neurological patients. Psychology science, 48(4), 451-462

Milner, A.D. & Goodale, M.A. (2008) Two visual systems re-viewed. Neuropsychologia, 46, 774-785.

Mishkin, M., Ungerleider, L.G. & Mako, K. A. (1983). Object vision and spatial vision: two cortical pathways. Trends in Neurosciences, 6(0), 414-417.

Newcombe, F. (1969). Missile wounds of the brain: a study of psychological deficits. Oxford U. Press, vi 145 pp.

Newcombe, F., Ratcliff, G. & Damasio, H. (1987). Dissociable visual and spatial impairments following right hemisphere posterior cerebral lesions: clinical, neuropsychological and anatomical evidence. neuropsychologia, 25(1B), 149-161.

Op de Beeck, H.P. & Baker, C.L. (2010). The neural basis of visual object learning. Trends in cognitive sciences, 14, 22-30.

Price, C. J., Moore, C. J., Humphreys, G. W., Frackowiak, R. S., & Friston, K. J., (1996). The neural regions sustaining object recognition and naming. Proceedings of the royal

society of London. Series B: Biological sciences, 263(1376), 1501-1507

(31)

31

Alzheimer's disease - the use of the visual object and space perception (VOSP) battery. PLoS ONE 8(7): e68398. DOI:10.1371/journal.pone.0068398.

Rapport, L.J., Millis, S. R. & Bonello, P. J. (1998). Validation of the Warrington theory of visual processing and the visual object and space perception battery. Journal of clinical and experimental neuropsychology, 20(2), 211-220.

Reinartz, W., Haenlein, M. & Henseler, J. (2009). An empirical comparison of the efficacy of covariance-based and variance-based SEM. International Journal of Research in

Marketing, 26, 332-344.

Rizzolatti, G. & Matelli, M. (2003). Two different streams from the dorsal visual system: anatomy and functions. Experimental Brain Research, 153, 146–157. DOI:

10.1007/s00221-003-1588-0.

Schintu, S., Hadj-Bouziane, F., Dal Monte, O., Knutson, K.M., Pardini, M., Wassermann, E.M., Grafman, J. & Krueger, F. (2014). Object and space perception – is it a matter of hemisphere? Cortex, 57, 244-253.

Schreiber, J.B., Nora, A., Stage, F.K., Barlow, E.A. & King, J. (2006). Reporting structural equation modelling and confirmatory factor analysis results: A review. The journal of Educational research, 99(6), 323-338, DOI: 10.3200/JOER.99.6.323-338.

Tinelli, F., Bulgheroni, S., Mazzotti, S., Vago, C., Groppo, M., Scaramuzzo, R.T., Riva, D., & Cioni, G. (2014). Ventral stream sensitivity in “healthy” preterm-born adolescents: Psychophysical and neuropsychological evaluation. Early human development, 90, 45- 49.

Ungerleider, L.G. & Haxby, J. V. (1994). “What” and “where” in the human brain. Current opinion in Neurobiology, 4(2), 157-165.

Ungerleider, L. G. & Mishkin, M. (1982). Analysis of visual behaviour. In D.J. Ingle, M. A. Goodale & R. J. W. Mansfield (Eds.), Two cortical visual systems. Cambridge: MIT

(32)

32

Press.

Warrington, E.K. (1982). Neuropsychological studies of object recognition. Philosophical transactions of the royal society in London. B. Biological sciences, 298, 15-33

Warrington, E.K. (2009). Warrington and Taylor’s 1978 paper. Autor’s update. Perception, 38, 933-947.

Warrington, E. K. & James, A. M. (1991). The Visual Object and Space Perception Battery. Amsterdam: Pearson Assessment and Information B.V.

Warrington, E.K. & Taylor, A.M. (1973). The contribution of the right parietal lobe to object recognition. Cortex, 9, 152-164.

Warrington, E.K. & Taylor, A.M. (1978). Two categorical stages of object recognition. Perception, 7, 695-705.

Zachariou, V., Klatzky, R., & Behrmann, M. (2013). Ventral and dorsal visual stream contributions to the perception of object shape and object location. Journal of cognitive neuroscience, 1-21.

Zelkowicz, B. J., Herbster, A. N., Nebes, R. D., Mintun, M. A., & Becker, J. T. (1998) An examination of regional cerebral blood flow during object naming tasks. Journal of the international neuropsychological society, 4(2), 160-166.

Zihl, J. (2011). Rehabilitation of visual disorders after brain injury. East Sussex. Psychology Press.

Validity of the visual object and space perception battery (VOSP) in stroke patients with posterior brain lesions