Brain Advance Access originally published online on December 14, 2007
Brain 2008 131(5):e94; doi:10.1093/brain/awm273
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
No reversal of the Oppel–Kundt illusion with short stimuli: confutation of the space anisometry interpretation of neglect and ‘cross-over’ in line bisection
1Fondazione Santa Lucia IRCCS, Roma, 2Dipartimento di Psicologia, Università degli Studi di Roma "La Sapienza", 3Dipartimento di Scienze Neurologiche – V a Cattedra Università degli Studi di Roma "La Sapienza", Italy and 4Institute of Cognitive Neuroscience, University College London, London, UK
Correspondence to: Prof. Fabrizio Doricchi, Centro Ricerche di Neuropsicologia, Fondazione Santa Lucia - IRCCS, Via Ardeatina 306 - 00179 Roma - Italy. E-mail: fabrizio.doricchi{at}uniroma1.it
Key Words: spatial neglect; line bisection; crossover; space anisometry; Oppel-Kundt illusion
Received October 5, 2007. Accepted October 24, 2007.
Right-brain damaged patients suffering neglect for the left side of space bisect long horizontal lines to the right of the true centre. This phenomenon is traditionally explained by pathological reduction of the attentional salience of the contralesional side of the line and enhancement of the attentional salience of its ipsilesional side due to brain damage (Pouget and Driver, 2000
; Bartolomeo and Chokron, 2002). At variance with this largely accepted explanation, Bisiach and co-workers (Bisiach et al., 1994; Bisiach, 1997; Bisiach et al., 2002) conjectured that in neglect patients the representation of horizontal space is continuously and progressively ‘relaxed’ toward the left and ‘compressed’ toward the right, in a logarithmic manner. By consequence the left side of a symmetrical horizontal line immersed in this ‘anisometrical’ representational gradient will be perceived as being shorter than its physically equivalent right side, causing rightward shift of the subjective line midpoint. It is clear that this hypothesis predicts that at any point in the gradient and for any length of a line set within this gradient, the line's left side will always be perceived as being shorter than its identical right counterpart and the subjective line midpoint will always be shifted rightward. Thus, we previously concluded (Doricchi et al., 2005) that the space anisometry hypothesis cannot account for the paradoxical leftward shift that can be observed when neglect patients are asked to mark the centre of very short horizontal lines, i.e. the cross-over effect.
Bisiach (1997) also originally suggested that the anisometrical distortion of space suffered by neglect patients is adequately represented by horizontal Oppel–Kundt (O–K) illusions with vertical segments progressively increasing in spatial frequency towards the right endpoint (see ‘Right Longer’ stimuli, Fig. 1). In their recent study Savazzi et al. (2007) wished to determine whether while viewing horizontal O–K illusions simulating the hypothetical distortion of space encountered in neglect syndrome, the performance of healthy subjects might mimic that of neglect patients, showing deviation in the illusion's direction in the bisection of long horizontal illusory gradients and deviation in the direction opposite to the illusion in the bisection of very short gradients (i.e. cross-over effect). Savazzi et al. argue that if the horizontal O–K illusion is able of producing both of these effects, then the space anisometry interpretation of neglect can account for, rather than being disproved by, the ‘cross-over’ effect. Compared to previous interpretations, the space anisometry interpretation would then provide only one functional mechanism to explain the patients’ performance both for long and short lines ‘... which is more consistent with the principle of economy, parsimony and simplicity in scientific theories ...’ (Savazzi et al., p. 2082). In the first experiment of their study, Savazzi et al. asked five healthy aged subjects to bisect horizontal illusory gradients with vertical line segments progressively compressed from right to left (‘Left Longer’ lines inducing subjective lengthening of their left side) or from left to right (‘Right Longer’ lines inducing subjective lengthening of their right side). Bisections of these stimuli were compared to bisections of control isometric gradients (‘Filled-isometrical’ lines with evenly spaced vertical segments) or empty gradients (‘Empty’ lines with no vertical segments; see examples of all stimuli in Fig. 1). The length of stimuli varied across 16, 8, 4 or 2 cm. Compared to Filled-Isometrical and Empty lines, Savazzi et al. found that anisometrical Left Longer items induced leftward deviation, while Right Longer items induced rightward deviation (thus mimicking left spatial neglect). The illusion in the conventional direction induced by anisometrical items diminished for progressively shorter 16, 8 and 4 cm stimuli. However, with very short 2 cm anisometrical stimuli the direction of the illusion reversed: Left Longer gradients induced rightward bisection shifts, while Right Longer gradients induced leftward bisection shifts thus mimicking, in this latter case, the cross-over effect in left spatial neglect. In the comment of these results, which represent the founding empirical premise on which the rationale and conclusions of their study is entirely based, Savazzi and co-workers claim (p. 2074) that their ‘... results are in keeping with psychophysical evidence (Bulatov et al., 1997; Bulatov and Bertulis, 1999) gathered using the traditional Oppel–Kundt illusion in a size comparison task ...’ and that ‘... These data and the ones found in the present experiment ... indicate that in healthy subjects the strength of the Oppel–Kundt illusion gradually decreases when using stimuli occupying little space. More importantly, they demonstrate that the effect of such an illusion reverses for very short lines: empty space appears larger (instead of smaller) than filled space. The present study provides the additional information that this distortion of space affects where the midline is perceived to be ...’.
|
Regrettably, we have found that these claims are refuted by the very studies by Bulatov and Bertulis (1999
, 2005) and Bulatov et al. (1997) quoted by Savazzi and co-workers, as these studies demonstrate no reversal of horizontal O–K illusion for any stimulus length [i.e. from 7' to 141' of the visual angle, see fig. 12 in Bulatov et al. (1997); see also Bertulis and Bulatov (2005) for stimuli encompassing 60' or 140' of the visual angle]. It is worth of note that the shortest stimulus used by Bulatov and co-workers (i.e. 7' of the visual angle), is much shorter than the 2 cm stimuli that Savazzi et al. presented to their subjects at reading distance. In their study, Bulatov et al. (1997) rather show (fig. 12) that the shorter the horizontal stimulus’ length, the greater, proportionally, the strength of the illusion in the conventional direction. This evidence replicates and expands results from a previous study by Long and Murthagh (1984) in which the length of O–K illusions was parametrically varied (from 0.65° to 1.43°) both in forced-choice comparison and absolute-size judgement tasks. Therefore, all of the findings reported in literature run against the rationale and conclusions of Savazzi et al.' s study. We supposed that several factors might have accounted for the discrepancy between the findings by Savazzi et al. and those reported by all of the other authors in literature: (i) the paucity of the sample of normal control subjects examined by Savazzi et al. (n = 5); (ii) the use of a visuomotor bisection task as opposed to the purely visual comparison tasks used in previous studies; (iii) the use of horizontal O–K illusions enclosed in a rectangular frame as opposed to classical O–K configurations with no frame enclosing the vertical segments inducing the illusion. To explore whether the findings reported by Savazzi et al. are representative of the visuomotor performance of normal subjects, we asked an adequately numbered sample of healthy subjects to perform bisections of horizontal O–K illusory gradients. Thirty young subjects (mean age = 24.53, SD = 2.29; mean education = 16.6, SD = 0.77) and 30 elderly subjects (mean age = 66, SD = 7.68; mean education = 9.2, SD = 5.7) bisected anisometrical O–K illusions (Left Longer, Right Longer) and control Filled-isometrical or Empty stimuli. Anisometrical illusions were created with the same exponential function used by Savazzi et al. (although we found that this function does not exactly fit the original size of 16 cm, fitting instead the size of 16.6 cm: therefore we graphically rescaled the initial 16.6 cm stimulus to 16 cm). Then, two types of stimuli were made. The first type (Closed Stimuli) exactly replicated stimuli used by Savazzi et al., consisting of 1 mm x 1 cm vertical segments enclosed in a 16 cm x 1 cm rectangular frame. These stimuli were subsequently downscaled to fit 8 cm x 1 cm, 4 cm x 1 cm and 2 cm x 1 cm rectangular frames. The same procedure was used to create Filled-isometrical and Empty-Closed control stimuli. The second type of O–K illusions (Open Stimuli) were created by using unframed 16 cm x 1 cm anisometrical gradients which were subsequently graphically rescaled to 8 cm x 1 cm, 4 cm x 1 cm and 2 cm x 1 cm stimulus size. The same procedure was adopted to create Filled-isometrical and Empty-Open control stimuli. Closed and Open stimuli were presented in separate consecutive blocks. In each block, stimuli of different sizes were completely randomized and each participant received a different sequence of stimuli. Half of the subjects (15 young, 15 elderly) first bisected Closed and then Open stimuli. The other half bisected items in reverse order. In both conditions, five trials were presented for each size x stimulus type (Left Longer, Right Longer, Filled-isometrical, Empty), making a total of 80 Closed and 80 Open trials. Using a magnifying lens, deviations of the subjects’ mark from the objective midpoint were measured to the nearest half millimetre: leftward deviations were coded as negative ones and rightward deviations as positive ones. Scorers were blind to the aim of the study.
Mean individual bisections were entered in an Age (Young, Elderly) x Block Order (First Closed, First Open), Stimulus Condition (Closed, Open) x Stimulus Type (Left Longer, Right Longer, Filled-isometrical, Empty) x Stimulus Length (16, 8, 4, 2 cm) Anova. The Anova showed significant main effects for Stimulus Type F(3,168) = 121, P < 0.0001, Stimulus Length F(3,168) = 39, P < 0.0001 and a significant Stimulus Type x Stimulus Length interaction F(9,504) = 76, P < 0.0001. In keeping with the observations by Savazzi et al., a rightward bias was observed in the bisection of longer (16 cm, 8 cm) Filled-isometrical and Empty control stimuli (Fig. 1 and Table 1). Left Longer stimuli determined leftward bisection deviations when compared to equivalent 16, 8 and 4 cm Filled-isometrical lines (planned comparisons, P < 0.001) and equivalent 16 and 8 cm Empty lines (planned comparisons, P < 0.001). Right Longer stimuli produced rightward bisection errors when compared to equivalent 16, 8 and 4 cm Filled-isometrical (planned comparisons, P < 0.001) and Empty lines (planned comparisons, P < 0.001). However, in contrast to the findings by Savazzi et al. and in keeping with results by Long and Murthagh (1984), Bulatov et al. (1997), Bulatov and Bertulis (1999) and Bertulis and Bulatov (2005), no reversal of the O–K illusion (with cross-over effect) was found in the bisection of short 2 cm anisometrical stimuli. More specifically no significant difference was observed between the bisection of 2 cm Left Longer lines in comparison to both equivalent Filled-isometrical and Empty control stimuli (planned comparison, P > 0.44) whereas, exactly replicating what was found for longer stimuli, 2 cm Right Longer stimuli induced a significant rightward deviation when compared both to equivalent Empty (planned comparisons, P = 0.001) and Filled-isometrical lines (planned comparisons, P < 0.05). These findings for short 2 cm stimuli were also replicated when, similarly to the statistical procedure used by Savazzi et al., separate Anovas were carried out to compare Right Longer with Empty lines [Stimulus type x Stimulus length interaction: F(3,168) = 106, P < 0.0001; planned comparison between 2 cm lines P < 0.001] or Left Longer with Empty lines [Stimulus Type x Stimulus Length interaction: F(3,168) = 25, P < 0.0001; planned comparison between 2 cm lines P = 0.6]. These data clearly demonstrate that the Right Longer O–K illusion supposedly resembling the horizontal space distortion occurring in left spatial neglect and inducing in normal subjects rightward shifts in the bisection of long horizontal stimuli does not reverse and does not cause cross-over towards the left of the true center in the bisection of short horizontal stimuli. We conclude that the findings reported by Savazzi and co-workers are not representative of the visuomotor performance of normal subjects. Most importantly, since horizontal O–K illusions faithfully represent the hypothetical anisometrical distortion of space suffered by neglect patients, as argued by Bisiach (1997) and by Savazzi et al. (2007), all of the findings reported in literature (Long and Murthagh, 1984; Bulatov and Bertulis, 1997; Bulatov et al., 1999; Bertulis and Bulatov, 2005) together with our visuomotor results unequivocally confute the space anisometry interpretation of spatial neglect behaviour and cross-over effect in line bisection.
|
We recently observed (Doricchi et al., 2005
) that, when compared to right-brain damaged controls without spatial neglect or hemianopia, cross-over in the bisection of 2 cm lines is quantitatively consistent in neglect patients with hemianopia (N+H+), whereas it is very infrequent and not quantitatively significant in neglect patients without hemianopia (N+H–). In keeping with many previous studies, we also found that N+H+ patients have stronger ipsilesional deviation in the bisection of long horizontal lines than N+H– patients. Though completely overlooked by the authors, the data by Savazzi et al. fully replicate our 2005 findings. According to individual data reported in table 2 from Savazzi et al. paper, it is evident that all of the N+H+ patients (cases N+H+ 1 to 4) had stronger ipsilesional deviation as compared to any of the N+H– patients (cases N+ 1 to 17) in the bisection of long 200 mm lines. This is confirmed by a simple statistical comparison performed between these two groups [average deviations: N+H– = 23.98 mm; N+H+ = 59.95 mm; F(1,19) = 35, P < 0.0001]. Most of all, N+H+ patients showed dramatic leftward cross-over in the bisection of 2 cm gradients independently of stimulus type (see experiment 4, table 5). Given any stimulus type, the leftward deviation showed by N+H+ patients was much higher than any leftward deviation showed by any of the other groups of patients considered in the study (N+H–, N–; experiment 2, table 3; experiment 3, table 4). This holds especially true for Filled-isometrical stimuli with evenly spaced inducers, which represent the most appropriate comparative baseline for stimuli with anisometrically spaced inducers. Short Filled-isometrical stimuli determined in N+H+ a leftward deviation (–5.20 mm; experiment 4) up to 13 times greater than that observed in N+H– patients (averaged deviation across experiments 2 and 3 = –0.4 mm) and N– patients (–0.4 mm; experiment 2).
To explain the strong cross-over effect found in N+H+ patients we recalled and developed the well-established notion that loss of retinotopic representation of space in hemianopia engenders compensatory fixations toward the blind contralesional hemifield with foveal or parafoveal areas of the seeing ipsilesional hemifield (Ishiai et al., 1987; Zihl, 1995; Trauzettel-Klosinsky and Reinhard, 1998): thus, a primary spatial deficit engenders a saccadic bias (in the discussion section of their study, Savazzi et al. entirely misunderstand and misrepresent this point as our explanation clearly does not attribute the primary cause of cross-over to eye movements). We argued further that when hemianopic patients with attentional neglect are engaged in the bisection of long horizontal lines, the contralesional bias caused by hemianopia is overcome and masked by the powerful pathological attentional bias induced by the long half of the line lying in the ipsilesional space. On the contrary, the very short ipsilesional half of short lines is unable of biasing attention so that the contralesional bias induced by hemianopia can find its full behavioural expression in the cross-over effect. In the third study from our investigation (Doricchi et al., 2005), recording of eye and arm movement during line bisection confirmed the plausibility of our simple mechanicistic explanation. Interestingly, similar oculomotor findings were reported in an independent study run by Ishiai et al. (2006) in four neglect patients suffering from concomitant visual field defects.
In conclusion, evidences from the very investigations quoted by Savazzi et al., together with the findings reviewed and reported in the present commentary, demonstrate that the space anisometry hypothesis cannot offer a parsimonious account of spatial neglect behaviour. Neglect-related phenomena like the cross-over effect can be better explained by precise and simple neurophysiological and attentional mechanisms outdating the purely metaphorical interpretation of clinical evidence offered by the space anisometry hypothesis.
 |
References |
|---|
 Top References |
|---|
Bartolomeo P, Chokron S. Orienting of attention in left unilateral neglect. Neurosci Biobehav Rev (2002) 26:217–34.[CrossRef][Web of Science][Medline]
Bisiach E. The spatial features of unilateral neglect. In: Parietal lobe contribution to orientation in 3D space.—Thier P, Karnath H, eds. (1997) Berlin: Springler Verlag.
Bisiach E, Rusconi ML, Peretti VA, Vallar G. Challenging current accounts of unilateral neglect. Neuropsychologia (1994) 32:1431–4.[CrossRef][Web of Science][Medline]
Bisiach E, Neppi-Modona M, Ricci R. Space anisometry in unilateral neglect. In: The cognitive and neural basis of spatial neglect.—Karnath H, Milner AD, Vallar G, eds. (2002) Oxford: Oxford University Press.
Bertulis A, Bulatov A. Distortions in length perception: visual field anisotropy and geometrical illusions. Neurosci Behav Physiol (2005) 35:423–34.[CrossRef][Medline]
Bulatov A, Bertulis A. Distortions of length perception. Biol Cybernetics (1999) 80:185–93.[CrossRef][Web of Science][Medline]
Bulatov A, Bertulis A, Mickiene L. Geometrical illusions: study and modelling. Biol Cybernetics (1997) 77:395–406.[CrossRef][Web of Science][Medline]
Doricchi F, Guariglia P, Figliozzi F, Silvetti M, Bruno G, Gasparini M. Causes of cross-over in unilateral neglect: between-group comparisons, within-patient dissociations and eye movements. Brain (2005) 128:1386–406.
Ishiai S, Furukawa T, Tsukagoshi H. Eye-fixation patterns in homonymous hemianopia and unilateral spatial neglect. Neuropsychologia (1987) 25:675–9.[CrossRef][Web of Science][Medline]
Ishiai S, Koyama Y, Seki K, Hayashi K, Izumi Y. Approaches to subjective midpoint of horizontal lines in unilateral spatial neglect. Cortex (2006) 42:685–91.[Web of Science][Medline]
Long GM, Murtagh MP. Task and size effects in the Oppel-Kundt and irradiation illusions. J Gen Psychol (1984) 111:229–40.[Web of Science][Medline]
Pouget A, Driver J. Relating unilateral neglect to the neural coding of space. Curr Opin Neurobiol (2000) 10:242–49.[CrossRef][Web of Science][Medline]
Savazzi S, Posteraro L, Veronesi G, Mancini F. Rightward and leftward bisection biases in spatial neglect: two sides of the same coin? Brain (2007) 130:2070–84.
Trauzettel-Klosinsky S, Reinhard J. The vertical field border in hemianopia and its significance for fixation and reading. Invest Ophthalmol Vis Sci (1998) 39:2177–86.
Zihl J. Eye movement patterns in hemianopic dyslexia. Brain (1995) 118:891–912.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||