Kiichiro Morita, MD, Cognitive and Molecular Research Institute of Brain Diseases, Kurume University, 67, Asahi-machi, Kurume 830-0011, Japan. Email: firstname.lastname@example.org
Abstract To characterize the left and right scanning function and the effect of affection in schizophrenia patients exploratory eye movements as biologic markers were recorded in 44 schizophrenia patients and 72 age-matched healthy controls. The total eye scanning length (TESL) and total number of gaze points (TNGP) in the left and right visual fields were calculated as subjects viewed neutral or affectively charged pictures. TESL of patients was shorter than that of controls when viewing pictures of smiling babies and open circles. TESL of patients was shorter for smiling faces than for crying babies, but TESL of controls was longer for smiling faces than for crying babies. Left TNGP for smiling faces and circles was lower in patients than in controls. In patients, left TNGP for crying babies was higher than for either smiling babies or circles. In controls, left TNGP for smiling babies was higher than crying babies. In patients, left TNGP for smiling babies and circles was smaller than the right TNGP. In controls, left TNGP was larger for smiling than for crying babies. When viewing smiling babies, both TESL and TNGP were negatively correlated with negative symptom scores in patients. Patients' eye movements in the left visual field were clearly different from controls', suggesting that visual cognitive function is impaired in schizophrenia patients. Exploratory eye movements are a useful marker of visual cognitive function, and are a useful tool to evaluate the influence of affection in schizophrenia patients.
Exploratory eye movements have been reported as biologic state or trait markers in patients with some psychiatric disorders, including schizophrenia.1–3 Subjects are examined with an eye-mark recorder as they view several pictures projected on a screen in front. Parameters of exploratory eye movement, including the eye scanning length, number of gaze points, and mean gaze time, have been analyzed to evaluate visual cognitive function.1–5 This method is believed to reflect abnormalities of eye movement and visual information processing in natural settings, and is a useful psychophysiologic marker of visual cognitive function.1–4 We previously suggested that exploratory eye movements in schizophrenia patients are more limited, with a longer maintenance of gaze, than in healthy controls when viewing simple figures such as a circle, being significant on comparison of task performance.6,7 Schizophrenia patients may gaze at a certain part of a picture aimlessly, rather than appropriately focusing their attention on a part of the picture.6,7 This finding suggests a possible impairment of visual information processing in schizophrenia patients.2,3,6 Phillips and David reported that schizophrenia patients had dysfunction of attention to the left field of stimuli, and indicated that the left specific eye scanning may be a state marker of attention processes in schizophrenia patients.8 One reason why such an analysis is important is because it can be used to investigate and clarify neurophysiologic data in order to diagnose schizophrenia and identify differences in brain function. Schizophrenia patients have also been found to perform less well in mental tests than controls, as well as in scanning the visual field.8
Affection is a critical part of human experience: negative affection reduces the quality of life while positive affection makes life feel worthwhile. The effect of affective stimuli has been shown to influence event-related potentials, especially P300 components9–11 and eye movements.12,13 Loughland et al. observed that schizophrenia patients have a deficit in positive emotion perception rather than negative perception, and suggested that this may reflect a failure to integrate visual information.13 However, the effect of affection on exploratory eye movements is not well known.
The present study compared exploratory eye movements in healthy controls and schizophrenia patients using an eye-mark recorder to determine whether or not there are any differences in left and right eye scanning on stimuli presentation. The second aim was to evaluate which stimuli or parts of stimuli elicited the greatest differences.
Seventy-two healthy paid volunteers (range, 18–64 years; mean age ± SD, 36.9 ± 13.4 years; male, n = 33; female, n = 39), and 44 schizophrenia patients diagnosed according to ICD-10 by two psychiatrists (male, n = 29; female, n = 15; age range, 18–62 years; mean age ± SD, 35.8 ± 14.1 years; 28 paranoid type; 16 non-paranoid) were enrolled (Table 1). All healthy controls and patients were right-handed, with normal visual and auditory functions. All controls had no family history and no history of psychiatric or neurologic diseases or drug addiction. All subjects gave written informed consent for study participation. The Ethics Committee of Kurume University approved the present study.
Table 1. Patient profile (n = 44)
Onset (years of age)
24.5 ± 5.6
Duration of illness (years)
11.5 ± 10.5
No. admissions (n)
1.1 ± 1.1
Duration of education (years)
12.7 ± 2.1
Eye mark recording
Eye movements were recorded using an eye-mark recorder (NAC, EMR-8, Tokyo, Japan) that consisted of two video cameras (left and right eye-mark-shooting units) fixed to the left and right sides of a headband and another camera (field-shooting unit) fixed to the top of a helmet. Infrared light (850 nm) sources were positioned in front of each lower eyelid. The side cameras recorded the infrared light reflected from the cornea of the eye. The camera on the top of the cap recorded the pictures shown on the screen. After a camera controller superimposed these recordings with a 0.01-s electronic timer, the combined recording was saved on videotape. Movement >1° with duration >100 ms was scored as an eye movement. Obayashi et al. reported that 100 ms was enough to evaluate visual stimuli and there was no significant difference between 100 ms and 200 ms.14 Furthermore, there was no significant difference between 100 ms and 200 ms in the present paper, and differences were obtained in responsive search score (RSS)1 only for 100 ms (Ishii et al., unpubl. obs. 2007). This technique enabled us to determine eye fixation points. Recorded data were assessed using a computer analysis system. In the present study exploratory eye movements were analyzed for two parameters: total eye scanning length for gazing points (TESL) and the total number of gazing points (TNGP), as reported previously.1–5 In the present study we subdivided the TNGP into the left and right halves of the visual field on the screen (left TNGP and right TNGP). Eye scanning length was calculated as the distance between two eye gazing points, as reported previously.1–5
Eye movement recording procedure
In a darkened room, where visual and auditory sensory stimuli were attenuated, eye movement was recorded using the eye-mark recorder. Subjects were instructed to identify each picture exactly as it was presented. All subjects were also instructed to fix their gaze at several corner points to check their eye movements to evaluate neurological deficits and low-level eye movement as a reflex organic saccade. No subjects had deficits under the present procedure. Exploratory eye movements and fixation points were recorded during viewing. The pictures were projected onto a screen to form images 120 cm wide and 90 cm high. Maximum angles of sight lines were 40° horizontally and 24° vertically. Each block consisted of a series of three pictures, each presented for 15 s.4–6 We used two symmetrical babies' faces to expand the eye scanning range and increase gaze points, and also to evaluate the differences between left and right scanning of the screen. A control study was performed using 100 healthy subjects; all subjects accordingly recognized affection and did not identify the gender difference. Furthermore, adult faces generated strong and sustained affective stimuli for patients and sometimes did not yield a fixation point. In the present study, all subjects accordingly recognized affection.
Three kinds of picture were used (Fig. 1): picture 1, two symmetrical smiling babies' faces with accompanying laughing sounds (70 dB, sound pressure level [SPL]) to study the influence of any possible affection; picture 2, two symmetrical simple circles as an index of motivation and compliance; and picture 3, two symmetrical crying babies' faces with crying sounds (70 dB, SPL) to study the influence of negative affection.
Laughing or crying sounds were used to increase the effects of affection and the subjects' arousal level. Indeed, the reaction (eye closing) time was significantly shortened when viewing babies' faces with laughing or crying sounds than without sounds in healthy controls (data not shown). Thus, in the present study we used laughing or crying sounds to clarify in detail the effect of emotion on exploratory eye movements.
The crying babies and smiling babies' faces were equally presented in different order. Recording was performed as follows using two kinds of babies' faces and simple circles.
Session 1: all subjects were instructed, “look at two symmetrical pictures in front of you in whatever way you want” (free task). Session 2: all subjects were instructed, “Look at the picture in front of you carefully and memorize it. Immediately after recording, I will ask you to describe the picture you saw” (memorization task). Session 3: all subjects were asked, “Are there any differences between the picture you are seeing now and the one you saw previously” to test confidence and attention (comparison task).
Data from the left and right eyes were analyzed and found to be similar, although the scanning was different in the left and right fields of the screen.
Clinical evaluation and medication
The clinical state of all patients was assessed using the Positive and Negative Symptom Scale (PANSS),15 which was administered individually by two well-trained psychiatrists within 1 week after eye movement recording. The positive symptom score was 25.4 ± 6.6 and the negative score was 21.3 ± 5.4. Higher scores of PANSS from two psychiatrists were incorporated into the present analysis as data. All patients were treated with neuroleptics with the mean daily dose (mg/day) of chlorpromazine equivalent being 338.8 ± 148.3.16
In the present study we used only data obtained from the comparison task (session 3) because the comparison task best reflects visual cognitive function.3–5 Two-way analysis of variance (anova; groups: patients or controls; stimuli: cry, smile, circle) for the main group effect was performed when we analyzed TESL and TNGP. Whether or not interactions were obtained, one-way anova was performed for each group (patients or controls) and stimuli (smiling babies, crying babies, or circles). When we analyzed the left and right TNGP, two-way anova (groups, stimuli) was performed for each side. Whether or not interactions were obtained, one-way anova was performed for each group for the main effect. Post-hoc analyses were conducted using Scheffe tests. The Pearson correlation coefficient was used to identify significant relationships between symptom scores, measures of eye movements, and dose of medication. P < 0.05 was accepted as significant.
Representative sequences of exploratory eye movements from a schizophrenia patient and one healthy control when viewing three pictures are shown in Fig. 1. The control eye movements were coherent and focused on the baby's eyes and mouth in both the left and right visual fields. The patient's eye movements appeared to be random and did not seem to reflect the presence of any organized strategy: they did not focus on the left and right baby's eyes and mouth.
Total eye scanning length
The TESL for patients was significantly shorter than for controls on two-way anova (F(1672) = 146.6, P < 0.001; Table 2; Fig. 2). There were interactions between the groups and stimuli. TESL in the two groups differed when viewing smiling babies (F(1224) = 84.6, P < 0.001) and circles (F(1224) = 117.3, P < 0.001), but not crying babies (F(1224) = 2.3, P = 0.13). TESL in controls was different for the three stimuli (F(2417) = 10.3, P < 0.001). TESL when viewing smiling babies was longer than for both crying babies (P < 0.001) and circles (P < 0.001). TESL in patients was also different for the three stimuli (F(2255) = 25.2, P < 0.001). TESL when viewing crying babies was longer than that for both smiling babies (P < 0.01) and circles (P < 0.001). In patients, TESL for circles was the shortest among the three stimuli.
Table 2. Measures of eye movements (mean ± SD)
TESL, total eye scanning length; TNGP, total no. gaze points.
585.6 ± 133.9
28.9 ± 5.1
501.3 ± 175.4
25.8 ± 6.1
526.1 ± 166.1
27.0 ± 6.0
389.6 ± 185.4
23.8 ± 5.4
467.9 ± 176.6
25.3 ± 6.0
276.4 ± 171.6
20.4 ± 6.4
Total number of gaze points
The TNGP for patients was significantly shorter than for controls on two-way anova (F(1672) = 25.5, P < 0.001; Table 2; Fig. 3). There were interactions between the groups and stimuli. TNGP differed for controls and patients when viewing smiling babies (F(1226) = 50.1, P < 0.001) and circles (F(1226) = 62.7, P < 0.001), but not crying babies (F(1226) = 2.21, P = 0.108). TNGP for controls was different for the three stimuli (F(2423) = 9.9, P < 0.001). TNGP when viewing smiling babies was greater than for both crying babies (P < 0.001) and circles (P < 0.001). TNGP for patients was also different for the three stimuli (F(2225) = 15.2, P < 0.001). TNGP when viewing circles was lower than for both smiling (P < 0.001) and crying (P < 0.001) babies.
Left and right visual fields of the screen
The left TNGP for patients was significantly smaller than for controls on two-way anova (F(1336) = 37.8, P < 0.001; Fig 4). There were interactions between the groups and stimuli. The left TNGP was different in controls and patients when viewing the smiling babies (F(1112) = 37.4, P < 0.01) and circles (F(1112) = 25.3, P < 0.01) but not crying babies (F(1112) = 2.3, P = 0.100). The left TNGP in patients for both smiling babies (P < 0.001) and circles (P < 0.0019 was smaller than in controls. No intergroup difference was observed in the right TNGP.
In controls, there were no differences between the left and right TNGP on two-way anova. The left TNGP was greater when viewing smiling babies than when viewing crying babies (P < 0.01). There was no significant difference for stimuli observed in the right TNGP.
Patients also showed differences between the three stimuli (F = 9.5, P < 0.001). The left TNGP was greater when viewing crying than when viewing smiling babies (P < 0.01) or circles (P < 0.001). The left TNGP in patients when viewing circles was the lowest value recoded. The right TNGP was similar for the three stimuli.
In patients there were significant differences between the left and right TNGP on two-way anova (F(1252) = 20.6, P < 0.001). The left TNGP was significantly lower than the right TNGP (P < 0.001). There were significant interactions between side and stimuli (P < 0.01). The left TNGP was significantly lower than the right TNGP when viewing smiling babies (F(1112) = 20.0, P < 0.001) and circles (F(1112) = 13.0, P < 0.001) but not when viewing crying babies (F(1112) = 2.6, P = 0.102).
Symptom scores and eye measures
The positive symptom score was 25.4 ± 6.6 and the negative symptom score was 21.3 ± 5.4. A significant correlation was obtained between TESL and negative symptom scores (r = −0.536, P < 0.001) when viewing smiling babies and circles (r = −0.568, P < 0.0019), and TNGP and negative symptom scores (r = 484, P < 0.001) when viewing smiling babies, as seen in Table 1. There were no significant correlations between eye measures and the positive symptom scores excepting the TESL when viewing circles (r = −0.328, P < 0.05) as shown in Table 3. There was also no significant correlation between the dose of antipsychotic and symptom scores.
Table 3. Relationship between eye measures and symptom scores
TESL, total eye scanning length; TNGP, total no. gaze points.
−0.536 (P < 0.001)
−0.484 (P < 0.01)
−0.568 (P < 0.001)
−0.328 (P < 0.05)
The most important finding of the present study was that exploratory eye movements clearly differed between schizophrenia patients and healthy controls, and the difference is evident when viewing simple circles and smiling babies only in the left field of the screen; this suggests that exploratory eye movement is a useful biologic marker for clinical evaluation. Kojima et al. investigated exploratory eye movements using specially constructed S-shaped geometric figures, demonstrating that exploratory eye movement was a useful biologic marker of visual cognitive function in patients with psychiatric disorders such as schizophrenia.1 The authors reported that exploratory eye movement might be a trait marker for schizophrenia patients. Streit et al. also reported that the visual scanning path may be a trait marker for facial affect recognition.12 In the present study we found that the exploratory eye movements of schizophrenia patients had properties similar to those reported previously.1–3,6 The present study also examined the effects of affective stimuli in different scanning fields in the left or right fields of the screen. Specific aspects are discussed here.
Total eye scanning length
According to one-way anova (group) under the confidence condition, TESL for patients was significantly shorter than for controls when viewing the smiling babies and simple circles, but TESL for patients when viewing the crying babies was not significantly different from that for controls. TESL when viewing the crying babies was longest among the three stimuli in patients, followed by smiling babies and then simple circles, while TESL when viewing the smiling babies was longest among the three stimuli for controls, followed by simple circles and then crying babies. Because significant differences in TESL were seen when viewing the smiling babies or simple circles, but not the crying babies between the two groups, the findings suggest that smiling babies or simple visual targets may be particularly useful for evaluating exploratory eye movement studies according to the clinical application, as reported previously.6,7 The findings might reflect a method of visual response assessment, but one should consider that no significant changes between patients and controls were seen when viewing the crying babies. Loughland et al. reported that schizophrenia patients had an impairment of scanning, and this was evident for happy and neutral faces rather than sad faces.13 Those authors concluded that the impairment of scanning in schizophrenia patients might reflect a failure to integrate facial stimuli due to dysfunction of the synchronization between local and global processing, especially when viewing positive (happy) or neutral faces.13 Thus, the present findings of significant differences between the positive (smiling babies) or neutral (circle) and negative (crying babies) visual stimuli may be due to differences in the level of attention given to facial features or figures; patients may pay less attention to smiling babies or simple circles.
Total number of gaze points
The TNGP in patients was significantly smaller than that in controls when viewing smiling babies or simple circles, but not when viewing crying babies, similarly to TESL. In patients, TNGP for crying or smiling babies was significantly larger than for simple circles. In controls, TNGP for smiling babies was significantly larger than for crying babies and simple circles. Both TNGP in patients for smiling babies and simple circles were significantly different from controls, similarly to TESL. These findings also indicate the impairment of visual information processing.
Patients may pay less attention to smiling babies, similarly to the simple circles, as aforementioned. Differences only in the left field of the screen strongly indicate major set impairment in schizophrenia patients. Indeed, Phillips and David reported that schizophrenia patients had difficulty in redirecting initial focus of attention to the left field of stimuli, and proposed that the left specific scan paths are markers of attention processes in schizophrenia patients.8 The right brain has been considered to be responsible for emotional functions, and the left brain symbolic functions.17–19 In the present study the difference of TNGP for affective stimuli is obvious only in the left field of the screen when viewing smiling babies or simple circles. However, this is not observed when viewing crying babies. These results suggest that patients might be influenced differently by the smiling face or simple circle compared to controls. The impairment of scanning in the schismatic decussating of optic nerve fibers from the retina in humans ensures that each primary visual cortex receives input exclusively from the contralateral visual field. Visual field asymmetries have been reported particularly favoring the left visual field.
Schwartz et al. reported that for emotion-linked images, the eyes tended to move leftward, in contrast to rightward for other images; they concluded that this supported links between the right brain and emotional function.18 Furthermore, Schwartz et al. reported that right-handed subjects tend to look to the left when answering affective questions, and concluded that the right hemisphere has a special role in emotion in the intact brain.18 We showed that TNGP was smaller when viewing smiling babies as well as simple circles only in the left field. Thus, the present observations, showing that the effects of affection are significant only on the left field of the screen, may be useful for evaluating the hemispheric function of the brain. Negative affective stimuli were symmetric between the left and right brain.
Symptom and eye movements
Negative psychiatric symptoms have been reported to correlate negatively with exploratory eye movements.6,7 A significant negative correlation (r < −0.4, P < 0.01) between TESL and negative symptom scores was observed in the present study. This may strongly suggest that exploratory eye movements are useful for clinical application and reflect negative symptoms. Correlations between scanning measures and symptoms showed that negative symptoms were related to scanning measures such as fixation numbers and the total scan path length of the face.8,20 The authors suggest that visual organization impairments may be related to cognitive inflexibility and frontal dysfunction.
The left visual field of the screen when viewing smiling babies was the most important for characterizing facial recognition in schizophrenia patients. This may result from dysfunction of the scanning property whereby human first scan the left field. This dysfunction may be due to impairment of the attention set to event processing, especially concerning positive facial recognition. Interestingly, the amygdale shows increased activation for viewing of smiling babies with a laughing sound, and the eye closing time is prolonged for viewing of smiling babies with a laughing sound only in patients with schizophrenia (Kurakake et al., unpubl. obs, 2007). Thus, we recorded exploratory eye movements with smiling or crying sounds to compare with other unpublished evidence. The present results indicate that schizophrenia patients may have dysfunction of the right frontal cortex, especially under pleasure conditions. This impairment may lead to difficulties in human social relationships. Rehabilitation concerned with pleasure should be the most important for schizophrenia patients for social rehabilitation.
Finally, specific differences were found for exploratory eye movements while viewing certain pictures between schizophrenia patients and healthy controls in the left visual field. Thus, exploratory eye movements appear to include clinical and hemispheric functional markers, suggesting their usefulness in exploring human visual cognition.