TY - JOUR
AU - Javitt, Daniel, C
AB - Abstract Schizophrenia (Sz) is associated with deficits in fluent reading ability that compromise functional outcomes. Here, we utilize a combined eye-tracking, neurophysiological, and computational modeling approach to analyze underlying visual and oculomotor processes. Subjects included 26 Sz patients (SzP) and 26 healthy controls. Eye-tracking and electroencephalography data were acquired continuously during the reading of passages from the Gray Oral Reading Tests reading battery, permitting between-group evaluation of both oculomotor activity and fixation-related potentials (FRP). Schizophrenia patients showed a marked increase in time required per word (d = 1.3, P < .0001), reflecting both a moderate increase in fixation duration (d = .7, P = .026) and a large increase in the total saccade number (d = 1.6, P < .0001). Simulation models that incorporated alterations in both lower-level visual and oculomotor function as well as higher-level lexical processing performed better than models that assumed either deficit-type alone. In neurophysiological analyses, amplitude of the fixation-related P1 potential (P1f) was significantly reduced in SzP (d = .66, P = .013), reflecting reduced phase reset of ongoing theta-alpha band activity (d = .74, P = .019). In turn, P1f deficits significantly predicted increased saccade number both across groups (P = .017) and within SzP alone (P = .042). Computational and neurophysiological methods provide increasingly important approaches for investigating sensory contributions to impaired cognition during naturalistic processing in Sz. Here, we demonstrate deficits in reading rate that reflect both sensory/oculomotor- and semantic-level impairments and that manifest, respectively, as alterations in saccade number and fixation duration. Impaired P1f generation reflects impaired fixation-related reset of ongoing brain rhythms and suggests inefficient information processing within the early visual system as a basis for oculomotor dyscontrol during fluent reading in Sz. fixation-related potential, magnocellular, visual, oculomotor, lexical, word skipping, reading, saccade Introduction Schizophrenia patients (SzP) show severe deficits in reading ability that contribute to impaired functional outcome.1–5 As opposed to individuals with developmental dyslexia, SzP show intact single-word recognition scores suggesting relatively normal reading development,3,5 but a decline in fluent reading ability during the years immediately surrounding illness, reflecting an acquired dyslexia.3,5–8 Schizophrenia patients most commonly show a “dual dyslexia” type pattern of reading impairment reflecting both auditory (phonological)3,8 and visual sensory-level impairment.2–5,9–11 Auditory level deficits are reflected in impaired tonal discrimination and mismatch negativity (MMN) generation.3,8 Visual sensory deficits include impaired contrast gain within the subcortical visual magnocellular pathway2,12 that affect both central and parafoveal processing13 and contribute to impaired oculomotor control.4,5,9–11 In addition, more general cognitive impairments may contribute. Here, we utilized a convergent eye-tracking, computational, and neurophysiological approach to evaluate the neural mechanisms or processes underlying visual/oculomotor aspects of reading impairment in schizophrenia (Sz). Fluent reading requires a sequence of saccades and fixations (figure 1) that can be evaluated to determine the basis for prolonged reading times.4,9 Determinants of reading rate include the number and pattern of saccades and the duration of the intervening fixations. Additional key parameters include the degree to which specific words are skipped (“skipping rates”), as well as rates of regressive (leftward) saccades to words that were previously fixated.14–19 In general, as reading difficulty increases, parallel increases are observed in fixation duration and saccade number, while skipping and saccade amplitude typically decrease.18,19 The relative change in these values can, therefore, be used to infer underlying mechanisms or processes. Fig. 1. Open in new tabDownload slide (A) Representative pattern of saccade and fixations (scanpath) from a healthy control (HC) (left) and a schizophrenia patient (SzP) (right) reading the same paragraph. (B) Patterns of fixation (“heatmap”) during reading representing the absolute number of fixations at each location of a representative text. (C) Normalized heatmaps (percentage fixations) to show relative patterns across groups. Fig. 1. Open in new tabDownload slide (A) Representative pattern of saccade and fixations (scanpath) from a healthy control (HC) (left) and a schizophrenia patient (SzP) (right) reading the same paragraph. (B) Patterns of fixation (“heatmap”) during reading representing the absolute number of fixations at each location of a representative text. (C) Normalized heatmaps (percentage fixations) to show relative patterns across groups. Patterns of alteration can be modeled using computational simulation programs such as E-Z Reader14–17,20,21 that assess the relative contributions of higher- (eg, lexical and postlexical processing, attention) and lower- (eg, visual/oculomotor function) level factors. For example, during development alterations in reading-related fixation patterns can be modeled simply by altering a specific model parameter—termed α 1—that reflects the speed of lexical processing.15 Other relevant model parameters include ε, which reflects the degree to which visual acuity decrements in the parafovea affect reading fluency, and λ, which reflects the degree to which refixation probability increases when the initial fixation location is outside of the center of a word.15 Neurophysiological approaches may be used to investigate underlying neural mechanisms. Reductions in posterior alpha activity during reading reflect the degree to which occipital visual regions are brought “on line” during reading.22 In parallel, fixation-related potentials (FRP) such as the fixation-P1 (P1f) or “lambda” potential23–29 reflect the efficiency with which information is processed during each fixation.30 FRPs are increasingly being used to investigate the effects of attention, load, and parafoveal preview during normal reading.23–28 As previously, we evaluated magnocellular function using contrast sensitivity to low spatial frequency stimuli12,31 and general cognitive ability using Wechsler Adult Intelligence Scale-III processing speed index (PSI). We hypothesized that SzP would show deficits in reading rate consistent with prior studies and that these would be associated with impairments in lower-, as well as higher-, level functions. Materials and Methods Participants Twenty-six SzP meeting DSM-IV criteria for Sz/schizoaffective disorder (22 males) and 26 healthy controls (HCs) (15 males) participated in this study. Subjects were recruited from the Nathan Kline Institute and affiliated clinics, and a signed informed consent was obtained following a full description of study procedures. The study was approved by the Institutional Review Board associated with the New York State Office of Mental Health. Subjects did not differ in age or parental socioeconomic status (SES)32 (table 1). All subjects had corrected 20/32 vision or better on the Logarithmic Visual Acuity Chart (Precision Vision), and there was no difference in handedness, as assessed by the Edinburgh Handedness Inventory.33 Schizophrenia patients had an average illness duration of 13.3 ± 8.1 years and were receiving antipsychotic medication (average 840.7 ± 805.4 chlorpromazine-equivalents/day). Symptoms were assessed using Positive and Negative Syndrome Scale (PANSS)34 with factor scores of Positive: 18.8 ± 5.6, Negative: 19.7 ± 5.0, and Global Psychopathology: 38.0 ± 8.1. Table 1. Mean (SD) Demographic, Psychophysiological, and Reading/Eye-Tracking Parameters Measure . HC (n = 26) . SzP (n = 26) . F . P . d . Demographics and psychophysiology measures  Age 37.1 (9.5) 35.5 (9.1) 0.41 0.5 0.2  Parental SES 42.8 (15.0) 38.6 (11.9) 1.04 0.3 0.1  Individual SES 44.0 (11.6) 24.5 (7.7) 47.75 <.0001 2.0  Years education 15.3 (2.1) 12.0 (2.3) 28.20 <.0001 1.5  PSI (WAIS) 105.5 (11.2) 89.0 (13.7) 21.44 <.0001 1.3  Contrast sensitivity (1 cpd) 188.0 (62.8) 142.5 (42.3) 9.12 .004 0.9 Reading and eye-tracking parameters  Reading rate (word/min) 233.7 (84.0) 162.3 (68.2) 11.1 .002 1.0  Time per word (msec/word) 270.3 (95.1) 445.4 (171.3) 20.16 <.0001 1.3  Fixation duration (msec) 249 (48) 287 (67) 5.24 .026 0.7  Saccades per word .81 (.26) 1.27 (.31) 31.81 <.0001 1.6  Mean regressive saccades 114.9 (65.8) 243.0 (113.7) 23.0 <.0001 1.4  Saccade amplitude (°) 2.57 (.71) 2.21 (.74) 3.13 .08 0.5  Regressive saccade amplitude (°) 2.16 (.75) 2.42 (.71) 1.50 .22 .22  Overall refixation probability .22 (.10) .37 (.11) 26.52 <.0001 1.4  Overall skipping probability .38 (.12) .25 (.10) 15.05 <.0001 1.2  First-pass refixation probability .10 (.05) .15 (.07) 7.67 < .01 0.8  First-pass skipping probability .48 (.11) .40 (.11) 7.51 <.01 0.7 Measure . HC (n = 26) . SzP (n = 26) . F . P . d . Demographics and psychophysiology measures  Age 37.1 (9.5) 35.5 (9.1) 0.41 0.5 0.2  Parental SES 42.8 (15.0) 38.6 (11.9) 1.04 0.3 0.1  Individual SES 44.0 (11.6) 24.5 (7.7) 47.75 <.0001 2.0  Years education 15.3 (2.1) 12.0 (2.3) 28.20 <.0001 1.5  PSI (WAIS) 105.5 (11.2) 89.0 (13.7) 21.44 <.0001 1.3  Contrast sensitivity (1 cpd) 188.0 (62.8) 142.5 (42.3) 9.12 .004 0.9 Reading and eye-tracking parameters  Reading rate (word/min) 233.7 (84.0) 162.3 (68.2) 11.1 .002 1.0  Time per word (msec/word) 270.3 (95.1) 445.4 (171.3) 20.16 <.0001 1.3  Fixation duration (msec) 249 (48) 287 (67) 5.24 .026 0.7  Saccades per word .81 (.26) 1.27 (.31) 31.81 <.0001 1.6  Mean regressive saccades 114.9 (65.8) 243.0 (113.7) 23.0 <.0001 1.4  Saccade amplitude (°) 2.57 (.71) 2.21 (.74) 3.13 .08 0.5  Regressive saccade amplitude (°) 2.16 (.75) 2.42 (.71) 1.50 .22 .22  Overall refixation probability .22 (.10) .37 (.11) 26.52 <.0001 1.4  Overall skipping probability .38 (.12) .25 (.10) 15.05 <.0001 1.2  First-pass refixation probability .10 (.05) .15 (.07) 7.67 < .01 0.8  First-pass skipping probability .48 (.11) .40 (.11) 7.51 <.01 0.7 Note: HC, healthy controls; SzP, schizophrenia patient; SES, socioeconomic status; PSI, processing speed index. The bolded ones are the ones that achieved statistical significance, per their P-values. Open in new tab Table 1. Mean (SD) Demographic, Psychophysiological, and Reading/Eye-Tracking Parameters Measure . HC (n = 26) . SzP (n = 26) . F . P . d . Demographics and psychophysiology measures  Age 37.1 (9.5) 35.5 (9.1) 0.41 0.5 0.2  Parental SES 42.8 (15.0) 38.6 (11.9) 1.04 0.3 0.1  Individual SES 44.0 (11.6) 24.5 (7.7) 47.75 <.0001 2.0  Years education 15.3 (2.1) 12.0 (2.3) 28.20 <.0001 1.5  PSI (WAIS) 105.5 (11.2) 89.0 (13.7) 21.44 <.0001 1.3  Contrast sensitivity (1 cpd) 188.0 (62.8) 142.5 (42.3) 9.12 .004 0.9 Reading and eye-tracking parameters  Reading rate (word/min) 233.7 (84.0) 162.3 (68.2) 11.1 .002 1.0  Time per word (msec/word) 270.3 (95.1) 445.4 (171.3) 20.16 <.0001 1.3  Fixation duration (msec) 249 (48) 287 (67) 5.24 .026 0.7  Saccades per word .81 (.26) 1.27 (.31) 31.81 <.0001 1.6  Mean regressive saccades 114.9 (65.8) 243.0 (113.7) 23.0 <.0001 1.4  Saccade amplitude (°) 2.57 (.71) 2.21 (.74) 3.13 .08 0.5  Regressive saccade amplitude (°) 2.16 (.75) 2.42 (.71) 1.50 .22 .22  Overall refixation probability .22 (.10) .37 (.11) 26.52 <.0001 1.4  Overall skipping probability .38 (.12) .25 (.10) 15.05 <.0001 1.2  First-pass refixation probability .10 (.05) .15 (.07) 7.67 < .01 0.8  First-pass skipping probability .48 (.11) .40 (.11) 7.51 <.01 0.7 Measure . HC (n = 26) . SzP (n = 26) . F . P . d . Demographics and psychophysiology measures  Age 37.1 (9.5) 35.5 (9.1) 0.41 0.5 0.2  Parental SES 42.8 (15.0) 38.6 (11.9) 1.04 0.3 0.1  Individual SES 44.0 (11.6) 24.5 (7.7) 47.75 <.0001 2.0  Years education 15.3 (2.1) 12.0 (2.3) 28.20 <.0001 1.5  PSI (WAIS) 105.5 (11.2) 89.0 (13.7) 21.44 <.0001 1.3  Contrast sensitivity (1 cpd) 188.0 (62.8) 142.5 (42.3) 9.12 .004 0.9 Reading and eye-tracking parameters  Reading rate (word/min) 233.7 (84.0) 162.3 (68.2) 11.1 .002 1.0  Time per word (msec/word) 270.3 (95.1) 445.4 (171.3) 20.16 <.0001 1.3  Fixation duration (msec) 249 (48) 287 (67) 5.24 .026 0.7  Saccades per word .81 (.26) 1.27 (.31) 31.81 <.0001 1.6  Mean regressive saccades 114.9 (65.8) 243.0 (113.7) 23.0 <.0001 1.4  Saccade amplitude (°) 2.57 (.71) 2.21 (.74) 3.13 .08 0.5  Regressive saccade amplitude (°) 2.16 (.75) 2.42 (.71) 1.50 .22 .22  Overall refixation probability .22 (.10) .37 (.11) 26.52 <.0001 1.4  Overall skipping probability .38 (.12) .25 (.10) 15.05 <.0001 1.2  First-pass refixation probability .10 (.05) .15 (.07) 7.67 < .01 0.8  First-pass skipping probability .48 (.11) .40 (.11) 7.51 <.01 0.7 Note: HC, healthy controls; SzP, schizophrenia patient; SES, socioeconomic status; PSI, processing speed index. The bolded ones are the ones that achieved statistical significance, per their P-values. Open in new tab Task Subjects were instructed to silently read 8 one-paragraph passages taken from the GORT-4,35 grade levels 5–8, presented on a computer monitor. Eight passages were presented in 4 different line spacing conditions (supplementary figures 1 and 2). The reading rate (msec per word) was calculated based on the number of words in each passage. Visual contrast sensitivity values were published previously2,12,31 (supplementary Methods). PSI was assessed using standard administration. Eye Tracking and EEG Recording Eye movements and EEG were recorded and analyzed off-line using standard approaches. Independent components analysis (ICA) was used to remove eyeblinks and other motor artifacts, as well as horizontal eye movements (supplementary figure 3). FRP were computed from −1000 to 2000 ms relative to fixation onset. Trials with artifacts remaining after ICA were rejected for the period of 0–500 ms. Positive peak amplitudes (P1f) were obtained over occipital electrodes (9R, 9z, 9L) for the 75–125-ms time range relative to the immediate (1–10 ms) post-fixation interval to minimize contributions from prior saccades. ERP spectral decomposition (time-frequency [TF]) analyses were implemented using the complex demodulation procedure implemented in Brain Electrical Source Analysis (BESA) (version 5.1, MEGIS Software GmbH) (see supplementary Methods). Computational Modeling Computational modeling was performed using E-Z Reader 10.2, which is implemented in Java.14,21 The model’s default parameters and assumptions were used unless otherwise stated. Word-based analyses of refixation and skipping probabilities were calculated for both overall and for first-pass fixations (see table 1; figure 2). These analyses were restricted to the widest spacing condition to minimize the impact of any vertical calibration error (see supplementary Methods). Fig. 2. Open in new tabDownload slide Comparison of empirical and E-Z Reader modeled data. (A and B) Empirical results for the overall skipping and refixation probabilities (left) and fixation durations (right); (C–H) Modeled results for slow lexical processing (α1 = 144), impaired visual/parafoveal processing (ε = 1.3; λ = .35), or both (“2-hit”) (see supplementary table 2 for full model parameters). **P < .01; ***P < .001. Fig. 2. Open in new tabDownload slide Comparison of empirical and E-Z Reader modeled data. (A and B) Empirical results for the overall skipping and refixation probabilities (left) and fixation durations (right); (C–H) Modeled results for slow lexical processing (α1 = 144), impaired visual/parafoveal processing (ε = 1.3; λ = .35), or both (“2-hit”) (see supplementary table 2 for full model parameters). **P < .01; ***P < .001. Similar to other E-Z Reader simulations,14 the simulated means were based on the stimuli from the Schilling et al36 corpus. We averaged the results over 10 000 simulated subjects to obtain stable results. Statistical Analysis Primary analyses were performed using between-group ANOVAs. Relationship of between-group deficits to potential predictors (eg, contrast sensitivity, P1f, PSI) were assessed initially with ANCOVA to verify the homogeneity of slope across groups, followed by either within-group or between-group (covaried for group) analyses as appropriate. Relationships among variables were assessed using simultaneous linear regression. For E-Z Reader, comparison between models was assessed by evaluation relative to empirical values and χ 2 analysis of Root-mean-square deviation values. Effect sizes (d) were interpreted in accordance with standard conventions.37 All tests were 2-tailed with a preset alpha level of significance of P < .05. Results Eye-movement and EEG data were obtained from 26 HC and 26 SzP. EEG data from 3 SzP and 3 HC were excluded due to excessive muscular and/or movement noise (see supplementary Methods). Behavior Reading Rate. As expected, there was a highly significant between-group difference in reading rate across all reading levels and line spacings (F1,50 = 21.8, P < .0001, d = 1.3) (supplementary figures 4A and 4B). Schizophrenia patients required significantly more saccade-fixation sequences to completely scan the passage (table 1; figure 1 B). Group differences in the number of saccades remained strongly significant even following control for PSI (F1,44 = 10.9, P = .002; residual d = 1.0). Fixation duration was also significantly increased in SzP, but the difference was no longer significant following control for PSI (F1,47 = .5, P = .47). The mean saccade amplitude and distribution were similar across groups (see table 1; supplementary figure 5). The spatial distribution of fixation locations for each passage was highly similar for SzP and HC (r = .85, P < .0001, figure 1C; supplementary figure 6). In a simultaneous regression analysis, saccade number (β = .66, P < .0001) and fixation duration (β = .51, P < .0001) independently contributed to reading rate across groups. Once these values were taken into account, the main effect of group was no longer significant (F1,44 = .01, P = .91). E-Z Reader Simulations Simulations assessed the degree to which manipulation of higher- vs lower-level parameters vs both in the E-Z Reader model reproduced the pattern of alterations in word-based eye-tracking measures14–17,21 in Sz. These included the α 1 parameter, which models high-level lexical processing speed, and the ε and λ parameters, which model the effects of visual acuity and saccadic programming, respectively.20 Manipulation of either set of factors alone to the Sz dataset significantly improved model fit relative to the default model (both χ 2 > 20, P < .0001) but did not fully capture the large reductions in skipping and increase in refixation probability vs smaller changes in first-fixation durations (figures 3A–C). A 2-hit model incorporating alterations in both high (α 1) and low (ε,λ) parameters uniquely captured the increase in refixation probability along with other gaze-related abnormalities and fit the data significantly better (P < .01) than the 1-hit model that incorporated only high-level (α 1 = 144) deficits (χ 2 = 10.4, P < .01) (figure 3D, supplementary table 2). Fig. 3. Open in new tabDownload slide (A and B) Fixation-related potentials (FRP) and head maps (inset) showing reduced P1f in healthy control (HC) vs schizophrenia patient (SzP). (C and D) time frequency (TF) plots showing reduced inter-trial phase-locking (ITPL) within the indicated tTF range. (E) Mean alpha power for passive vs reading. (F) Difference values across groups for passive reading; *P < .05; **P < .01. Fig. 3. Open in new tabDownload slide (A and B) Fixation-related potentials (FRP) and head maps (inset) showing reduced P1f in healthy control (HC) vs schizophrenia patient (SzP). (C and D) time frequency (TF) plots showing reduced inter-trial phase-locking (ITPL) within the indicated tTF range. (E) Mean alpha power for passive vs reading. (F) Difference values across groups for passive reading; *P < .05; **P < .01. Physiological Measures Contrast Sensitivity. Schizophrenia patients showed a significant deficit (F1,49 = 9.12, P = .004, d = .86) in visual contrast sensitivity (table 1) that predicted slower reading rate over and above the effect of group (F1,46 = 10.3, P = .002). In addition, the group × contrast sensitivity interaction was significant (F1,46 = 6.99, P = .011), reflecting a highly significant correlation in SzP (r = −.63, P < .001) but not HC (r = −.14, P = .5) (supplementary figure 4F). Fixation-Related Potentials. A prominent P1f was observed (figure 2A) that showed the anticipated occipital distribution. The amplitude of the P1f was significantly reduced in SzP relative to HC (F1,44 = 4.84, P = .033, d = .66) (figure 2B). In TF analyses, the P1f corresponded to increased inter-trial phase-locking (ITPL) within the alpha-theta (6–12 Hz) frequency range that was also significantly reduced in SzP across hemispheres (F1,41 = 6.76, P = .013), with no significant effect of hemisphere or group × hemisphere interaction (both P > .5) (figures 2C and 2D). Fixation-P1 amplitude correlated significantly with increases in the number of saccades (r = −.42, P = .004), and this correlation remained significant even following covariation for group status (rp = −.30, P = .048). The correlation between P1f amplitude and regressive saccades was also independently significant (r = −.35, P = .019). By contrast, no significant correlation was observed between P1f amplitude and fixation duration (r = −.11, P = .48), further suggesting that reduced P1f primarily indexes dysfunction of low-level visual oculomotor control pathways. As opposed to FRP-related activity, ongoing, nonphase-locked alpha activity was not significantly different between groups (F1,33 = .99, P = .3), with both groups showing highly significant ongoing alpha power suppression over visual cortex during reading vs nonreading (F1,33 = 61, P < .001) (figures 2E and 2F, supplementary Methods). Correlation With Clinical Features As expected, SzP showed significantly lower individual SES and reduced years of education vs HC (table 1). In ANCOVA analyses, fixation duration (F1,46 = 12.9, P = .001) was significantly related to years of education over and above the effect of group, and between-group differences were no longer significant following covariation (F1,46 = .16, P = .7). By contrast, the higher saccade number observed in Sz remained highly significant even following control for years of education (F1,43 = 9.99, P = .003). The reduction in P1f amplitude in SzP also remained significant following covariation for years of education (F1,40 = 4.33, P = .044; residual d = .66) and did not correlate with education (F1,40 = .14, P = .7). Control Analyses. Correlations between reading and neurophysiological tests, and medication dose and illness duration, were all nonsignificant (all r < .2, P > .3). Although SzP had significantly worse vision than control, all participants had at least 20/32 vision. Further, the between-group differences in reading rate remained highly significant even following covariation for vision (F1,46 = 22.4, P < .0001). Reading rate decreased (r = .42, P = .002) and saccade number increased (r = .33, P = .018) with age across groups, but the group effects remained significant even following covariation (both P < .001) There were no significant gender effects on any measure, and all results remained significant following the inclusion of gender as a covariate. Discussion The ability to read is essential for effective functioning in modern societies and is strongly linked to economic, occupational, and social success.3,5,8,38 Over the past decade, severe deficits in reading ability have been documented in Sz that arise even prior to the illness onset and that represent a decline from an earlier, higher level of function.3,5 Here, we demonstrated the utility of combined eye-tracking, neurophysiological, and computational modeling analysis to evaluate mechanistic models of reading impairments in Sz. As summarized below, we report a variety of findings that, taken together, support a “2-hit” model of reading dysfunction in Sz that includes both top-down and bottom-up contributions. Eye Tracking and Reading Measures In replication of prior studies of reading in Sz,5 we demonstrated reduced reading rate in Sz, which was reflective of moderate increases in fixation duration and dramatic increases in saccade number (table 1). As support for the “2-hit” account of reading deficits in Sz, we observed marked differences in the pattern of correlations for the fixation duration vs saccade number measures. Specifically, as evidence for a higher-level (lexical) deficit in Sz, the increases in fixation durations for Sz were strongly related to cognitive ability, which is consistent with previously reported activation deficits within semantic networks including inferior frontal gyrus during reading in Sz.12 Consistent with this interpretation, prior studies have demonstrated strong lexical influences on fixation duration during normal reading.18,19,39,40 In contrast, as evidence that increases in saccade number reflected lower-level visual and/or oculomotor dysfunction, as has also been previously reported in Sz,4 the increases in saccade number for Sz were unrelated to educational achievement, and the differences in saccade number across groups remained significantly different even following control for education (P = .003) or overall cognition (P = .005). Further, the distributional analyses of saccade amplitude (supplementary figure 5) indicated that the patients were primarily showing increases in short (2–8 characters) regressive saccades. As reviewed by Dias et al and Reichle et al,13,14 short amplitude saccades could potentially reflect low-level visual and oculomotor factors (in addition to high-level factors), whereas long-range saccades are instead typically assumed to reflect comprehension difficulty. Thus, the fact that regression length was unaffected in Sz, whereas the regression number was greatly elevated, suggests that disturbances in oculomotor programming may contribute to reduced reading fluency above and beyond potential effects of disturbances in comprehension. Computational Modeling Models such as the E-Z Reader provide a theoretical framework for simulating the impact of both high-level factors (lexical, postlexical, and attentional factors) and lower-level factors (visual/oculomotor factors) on eye movements during reading.14,15,21 The model assumes first that word recognition occurs serially, and, second, that word recognition consists of 2 stages, termed L1 and L2. The completion of the early stage of lexical processing (ie, L1) sets in motion the programming of a saccade to the next word, whereas the completion of the later stage of processing (ie, L2) triggers a shift in attention to the next word (rev. in 14–17,21). The mean duration of these 2 stages (ie, L1 and L2) increases when lexical processing is impaired, which is modeled by an increase in the α 1 parameter. Also, as the initial fixation location moves farther from optimal viewing position in the center of a word,41 the mean duration required to complete the early stage of lexical processing (ie, L1) increases due to the drop-off in visual acuity that occurs within parafoveal vs foveal vision. The probability of a refixation also increases. To the extent that the L1 and L2 stages take longer to complete, the model predicts longer fixation durations as well as a higher number of saccades. The model has previously been shown to account parsimoniously for changes in reading ability across the life span. Specifically, children not only make more and longer fixations while learning to read, similar to what we observed here in Sz, but they also show shorter saccades, whereas saccade amplitude in Sz was unaffected (table 1). In E-Z Reader, changes in eye-movement patterns during reading in children were adequately modeled by alteration in the speed-of-lexical-processing (α 1) parameter.20,42 By contrast, older readers not only show increased regressions and refixations compared with middle-aged adults, similar to what we observed here in Sz, but they also show higher skipping rates, reflecting a “risky” reading strategy43 as opposed to the lower rates observed here (table 1, figure 3).20 In E-Z Reader, these changes were effectively modeled by both slowing lexical processing (α 1) and increasing the propensity to guess words (κ).44 Here, we tested the hypothesis that manipulation of both lexical processing and low-level measures representing visual and oculomotor function would be required to adequately model the pattern of eye movement disturbance in Sz. We selected an α 1 parameter of 144 for our modeling because it best equates to mean ~8th-grade comprehension levels that we have observed previously in Sz.2,3 This manipulation significantly improved model fit vs the default model (P < .001). Nevertheless, it continued to underestimate the patients’ skipping and refixation probability deficits relative to their fixation duration deficits (figure 3). A more extreme value of 208 (which has been used to model a 2–6th-grade normative reading level)20 produced only small additional improvement and also did not adequately model the increased probability of refixation (supplementary table 2). Manipulation of the ε and λ parameters, which relate to aspects of visual sensory function and low-level saccadic programming, also significantly improved goodness-of-fit relative to default (P < .001) and relative to the lexical-alone manipulation (P < .01). Manipulation of both parameter sets (ie, the “2-hit” model) produced a model fit that uniquely captured all aspects of the deficit in Sz (supplementary table 2). In addition, it was superior to the lexical-alone model (P < .01), although not significantly different from the visual/oculomotor deficit-alone model (P > .1). Of note, it has previously been reported that manipulation of the visual acuity (ε) parameter does not improve model fits for older readers,44 whereas manipulation of the “risky reading” (κ) parameter does not explain the pattern of deficit observed in Sz, reflecting a multi-way dissociation in reading pattern across groups. Deficits in postlexical processing also typically produce a reduction in saccade size20 vs the intact saccade size observed here in SzP and thus do not parsimoniously account for the present findings. Fixation-Related Potentials Our FRP findings also support low-level contributions to impaired reading. Unlike standard ERP studies in which responses are time-locked to external events, FRP are calculated relative to the end of saccade and initiation of fixation as detected using combined eye tracking and EEG recording and thus reflect physiological processes related to reading. The P1f (“lambda” potential) indexes phase reset of ongoing theta rhythms within the visual cortex and is thought to “prime” the visual system to efficiently process newly fixated information.30 Unlike externally driven visual ERP, the amplitude of the P1f is relatively unaffected by the physical properties of the stimulus fixated either pre- or post-saccade.29 Both the P1f25 and the associated theta-phase reset45 appear to be unaffected by the semantic or cognitive load during reading, supporting its use to assess low-level function. This physiological approach is critical to understanding the reading rate deficits in Sz because it permits the study of both oculomotor and parafoveal processing, which are important components of reading that are not fully captured by typical visual ERP approaches that only present a single word at a time. As reviewed by,13,14 skilled reading requires an intricate coordination of oculomotor programming with ongoing lexical and linguistic processing. It is known that skilled readers rely heavily on parafoveal processing to begin to preprocess upcoming words in their parafovea prior to fixating on them, as well to extract information about word length and word boundaries to facilitate saccadic programming. The P1f is diminished during reading tasks that do not require eye movements (ie, reading of crawling texts),46 supporting its link to the physiological processes involved in the planning and execution of saccades related to reading. As predicted, the amplitude of the P1f and associated theta response were significantly smaller in Sz (P = .033) (figure 3) and correlated significantly with increased saccade number, but not increased the fixation duration or general cognitive function. These findings thus support the results of our simulations suggesting that the disproportionate increase in saccade number relative to fixation duration reflects the presence of low- as well as high-level deficits in visual function. These findings parallel a recent study in which reduced P1f/theta phase reset significantly predicted impaired serial search ability, further demonstrating functional consequences of impaired efficiency of fixation-related early visual processing in Sz.13 Alpha Activity To further evaluate potential top-down influences, we also analyzed the amplitude of ongoing alpha activity over visual occipital regions during reading. Ongoing alpha activity is significantly modulated by cross-modal attention, such that levels are high during neutral or attend-auditory conditions but suppressed during visual attention.47 This is thought to reflect bringing the occipital regions “online” during visual task performance. Reductions in posterior alpha activity may also reflect cognitive load during reading.48 Highly significant reductions in posterior alpha activity were observed in both groups during reading, with no significant between-group difference (figures 3E and 3F), arguing against differential attentional allocation or subjective cognitive load as explanations for the observed reading deficits. Limitations E-Z Reader simulations for this study were performed at the group level, preventing the assessment of the interrelationship between model parameters and physiological measures on a per-subject basis. Furthermore, we tested only selected parameters related to either lexical processing or visual sensory function. Future studies using a wider range of visual materials are required to permit more extensive simulations. Also, subjects read GORT passages silently so as not to introduce artifact into the EEG recordings. The accuracy of reading thus could not be assessed in this study, although we have extensively documented impaired reading fluency in prior studies, as have others,2–5,12,49 including a subset of subjects participating in this study (supplementary Results). All SzP were receiving medication; however, no correlations with antipsychotic dose were observed. Here, we propose 2 “clusters” of disturbance in Sz. First, a high-level lexical deficit associated with reduced educational achievement and, potentially, with impaired auditory phonological processing and MMN generation,3 and manifesting as increased fixation duration. Second, a lower-level deficit in oculomotor control and associated with reduced FRP generation, magnocellular dysfunction,2 and reduced parafoveal processing,5 as manifested in increased short regressive saccades and within-word refixations. This pattern differs from patterns documented previously for both beginning readers,20 who have not only increased saccade number but also reduced saccade amplitude, and healthy elderly readers, who adopt a risky strategy but show relatively intact low-level visual contributions.18,44 Increases in both fixation duration and saccade number contributed independently to prolonged reading times. Thus, neural mechanisms underlying these 2 separate deficits can be analyzed independently. Conclusions Reading dysfunction remains an underappreciated and undertreated aspect of Sz, and a significant contributor to poor social and role function in both prodromal8 and established3,5,49 Sz subjects. Patterns of dysfunction suggest deficits in both high-level lexical processing and lower-level oculomotor control. Oculomotor control deficits are documented using eye-tracking, computational modeling, and neurophysiological (FRP) measures. Detailed computational modeling combined with multimodal imaging may be used to further examine underlying pathophysiological mechanisms. Funding The funding for this study was provided by National Institute of Mental Health grants (MH049334 and MH121449 to D.C.J.) Acknowledgments We thank Drs. Filipe Braga and Julianne Ammirati for their support in programming the task. The results were presented in part at the Meeting of the Society for Neuroscience, Washington, DC, November 2017. Dr Javitt reports personal fees from Cadence, Autifony, SK Life Science, Biogen, and Boehringer-Ingelheim outside the submitted work. Dr Javitt holds significant equity in Glytech, AASI, and NeuroRx. He serves on the scientific advisory boards of Promentis and NeuroRx, and a Biogen data and safety monitoring board. He also holds intellectual property for use of N-Methyl-D-aspartate receptor (NMDAR) antagonists in depression and other disorders, NMDAR agonists in schizophrenia, and has pending disclosures for magnetic ressonance-based prediction of electroconvulsive therapy and transcranial magnetic stimulation response and EEG-based diagnosis of mental disorders. Dr Hoptman consults for Kessler Research Foundation. Drs Dias, Sehatpour, Martinez, Sheridan, Butler, and Revheim and Mss. Silipo, Rohrig, and Hochman reported no biomedical financial interests or potential conflicts of interest. References 1. Gold JM , Goldberg RW, McNary SW, Dixon LB, Lehman AF. Cognitive correlates of job tenure among patients with severe mental illness . Am J Psychiatry. 2002 ; 159 ( 8 ): 1395 – 1402 . Google Scholar Crossref Search ADS PubMed WorldCat 2. Revheim N , Butler PD, Schechter I, Jalbrzikowski M, Silipo G, Javitt DC. Reading impairment and visual processing deficits in schizophrenia . Schizophr Res. 2006 ; 87 ( 1-3 ): 238 – 245 . Google Scholar Crossref Search ADS PubMed WorldCat 3. Revheim N , Corcoran CM, Dias E, et al. Reading deficits in schizophrenia and individuals at high clinical risk: relationship to sensory function, course of illness, and psychosocial outcome . Am J Psychiatry. 2014 ; 171 ( 9 ): 949 – 959 . Google Scholar Crossref Search ADS PubMed WorldCat 4. Whitford V , O’Driscoll GA, Pack CC, Joober R, Malla A, Titone D. Reading impairments in schizophrenia relate to individual differences in phonological processing and oculomotor control: evidence from a gaze-contingent moving window paradigm . J Exp Psychol Gen. 2013 ; 142 ( 1 ): 57 – 75 . Google Scholar Crossref Search ADS PubMed WorldCat 5. Whitford V , O’Driscoll GA, Titone D. Reading deficits in schizophrenia and their relationship to developmental dyslexia: a review . Schizophr Res. 2018 ; 193 : 11 – 22 . Google Scholar Crossref Search ADS PubMed WorldCat 6. Fuller R , Nopoulos P, Arndt S, O’Leary D, Ho BC, Andreasen NC. Longitudinal assessment of premorbid cognitive functioning in patients with schizophrenia through examination of standardized scholastic test performance . Am J Psychiatry. 2002 ; 159 ( 7 ): 1183 – 1189 . Google Scholar Crossref Search ADS PubMed WorldCat 7. Ho BC , Andreasen NC, Nopoulos P, Fuller R, Arndt S, Cadoret RJ. Secondary prevention of schizophrenia: utility of standardized scholastic tests in early identification . Ann Clin Psychiatry. 2005 ; 17 ( 1 ): 11 – 18 . Google Scholar Crossref Search ADS PubMed WorldCat 8. Carrión RE , Cornblatt BA, McLaughlin D, et al. Contributions of early cortical processing and reading ability to functional status in individuals at clinical high risk for psychosis . Schizophr Res. 2015 ; 164 ( 1-3 ): 1 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 9. Roberts EO , Proudlock FA, Martin K, Reveley MA, Al-Uzri M, Gottlob I. Reading in schizophrenic subjects and their nonsymptomatic first-degree relatives . Schizophr Bull. 2013 ; 39 ( 4 ): 896 – 907 . Google Scholar Crossref Search ADS PubMed WorldCat 10. Fernández G , Sapognikoff M, Guinjoan S, Orozco D, Agamennoni O. Word processing during reading sentences in patients with schizophrenia: evidences from the eyetracking technique . Compr Psychiatry. 2016 ; 68 : 193 – 200 . Google Scholar Crossref Search ADS PubMed WorldCat 11. Fernández G , Guinjoan S, Sapognikoff M, Orozco D, Agamennoni O. Contextual predictability enhances reading performance in patients with schizophrenia . Psychiatry Res. 2016 ; 241 : 333 – 339 . Google Scholar Crossref Search ADS PubMed WorldCat 12. Martínez A , Revheim N, Butler PD, Guilfoyle DN, Dias EC, Javitt DC. Impaired magnocellular/dorsal stream activation predicts impaired reading ability in schizophrenia . Neuroimage Clin. 2012 ; 2 : 8 – 16 . Google Scholar Crossref Search ADS PubMed WorldCat 13. Dias EC , Van Voorhis AC, Braga F, et al. Impaired fixation-related theta modulation predicts reduced visual span and guided search deficits in schizophrenia . Cereb Cortex. 2020 ; 30 ( 5 ): 2823 – 2833 . Google Scholar Crossref Search ADS PubMed WorldCat 14. Reichle ED , Pollatsek A, Rayner K. Using E-Z Reader to simulate eye movements in nonreading tasks: a unified framework for understanding the eye-mind link . Psychol Rev. 2012 ; 119 ( 1 ): 155 – 185 . Google Scholar Crossref Search ADS PubMed WorldCat 15. Reichle ED , Rayner K, Pollatsek A. The E-Z reader model of eye-movement control in reading: comparisons to other models . Behav Brain Sci. 2003 ; 26 ( 4 ): 445 – 476 ; discussion 477–526 . Google Scholar Crossref Search ADS PubMed WorldCat 16. Reichle ED , Sheridan H. E-Z reader: an overview of the model and two recent applications . In: Pollatsek A, Treiman R, eds. The Oxford Handbook of Reading. Oxford, UK : Oxford University Press ; 2015 : 277 – 290 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 17. Reichle ED , Warren T, McConnell K. Using E-Z reader to model the effects of higher level language processing on eye movements during reading . Psychon Bull Rev. 2009 ; 16 ( 1 ): 1 – 21 . Google Scholar Crossref Search ADS PubMed WorldCat 18. Rayner K . Eye movements in reading and information processing: 20 years of research . Psychol Bull. 1998 ; 124 ( 3 ): 372 – 422 . Google Scholar Crossref Search ADS PubMed WorldCat 19. Rayner K . Eye movements and attention in reading, scene perception, and visual search . Q J Exp Psychol (Hove). 2009 ; 62 ( 8 ): 1457 – 1506 . Google Scholar Crossref Search ADS PubMed WorldCat 20. Reichle ED , Liversedge SP, Drieghe D, et al. Using E-Z reader to examine the concurrent development of eye-movement control and reading skill . Dev Rev. 2013 ; 33 ( 2 ): 110 – 149 . Google Scholar Crossref Search ADS PubMed WorldCat 21. Reichle ED , Pollatsek A, Fisher DL, Rayner K. Toward a model of eye movement control in reading . Psychol Rev. 1998 ; 105 ( 1 ): 125 – 157 . Google Scholar Crossref Search ADS PubMed WorldCat 22. Kornrumpf B , Dimigen O, Sommer W. Lateralization of posterior alpha EEG reflects the distribution of spatial attention during saccadic reading . Psychophysiology 2017 ; 54 ( 6 ): 809 – 823 . Google Scholar Crossref Search ADS PubMed WorldCat 23. Baccino T , Manunta Y. Eye-fixation-related potentials: insight into parafoveal processing . J Psychophysiology. 2005 ; 19 : 204 – 215 . Google Scholar Crossref Search ADS WorldCat 24. Dimigen O , Sommer W, Hohlfeld A, Jacobs AM, Kliegl R. Coregistration of eye movements and EEG in natural reading: analyses and review . J Exp Psychol Gen. 2011 ; 140 ( 4 ): 552 – 572 . Google Scholar Crossref Search ADS PubMed WorldCat 25. Dimigen O , Kliegl R, Sommer W. Trans-saccadic parafoveal preview benefits in fluent reading: a study with fixation-related brain potentials . Neuroimage 2012 ; 62 ( 1 ): 381 – 393 . Google Scholar Crossref Search ADS PubMed WorldCat 26. Vanyukov PM , Warren T, Wheeler ME, Reichle ED. The emergence of frequency effects in eye movements . Cognition 2012 ; 123 ( 1 ): 185 – 189 . Google Scholar Crossref Search ADS PubMed WorldCat 27. Henderson JM , Luke SG, Schmidt J, Richards JE. Co-registration of eye movements and event-related potentials in connected-text paragraph reading . Front Syst Neurosci. 2013 ; 7 : 28 . Google Scholar Crossref Search ADS PubMed WorldCat 28. Kornrumpf B , Niefind F, Sommer W, Dimigen O. Neural correlates of word recognition: a systematic comparison of natural reading and rapid serial visual presentation . J Cogn Neurosci. 2016 ; 28 ( 9 ): 1374 – 1391 . Google Scholar Crossref Search ADS PubMed WorldCat 29. Kazai K , Yagi A. Comparison between the lambda response of eye-fixation-related potentials and the P100 component of pattern-reversal visual evoked potentials . Cogn Affect Behav Neurosci. 2003 ; 3 ( 1 ): 46 – 56 . Google Scholar Crossref Search ADS PubMed WorldCat 30. Rajkai C , Lakatos P, Chen CM, Pincze Z, Karmos G, Schroeder CE. Transient cortical excitation at the onset of visual fixation . Cereb Cortex. 2008 ; 18 ( 1 ): 200 – 209 . Google Scholar Crossref Search ADS PubMed WorldCat 31. Butler PD , Zemon V, Schechter I, et al. Early-stage visual processing and cortical amplification deficits in schizophrenia . Arch Gen Psychiatry. 2005 ; 62 ( 5 ): 495 – 504 . Google Scholar Crossref Search ADS PubMed WorldCat 32. Hollingshead AB . Two factor index of social position ; Yale University Press , New Haven 1957 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 33. Oldfield RC . The assessment and analysis of handedness: the Edinburgh inventory . Neuropsychologia 1971 ; 9 ( 1 ): 97 – 113 . Google Scholar Crossref Search ADS PubMed WorldCat 34. Kay SR , Opler LA, Fiszbein A The positive and negative syndrome scale (PANSS) for schizophrenia . Schizophr Bull. 1987 ; 13 ( 2 ): 261 – 276 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC 35. Wiederholt JL , Bryant BR Gray Oral Reading Tests (GORT-4): Examiner’s Manual. 4th ed. Austin, TX : Pro-Ed ; 2001 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 36. Schilling HH , Rayner K, Chumbley JI. Comparing naming, lexical decision, and eye fixation times: word frequency effects and individual differences . Mem Cognit. 1998 ; 26 ( 6 ): 1270 – 1281 . Google Scholar Crossref Search ADS PubMed WorldCat 37. Cohen J Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ : Lawrence Erlbaum Assoc. ; 1988 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 38. Kirsch IS , Jungeblut AJ, Jenkins L, Kolstad A Adult Literacy in America: A First Look At Findings of The National Adult Literacy Survey. US Department of Education Office of Educational Research and Improvement , ed. 3rd ed. Washington, DC ; 2002 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 39. Reingold EM , Reichle ED, Glaholt MG, Sheridan H. Direct lexical control of eye movements in reading: evidence from a survival analysis of fixation durations . Cogn Psychol. 2012 ; 65 ( 2 ): 177 – 206 . Google Scholar Crossref Search ADS PubMed WorldCat 40. Sheridan H , Reingold EM. The time course of predictability effects in reading: evidence from a survival analysis of fixation durations . Vis Cogn. 2012 ; 20 : 733 – 745 . Google Scholar Crossref Search ADS WorldCat 41. O’Regan JK , Lévy-Schoen A. Eye-movement strategy and tactics in word-recognition and reading . In: Coltheart M, ed. Attention and Performance . Vol XII . Hillsdale, NJ : Erlbaum ; 1987 : 363 – 384 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 42. Mancheva L , Reichle ED, Lemaire B, Valdois S, Ecalle J, Guérin-Dugué A. An analysis of reading skill development using E-Z reader . J Cogn Psychol (Hove). 2015 ; 27 ( 5 ): 357 – 373 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 43. Rayner K , Reichle ED, Stroud MJ, Williams CC, Pollatsek A. The effect of word frequency, word predictability, and font difficulty on the eye movements of young and older readers . Psychol Aging. 2006 ; 21 ( 3 ): 448 – 465 . Google Scholar Crossref Search ADS PubMed WorldCat 44. McGowan VA , Reichle ED. The “risky” reading strategy revisited: new simulations using E-Z Reader . Q J Exp Psychol (Hove). 2018 ; 71 ( 1 ): 179 – 189 . OpenURL Placeholder Text WorldCat 45. Vignali L , Himmelstoss NA, Hawelka S, Richlan F, Hutzler F. Oscillatory brain dynamics during sentence reading: a fixation-related spectral perturbation analysis . Front Hum Neurosci. 2016 ; 10 : 191 . Google Scholar Crossref Search ADS PubMed WorldCat 46. Kliegl R , Dambacher M, Dimigen O, Jacobs AM, Sommer W. Eye movements and brain electric potentials during reading . Psychol Res. 2012 ; 76 ( 2 ): 145 – 158 . Google Scholar Crossref Search ADS PubMed WorldCat 47. Rohenkohl G , Nobre AC. Alpha oscillations related to anticipatory attention follow temporal expectations . J Neurosci. 2011 ; 31 ( 40 ): 14076 – 14084 . Google Scholar Crossref Search ADS PubMed WorldCat 48. Scharinger C , Kammerer Y, Gerjets P. Pupil dilation and EEG alpha frequency band power reveal load on executive functions for link-selection processes during text reading . PLoS One. 2015 ; 10 ( 6 ): e0130608 . Google Scholar Crossref Search ADS PubMed WorldCat 49. Dondé C , Martinez A, Sehatpour P, et al. Neural and functional correlates of impaired reading ability in schizophrenia . Sci Rep. 2019 ; 9 ( 1 ): 16022 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2020. Published by Oxford University Press on behalf of the Maryland Psychiatric Research Center. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Neurophysiological, Oculomotor, and Computational Modeling of Impaired Reading Ability in Schizophrenia
JF - Schizophrenia Bulletin
DO - 10.1093/schbul/sbaa107
DA - 2021-01-23
UR - https://www.deepdyve.com/lp/oxford-university-press/neurophysiological-oculomotor-and-computational-modeling-of-impaired-oVQoIYseXy
SP - 97
EP - 107
VL - 47
IS - 1
DP - DeepDyve
ER -