Erp Evidence Of Abnormal Context-dependent Modulation Of Global Semantic Processing In Schizophrenia Patients With Positive Thought Disorder

ERP Evidence of abnormal 1 Running head: EVIDENCE OF ABNORMAL SEMANTIC PROCESSING

ERP evidence abnormal context-dependent modulation of global semantic processing in schizophrenia patients with positive thought disorder Brian C. Woolfrey University of Minnesota

ERP Evidence of abnormal 2 Abstract Previous behavioral and psychophysiological studies of word-pair associations and sentence comprehension in schizophrenic individuals with positive thought disorder (TD) have revealed abnormal patterns of semantic processing at a local level of text comprehension. More recent studies using ERP methods to study on-line processing of single sentences provide preliminary evidence that TD patients may also fail to use context properly during discourse to integrate semantic information into a coherent idea. We propose to explore the latter idea in more detail using a global context manipulation paradigm introduced by St. George (1994). Specifically, we hypothesize that TD subjects will show an abnormally small global N400 attenuation effect for all words in titled paragraphs vs. untitled paragraphs when a context-setting title is required for paragraph coherence. If the results support the hypothesis, it would support the idea that both local semantic abnormalities and global context modulation difficulties are key components in positive thoughts disorder.

ERP Evidence of abnormal 3 ERP evidence of abnormal context-dependant modulation of global semantic processing in schizophrenia patients with positive thought disorder Schizophrenia is a severe and often debilitating brain disorder that affects nearly 1% of the US population (NIMH). Although a variety of distinct subtypes of schizophrenia exist, many of them involve either disorganized speech patterns or positive thought disturbances such as delusions, auditory hallucinations and paranoia. Because of this, researchers have long speculated that the etiology of certain types of schizophrenia may have something to do with systematic abnormalities in language processing. Preliminary evidence of semantic processing abnormalities some schizophrenic patients appear to support this model. Specifically, researchers have capitalized on the classic lexical decision paradigm to show reduced semantic priming effects in schizophrenic patients with thought disorder. In this paradigm a prime word is presented a moment before a target word, and the participant's task is to choose whether the target word is a real word or a nonsense word. Current cognitive theory suggests that the prime word automatically activates semantically related concepts though a process known as "spreading activation", and that this activation facilitates neural communication between activated "nodes" in long-term memory (LTM) and working memory (WM) such that they require less bottom-up processing to access in the immediate future. Consistent with this idea, normal subjects typically make faster decisions when a target word is immediately preceded by a prime word that is strongly related in terms of meaning (Wentura, 2007). This semantic priming process, however, seems to be disrupted in schizophrenic patients with positive thought disorder. In a lexical-decision study by Manschreck and colleagues (1988), semantic priming effects were shown to be significantly smaller in TD patients than in healthy control subjects. Subsequent priming studies using strongly-relate word

ERP Evidence of abnormal 4 pairs have corroborated these findings, supporting the notion that certain types of schizophrenia involve dysfunctional semantic processing abilities (Besche, 1997). Interestingly, a completely different effect has been found for word pairs that are weakly related in terms of meaning. Rather than showing a lack of a priming effect, TD patients have actually been found to show a stronger than normal priming effect (Wentura, 2007). This apparent hypo-facilitation of strongly-related word pairs and hyper-facilitation of weakly-related word pairs has been interpreted by many to suggest that abnormalities in semantic processing may involve deficient inhibition mechanisms that result in diffuse, unfocused and uncontrolled semantic activation. Many have even speculated that this may be an etiological factor in the development of discourse incoherence and even delusion (Kumar, 2004). Further support for this inhibition dysfunction model of thought disorder comes from studies involving sentence reading. Using a lexical decision task after the auditory presentation of a sentence, Titone (2000) found that schizophrenic patients failed to properly inhibit the dominant meanings of certain critical words when immediate sentence context moderately favored their subordinate meaning. For example, after hearing a sentence about an animal "pen" (the subordinate meaning of pen for someone who does not live on a farm), schizophrenia patients showed lexical decision priming effects to pictures of both animal enclosures and objects related to the dominant meaning of "pen" (e.g. "paper" or "pencil"). Normal subjects, in contrast, properly inhibited the dominant meaning of pen and only showed a priming effect to animal enclosures. These results were interpreted to suggest that inhibitory deficits may lead to the failure of proper context integration in schizophrenics. Further evidence of this came from Kuperberg and colleagues (2006), who showed that schizophrenic subjects were relatively insensitive to verbs that violated the context of a sentence when those verbs were semantically

ERP Evidence of abnormal 5 associated with any of the preceding words. Healthy control subjects, in contrast, were sensitive to all verb violations. This lends strong support to the idea that schizophrenia patients may not be able to properly inhibit semantic activation. Additional evidence for the semantic dysfunction model of schizophrenia comes from investigations employing psychophysiological methods. As far back as 1991 schizophrenia has been shown to be associated with abnormalities in a negatively-deflecting ERP component that occurs around 400ms after the onset of a target word (Kostova 2005). In cognitive neuroscience literature the amplitude of the N400 component has been established as a fairly reliable measure of semantic expectancy. When a word is presented to a participant, semantically unexpected words typically produce a larger N400 wave than semantically primed words (Kutas & Hillyard, 1980). The size of the N400 attenuation response to primed words, then, is thought to provide a reliable psychophysiological measure of semantic priming and is analogous to faster lexical decision reaction times. The main advantage of using this measure of priming over lexical decision tasks is that it allows researchers to capture rapid on-line processing of information at any point in a discourse without interruption or decay of effect due to delay in measurement (Ditman, 2007). Using this N400 paradigm as a measure of semantic activation, Grillon (1991) found a reduced priming effect in a subgroup of schizophrenics during a word pair priming task. This lack of N400 effect is analogous to lexical decision measures of strongly related word pair priming, and has been replicated many times. Like the lexical decision findings, the N400 effect has also been linked to the severity of positive thought disorder symptoms (Kostova, 2005). Some studies have also found a stronger than normal N400 effect in response to weakly-related words (Kreher, 2007, as cited in Ditman, 2007), although others have only found changes in

ERP Evidence of abnormal 6 latency measures (Kiang, 2008). Further study on this issue may be required to clarify this relationship. If the effect is indeed stronger than normal as Kreher suggests, this would be in line with behavioral findings of hyperactivation of weakly related concepts, and thus lend further support to the model of diffuse and unfocused semantic activation. Other studies have used the N400 paradigm to examine the on-line dynamics of semantic activation during whole-sentence discourse. These studies have yielded several interesting results. First, Adams (1993) found that schizophrenic patients showed significantly smaller N400 amplitudes than controls in response to semantically anomalous endings to otherwise coherent sentences. This provides general support for the idea that schizophrenia is associated with a deficiency in the context integration. More interestingly, Sitnikova (2002) found N400 evidence of failure to override the dominant meaning of a homonym in a sentence where context suggests that the subordinate meaning is more appropriate. All participants in this study were given a twoclause sentences with the first clause ending in a homograph (word with multiple meanings). The second clause of the sentence always started with a target word that was semantically related to the dominant meaning of the homograph and were either congruent or incongruent with prior sentence context. The average evoked N400 component was then analyzed for each group and target word, and it was found to discriminate between congruent and incongruent target words in the control group but not schizophrenic group. Consistent with behavioral findings described earlier, this suggests that abnormal sentence processing in schizophrenia may be related to a failure to inhibit context-inappropriate meanings of words. These word-pair and single-sentence studies have provided valuable insights into the local mechanisms of semantic and contextual processing in thought disorder patients. A fuller and more fruitful understanding of thought disorder, however, can only be achieved if these basic

ERP Evidence of abnormal 7 problems are found to be related to the more global, controlled level of everyday discourse where psychological intervention is actually possible. With this in mind, Ditman (2007) conduced an ERP investigation into the global build up meaning in schizophrenia patients. Specifically, he examined how the N400 is modulated while schizophrenics integrate critical words into preceding discourse context. To do this he used a 3-sentence discourse scenario design in which all sentences were internally coherent but the last sentence varied in terms of its causal relatedness to the preceding sentences. In the control group, the amplitude of the N400 response to critical words in the final sentence decreased when the sentences were more causally related. In subjects with high severity of thought disorder symptoms, however, this effect was significantly reduced. Based on these results it was concluded that TD patients have difficulties integrating both context cues and semantic information across sentences to build up global discourse coherence. To further explore this claim, I propose to employ a context manipulation paradigm introduced by St. George (1994) that has been shown to be sensitive to global semantic expectancy in whole paragraphs. This paradigm capitalizes on the unique strengths of N400 measurements in that it is designed to assess real-time on-line processing of semantic expectancy during extended discourse. Essentially, it involves recording ERP responses to specially designed paragraphs that are presented one word at a time. These paragraphs, which were taken from several well-known studies on the effects of context on comprehension, are perfectly coherent in terms of grammar and structure but require the presence of a context-setting title to be coherent in terms of meaning. Without context, they seem to be just a series of vague, disconnected propositions. In normal subjects it was shown that the average N400 peak values for every word in the untitled paragraphs, across all electrodes, were significantly greater than in the titled

ERP Evidence of abnormal 8 paragraphs (St. George, 1994). According to the authors, this reflects the modulation of semantic processing and integration by contextual cues. Because this process may be dysfunctional in TD patients, I hypothesize that schizophrenic patients will show a less pronounced N400 effect for the same untitled paragraphs. Method Participants For this study I propose to recruit 20 (10 male) schizophrenia patients with clinically diagnosed positive thought disorder from the inpatient psychiatric ward at a local hospital. All will receive an appropriate monetary compensation for their time. Patients with neurological disorders or EEG abnormalities as detected in the routine examination will be excluded. No change will be made to patient's current course of medication, and all clinical patients will complete a Positive and Negative Syndrome Scale at the onset of the study (Kay, Fiszbein, & Opler, 1987, as Cited in Wagner, 2006). Additionally, 20 (10 male) healthy controls will be recruited and matched for age and level of education. All participants must be native English speakers and have normal or corrected-to-normal vision. Additionally, all participants will be right-handed as determined by scores on the Edinburgh Handedness Inventory. Apparatus As in the St. George study (1994), five Ag/AgCl electrodes will be placed on the scalps of participants. Three electrodes will be located along the midline at Fz, Cz and Pz according to the international 10-20 system (Harner & Sannit, 1974, as cited in St. George, 1994). One of the electrodes will be placed approximately over Wernike's area in the left hemisphere (lateral to the vertex by 30% of interaural distance and posterior by 12.5% of the nasion-inion distance), and the other will be placed symmetrically on the right hemisphere as a comparison. A linked

ERP Evidence of abnormal 9 mastoid reference will be used and the ground site will be placed on the forehead. Additionally, a vertical electro-oculogram (EOG) electrode will be used to detect and discard trials with ocular artifacts. During the procedure participants will be seated in a comfortable chair in a temperaturecontrolled laboratory room. All words will be presented on 36-point red font against a solid gray background on a high-definition, high-contrast computer screen situated approximately 90 centimeters from the participants' eyes (as ensured by a chin-rest). ERPs will be recorded for every word in each paragraph starting 50ms before each word is presented. All fluorescent lights will be turned off to reduce 60hz noise, and all electrode impedances will be kept below 5Ω. The same four paragraphs that were used in St. George's (1994) study will be used in the present study. As described earlier, these paragraphs were specially designed so that each sentence is locally coherent, but a context-setting title is required to establish global coherence. An example of this is the paragraph entitled Making and Flying a Kite. The first few sentences of the body of the paragraph are as follows: "A newspaper is better than a magazine. A seashore is a better place than the street. At first it is better to run than to walk. You may have to try several times….". Clearly the propositions in this paragraph seem meaningless and unconnected without context. Other paragraph titles include The Procedure for Washing Clothes, Space Trip To the Moon and Discovering America. Procedure As in the St. George (1994) study, each participant will be run individually. After filling out consent forms, the procedure will be explained in detail so there is no confusion. Additionally, to ensure attention is being paid to the whole paragraph and not the individual

ERP Evidence of abnormal 10 pieces, participants will be told that there will be a recall task regarding the meaning of the paragraph later in the study. At the onset of the experiment, participants will be given a demonstration paragraph to ensure they understand the task. The four stimuli paragraphs will be then presented to participants. As in the St. George (1994), presentation of these paragraphs will be counterbalanced such that all subjects will see the same four paragraphs in one of four different orders. Half of the participants will be randomly assigned to receive the titles of the paragraphs before they are presented, and the other half will not be provided with titles. After all four paragraphs are presented, participants will be debriefed and compensated. Data Analysis Ocular artifacts will be corrected for before the data is analyzed using Gratton, Coles, and Donchin's (1983) eye movement correction paradigm (as cited in St. George, 1994). Also, a 14HZ low-pass filter will be applied to the data and the resulting ERP waveforms will be averaged across all words in all paragraphs for each participant across all electrodes. N400 waveforms will be quantified by taking the mean amplitude within the 300ms to 500ms time window. These raw N400 mean values will then be entered into a 2 (clinical status: TD and no TD) X 2 (title manipulation: title and no title) between-subjects analysis of variance for statistical analysis. Results If my hypothesis is correct, statistical analysis should result in the following outcomes. First, analysis of variance should show a main effect for title manipulation such that participants who were presented with titled paragraphs have smaller N400 waves than those who were presented with untitled paragraphs F(1,39) = X, p < .05. There should also be a main effect of

ERP Evidence of abnormal 11 clinical status such that TD patients elicit a greater N400 wave overall than non TD patients, F(1,39) = X, P < .05. Further analysis should reveal an interaction between TD and title such that the effect of title is significantly weaker in TD patients (See Figure 1). Moreover, there should be no N400 difference between TD and non TD groups in the non-titled condition because there is no context to process differently. If this pattern is found, it would lend support to the idea that global semantic processing is modulated differently by context cues in TD patients than in healthy individuals. An alternative to these results may be that TD and NTD subjects react the same way to the title manipulation. If so, this would suggest that the modulation of semantic processing is a more local process that is affected by more specific conditions. Discussion In this study changes in N400 waves were measured during the processing of whole paragraphs to test the hypothesis that, at a global coherence level (linking multiple propositions to form a coherent idea), context modulates semantic processing differently in schizophrenic patients with thought disorder than in healthy controls. This paradigm was proposed because it is able to capture neural correlates of real-time on-line semantic processing that would be impossible to measure with behavioral methods. If the results support my hypothesis, it would be consistent with previous behavioral and psychophysiological findings of local disturbances in semantic processing as well as preliminary N400 findings of context-modulated abnormalities in global semantic processing. To be interpreted correctly, however, several weaknesses of this study bust be addressed. First, an important confound would exist in the form of anti-psychotic medication. Since it would be unethical to instruct schizophrenic patients cease their regular medication for this study, we can not be sure whether changes in N400 patters are due to schizophrenia itself or due to the

ERP Evidence of abnormal 12 effects of psychoactive medication. Also, the EEG apparatus proposed in this study is a simple one with only 5 electrodes and no differentiation between electrodes in the final analysis. Effects of context manipulations may be stronger if a more complex electrode montage is used and the final statistical analysis employs weighted principle components rather than a grand average. Future studies may also benefit form a more graduated categorization of TD and TD symptoms. If a lack of N400 effect could be shown to be correlated with severity of TD symptoms this would give us even more information. Despite these few weaknesses, this study would be an important step in exploring the neurophysiological correlates of global semantic processing and context use in positive thought disorder. Better knowledge of these language comprehension processes may eventually lead to development of clinical applications that capitalize on new behavioral approaches to better the clinical treatment of schizophrenia. If nothing else, psychophysiological measures of schizophrenia markers like abnormal modulation of the N400 wave may eventually provide an alternative or converging method of diagnosis or subtype differentiation. Future studies would be wise to investigate these and other potentially useful applied clinical applications.

ERP Evidence of abnormal 13 References Adams, J., Faux, S. F., Nestor, P. G., Shenton, M., Marcy, B., Smith, S., McCarley, R. W. (1993) ERP abnormalities during semantic processing in schizophrenia. Schizophrenia Research, 10, 241-251 Besche, C., Passerieux, C., Segui, J., Sarfati, Y., Laurent, J., Hardy-Bayle, M. (1997). Syntactic and semantic processing in schizophrenic patients evaluated by lexical-decision tasks. Neuropsychology, 11, 498-505. Ditman, T., Kuperberg, G.R. (2007) The time course of building discourse coherence in schizophrenia: An ERP investigation. Psychophysiology, 44, 991-1001. Grillon, C., Ameli, R., Glazer, W. M., (1991). N400 and semantic categorization in schizophrenia. Biological Psychiatry, 29, 467-480. Kiang, M., Kutas, M., Light, G.A, Braff, D.L. (2008) An event-related brain potential study of direct and indirect semantic priming in schizophrenia. American Journal of Psychiatry, 165, 74-81. Kostova, M., Passerieux, C., Laurent, J., Hardy-Bayle, M.C. (2005) N400 anomalies in schizophrenia are correlated with the severity of formal thought disorder. Schizophrenia Research, 78, 285-291. Kuperberg, G. R., Sitnikova, T., Goff, D., Holcomb, P. J. (2006). Making sense of sentences in schizophrenia: Abnormal interactions between semantic and syntactic processing. Journal of Abnormal Psychology, 115, 251–265. Kumar, N., Debruille, J.B. (2004) Semantics and N400: Insights for schizophrenia. Journal of Psychiatry & Neuroscience, 29, 89-98. Kutas, M., Hillyard, S.A. (1980). Reading senseless sentences: Brain potentials reflect semantic incongruity. Science, 207, 203–205. Manschreck, T. C., Maher, B. A., Milavetz, J. J., Ames, D., Weisstein. C.C., Schneyer, M.L., (1988). Semantic priming in thought disordered schizophrenic patients. Schizophrenia Research, 1, 61-66 National Institute of Mental Health (2008). What is schizophrenia? Retrieved October 3, 1999, from http://www.nimh.nih.gov/health/publications/schizophrenia/what-isschizophrenia.shtml Sitnikova, T., Salisbury, D., Kuperberg, G., Holcomb, P. (2002) Electrophysiological insights into language processing in schizophrenia. Psychophysiology, 39, 851-86

ERP Evidence of abnormal 14 St. George, M., Mannes, S., Hoffman, J.E. (1994). Global semantic expectancy in language comprehension. Journal of Cognitive Neuroscience, 6, 70-83. Titone, D., Levy, D., Holzman, P. (2000). Contextual Insensitivity in Schizophrenic Language Processing: Evidence From Lexical Ambiguity. Journal of Abnormal Psychology, 109, 761-767 Wagner, M., Baving, L., Berg, P., Cohen, R., Rockstroh, B. (2006) An ERP Investigation of Semantic Priming, Repetition Priming, and Negative Priming in Schizophrenic Patients. Journal of Psychophysiology, 20, 195-211. Wentura, D., Moritz, S., Frings, C. (2007). Further evidence for “hyper-priming” in thoughtdisordered schizophrenic patients using repeated masked category priming. Schizophrenia Research, 102, 69–75

ERP Evidence of abnormal 15 Figure Caption Figure 1. Hypothetical results showing an interaction between thought disorder (TD and NTD) and paragraph condition (titled and untitled) on the mean of evoked N400 waves.

Figure 1

-

TD N400 Mean NTD + Non Titled Titled Paragraph