. Using positron emission tomography (PET), Howard, et al (1992) were able to identify a number of physically separate brain structures that become active during various language processing tasks. They found that physically discrete areas are responsible for auditory and visual word recognition and that these areas are separate from those involved in either word comprehension or production. Similarly, measurements of N400 sensory evoked brain potentials have found that different areas of the brain are active during processing of coherent and incoherent sentences, suggesting that language processing may be a more global activity than has been generally believed.
Fried, Ojemann and Fetz (1981) suggest that activity in Broca’s area and Wernicke’s area is simultaneous, communication between these areas is both bi-directional and interactive, and a parallel rather than serial processing strategy is used. Mesulan (1990) characterizes Wernicke’s area not as a discrete memory store of individual words but as: “a nodal bottleneck for accessing a distributed grid connectivity that contains information about sound-word-meaning relationships.” Similarly, Broca’s area is thought to be a parallel activation network which is primarily responsible for the translation of word representations produced in Wernicke’s area into corresponding action potential. Other language processing activities occur throughout the brain, including the right hemisphere.
Studies involving individuals with various localized brain injuries support this notion. Sartori, Miozzo and Job (1993) report cases in which the ability to name and describe inanimate objects is intact while the production of animal names and descriptions is significantly impaired. A similar dissociation is seen in the ability to correctly comprehend and define words which cannot be accurately pronounced, spelled or written. For instance, Ellis, Miller and Sin (1983) describe individuals who are able to recognize and write the names of objects but cannot pronounce the names or read out loud the descriptions of those objects which they have just written. Conversely, Goodwin and Carmazza (1986) cite cases of individuals who are able to correctly pronounce and define words orally, but cannot write or type them.
The localization and identification of underlying systems which support speech production also provides support for a distributed processing model. Mandarin-speaking Broca’s aphasics have been found to be unable to correctly produce or interpret linguistic tonal nuances (Packard, 1986) while English-speaking aphasics are unable to produce and evaluate word and sentence timing. Ross, Edmondson, Seibert and Chan (1992) report that patients with inferior frontoparietal damage are unable to recognize the emotional content of spoken language. Shedlack, et al. (1991) propose that the tendency for left hemispheric dominance may arise out of a need for complex inhibitory mechanisms to moderate and coordinate the brain’s various processing units. If language processing is in fact accomplished by the activation of discrete processing units in various parts of the brain, and the level of activation of these units is modulated by excitatory and inhibitory mechanisms, then a distributed processing architecture is strongly indicated.
Dennett (1991), however, maintains that no particular brain system exists which acts as interpreter or integrator of the output from these various processing elements, but rather that it is their interaction which produces a unified output. Recent advances in the computer sciences offer support for a parallel processing model of cognition. A number of computer based neural network models have been constructed which exhibit behavioral characteristics similar to human cognitive functions. An example of this is seen in the work of Herrmann, Ruppin and Usher (1993). Their network, based on computer simulated neurons, successfully learned concepts associated with the separation of information into hierarchical semantic classes and acquired the ability to form and use episodic associations. Other neural network programs have been successful at such characteristically intelligent behaviors as speech perception, letter and word recognition and even syntactic and semantic processing.
Computer simulated neural networks can also can also be made to exhibit neuropathological behavior patterns such as paranoia and schizophrenia. While none of these computer simulations even begin to approach the complexity of human cognitive processing, they may offer an insight into the kind of architecture required to accomplish mind. Tooby and Cosmides (1995) assert: “the brain is more than a physical system: It is both a computational system and an evolved biological system.” The key word in this assertion is system. The authors suggest that the brain not a single-function whole, but an organism, an array of very specific processing units, each evolving in response to an equally specific ecologically relevant problem. And these processors do not follow any forward-looking blueprint in their ongoing developmental progress. Each processing system, at every point along its evolutionary trajectory, exists and is passed on to future generations only on the basis of its immediate survival advantage. Therefore, Tooby and Cosmides argue, the concept that the brain evolved as a holistic, general-purpose processing unit is biologically implausible.
Consciousness, the most cherished human mental capacity, has long been thought of as too complex a phenomena to be produced through neural activity alone. Recent investigations in neuroscience, however, do not support this notion (Kinsbourne, 1995). When considered as an information processing system, consciousness arises through the integration and organization of the output from preconscious representational systems. The adaptive function of such a process, Kinsbourne (1988) suggests, is context. The ability to compare the output of various sensory systems allows the individual a greater understanding of how these various representations are interrelated. Further, the ability to compare current representations with those based on previous experience and stored in memory allows the individual to transcend the “concrete present stimuli = immediate response” barrier.
Recent biological adaptations have generated improvements to this basic system allowing the individual to compare current representational information with previous representational memories and to combine them in new ways, creating awareness not only of what is happening now and what happened previously, but what might happen in the future. It is through these basic neurological activities that consciousness arises. Similar research has demonstrated that language begins not as a means of communication, but as a representational system. The same neurological architecture which allows animals to perceive the presence of predators, Bickerton suggests, provides them with the ability to communicate that presence to others. The interrelation between consciousness and symbolic communication begins when animals learn to lie.
Survival advantage accrues to those animals able not only to perceive the presence of food and recognize that other animals are present who would prevent them from obtaining that food, but can also realize that if they produce a “false” alarm call, indicating the presence of danger, the animal them from obtaining food will be tricked into flight. The transition of language from a representational to a communication system begins, says Jolly, when a species learns to mis-represent. Alexander (1989) suggests that the rapid brain growth which eventually produced Homo Sapiens’s most characteristic feature took place after the species achieved ecological dominance and that the selective pressure responsible for that growth was intra-species social competition. Given the inter-relation between social complexity and language skills, the same social pressures responsible for rapid brain growth may well have also selected for maximally efficient language processing systems.
Mandler and Johnson (1977) suggest that verbal information is encoded, structured and stored in memory in specific ways as this tends to increase the efficiency of storage and retrieval. Typically, the information received through the presentation of words, ideas and sentences is integrated into narrative stories. Depending on the nature and context of the verbal information which is to be integrated, narratives can be structured to emphasize spatial relationships, emotional content, or grammatical features. Also, narrative production is an important language processing system in that it provides the context for human social discourse and interpersonal relations. Through the use of narratives, humans have acquired the ability to build mental models of their physical and social environment and use these to consider, predict, plan and explain the events which occur within that environment. In Human Communication As Narration: Toward A Philosophy Of Reason, Values And Action (1987), Fisher states: “Far from being one code among many that a culture may utilize for endowing experience with meaning, narrative is a metacode, a human universal on the basis if which trans-cultural messages about their shared reality can be transmitted.”
In formulating his narrative paradigm, Fisher (1987) proposes that rational thought and logical decision making are based on the inherent qualities of a narrative story. The truth of a thing is based on its narrative probability (how likely a story it tells) and fidelity (how internally consistent that story is). So valuable have narratives become, that it would be difficult for the human species to exist without them. So fundamental is narrative production to human nature, Fisher (1977) suggests, the species might best be classified as “Homo Narrans”, man the narrator. Whether or not mankind’s definitive characteristic is narrative production, the importance of this activity is clearly evident given the strength of the output it produces. Readers often remember and recall their narrative constructions of textual materials rather than the text itself. That is, individuals are more likely to recall the organizing themes and salient features of a story rather than its grammatical details.
Input to this system can be further modified by learned expectations. Fischler and Goodman (1978) demonstrate that knowledge about relative sizes and spatial relationships between objects affects interpretation and even perception of verbal information related to them. Examples of this are seen in top-down processing strategies such as letter and word recognition and the interpretation of anomalous sentences. The accuracy of response and response selection time required to correctly identify incomplete or ambiguous stimuli are significantly improved through the use of context-relevant information. These two processes, whereby incoming information is integrated into a Gestalt and moderated by context-relevant expectations, help to improve the overall efficiency of the language processing system, provided of course that both systems are functioning properly and communicating interactively.
While many clinical populations experience linguistic deficits, schizophrenics exhibit a wide range of problems associated with the processing, production and interpretation of language. Hoffman and McGlashan (1993) suggest that this may result from a reduction in the overall number if pathways connecting various neural processing systems within the brain. It is suggested that neuronal decoupling results from the overpruning of synaptic connections during critical developmental periods. This phenomena would prevent the individual processing units from interacting effectively and producing a coherent output. While this does not account for the mediating effects of antipsychotic medications, it does provide a theoretical framework capable of explaining many of the phenomena associated with schizophrenia.
Of particular interest is the schizophrenic’s inability to use top-down processing strategies or form Gestalts. While many studies indicate that schizophrenics are not able to employ higher level cognitive processing, John and Hemsley (1992) find that, given sufficient time, similar results could be obtained using data driven or “bottom up” processing strategies. This is consistent with the findings reported by Hallford (1994, 1993). On a facial recognition task, Hallford finds that schizophrenics are not inferior to community controls in the number of correct responses, but require significantly longer to choose the correct response. Following up on this preliminary work, Nixon, Hallford and Tivis (1996) report that schizophrenics were not inferior to community controls on other standard neuropsychological tests, such as the Trail-Making test.
This would seem to indicate that while schizophrenics may preserve at least some kinds of cognitive skills, those involving integrative information processing are more likely to be affected. From this deficit pattern, the authors conclude that investigations which attempt to establish processing deficits in schizophrenics must consider not only the accuracy of the processing system in question but the speed at which this processing occurs as well. A second processing deficit common in schizophrenics is the tendency to exhibit poorer overall linguistic processing than community controls. Gold, Randolph, Carpenter, Goldberg and Weingerger (1992) find that schizophrenics demonstrate impairment in the learning, recall, recognition and semantic encoding of verbal information. Also, schizophrenics are found to form poorer social schemata.
These two deficit patterns, when considered together, suggest that schizophrenics would exhibit poorer narrative formation skills. Such an explanation would, Following Fisher’s narrative paradigm (1987), tell a very likely and internally consistent story. The quality of narrative formation would be further compromised by the fact that schizophrenics tend to form associations based on irrelevant and delusional information. While studies have been conducted which consider the various components of narrative construction in schizophrenics, none have addressed both the accuracy and speed of this linguistic processing system. Might schizophrenics be able to achieve linguistic parity, exercise equal social conceptualization and create narrative stories, if given sufficient time to do so? The purpose of this investigation is to begin the task of considering this important issue.
The current research project is based on the protocol designed and employed by Bransford and Franks (1971). Their work focuses on the tendency of verbal information contained in individual narrative sentences to become integrated into a more complex linguistic structure, called an idea set. They propose that once an idea set is formed, individuals will be unable to recall the exact structure and content of the individual sentences from which it is formed. To test this theory, Bransford and Franks constructed four narrative stories, each containing four basic story elements. Each story was expressed in the form of a single idea set sentence containing all four story elements. From these idea set sentences were constructed test sentences containing either one, two or three story elements.
The test sentences were then divided into two sets. The first set OLD, consisting of one, two and three element sentences from each of the four idea sets, was used to present these idea sets to the participants. The second set NEW, containing the remaining test sentences, were not presented. Participants were read one of these test sentences and then given a ten second distracter (color naming) task. At the conclusion of the distracter task, participants were asked a question about the test sentence. This procedure was followed until all test sentences in the OLD set were presented. Sentences from the same idea set were not presented consecutively to ensure that idea set formation was not simply an artifact of serial presentation.
After a five minute delay, participants were shown all the test sentences (OLD and NEW) and the idea set sentence and asked to identify which of those sentences had been previously presented. In their study, Bransford and Franks asked the following empirical questions: Would participants be able to distinguish between sentences that had been presented OLD and NEW sentences that were structurally similar and contained elements from the same idea set? Would participants be more likely to report having previously seen sentences containing more story elements from the idea set, than those sentences containing fewer elements?
The authors report that the four element Idea Set sentences are most often identified as having been previously seen even though these sentences were never actually presented. Also, three element sentences, whether previously presented or not, are more often recognized than sentences containing only one or two sentence elements. This study suggests that information from contextually related sentences, even when those sentences are not presented consecutively, are stored and processed holistically.
The current project attempts to apply a similar research protocol to schizophrenics and community controls. Some changes were made, however, to address the question of how information processing takes place in schizophrenics. Although schizophrenics have been shown to be inferior to controls in holistic and Gestalt formation (John & Hemsley, 1992; Knight, 1984), they are able to achieve similar results using bottom-up processing strategies if given enough time. Therefore, a measure of response selection time was obtained in addition to the number of Idea Set and three element NEW sentences selected. In the recognition phase of this task, when subjects are asked to indicate whether or not they have seen the sentence presented, the time required for making their response will be collected.
Also, as schizophrenics are more likely to incorporate irrelevant or incongruous information Vinogradov, King, & Huberman, 1992), the current protocol includes two idea set sentences constructed so that when their four story elements are integrated into a unified whole, an ambiguous construct will be produced. In this study, the following hypotheses were addressed: It is expected that schizophrenics will be inferior to controls with respect to holistic processing. Specifically, on the Narrative Recognition Task, schizophrenics will be less likely to indicate that the four element Idea Set sentence or the three element sentences had been seen previously. Groups will not differ on the number of identified one and two element sentences.
Because of poorer top-down processing skills, it is expected that should schizophrenics report seeing the Idea Set sentence and sentences incorporating three story elements, the response selection time required to make these identifications will be longer than those of community controls. Groups will not differ, however, on the time to respond to one and two element sentences, which do not derive from the integration of information. It is expected that community controls will be less likely to form idea sets based on incongruent sentence elements than schizophrenics. Schizophrenics will not differ in the number of congruent and incongruent constructs formed, nor on the time required for them to do so. Specifically, schizophrenics will be just as likely to indicate seeing incongruent Idea Set sentences as congruent ones. Also, while controls will tend to respond more quickly when identifying congruent Idea Set sentences, schizophrenics will not perform differentially with respect to response selection time.
Participants included 15 schizophrenics (10 males, 5 females) and 15 community controls (9 males, 6 females). Schizophrenics were recruited from both state funded and private inpatient facilities. Based on clinical chart review, all clinical patients were found to meet DSM-IV (American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, 1994) criteria for schizophrenia. Only those with a stable diagnosis, having been diagnosed, medicated and treated as a schizophrenic for at least five years, were asked to participate. Individuals receiving below normal IQ or mini-mental scores, or other chart-based indications of mental impairment were not included in this study. Additionally, only those patients between the ages of 21 and 55, who had been in an inpatient facility long enough to be stabilized, that is, no longer experiencing hallucinations or exhibiting floridly psychotic behavior, but with less than 45 days of inpatient care were asked to participate.
Information concerning the patient’s current state was obtained directly from a psychiatric treatment professional responsible for that patient. These criteria were used to insure that any deficits identified in the clinical group resulted from the relatively stable symptomology associated with schizophrenia and not from the effects of the normal process of aging, short term acute psychosis, or social deprivation resulting from long-term hospitalization. Community controls were recruited from the Oklahoma City catchment area using newspaper ads, fliers, and word-of mouth. Every effort was made to insure that community controls did not differ significantly in age or years of completed education from the clinical group as these factors would be expected to correlate with most neuropsychological measures. Because of the rather limited subject pool, subjects were not directly matched with respect to gender and ethnic membership. The gender and racial distribution of each group is presented in Table 1 below.
Gender and Ethnic Distribution By Group. Participants were excluded on the basis of medical, physical, neurological, or psychiatric problems as these might affect performance. Individuals indicating learning disabilities or poor academic performance were also excluded. Also, community controls were excluded if currently taking any medications that might interact with cognitive function. Schizophrenics were not excluded because of current medication.
All participants provided informed, written consent. All were assured that the information they provided would be numerically coded to insure confidentiality, and would not be published, released or used in any way that might harm, embarrass or affect the individual providing that information. Inpatient participants were assured that the information they provided was for research purposes only, that it would not be made available to anyone, including the clinical facility in which they resided, and that participation would not affect their treatment in any way. It was made clear that should a participant become upset, uncomfortable or otherwise concerned, he or she might withdraw from the study at any time. Further, if the participant had any questions or concerns as a result of the experimental protocol, appropriate responses would be provided by the test administrator.
The Cognitive Studies Laboratory Group Screen Packet was used to obtain basic demographic information and measures likely to be correlated with cognitive function and determine appropriateness of potential participants for further testing. This packet is included in Appendix 1. Specific questions were asked which relate to the following areas of concern: Basic Demographics. The age, race, gender and number of years of completed education of each prospective participant was obtained. These measures were important as they were to be controlled between groups. Substance Abuse. Objective measures of alcohol and substance use and abuse were gathered. These were used to evaluate appropriateness for continued participation. Individuals expressing a significant problem, or with heavy or frequent use were excused from further participation.
Physiological and Psychiatric Status. Information concerning periods of unconsciousness, overdoses, suicide attempts, illnesses, injuries or other events which might affect the proper functioning of the central nervous system were obtained. Neurologically impaired individuals and those indicating significant psychiatric problems, other than those relating to schizophrenia in the clinical group, were not tested. As overall verbal functioning and affective state might have a bearing on narrative construction, standardized tests were given to assess these issues and prevent possible confounds. Because the Shipley Vocabulary Test has been widely used with many clinical and control populations and is quickly and easily administered, it was used as a measure of overall verbal functioning. The Beck Depression Inventory and the Spielberger State Anxiety Inventory were given so that an assessment of affective state at the time of testing can be obtained.
Narrative Acquisition Task. For this task, four narrative stories were constructed. Each of these narrative stories consisted of four simple relationships between concrete objects, which are to be integrated into a complete four element whole. These narrative stories are based on the idea sets and constructed by Bransford and Franks (1971). Two of the narrative stories used here are identical to the sentences used by Bransford and Franks (1971) and two were modified so that the four individual story elements, when integrated into a whole, would form an ambiguous idea set.
This experiment employs a pseudo-random design. Individuals were assigned to groups based on the presence or absence of a clinical diagnosis of schizophrenia. The first fifteen volunteers meeting the selection criteria for either the community control or clinical group were tested. As the clinical participants were tested on the unit from which they were recruited, it was not possible for testing to take place without the experimenter being aware of the participant’s group membership. To insure that the testing protocol and environment were as similar as possible for both groups, community controls were tested at the Cognitive Studies Laboratory facility. Identical materials, instructions, and test administration procedures were employed for both groups. Dependent measures collected on this experimental protocol included accuracy and response selection time for responses to the sentences presented in the narrative recognition task.
From each narrative story, twelve sentences were constructed. One sentence was constructed which contained all four story elements, three that contained three story elements, four that contained only two story elements, and four that contained only one story element. A complete listing of the four narrative stories and the sentences which comprise them can be found Appendix 2. From these sentences were constructed the sets of sentences used in the acquisition and recognition portions of the task. Narrative Acquisition Task. For the acquisition task, six sentences were randomly selected from each narrative story; two three element sentences, two two element sentences, and two single element sentences. These 24 sentences were divided into four sets. The construction criterion for these sets was as follows: Sentences from the same narrative story can not be presented consecutively.
No more than two sentences from the same narrative story can occur in any set. Only one three element sentence from the same story can occur in any set. All sets must begin and end with sentences from different narrative stories. Given these constraints and the limited number of unique sentences generated in the four narratives, only one presentation order was possible. These sentences and their sequential ordering is presented in Appendix 2. Narrative Recognition Task. The recognition task was constructed using a combination of sentences presented in the narrative acquisition task OLD, similar sentences which were not presented in the acquisition task NEW, and of the Idea Set sentences. This task consisted of 28 sentences in all. These sentences included: the four Idea Set sentences containing all four elements of each story that were not presented in the acquisition task;
One of the two OLD sentences containing three, two and only one story element from each story which was presented in the acquisition task, and an equivalent three, two and single element NEW sentence from each narrative story that was not used in the acquisition task. These sentences are presented in Appendix 2. Dependent Measures. As indicated earlier, it was expected that community controls would be more likely to believe that they have previously seen the Idea Set sentence, and NEW sentences containing three story elements, than OLD sentences containing only one or two story elements. Schizophrenics, exhibiting decreased Gestalt formation ability, would therefore be less likely to indicate seeing the Idea Set sentence and NEW sentences which were not presented in the acquisition task than controls. Therefore, schizophrenics were expected to be more accurate at correctly recognizing OLD sentences and rejecting both the Idea Set sentence and NEW sentences. In order to test this hypothesis, three measures were needed: the number of correct OLD sentences identified, the number of Idea Set sentences correctly rejected, and the number of NEW sentences correctly rejected.
It has also been suggested that community controls would be less likely to form Idea Sets when the individual story elements form an ambiguous whole. Schizophrenics, however, are expected to be no less likely to form Idea Sets containing incongruent story elements than congruent ones. In order to test this hypothesis two measures are needed: the number of congruent Idea Set sentences correctly rejected, and 2) the number of incongruent Idea Set sentences rejected. Lastly, it is suggested that schizophrenics may be able to compensate for poorer Gestalt formation if given sufficient time in which to employ bottom up processing strategies. Thus, it is predicted that paranoid schizophrenics will exhibit longer response selection times when indicating that the Idea Set sentence and NEW three element sentences, requiring information integration, were seen previously.
Further, the time required to correctly identify OLD and reject NEW sentences is not expected to differ as a result of group membership. To test this hypothesis, mean response selection times for correctly identifying OLD and correctly rejecting Idea Set and NEW sentences will be required. Group differences will be considered as significant if an alpha level of .05 is obtained. Participants were tested individually and were paid for their participation. After providing written, informed consent, each volunteer was given the screening packet to determine eligibility for continued participation in the study. Before beginning the test protocol, each participant was given the Shipley Vocabulary Test (Shipley, 1940) to determine overall verbal ability, and the Beck Depression, and Spielberger State Anxiety Inventories as these tests are useful in determining the participant’s affective state at the time of testing.
Prior to the administration of the test protocol, participants were given instructions for a typical serial recall test. Participants were told that they would be shown four different sets of cards, each made up of a series of six cards. On each card would be a short sentence and the cards would be presented one at a time for ten seconds each. Participants were told to read each sentence out loud and to try and remember both the order in which the sentences were presented and the information contained in each sentence as they were to be tested later. They are instructed not to touch the cards at this time, but only to look at the card which is currently being presented. After all six sentences had been presented, the sentence cards were collected, shuffled, and placed face down on the table in front of the participant who is then asked to turn them over and arrange them in the order in which they were presented. The participants were also informed that later they would be asked to recall the sentences. The same protocol was followed for each of set of cards.
Participants were then presented with one of the four card sets. Cards were presented individually at the rate of ten seconds per card. The experimenter placed each card on the table before the participant so that it was clearly visible and only the information from the card that was currently being presented was visible. Participants were asked not to handle the cards during the presentation period. After all six cards were presented, they were collected by the experimenter and shuffled. The cards were then placed on the table in front of the participant, face down. Participants were instructed to arrange the cards so that they were in the same order as they were when originally presented. The process was then repeated with the remaining card sets. Both sequencing accuracy and sort time measures were collected. These measures were employed to determine if observed between group differences in narrative construction were the result of poorer acquisition and encoding rather than narrative formation abilities. It is suggested that if groups were not statistically different on either the number of cards correctly sorted, or the time required to reorganize them, then the information presented within those sentences was at least equally available for integration into Idea Sets.
Following the final serial arrangement, participants were asked to recall as many items as possible, in any order, from the sentences presented. Participants were given a maximum of five minutes in which to respond. Recall data was collected with an audio cassette recorder. The recall task was terminated when 45 seconds passed after the participant’s last response. This task was primarily employed so as to establish a uniform time interval between the narrative acquisition and recall tasks and to ensure that participants spent that time thinking about the narrative sentences. Given the inherent difficulties in interpreting the often bizarre responses provided by schizophrenic patients in a free recall test, statistical analysis of those responses would be likely to produce questionable, if not invalid results. Therefore, these data were not subjected to statistical analysis.
When the five minute interval allowed for the free recall task had elapsed, participants were given the narrative recall task. This task consisted of a set of 28 sentence cards, some presented earlier OLD and some NEW. Participants were told that they would be shown a series of sentences, some that had been seen previously and some very similar, but not identical to, sentences seen previously. Their task, therefore, was to identify which sentences were OLD and which were NEW. Sentences were presented one at a time for a period of ten seconds each. Participants were asked to read each sentence out loud. After the ten second presentation time had passed, participants were asked to indicate whether or not that exact sentence had been presented earlier in the card sort task. Response accuracy and response selection time measures were recorded for all sentences.
The purpose of this project was to assess the ability of schizophrenic patients to form Idea Sets from narrative story elements presented in non-consecutive sentences. In order to test this ability, Schizophrenics were compared with community controls. Differences in Idea Set formation were to be established on the basis of their performance on the Narrative Recognition Task. To ensure that observed group differences were, in fact, the result of real differences in the ability of these groups to form narratives, it must be clearly shown that groups were not different on a number of other variables which might be related to, or account for, differences on the Narrative Recognition task. Variables likely to affect Idea Set formation are of several types. These include measures relating to: basic demographic factors, affective state, and overall verbal processing abilities. Only when schizophrenics and community controls have been shown to be equivalent with respect to these variables, or when significant differences are found, these differences corrected for using appropriate analyses of covariance, will it be legitimate to look at the question of narrative construction.
As a negative affective state would be likely to influence an individual’s motivation and concentration, measures of depression and anxiety were collected. Groups were found to differ significantly with respect to both measures. When given the Beck Depression Inventory, schizophrenics indicated higher levels of depression than did community controls. Although groups were significantly different with respect to reported levels of depression, the levels reported in all cases were clinically insignificant according to standard scoring guidelines. Group means, ranges and standard deviations are presented in Table 3 below. Similarly, community controls reported significantly lower levels of anxiety, as measured with the Spielberger State Anxiety Inventory, than schizophrenics, at the time of testing. Again, the scores obtained by all participants represented subclinical levels of current anxiety, based on the guidelines reported by Spielberger (1983). Group means, ranges and standard deviations are also presented in Table 3 below.
Beck Depression Inventory and Spielberger State Anxiety Inventory Distributions By Group. Controls Mean Standard Deviation Min Max Correlations Correlational analyses (Pearson’s r) between the affective measures and the total number of correct responses indicated no significant relationships between either depression and response accuracy. Correlational analysis (Pearson’s r) when conducted on the entire sample (N=30) indicated no significant correlation between depression and mean sort completion time. The Spielberger State Anxiety Inventory score was significantly correlated with mean sort completion time. When considered by group, however, a significant correlation was no longer present for either controls or schizophrenics .
The Narrative Acquisition Task. As this task required participants to read and remember the information presented on each card within the four card sets, it serves as a gross measure of reading ability and verbal encoding, the number of correctly sorted cards and the time required to complete the sort were recorded for each of the four trials. As indicated earlier, this study was designed to assess the ability of schizophrenic patients to form Idea Sets from narrative story elements presented in non-consecutive sentences. The specific empirical questions asked were: Would schizophrenics be less likely than controls to identify Idea Set and three-element narrative sentences, and would groups differ with respect to their recognition of the less complete one-element sentences? Would schizophrenics respond more slowly than controls when identifying Idea Set and three-element sentences than when identifying the simpler sentences? and Would community controls be less likely to form Idea Sets based on incongruent sentence elements and take longer to do so than schizophrenics?
Sentence Complexity. In the Narrative Recognition Task, sentences were presented containing either one, two or three elements from one of the four narrative stories. If, in fact, narrative construction occurs, it is expected that individuals will be more likely to identify sentences containing more narrative elements than those containing fewer elements. To provide for maximum resolution, analysis was conducted only on the most limited (single-element) and most complete (three-element) sentences. The total number of recognized OLD and NEW sentences was determined for each level of sentence complexity. These scores were then corrected for guessing, following the methodology suggested by Nunnley (1959), and analyses were performed on the guessing-corrected scores. A repeated measures analysis of variance indicated an overall preference for three-element over one-element sentences.
Response Selection Time. A second measure expected to differentiate schizophrenics from controls in the Narrative Recognition Task was response selection time. The times required for subjects to recognize single-element, three-element and Idea Set sentences were recorded. Mean response selection times were calculated for each of these sentence types. A repeated measures analysis of variance looking only at single-element and three-element sentences revealed no significant difference in the time required for individuals to recognize three, or one element sentences, nor a group main effect on response selection time. The group by response time interaction, however, was significant. Group means, ranges and standard deviations are presented in Table 8 below.
As the response selection time for Idea Set sentences is based entirely on sentences that had not been seen previously, whereas the means reported for three-element and one-element recognitions includes both OLD and NEW sentences, this variable was considered separately. A one way analysis of variance revealed no significant group differences in response selection time for Idea Set sentences. Group means, ranges and standard deviations are presented in Table 8 below. Formation of Ambiguous Idea Sets. Of the narrative story sentences presented in the Narrative Acquisition Task, half contained information that, when combined into Idea Sets, told an ambiguous story. In the Narrative Recognition Task, the four-element Idea Set sentences, comprised of either consistent or ambiguous narrative story elements were presented.
The number of consistent and ambiguous Idea Set sentences recognized, and the mean response selection time required for each sentence type was recorded. A repeated measures analysis of variance indicated that, when corrected for guessing, consistent Idea Set sentences were more likely to be recognized by both schizophrenics and controls . A significant group effect was also found. The group by sentence coherence interaction, however, was not significant. Analysis of simple main effects revealed that schizophrenics were less likely to recognize Idea Set sentences regardless of whether those sentences were internally consistent or ambiguous. Group means, ranges and standard deviations of uncorrected raw scores are presented in Table 9 below. This study was conducted to determine whether schizophrenics were less able than community controls to form Idea Sets based on narrative story elements presented in non-consecutive sentences. It was suggested that schizophrenics would differ from controls in at least three ways:
Schizophrenics would be inferior to controls with respect to holistic processing. Specifically, schizophrenics would be less likely to indicate that the four-element Idea Set sentence or the three-element sentences had been seen previously. Groups were not expected to differ on the number of identified one and two element sentences. Because of poorer top-down processing skills, it was expected that should schizophrenics report seeing the Idea Set sentence and sentences incorporating three story elements, the response selection time required to make these identifications would be longer than those of community controls. Groups were not expected to differ, however, on the time to respond to one and two element sentences, which do not derive from the integration of information.
It was expected that community controls would be less likely to form Idea Sets based on incongruent sentence elements than schizophrenics. Schizophrenics were not expected to differ in the number of congruent and incongruent constructs formed, nor on the time required for them to do so. Specifically, schizophrenics would be just as likely to indicate seeing incongruent Idea Set sentences as congruent ones. Also, while controls would tend to respond more quickly when identifying congruent Idea Set sentences, schizophrenics were not expected to perform differentially with respect to response selection time.
The Narrative Recognition Task was specifically designed to collect data that would provide information relevant to each of these hypotheses. Though not broad enough in scope to provide a final resolution of these issues, it was hoped that the data collected here would extend the current knowledge base. Pursuant to this objective, analyses were conducted on variables relating to three specific issues relating to Idea Set Formation. These issues are sentence complexity, response selection time and internal consistency. Before these issues could be addressed, however, it was critical to establish that the two groups were comparable with respect to overall verbal functioning, as this would be expected to affect narrative functioning. As community controls and schizophrenics were equivalent in the number of years of completed education and performed with similar levels of success on the Shipley Vocabulary Test, there is no compelling reason to assume that the groups differ significantly with respect to either overall intellectual capacity or verbal functioning.
This assumption is given additional weight by the fact that groups performed similarly on the Narrative Acquisition Task. All participants were able to read all the sentences well within the allotted ten second presentation period. Also, groups achieved equivalent levels of sort accuracy, 65 percent for controls and 69 percent for schizophrenics. Though this level of accuracy would, at first, appear low, it is surprising given the similarity of the sentences being sorted across the four sentence presentation trials. Clearly, the Narrative Acquisition Task is not designed to be an accurate measure of verbal performance, but the lack of significant group differences on this task offers an indication of verbal parity. As schizophrenics have been shown to exhibit significant deficits on verbal processing tasks elsewhere , the current finding suggests that this sample of schizophrenics might perform better (i.e. more like controls) than one might expect on other measures of linguistic processing as well. Therefore, a comparison of controls and schizophrenics with respect to Idea Set formation would seem justified.
Hypothesis one predicted that schizophrenics would: 1) not differ from controls in the recognition of one-element sentences, 2) be less likely than controls to recognize three-element sentences, and 3) be less likely than controls to indicate having seen the Idea Set sentence previously. Data collected in the Narrative Recognition Task indicate, however, that schizophrenics reported recognizing fewer one-element sentences than did controls. As the probability of this finding arising as a result of chance is very close to the .05 alpha level, this difference, though statistically significant, is not strong. It is likely that future studies would fail to replicate this finding. Also, groups did not differ on the reported frequency of recognized three-element sentences. These findings are inconsistent with the predictions of hypothesis one.
A stronger argument for Idea Set formation is the tendency of individuals to believe that the four-element Idea Set sentences had been presented in the Narrative Acquisition Task. Schizophrenics recognized significantly fewer Idea Set sentences than did controls. On average, community controls reported seeing 75 percent of the Idea Set sentences while schizophrenics recognized only 33 percent of these four-element sentences. This difference does offer support the predictions of hypothesis one.
Hypothesis two predicted that schizophrenics would: not differ from community controls on the time required to recognize one-element sentences, 2) take longer than controls to recognize three-element sentences, and take longer than controls to recognize the four-element Idea set sentences. Analysis of the response selection time data collected in the Narrative Recognition Task revealed a group by sentence complexity interaction. That is, response selection time tended to decrease with sentence complexity for controls while schizophrenics responded more quickly to less complex sentences. Though response selection times for Idea Set sentences were not included in this analysis, this pattern of directionality can be seen in them as well (see table 8 for means and standard deviations). Figure 1 below illustrates this interaction, including the response selection times for idea set sentences. This pattern is consistent with the notion that schizophrenics, when processing sentences incorporating more of the Idea Set information, would be able to integrate that information using bottom-up processing strategies, but that they would require more time to do so.
Hypothesis three predicted that schizophrenics would: be more likely to recognize Idea Set sentences which incorporate ambiguous information than community controls,not demonstrate a differential recognition rate for congruent and ambiguous Idea Set sentences, and would not respond any more quickly to congruent Idea set sentences than to ambiguous ones. The analysis of response rates for congruent and ambiguous Idea Sets revealed that all participants were more likely to recognize consistent sentences than ambiguous ones, and that schizophrenics recognized fewer of both consistent and ambiguous Idea Set sentences than did controls. These findings support hypothesis three in that controls were found to recognize fewer ambiguous Idea Set sentences than schizophrenics, but fails to support the prediction that schizophrenics would not respond differentially to consistent and ambiguous sentences.
Analysis of response selection time for consistent and ambiguous Idea Set sentences demonstrated that schizophrenics took longer to recognize consistent sentences than did controls, but that groups did not differ in the response selection time required for ambiguous sentences. This finding, in addition to supporting hypothesis three, adds support to hypothesis one, in that when ambiguous Idea Set sentences are excluded, Schizophrenics take significantly longer than controls to recognize the Idea Set sentences. The overall worth of this project of resolving the issue of Idea Set formation in schizophrenics is limited at best. While demonstrating that schizophrenics are less likely to recognize Idea Set sentences than community controls, it is not clear that sentence complexity is related to recognition. Groups were not different in the number of recognized three-element sentences .
The most powerful support for inferior Idea Set formation in schizophrenics was the reaction time measures. This measure, however, is perhaps the most complicated to interpret. Virtually all of the existing neuropsychological literature would predict a generally slower response rate for schizophrenics (Prelick, Mattis, Stastny, & Teresi, 1992). Indeed, the sort times for schizophrenics on the Narrative Acquisition task were longer than those for controls. Though these response rates were not correlated in any consistent way with the response selection rates in the Narrative recognition Task, the group difference is troubling.
The issue is further complicated by the inclusion of ambiguous Idea Set sentences. Since two of the four Idea Set sentences were ambiguous, and controls were less likely to recognize inconsistent sentences, the inclusion of these sentences in this experiment may well have reduced its overall worth. This is supported by the fact that when response selection times for ambiguous Idea Set sentences were excluded, the overall pattern of response selection time and sentence complexity was changed. The integration of ambiguous information into more complex sentences may well have affected the response patterns for three-element sentences as well.
As a beginning, however, this project is of value. It, at the very least, can recommend several possible avenues for future research. Initially, the concerns relating to the inclusion of ambiguous sentences could easily be resolved by conducting two studies, one using all consistent and the other all ambiguous Idea Sets. Such a design would greatly clarify the relationship between these types of Idea Sets. Secondly, the significant group difference on the recognition of one-element sentences must be clarified. A greater number of subjects would be required to establish whether or not this finding was obtained in error. Finally, the ultimate worth of any research project is its replicability and generalizability. A true assessment of this project must be made by those who attempt to replicate it. If others achieve similar results, then perhaps the inherent problems within this study can be overlooked. If, however, those who apply the research methodology used here fail to reproduce these findings, then perhaps the author’s reservations will have been supported.
- American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Mental Disorders (4th ed.). Washington, DC: Author.
- Alexander, R. (1989). The evolution of the human psyche. In: Mellars, P. & Stringer, C. (Eds.) The Human Evolution. Princeton, NJ: Princeton University Press.
- Beck, A.T., Steer, R.A., & Garbin, M.G. (1988). Psychometric properties of the Beck Depression Inventory: twnety-five years of evaluation. Clinical Psychology Review, 8, 77-100.
- Bickerton, D. (1992). Language and Species. Chicago, IL: The University of Chicage Press.
Bower, G.H. & Morrow, D.G. (1990). Mental models in narrative comprehension. Science, 247(January 5th), 44-48.
- Bransford, J.D. & Franks, J.J. (1971). The abstraction of linguistic ideas. Cognitive Psychology, 2, 331-350.
- Chaika, E. & Lambe, R. (1985). The locus of dysfunction in schizophrenic speech. Schizophrenia Bulletin, 11, 8-15.
- Corrigan, P.W., Wallace, C.W., & Green, M.F. (1992). Deficits in social schemata in schizophrenia. Schizophrenia Research, 8, 129-135.
- Danly, M. & Shapiro, B. (1982). Speech prosody in Broca’s aphasia. Brain and Language, 16, 171-190.
- Dennett, D.C. (1991). Consciousness Explained. Boston, MA: Little Brown & Co.
- Dennett, D.C. & Kinnsbourne, M. (1992). Time and the observer: the where and when of consciousness in the brain. Behavioral Brain Sciences, 15, 183-247.
- Ellis, A.W., Miller, D., & Sin, G. (1983). Wernicke’s aphasia and normal language processing: a case study in cognitive neuropsychology. Cognition, 15, 111-114.
- Fisher, W.R. (1987). Human Communication as Narration: Toward a Philosophy of Reason, Value, and Action. Columbis, SC: University of South Carolina Press.
- Fishler, I. & Bloom, P.A. (1985). Effects of constraint and validity of sentence context on lexical decisions. Memory and Cognition, 13, 128-139.
- Fishler, I. & Goodman, R. (1978). Semantic and episodic constraints on the use of surface-order cues in the comprehension of locative sentences. Journal of Psycholinguistic Research, 7(2), 121-134.
- Fried, I., Ojemann, G.A., & Fetz, E.E. (1981). Language related potentials specific to human cortex. Science, 212, 353-356.
- Garrett, M. (1995). The structure of language processing: neuropsychological evidence. In: Gazzaniga, M. (Ed.), The Cognitive Neuroscience. Cambridge, MA: MIT Press.
- Gold, J.M., Randolf, C., Carpenter, C.J., Goldberg, T.E., & Weinberger, D.R. (1992). Forms of memory failure in schizophrenia. Journal of Abnormal Psychology, 101(3), 487-494.
- Goldman-Rakic, P.S. (1988). Changing concepts of cortical connectivity:parallel distributed cortical networks. In: Rakic, P. & Singer, W. (Eds.), Neurobiology of Neocortex (pp. 177-202). New York, NY: J Wiley.
- Goodman, R.A. & Caramazza, A. (1986). Aspects of the spelling process: Evidence from a case of acquired dysgraphia. Language and Cognitive Processes, 1(4), 263-296.
- Hallford, H.G. (1994, April). Is a face a face? In: Nixon, S.J. (Chair), Alcoholism, Schizophrenia, and Dual Diagnosis. Symposium conducted at the Southwestern Psychological Association Annual Conference, Tulsa, OK.
- Hallford, H.G. (1993). Facial recognition skills in schizophrenic and normal populations. Paper presented at the meeting of the Oklahoma Psychological Society Annual Conference, Stillwater, OK.
- Herrmann, M., Ruppin, E., & Usher, M. (1993). A neural model of the dynamic activation of memory. Biological Cybernetics, 68, 455-463.
- Hemsley, D.R. (1987). An experimental psychological model for schizophrenia. In: Hafner, R., Gattaz, W.F., & Janzarik, W. (Eds.), Search for the causes of schizophrenia. New York, NY: Springer.
- Hillis, A.E. & Caramazza, A. (1991). Mechanisms for accessing lexical representations for output: evidence from a case with category-specific semantic deficit. Brain and Cognition, 40, 106-144.
- Hoffman, R.E.& Dobscha, S.K. (1989). Cortical pruning and the development of schizophrenia: a computer model. Schizophrenia Bulletin, 15, 477-490.
Hoffman, R.E. & McGlashan, T.H. (1993). Parallel distributed processing and the emergence of schizophrenic symptoms. Schizophrenia Bulletin, 19(1), 119-140.
- Howard, D., Patterson, K., Wise, R., Brown, W.D., Friston, K., Weller, C., & Frackowiak, R. (1992). The cortical localization of the lexicons. Brain, 115, 1769-1782.
- Huttenlocher, P.R. (1979). Synaptic density in the human frontal cortex: developmental changes and effects of aging. Brain Research, 163, 195-205.
- Hyde, T.M., Zeigler, J.C., & Weinberger, D.R. (1992). Psychiatric disturbance in metachromatic leukodystrophy: insight into the neurobiology of psychosis. Archives of Neurology, 49, 401-406.
- Johnson-Laird, P.N. (1983). Mental Models. Cambridge, MA: Harvard University Press.
- John, C.H.& Hemsley, D.R. (1992). Gestalt perception in schizophrenia. European Archives of Psychiatry and Clinical Neuroscience, 241, 215-221.
- Jolly, A. (1996). Primate communication, lies and ideas. In: Lock, A. & Peters, C.R. (Eds.), Handbook of Human Symbolic Evolution. New York, NY: Clarendon Press.
- Kinsbourne, M. (1988). Integrated field theory of consciousness. In: Marcel, A.J. & Bisiach, M. (Eds.), Consciousness in Contemporary Science. New York, NY: Erlbaum.
- Knight, R.A. (1984). Converging models of cognitive deficit in schizophrenia. In: Spaulding, W.E. & Cole, J.E. (Eds.), Theories of Schizophrenia and Psychosis. London, NB: University of Nebraska Press.
- Kutas, M. & Hillyard, S.A. (1980). Reading senseless sentences: brain potentials reflect semantic incongruity. Science, 207, 203-205.
- Lock, A. & Symes, K. (1996). Social relations, communication and cognition. In: Lock, A. & Peters, C.R. (Eds.), Handbook of Human Symbolic Evolution, New York, NY: Clarendon Press.
- Mandler, J.M. & Johnson, N.S. (1977). Rememberance of things parced: story structure and recall. Cognitive Psychology, 9, 111-151.
- McClelland, J.L. & Elman, L.J. (1986). The trace model of speech perception. Cognitive Psychology, 18, 1-86.
- Mello, N.K. (1989). Some implications of motivated performance for neuropsychological assessments. In: (Parsons, O.A., Butters, N., & Nathan, P.E. (Eds.), Neuropsychology of Alcoholism: Implications for Diagnosis and Treatment. New York, NY: The Guilford Press.
- Mesualm, M-M. (1990). Large-scale neurocognitive networks and distributed processing for attention, language, and memory, Annals of Neurology, 28(5, November), 597-613.
- Neisser, U. (1976). Cognition and Reality. San Francisco, CA: Freeman Press.
- Nixon, S.J. Hallford, H.G., & Tivis, R. (1996).
Nixon, S.J. Tivis, R., Varner, J.L., & Rohrbaugh, J. (In press). Neurophysiological efficiency in male and female alcoholics.
- Nunnally, J.C. (1959). Tests and Measurements: Assessment and Prediction. New York, NY: McGraw Hill.
- Oades, R.D., Blunk, D., & Eggers, C. (1992). Paranoid schizophrenics may not use irrelevent signals: the use of measures of blocking and of urinary dopamine. Acta Paedopsychiatrica, 55, 183-184.
- Packard, J. (1986). Tone production deficits in nonfluent aphasic Chinese speech. Brain and Language, 29, 212-296.
- Patterson, K.E. (1986). Lexical but nonsemantic spelling. Cognitive Neuropsychology, 3, 341-367.
- Penrose, R. (1994). Shadows of the Mind: A Search for the Missing Science of Consciousness. New York, NY: Oxford University Press.
- Perlick, D., Mattis, S., Stastny, O., & Teresi, J. (1992). Neurophysiological discriminators of long-term inpatient or outpatient status in chronic schizophrenia. Journal of Neuropsychiatry, 4, 428-434.
- Rakic, P., Boutgeois, J.P., Eckenhoff, M.F,. Zecevic, N., & Goldman-Rakic, P.S. (1986). Concurrent overproduction of synapses in diverse regions of the primate cerebral cortex. Science, 232, 232-235.
- Rehak, A., Kaplin, J.A., Weylman, S.T., Kelley, B., Brownell, H.H., & Gardner, H. (1992). Story processing in right-hemisphere brain-damaged patients. Brain and Language, 42, 320-336.
Ross, E., Edmondson, J., Seibert, G., & Chan, J-L. (1992). Affective exploitation of tone in Taiwanese: an acoustical study of tone latitude. Journal of Phonetics, 20, 441-456.
- Rumelhart, D.E. (1977). Toward an interactive model of reading. In: Dornic, S. (Ed.), Attention and Performance VI. Hillside, NJ: Erlbaum.
- Sartori, G., Miozzo, M., & Job, R. (1993). Category-specific naming imparement? Yes. Quarterly Journal of Experimental Psychology, 46(A), 489-504.
- Searle, J.R. (1992). The Rediscovery of Mind. Cambridge, MA: MIT Press.
- Shedlack, K.J., Hunter, R., Wyper, D., McLuskie, R., Fink, G., & Goodwin, M.G. (1991). The pattern of cerebral activity underlying verbal fluency shown by split-dose single photon emission tomography (SPET of SPSCT) in normal volunteers. Psychological Medicine, 21, 687-696.
- Shipley, W.C. (1940). A self-administering scale for measuring intellectual impairment and deterioration. The Journal of Psychology, 9, 371-377.
- Simpson, G.B., Peterson, R.R., Casteel, M.A., & Burgess, C. (1989). Lexical and sentence context effects in word recognition. Journal of Experimental Psychology: Learning, Memory and Cognition, 15, 541-549.
- Spielberger, C.D. (1983). Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press.
Thorndyke, P.W. (1977). Cognitive structures in comprehension and memory of narrative discourse. Cognitive Psychology, 9, 77-110.
- Tooby, J. & Cosmides, L. (1995). Mapping the evolved functional organization of mind and brain. In: Gazzaniga, M. (Ed.), The Cognitive Neuroscience. Cambridge, MA: MIT Press.
- Vinogradov, S., King, R.J., & Huberman, B.A. (1992). An associationist model of the paranoid process: application of phase transitions in spreading activation networks. Psychiatry, 55, 79-94.
- Warren, R.M. & Warren, R.P. (1970). Auditory illusions and confusions. Scientific American, 223, 393-395.
- Weinberger, D.R. (1987). Implications of normal brain development for the pathogenesis of schizophrenia. Archives of General Psychiatry, 44, 660-669.
- Weintraub, S.W., Mesualm, M-M, & Kramer, L. (1981). Disturbances in prosody: a right hemisphere contribution to language. Archives of Neurology, 38, 742-744.
- Zipser, D. & Anderson, R.A. (1988). A back propagation programmed network that simulates response properties of a subset of posterior parietal neurons. Nature, 311, 679-684.