Holist Theories

Black Belt Memory

10x Your Memory Power

Get Instant Access

Unlike narrow localizationist theories, there is no one holist model that has emerged as the major example of this class of theories. However, several lines of evidence are adduced as evidence for holist theories, and all holist theories suffer from similar inadequacies in accounting for certain empirical findings.

The first line of evidence supporting holist theories consists of the ubiquity of general factors in accounting for the performance of aphasic patients. For instance, factor analyses of the performances of groups of patients both on general aphasia tests and on tests of specific language abilities almost always result in first eigenvectors (usually accounting for more than half of the variance in performance) that are approximately equally weighted for most of the subtests used to test the population. Such vectors are usually taken to reflect disruption of a single factor that affects performance on all measures, such as a limited amount of mental resources available for psycholinguistic computations. The existence of such factors would be the immediate consequence of a system in which functions were disruptable by lesions in a variety of locations, and they have therefore been widely taken as evidence for a distributed basis for language functions. A second finding supporting holist theories is the frequent observation of so-called "graceful degradation" of performance within specific language domains after brain damage. An example of such degradation is the strong tendency of certain dyslexic patients to read irregularly spelled words according to a regularization strategy (e.g., "pint" is read with a short "i"), a tendency that is inversely proportional to the frequency of the word. Graceful degradation reflects the preservation of the simplest (in many cases, the most commonly occurring) aspects of language processing after brain damage. Modern work with parallel distributed processing models, which provide formal models of holist concepts, indicates that such patterns of performance can arise following focal lesions in systems in which information is represented and processed in massively parallel, distributed forms. A third source of empirical support for holist theories comes from the finding of an effect of lesion size on the overall severity of functional impairments in several language spheres. This would follow from the principle of mass action. Therefore, these results therefore are consistent with some form of holism in the neural basis for linguistic representations and processes.

Against the complete adequacy of any holist model is the finding that multiple individual language deficits arise in patients with small perisylvian lesions, often in complementary functional spheres. For instance, studies of acquired dyslexia have documented patients who cannot read by a whole-word route (i.e., by using the entire form of a written word to gain access to the mental representation of that word) and others who cannot read by the application of spelling-sound correspondences at the letter and grapheme level. The existence of these isolated complementary deficits in different single cases indicates that at least one abnormal performance cannot result from the relative complexity of processing required by one of these tasks. Double dissociations of this sort abound in the contemporary psycholinguistic aphasiological literature. They indicate that the mode of organization of language in the brain must be one that allows focal lesions to disrupt specific aspects of psycholinguistic processing, not simply a mode of organization that produces complexity effects and degrades gracefully. Although some selective disruptions of function can occur when "lesions" are produced in simulated language processing systems that operate in parallel and distributed fashion, to date no mechanism of lesioning a distributed neural system has been shown to produce the range of specific patterns of language breakdown observed in patients.

B. Localizationist Theories

Although many localizationist models exist, the con-nectionist model of language representation and processing in the brain, revived by Norman Geschwind and colleagues in the 1960s and 1970s, is probably the best known localizationist model of the functional neuroanatomy of language, at least in medical circles in North America. This model is based on observations of aphasic patients and the interpretation of those observations that were first made more than a century ago.

Figure 3 represents the basic connectionist model of auditory-oral language processing and its relation to areas within the dominant perisylvian cortex. This model postulates three basic "centers" for language processing, all in cerebral cortex. The first (Fig. 3A), located in Wernicke's area, stores the permanent representations for the sounds of words (what psycholinguists would now call a ''phonological lexicon''). The second (Fig. 3M), located in Broca's area, houses the mechanisms responsible for planning and programming speech. The third (Fig. 3C), diffusely localized in cortex in the 19th-century models, stores the representations of concepts. A major innovation proposed by Geschwind is in the location of one aspect of the concept center. Geschwind proposed that the inferior parietal lobule—the supramarginal and angular gyri—is the location at which the fibers projecting from somesthetic, visual, and auditory association cortices all converge and that as a consequence of this

Figure 3 The classical connectionist model. A represents the auditory center for the long-term storage of word sounds. M represents the motor center for speech planning, and C represents the concept center. Information flow is indicated by arrows. The location of these centers in the brain is described in the text.

Figure 3 The classical connectionist model. A represents the auditory center for the long-term storage of word sounds. M represents the motor center for speech planning, and C represents the concept center. Information flow is indicated by arrows. The location of these centers in the brain is described in the text.

convergence, associations between word sounds and the sensory properties of objects can be established in this area. Geschwind argued that these associations are critical aspects of the meanings of words and that their establishment is a prerequisite of the ability to name objects.

Language processing in this model involves the activation of linguistic representations in these cortical centers and the transfer of these representations from one center to another, largely via white matter tracts. For instance, in auditory comprehension, the representations of the sound patterns of words are accessed in Wernicke's area following auditory presentation of language stimuli. These auditory representations of the sounds of words in turn evoke the concepts associated with words in the ''concept center.'' Accessing the phonological representation of words and the subsequent concepts associated with these representations constitutes the function of comprehension of auditory language. In spoken language production, concepts access the phonological representations of words in Wernicke's area, which are then transmitted to the motor programming areas for speech in Broca's area. In most versions of this model, the proper execution of the speech act also depends on Broca's area receiving input directly from the concept center. Repetition, reading, and writing are modeled as involving similar sequences of activation of centers via connections.

Recently, these aphasic syndromes have been related to the brain using a series of neuroimaging techniques—first T99 scanning and then computed tomography MRI, and PET. All have confirmed the relationship of the major syndromes to lesion locations. Broca's aphasia is associated with anterior lesions; Wernicke's aphasia is associated with posterior lesions, centered in the temporal-parietal juncture; pure motor deficits of speech are associated with subcortical lesions; pure word deafness is associated with lesions in the auditory association areas and surrounding white matter tracts, often bilaterally; transcortical motor and transcortical sensory aphasia are associated with watershed infarcts between the anterior and middle cerebral arteries and middle and posterior cerebral arteries; and conduction aphasia is associated with smaller lesions that often appear to affect the arcuate fasciculus. However, despite these general correlations, the classical aphasic syndromes are not as well correlated with lesion sites as the theory claims they should be. Virtually all studies exclude many types of lesions, such as various sorts of tumors, degenerative diseases, and others.

The classical syndromes are best related to lesion sites in cases of rapidly developing lesions, such as stroke. Even in these types of lesions, the syndromes are never applied to acute and subacute phases of the illness and, in the chronic phase of diseases such as stroke, between 15 and 40% of patients have lesions that are not predictable from their syndromes.

Lesion-deficit correlations have been studied in patients with more specific functional impairments than are captured by the classic aphasic syndromes (e.g., semantic memory, whole-word writing and the conversion of sounds to their corresponding orthographic units, short-term memory, and word comprehension). For the most part, these studies have involved relatively small numbers of subjects because of the difficulty in obtaining large numbers of subjects with specific deficits. For instance, one study of disorders affecting semantic memory in stroke patients could only identify three patients with a selective deficit in this function from the many patients that were screened; many other patients had problems in naming objects or in matching spoken words to pictures but not with semantic memory.

These more focused studies have provided evidence for localization of function, but the picture that emerges is complex. The localizations found have often not been consistent with the classical connec-tionist model. For instance, semantic memory (word meaning) appears to be disrupted after temporal, not inferior parietal, damage, and auditory-verbal short-term memory appears to be disrupted after parietal, not temporal, lesions.

An important aspect of the database relating specific language functional deficits to lesions is the finding that individual components ofthe language processing system can be either affected or spared following lesions in particular parts of the perisylvian association cortex. This variability in the effects of lesions at particular sites ranges across all areas of the perisyl-vian association cortex and is true of components of the language processing system responsible for activating any ofthe linguistic representations described in Section I (i.e., lexical, morphological, and sentential representations). For instance, most studies have reported phoneme discrimination deficits after both anterior and posterior lesions, in both real words and in computer-synthesized stimuli. At the same time, some studies have found that most lesions producing these impairments occur with posterior lesions. Similar individual variability has been documented in the localization of the lesions responsible for the comprehension of single words, the production of morpholo gical forms, aspects of sentence comprehension, the production of function words in sentence production, and the production of phonemic errors in spontaneous speech, picture naming, repetition, and reading, in some cases with "central tendencies" toward deficits following lesions in specific locations. This pattern has led some rsearchers to view the entire perisylvian cortex as a neural net that supports all aspects of language processing, with some degree of specialization of parts of this net for specific functions. Models of this type cannot account for severe impairments of a single langauge operation following small lesions in different parts of this region in different individuals, however.

The finding of variability in the effects of lesions on language functions also emerges from electrocortical stimulation studies, which have studied phoneme discrimination, picture naming, sentence comprehension, and other language functions. Each of these language tasks was most likely to be interrupted with stimulation in particular regions of the perisylvian cortex across all patients, but considerable variation in sites associated with interruption of each task was also noted. For instance, phoneme discrimination was interrupted by stimulation throughout the entire central region of the perisylvian association cortex in approximately 80% of cases and in sites within the language zone that are further removed from the Sylvian fissure (such as the more dorsal regions of the inferior parietal lobule) in the remaining 20% of cases.

This variability suggests that language operations are localized in small parts of the perisylvian association cortex, but that the areas in which they are localized are different in different people. The fact that lesions restricted to specific parts of the perisylvian cortex are each associated with all levels of performance of a language processing operation, from normal performance to the worst performances seen in aphasic patients, implies that for some individuals a lesion in each of these lobar regions does not affect the operation at all, whereas for others some or all of the operation is impaired by a lesion in the same region. It is difficult to understand how this could result from anything other than individual variability in the premorbid location of these operations.

On the other hand, to the extent that specific lesions tend to be associated with certain deficits, as may be the case for some impairments, these studies also support the view that there are central tendencies in the localization of these language processing components within the perisylvian cortex. A reasonable conjecture is that the functional neuroanatomy of language in the perisylvian association cortex has three features: (i) localization of language functions in individuals, (ii) tendencies for functions to be localized in specific portions of the perisylvian association cortex, and (iii) at least 20% of normal adults (and often many more) showing significant deviations from these central localizationist tendencies for each language processing component.

In the past 10 years, a considerable number of studies of the neural correlates of language processing have been published that employed PET, fMRI, and other "activation" techniques. As noted previously, some of these studies have led to the appreciation of possible roles for brain regions outside the perisylvian cortex in carrying out language operations. These studies also provide evidence about the organization of the perisylvian cortex for language. Many studies claim to find evidence for the localization of particular language representations, such as the phonological forms of words used in speech recognition, or language operations, such as transforming letters into their corresponding sounds, in particular parts of this cortex. Some of these localizations depend on specific tasks; for instance, the exact part of the perisylvian cortex in which blood flow changes during tasks that require comparison of sequences of phonemes has varied considerably in different studies in which different tasks were used. In other cases, there is more consensus about the areas that increase their vascular responses as a function of particular operations. Verbal rehearsal, which appears to involve Broca's area and adjacent parts of the frontal lobe, and maintenance of phonological representations in verbal short-term memory, which seems to activate the inferior parietal lobe, are examples of such functions; however, even here, different studies show some variation in the areas activated. Very few studies have examined patterns of vascular responsivity in individual subjects, which will be necessary to address the issue of variability. Much more work using these techniques can be expected, with consequent increases in the database relevant to our understanding of these issues.

Was this article helpful?

0 0
Breaking Bulimia

Breaking Bulimia

We have all been there: turning to the refrigerator if feeling lonely or bored or indulging in seconds or thirds if strained. But if you suffer from bulimia, the from time to time urge to overeat is more like an obsession.

Get My Free Ebook

Post a comment