Basic Studies: The long-term goal of these studies is to understand the neural mechanisms underlying cognitive processes. To this end, we have been examining the functional areas that comprise association cortex in macaque monkeys and exploring their interconnections by the use of anatomical tracing techniques in combination with physiological recording of neural activity. Parallel brain imaging studies in humans are aimed at identifying the multiple areas in the human cortex that have been differentiated in the monkey and delineating the neural systems that are engaged during perception, attention, learning, and memory.
Parallel processing within the occipitotemporal pathway for object vision.
Our earlier behavioral work had demonstrated the existence of two cortical visual systems in monkeys: an occipitotemporal pathway for the visual recognition of objects and an occipitoparietal pathway for the appreciation of the spatial relationships among objects. Further, our anatomical studies had indicated that area V4 plays a crucial role in relaying information from the early stations in the occipitotemporal pathway to the higher-order area in the inferior temporal cortex. To test this, we studied the effects of V4 lesions affecting a single quadrant of the visual field in monkeys trained to fixate; the other quadrants served as within-animal controls. We found that color, orientation, pattern, and shape perception are all impaired in the quadrant affected by the lesion, but motion perception is not, consistent with the anatomy, However, there remains some residual object vision capacity in the monkeys with V4 lesions. Correspondingly, inferior temporal neurons continue to respond to stimuli in the affected quadrant, suggesting that there must be an alternate anatomical route into the temporal cortex. Our recent work has demonstrated a pathway from V2 directly to area TEO (bypassing V4) and thence to area TE in the temporal cortex. This newly discovered pathway suggests that V4 and TEO may operate in parallel for some visual functions, a possibility now being tested in studies of the effects of TEO lesions.
Bottom-up versus top-down mechanisms in vision.
Our anatomical work has shown that all feedforward projections from lower-order to higher-order visual areas in cortex are reciprocated by feedback projections from higher-order to lower-order ones. Although feedforward projections are obligatory for the subsequent activation of higher-order areas, the role of feedback projections remains obscure. To investigate the possible contribution of these projections to visual perception, we recorded single-cell activity from area V3 in fixating monkeys as they viewed dynamic stimuli that "fill inperceptually but not physically. We found that cells in V3 respond to such stimuli with climbing activity, which corresponds in time to the time it takes human observers to report filling-in. This activity could reflect either intrinsic network properties within V3 itself or feedback projections from a higher-order area. To distinguish between these two possibilities, future studies will examine the responses of V3 cells before and after pharmacological inactivation of V4. Functional MRI studies are also being planned to examine activations in human visual cortex during perceptual filling-in.
Functional specialization in ventral extrastriate cortex of humans.
Previously, we demonstrated with positron emission tomography (PET) the existence in humans, as in monkeys, of two functionally specialized visual cortical pathways: an occipitotemporal pathway, or 'ventral stream', for object vision, and an occipitoparietal pathway, or 'dorsal stream', for spatial vision. With functional magnetic resonance imaging (fMRI) it is possible to identify the precise location of cortical areas activated in individual subjects, as compared to the group average locations identified by PET studies. The cortical areas identified by fMRI in the ventral object vision pathway were consistently located in the same brain areas on the fusiform gyrus and within the occipitotemporal sulcus that we had identified with PET, but they demonstrated smaller spatial extent, had discrete borders, and demonstrated some individual variability in their locations relative to sulcal anatomy.
Evidence from previous, separate PET studies by us and others suggested that cortical areas for perceiving color and object identity are dissociable. To test whether such a dissociation exists, we conducted an fMRI study of selective attention to the color or identity of color-washed faces. Results revealed a region in the collateral sulcus, medial to the region activated by face and object perception, that was activated by selective attention to color, demonstrating that functional specialization exists within the ventral object vision pathway and that selective attention can modulate activity of different areas within this pathway independently. In a second study of functional specialization, we are testing the hypothesis that the pathway activated by face perception contains regions that participate in both face and object perception and regions that are specialized for face perception alone. Preliminary results indicate the existence of small cortical regions specialized for face perception as compared to perception of other classes of objects, in this case houses, that, like faces, contain a large number of exemplars that we routinely perceive as unique.
Working memory and frontal lobe participation in visual processing.
Working memory refers to processes that maintain an active representation of information needed for ongoing information processing. PET and fMRI studies tested the hypothesis that visual working memory is mediated by prefrontal cortical areas, and, further, that different prefrontal areas are specialized for different aspects of visual working memory. In a PET study of face working memory, we found that participation of right and left prefrontal areas varies depending on the retention delay interval, with right prefrontal areas mediating retention over brief intervals and left prefrontal areas mediating retention over longer intervals, suggesting that the working memory representations of faces in right and left prefrontal areas are encoded differently. A PET study of object and spatial working memory demonstrated that different regions in prefrontal cortex were associated with these types of visual working memory, indicating separate frontal projection zones for the ventral object vision and dorsal spatial vision streams.
With the temporal resolution afforded by fMRI, it has been possible to distinguish the functional roles played in face working memory by posterior, visual extrastriate regions and prefrontal regions. Multiple regression analysis enabled us to decompose a working memory task into three functions, namely a nonselective perceptual response to visual stimuli, a selective perceptual response to faces, and sustained activity during memory delays. We found that the response in posterior, ventral occipital areas was dominated by a nonselective, perceptual response to visual stimuli. More anterior, ventral temporal extrastriate regions demonstrated a more selective response to faces than to non-face control stimuli, and also demonstrated a small, but significant sustained response over memory intervals. Three distinct prefrontal regions were identified that all demonstrated greater levels of sustained activity during memory delays. Moreover, the relative contribution of perception-related and memory-related responses differed significantly for these three regions, suggesting they play different functional roles in working memory. Our results are the first direct demonstration of memory-related sustained activity in discrete regions of human prefrontal cortex and are also the first demonstration of multiple, functionally-specialized prefrontal areas that participate in working memory function. Moreover, activity related to the different components of the working memory task was distributed across multiple extrastriate and prefrontal regions, suggesting that the cognitive operations associated with these components may not be discretely localized to single regions but, rather, are realized by the concerted activity of multiple regions.
Long-term episodic memory for faces.
Neural systems that participate in the encoding and retrieval of new long-term memory for faces were investigated in a PET study, testing the hypothesis that the hippocampus is involved, especially in the encoding of new long-term memories. Non-overlapping areas of activation were associated with face memory encoding and recognition. The right hippocampus and parahippocampal gyrus were activated during encoding but not during recognition testing, supporting the hypothesis. A dissociation of neocortical systems was also found. Whereas encoding also activated left prefrontal cortex, recognition activated right prefrontal cortex, as well as left anterior cingulate, right parietal, and right occipital areas and deactivated left anterior temporal cortex. In a PET study of healthy aging and memory, older subjects showed no significant activation in the right hippocampus or left prefrontal cortex during encoding, but did show right prefrontal activation during recognition, suggesting that normal, age-related impairments of memory may be due to a failure to encode the stimuli adequately.
The organization of stored knowledge about objects.
We used PET to determine where in the brain information is stored and how this information is organized. The results indicated that knowledge about concrete objects is represented in the brain as a network of discrete cortical areas in which the attributes that define an object are stored near the regions that mediate perception of those attributes. These findings provide direct evidence that the organization of stored information may parallel the organization of sensory and motor systems in the human brain. In a related series of PET studies, we showed that object knowledge is activated automatically during object identification and naming. In addition, we demonstrated that the regions activated are dependent, in part, on the intrinsic properties of the object to be identified. These studies provide additional insight into the cortical organization of information storage and the systems that mediate retrieval of this information.
Modulation of neural activity during learning.
Funtional MRI was used to explore changes in neural activity associated with a specific form of learning called repetition priming. For example, we can name an object faster the second time we see it relative to the first time it is seen. This form of learning is of considerable interest because it is preserved in patients with amnesia due to damage to the medial temporal lobes, and thus is not dependent on the functioning of this region of the brain. Repeated presentation of a series of object pictures for naming was associated with faster naming times, and decreased neural responses throughout all regions of the neocortex that are normally active during object naming. This finding is consistent with neurophysiological studies of monkeys showing decreased neuronal activity as objects become increasingly familiar to the animal. Thus, increased processing efficiency may be mediated by decreased neural activity throughout a specific processing system.
Neural systems subserving perceptual and motor skill learning.
Monkeys and humans with amnesia due to medial temporal lobe damage are also able to learn and retain certain types of perceptual and motor skills. Moreover, psychophysical data in normal subjects indicate that such learning involves neural mechanisms located in very early stages of cortical processing, possibly at the level of primary visual (V1) and motor (M1) cortex. To test this possibility, we used fMRI to measure hemodynamic changes associated with the acquisition of a simple motor skill, namely, a sequence of finger-to-thumb opposition movements. Subjects practiced this task daily over a period of 4-6 weeks, which led to a significant improvement in the speed at which the task could be performed accurately. Coincident with this improvement, there was an increase in the area of cortical activity within M1, which was highly specific for the sequence and not the component movements, suggesting that practice recruits M1 units into the representation of the learned sequence. To test whether similar neural mechanisms subserve skill learning throughout the cortex, we have initiated an investigation of perceptual learning of a simple visual detection task; parallel experiments are being conducted in humans with fMRI and in monkeys with single-cell recording.
Magnetic resonance imaging of perfusion in brain.
Many of the fMRI studies described above have successfully used blood oxygenation related contrast changes as a measure of brain activity. Nevertheless, a better understanding of the nature and origins of the signal changes seen with the technique is essential to fully utilize this technology. To this end, we have been involved in experiments that seek to tease apart the complicated interplay of contributions that make up the signal changes that are seen in MRI data during neuronal stimulation. We have begun to develop an animal model of blood oxygenation dependent image contrast accompanying neurological stimulation of visual cortex. For cross-validation purposes, the animals are also being mapped using another neuroimaging modality, optical imaging, in collaboration with Georgetown University. In addition to animal model preparations, biophysical modeling in human subjects is also being undertaken. The ability to obtain quantitative cerebral perfusion measurements in humans or animals will further improve the understanding of the technique, and may in itself provide a way of mapping brain function through local perfusion changes.
Development of software for image analysis.
Functional MRI and PET studies generate huge data sets that require sophisticated methods of analysis. We have developed an integrated analysis platform for UNIX workstations that we have named the NIH Functional Imaging Data Analysis Platform (NIH FIDAP). The platform consists of programs, written in C, that are accessed using a command line interface and a C-shell script. Analysis consists of a sequence of steps. Image preprocessing includes image reconstruction from and realignment to correct for between-scan head movement. Analysis of significant changes associated with experimental manipulations includes ANOVA, ANCOVA, and multiple regression. The latest addition to analysis programs, multiple regression, allows us to analyze fMRI data from multiple runs, factor out between-run mean shifts and within-run linear drifts, and decompose the experimental manipulations, that is cognitive tasks, into different functions and identify changes in signal associated with each function. Statistical significance is determined by the spatial extent of the area of activation, defined as the set of contiguous voxels that exceed a threshold statistical value. Results are output as images that can be used for graphic display or can be used for further analysis. We have made NIH FIDAP available to other fMRI research teams at NIH, and it is currently being used by multiple laboratories in NIMH and other institutes.
Clinical Studies: In addition to basic experiments in monkeys and humans investigating the neural mechanisms underlying normal perception, attention, learning, and memory, studies have been undertaken to investigate how these cognitive processes go awry in clinical populations.
HIV-related neurobehavioral dysfunction.
In collaboration with investigators at the Henry M. Jackson Foundation, we have continued our studies of the neurobehavioral consequences of Human Immunodeficiency Virus infection (HIV). This longitudinal investigation has documented subtle psychomotor and cognitive slowing, impaired motor-skill learning, and other deficits consistent with involvement of subcortical regions in a subgroup of HIV-infected individuals. These results suggest that subtle neuropsychological impairment can occur during the relatively early stages of infection. This conclusion has been strengthened by findings of a relationship in HIV+ subjects between reaction time and biological parameters, <°ng: 1) concentrations of an endogenous neurotoxin, quinolinic acid, in the cerebral spinal fluid (CSF); 2) the presence of HIV in CSF; and 3) MRI abnormalities.
Cognition in patients with obsessive compulsive disorder.
Two studies of cognition in patients with obsessive compulsive disorder (OCD) were completed. Contrary to recent claims of memory deficits and frontal lobe-related impairments, it was found that patients with OCD performed normally on tests of incidental memory and working memory. However, the working memory task was completed at an abnormally slow rate, suggesting a problem with resource allocation or with the regulation of attention. Additional studies suggested that the attention disorder in OCD may be characterized by an excessively narrow, or hypervigilant, focus. Heightened arousal or hypervigilance may provide a useful framework for understanding cognitive dysfunction in OCD.
Familial-genetic influences on cognitive (attentional) behavior in persons with absence epilepsy and schizophrenia
We are in the course of a large-scale study of probands and first-degree relatives of patients with absence epilepsy. This work has been conducted in collaboration with investigators in seizure clinics in Jerusalem and at the Montreal Neurological Institute. The study has involved the intensive neuropsychological assessment of approximately 75 families that include a proband with a childhood-onset seizure disorder (absence-type disorders, in addition to juvenile myoclonic epilepsy) as well as seizure controls (partial seizures) and normal control families. We hope to test the proposition (suggested by preliminary results) that certain subclinical behavioral traits found in families that contain a proband with seizures are inherited in the maternal line.
The use of event-related brain potentials (ERP) to assay cognitive deficits associated with neuropsychiatric disorders
The ERP studies have illuminated seasonal changes in information processing that differ between men and women. Moreover, these studies have characterized persistent changes in attention-related ERP components (N200, P300) in persons who have sustained, and presumably recovered from, the diffuse frontal-temporal lobe damage found in closed head injuries. These latter studies serve, as well, to provide brain-injury controls for the ERP investigations of persons with seizure disorders and schizophrenia.
The characterization of impaired attentional functions in normal and disordered populations of children and adults
The characterization of attention in normal and disordered children and adults has a number of facets. This includes the description of the development of attentional functions in populations of school children from large urban areas; the delineation of normal and abnormal attentional functions in autistic children; the development of new tests to assay various aspects or elements of attentional function; the study of multimodal effects in attention; a new PET/fMRI protocol to describe cerebral metabolic characteristics of attention in probands and first-degree relatives of patients with disordered attention.
Autonomic nervous system activity and attention in psychopathology.
The central purpose of this research is to determine the role of attentional processes and autonomic nervous system (ANS) functioning in adult and child psychopathology by studying their relationships to diagnosis, symptoms, and other biological markers. One group of studies focuses on schizophrenia. Unmedicated adolescents with childhood-onset schizophrenia, studied in collaboration with investigators in the Child Psychiatry Branch, NIMH, show similar abnormalities in resting arousal indices, attenuated reactivity to nonsignal and signal stimuli, and retarded habituation and adaptation as shown by adult schizophrenic patients. In two reaction time (RT) paradigms, the childhood-onset patients also resembled adult cases in profoundly retarded RT and exaggerated foreperiod and sequence effects. The data suggest a fundamental similarity and thus continuity between childhood- and adult-onset schizophrenia. There seems also to be a clear diagnostic specificity of these findings compared with other types of child psychopathology such as attention deficit hyperactivity disorder and OCD. We were also able to show that 'multidimensionally impaired' children who exhibit some schizophrenia symptoms but who were differentiated clinically from child schizophrenia were much different psychophysiologically from the schizophrenia patients. However, they had similar profound attention deficits. A large sample of adult schizophrenia patients recruited by the Experimental Therapeutics Branch, NIMH, tested both on placebo and on a conventional neuroleptic, showed large but selective attenuations of ANS responsiveness by neuroleptics on ANS activity. This finding has important implications because many studies on the ANS in schizophrenia have been done on medicated patients.
In studies on nonschizophrenic psychopathology in collaboration with the Child Psychiatry Branch, we found very few differences in ANS activity between children and adolescents with OCD and controls, but within the OCD group electrodermal reactivity to both innocuous and signal stimuli was correlated positively with ratings of the severity of OCD symptoms. Thus although ANS activity may not differentiate OCD children from controls, it seems to be a significant marker for the severity and outcome of the illness. We also studied relationships between monoamine metabolites from cerebrospinal fluid and ANS activity in subjects with OCD and Disruptive Behavior Disorders. Metabolites of dopamine and serotonin were positively related to a number of variables reflecting ANS responsivity in the OCD group. In addition, ACTH, measured only in the OCD group, showed a very consistent pattern of correlations, being positively related to autonomic base levels and responsivity throughout the session. These findings are consistent with pharmacological studies in humans but conflict with some animal studies of the neurochemical correlates of ANS activity.
Etiological factors in schizophrenia.
Studies of the occurrence of mental illness in families have been useful in identifying familial forms of the illnesses but incompletely specify the nature and strength of genetic and environmental factors in etiology. We have previously reported that chronic schizophrenia and schizophrenia-like disorders occur at a significantly elevated rate in the biological relatives of chronic schizophrenic adoptees and not in their adoptive relatives. Our recent replication of this finding in a national register of nearly 15,000 Danish adoptees provides compelling evidence for the significant operation of genetic factors in the etiology of schizophrenia. Furthermore, our demonstration that the increased prevalence of schizophrenia in the biological families of adoptees does not differ significantly from that in the natural families of schizophrenic individuals indicates that the familial tendency in this disorder is an expression of genetic factors, and provides justification for the use of family studies to examine the modes of genetic transmission and to search for genetic linkages.
Genetic effects in families with high incidence of schizophrenia
This work, conducted in conjunction with investigators from the Karolinska Institute in Sweden, involves combined neuropsychological and genotypic study of the Genain quadruplets. The study is designed to discover whether a particular genetic anomaly found in a group of schizophrenic Swedish triplets will also characterize the Genains. The latter are a well-known group of identical quadruplets, all of whom developed schizophrenia during their early years, although to differing degrees of severity.
Last modified April 6, 2001.