Working Memory Deficits


Neurocognitive deficits are core features of schizophrenia that may determine functional outcome. Working memory (WM) deficit has been identified as one of the most important components of the neurocognitive deficits in schizophrenia. WM is a limited-capacity, active short-term memory system that guides and controls behavior. A majority of patients with schizophrenia and about half of their healthy first-degree relatives have WM deficits and these deficits are linked to poor social and adaptive functioning yet the origins and consequences of WM deficit are unclear. Thus it has not been possible to develop targeted interventions that might ameliorate WM deficits and improve adaptive functioning.

We are interested in to identifying mechanisms that are central to WM deficit and their neural correlates. We also want to examine the effects of WM deficit on social functioning. Cognitive and neuroanatomical data suggest that selective attention and affect modulate encoding and maintenance in WM but schizophrenic patients are unable to effectively integrate attention and affect to guide goal-directed behavior.

  1. We investigate the roles of perceptual, attentional and affective factors in encoding to specify both optimal and detrimental encoding conditions. Encoding affects all forms of memory and thus has far-reaching consequences. In addition to impaired encoding there is good evidence for abnormal maintenance in schizophrenia.
  2. The effects of attentional control and affect on maintenance are examined to identify where vulnerabilities lie. We are particularly interested in studying the role mental imagery in the manipulation of memory representations. A series of experiments on manipulation and maintenance interactions are currently being conducted.
  3. The roles of prefrontal and parietal cortices in normal and abnormal WM are being investigated with event-related fMRI and near infrared optical imaging experiments. The neural correlates of correct and incorrect performance are observed to elucidate the difference between remembering and forgetting in the brain.

To summarize, we aim to identify the key components of WM deficits in schizophrenia. We recruit unaffected siblings of schizophrenic patients to observe potential deficits as well as spared abilities. We also study bipolar individuals to compare the two psychotic disorders. Identification and elucidation of core neurocognitive deficits of schizophrenia will contribute towards the understanding of the complex interplay between cortical functions and cognitive deficits in schizophrenia. Moreover, specifying abnormal mechanisms within WM in relation to attention, affect and brain activation patterns could lead to targeted strategies that might ameliorate WM deficits and improve adaptive functioning in schizophrenia.

Parkinson's Disease

Parkinson's Disease results in a wide range of motor, cognitive and affective deficits. With Dr. Joseph Neimat in Neurosurgery, we are investigating working memory deficits in Parkinson's disease patients who have deep brain stimulation (DBS) device implanted to stimulate the subthalamic nucleus. DBS is a relatively new treatment option when drugs no longer work. When DBS is switched on, there is a remarkable improvement in motor behavior. Although the efficacy of DBS for movement initiation and control is excellent, there is some evidence to suggest that cognitive functioning may get worse. We are examining working memory on and off DBS in Parkinsonian patients. In addition, we collect near infrared optical imaging data during a spatial working memory task in Parkinson's disease patients to examine frontal cortical activity in relation to DBS stimulation in the subthalamic nucleus.

Control of Mental Representations

Successful interaction with the world depends on how well we are able to adjust our ongoing behavior to meet the contextual and situational demands. On-line adjustments of behavior require a certain degree of self and response monitoring. Individuals with psychosis are thought to be impaired in their ability to monitor, inhibit and control ongoing behavior.

To study control of action, we examine response monitoring in schizophrenia and bipolar disorder with the countermanding paradigm (stop signal reaction time task). To study control of mental representations, a series of experiments involving mental rotation, perspective-taking and simulation are currently in progress. The ability to mentally transform visuospatial perspectives and internally simulate external events may be very important for understanding the social world.

Control of action and mental representation also depend on internally stored regularities and predictions. We appear to have an implicit understanding of probabilities of events in the world, which we use to guide our behavior. But it is unclear how we come to acquire these probabilities. One possibility is that working memory and long-term memory interactively glue sequential events over time. We are examining how healthy individuals and psychosis patients deduce probabilities of simple events over time. Some people need to accumulate a lot of data whereas others jump to conclusion with very little evidence. Again, individual differences interact with cognitive machinery in complex ways.

Neurocognitive Origins of Social Deficits

Cognitive deficits, such as impairments of working memory and attention are associated with reduced social functioning and poor outcome in schizophrenia but it is unknown how they cascade into abnormal social behavior. Schizophrenic patients have a wide range of socio-affective deficits including abnormal emotion recognition, social cue perception, "theory-of-mind" (TOM) and attribution styles but the origins of these deficits and their neural underpinnings are not yet clearly understood. We are studying the relationship between socio-affective and cognitive functions in schizophrenia using behavioral and functional neuroimaging methods.

Successful social interactions depend on fast and accurate interpretation of actions, intentions and emotions of others. We hypothesize that deficits in attentional orienting and working memory may be responsible for abnormal perception and interpretation of social stimuli in schizophrenia. The neural network that supports attention and working memory largely overlap with regions that are heavily implicated in the pathophysiology of schizophrenia, including the prefrontal, the posterior parietal, the anterior cingulate cortices and basal ganglia. Abnormalities of this network also result in social/affective deficits.

Cognitive roots of social deficits in schizophrenia may be characterized by:

  1. Inability to attend to socially or emotionally relevant features
  2. Inability to generate internal representation to guide behavior
  3. Inability to maintain social/emotional context in working memory

We study perceptual and cognitive abnormalities that may lead to inaccurate interpretation of external events. To better understand possible perceptual origins of ToM problems, we have examined eye gaze perception and biological motion processing in schizophrenic patients. Accurate decoding of social attention in real time is important for grasping the significance of social behavior. In the case of biological motion processing, we find that schizophrenic patients are impaired in the detection and discrimiation of biological motion (i.e., movements generated by living things. See also Point Light Display Movies) and that these deficits arise from high 'false alarm' rate in the patients; they tend to see scrambled, nonsense motion as 'living'. In healthy people, biological motion stimuli are clearly associated with increased activation of the superior temporal sulcus whereas in schizophrenic subjects, this association is not observed. An inability to detect and discern socially relevant stimuli (e.g. people) could lead to subtle deficits in social behavior and indeed we observe a correlation between biological motion deficit and social functioning scores in schizophrenic subjects.

We are also interested in understanding the relationship between spatial and affective functions. We and other labs have observed the role of affect in attentional orienting, spatial localization and other tasks. It is not surprising that spatial attention and affective arousal should be closely linked. The affective nature of the stimulus determines whether the organism moves towards or away from the stimulus. In addition, neurological data indicate that right hemisphere deficits are associated with emotional and social problems. However, the role of the right parietal cortex in spatial and affective processes is not well understood. We are currently investigating individual differences in empathy in relation to spatial information processing (perspective taking, mental rotation, pseudoneglect).

To elucidate generation of internal representation of the social world, we are examining a wide range of imitation abilities in relation to social functioning. Schizophrenic patients seem unable to imitate very simple manual gestures, mouth movements or facial expressions even thought they can correctly identify these acts. Imitation ability is related to simulation of external events is linked to the "mirror neuron" mechanism supported by the left inferior frontal region that includes Broca's area. We are currently conducting a fMRI study of language and imitation in healthy and schizophrenic subjects to relate behavior, structural findings and functional activation patterns.

To summarize, we aim to better understand how cognitive deficits may lead to social deficits and to elucidate functional neuroanatomy of social cognition.

Cognitive Neuroscience of Belief

Thoughts, Delusions, and Beliefs

Theories of delusion formation can be roughly divided into more perceptual vs cognitive accounts. The perceptual account proposed by Brendan Maher and others posit that delusions are natural product of rational explanation of abnormal sensory experiences. Cognitive accounts take a more top-down approach and point to the roles of cognitive biases, abnormal hypothesis testing, and "jumping to conclusions" in delusion formation (e.g., Helmsley, Bentall, Garety, David). Interacting with cognitive and perceptual theories, are the motivational and emotional factors (e.g. Freud, Bentall). Delusions are rarely non-emotional and may often reflect projections of unconscious unresolved conflicts to external targets or attempts to protect self. There are also hybrid models such as the two process theory (Langdon, Davis) and similarly, the jump-to- perception-and-conclusion theory (Fleminger). While delusions held by individuals with psychosis do not necessarily have specific neuroanatomical correlates, there are neurological lesion cases that offer fascinating insight into specific delusions. Delusional misidentification syndromes, autoscopic phenomena, paranormal beliefs have been studied extensively in lesion patients (Brugger). Last but not least, there are sociocultural and ecological influences on delusion formation (see Van Os).

We study the possible interactions between in pattern recognition and personality traits as major factors in delusion formation. Our studies of biological motion perception, eye gaze direction perception and other perceptual tasks suggest that increased false alarm rates (seeing something where there is none) may be related to the presence of delusions. This may be due to too much top-down control, poor quality sensory data and an interaction between the two. Noisy input from abnormal sensory systems necessitate increased top-down, cognitive processing, which may then lead to faulty conclusions. Over time, these experiences may result in an accumulation of inaccurate information and probabilities about the world. We are currently investigating visual and auditory detection of living and nonliving stimuli in relation to delusion formation in healthy individuals and psychosis patients.

Read the summary of a workshop on delusions held at the International Congress on Schizophrenia Research in San Diego, March 2009 here.

Music and the Brain

The effects of music lesson on enhancing cognition and school performance in low-income children at the W.O. Smith Music School

Collaborators: Gabrielle Chapman Ph.D. and Jan Morrison, M.Ed of the Peabody Research Institute (PRI)
Funding: The Curb Center, Gertrude Conaway Vanderbilt Endowment and Littlejohn Faculty Fellow Award

Since 1984, the W.O. Smith Music School has provided one-on-one music instructions to children from low-income families in Nashville for the nominal cost of 50 cents per lesson. Today, enrollment exceeds 400 students per year. Qualified and vetted volunteers provide individual music lessons, and musical instruments are loaned to the students at no cost. Students may also participate in ensembles. Ensembles perform throughout the year at W.O. Smith recitals, senior centers, civic organizations, and special events. Anecdotal reports from teachers, parents and children suggest that these individual music lessons may have beneficial effects on cognitive and social development as well as on their participation and achievement in school but no empirical evaluation has been conducted to date. Thus, there is a need to qualify and quantify the program’s impact on students. Music training boosts cognitive abilities Public schools have been steadily reducing music classes especially in low income areas but rigorous research indicates that music training greatly enhances neurocognitive functions. Music training has been shown to increase cognitive abilities (Costa-Giomi et al, 2004; Portowitz, et al., 2009), verbal memory (Ho et al, 2003), IQ (Gromko & Poorman, 1998), prosocial behavior (Kirschner & Tomasello, 2010) and empathy (Kalliopuska & Ruókonen, 1993). Only some of the studies focused on disadvantaged children with little opportunity otherwise to experience sustained music instruction. Also, few of these studies examined both academic and cognitive outcomes. It is particularly important to know if the W.O. Smith program has the capability to boost the academic performance of economically disadvantaged students.

We are investigating the effects of music training on cognition and academic performance in children at W.O. Smith School. Each fall the school receives over 100 new applications from families interested in starting the music program, for 60 openings. New students are randomly selected for the fall program session or placed on a waiting list for the spring session. The music training group study with music teachers to learn about reading music, counting/tempo, instrument sounds, and so forth before selecting an instrument of their own choice and beginning weekly individual lessons. The waitlist group do not participate in musical activities. After obtaining informed consent approved by Vanderbilt IRB, baseline data were collected in the fall semeseter. The baseline measures included interview questions about demographic and family background information, cognitive assessments, and TCAP scores from the previous academic year. For cognitive assessments, we use CogState, a standardized computerized program that has excellent reliability and validity and is child-friendly.

At the end of the academic year, cognitive assessments are administered to both the music training and control groups along with follow-up questions about school performance and behavior during the prior year. These responses are compared to the baseline data for the two groups to determine whether there are greater improvements on any of these measures for the music-training group relative to the control group. The data collection during this phase also include a music teacher survey that is already in place at W.O. Smith. This teacher survey grades the student’s effort and records attendance and will help determine the level of program exposure experienced by children in the experimental group. Then at the end of the school year, we will also obtain the TCAP data to compare with the scores from the previous year. We expect those who completed music training to show better cognitive and academic performance by the end of the academic year.

Social and Cognitive Intervention and Enhancement

Physiology-based emotion sensing robotic technology for social cognitive intervention in schizophrenia

Collaborator: Nilanjan Sarkar, Professor of Mechanical Engineering
Funding: NARSAD Distinguished Investigator Award

Social impairment is a core feature of schizophrenia, present throughout the course of illness from the prodromal stage. Social functioning is essential to determining quality of life and functional outcome, but difficulties in conceptualizing, assessing and targeting social deficits have impeded the implementation of efficacious interventions. These challenges could be tackled by parsing social deficits into neurobiologically constrained components that can be systematically tested.

Successful social interactions depend on fast and accurate perception and interpretation of actions, intentions and emotions of self and others; these processes that support a “theory of mind” are severely impaired in schizophrenia. Having a “theory of mind” involves accessing a stored body of knowledge or rules about mental states to make inferences about others’ behavior but importantly, it also involves internally modeling or running a mental simulation of the observed behaviors to understand intentions of others. Recent work from our lab indicates that schizophrenia patients are significantly impaired in their ability to imitate or simulate another person’s actions that are associated with social competence and abnormal recruitment of the social brain network including the ‘mirror neuron’ area. Such simulation and imitation deficits impede social learning, which present obstacles to social behavioral remediation, but these impairments are readily responsive to targeted exercises; simulated training has been shown to improve real life skills.

Much progress has been made in identifying key cognitive deficits of schizophrenia and evidence indicates that neuroplasticity-based, computerized training can lead to changes in cortical functions and solid improvements in attention and memory, but cognitive remediation alone does not normalize social deficits. We hypothesize that social cognitive exercises that target the social brain network and support mental simulation should improve social deficits in schizophrenia. We aim to implement a user-friendly, high-compliance method based on innovative physiology-based, emotion sensitive virtual reality (VR) and robotics technology that dynamically adjusts exercise complexity according to continuously monitored affective state of the participant. Such efforts are already on the way in autism and physical rehabilitation. VR-based methods would circumvent some of the difficulties present in existing social remediation programs, such as low compliance, the lengthy duration and therapist-hours needed to reach improvements. We propose to personalize social training using emotion-sensing VR in two steps.

In Study 1, we will compile individual profiles of physiological responses to standardized affective stimuli (International Affective Picture System: IAPS) and to interactive social vignettes involving realistic ‘avatars’ whose facial expressions change depending on the stories they tell. Multichannel physiological measures of affective responses to these stimuli are monitored continuously to model mathematical algorithms of emotional states for each subject as he/she makes decisions about each event. This method allows us to directly compare self-reported explicit emotional responses of the participant and the real-time predictions of the machine-learning algorithms generated by the participant’s changing psychophysiological profile to determine the disparity between explicit and implicit measures of emotional responses to social stimuli. This information collected from each subject can then be used in real time to dynamically alter the task configurations in VR to increase the participant’s level of engagement in social skills exercises program. In Study 2, we will use VR based, adaptive behavior tasks that directly influence the quality of life: cooking and driving. While engaged in these tasks, physiological data will be monitored online to gauge dynamic affective states of the subjects to dynamically adjust the task difficulty online. We will use naturalistic VR scenarios in which one may move through a sequence of steps to complete a series of driving or cooking exercises. At each step, a virtual instructor is ready to step in if errors are made or subjects request help. As the subject’s proficiency increases, emotional arousal and dependence on the instructor should decrease and higher level of mastery achieved. At baseline, we will assess social functioning, cognitive functions and symptoms severity, and examine the social brain network activity with a fMRI scan of an action imitation task. After 15 training sessions, we will re-assess social functioning, symptoms and cognition and repeat the fMRI scan.

These studies are a first step toward individualized and modular (e.g. job interview, cooking) social skills training VR program that uses simulated exercises to help restore the social brain network in individuals with schizophrenia. If successful, our long-term goal is to develop individualized ‘Apps’ that can be used easily in daily life to support social cognition and skills training online to augment existing therapies.

Neuroplasticity-Based Cognitive Remediation for Pediatric Brain Tumor Survivors

Collaborator: Bruce Compas, Professor of Psychology and Human Development, Peabody
Funding: Hobbs Discovery Grant, R21 CA175840

Every year over 4,000 children and adolescents (and their families) in the U.S. face the challenging path from brain tumor diagnosis to survivorship that can require neurosurgery followed by radiotherapy and intensive chemotherapy. Fortunately, the survival rates for brain tumor has increased to over 70% of children living at least 5 years post-diagnosis. However, treatment comes at a cost, as the very therapies (surgery, radiation, chemotherapy) that lead to these high survival rates have significant adverse effects on brain development and neurocognitive function. The goal of our work is to develop non-invasive and usef-friendly interventions to improve the quality of life of pediatric brain tumor survivors.

There are three salient features of neurocognitive late effects in survivors of childhood brain tumors. First, pediatric brain tumor survivors show deficits in a wide range of cognitive functions including executive function, working memory, and attention (Robinson et al., 2010a, 2013). Second, cognitive functions show a progressive decline over the course of treatment and continuing through recovery. Scores on standardized tests continue to decline years after treatment ends (Merchant et al., 2009; Spiegler et al., 2004; Stargatt et al., 2007). Third, remediation of these cognitive impairments is a critical target for research. However, the development of cognitive remediation interventions with this population has been slow.

We are currently conducting a pilot randomized controlled trial (RCT) to document the feasibility, acceptability, and initial proof of concept of a neuroplasticity-based adaptive cognitive training program (Cogmed) to train WM and attention in newly diagnosed children and adolescents with a brain tumor compared with a matched group of patients in a control condition (non-adaptive version of Cogmed). Further, we are testing the feasibility of this intervention in a prospective design beginning pre-surgery to examine the effects of Cogmed in deflecting the downward trajectory of cognitive function in children with brain tumors. We use functional near-infrared spectroscopy (NIRS) to examine the efficacy of Cogmed in normalizing frontal cortical activity associated with reduced cognitive functions. Findings from this pilot study will inform the development of a large multi-site randomized efficacy trial to test an individualized cognitive training program.