Final Report Summary - TRW SCHIZO (Temporal dimensions of information processing as a functional marker of mental state: evidence from schizophrenia)
Introduction
The ability to store, integrate and manipulate information is crucial to our functioning in a ever-changing world. In the most real-life situations (e.g. watching a movie, engaging in a conversation), the information gathered at each particular moment becomes meaningful only in the context of previous events. Thus, process a continuous flow of information is required for comprehension of complex situations, as processing occurs at different time scales and relies on accumulation of information over time.
Hierarchies of information processing have been long expected to reflect organization levels in brain processing (Gazzaniga, Ivry & Mangun, 2008). It is a well-established organizing principle of the visual system that neurons along visual cortical pathways have increasingly larger spatial receptive fields, such as neurons in the higher-level visual areas receive input from the low-level neurons with smaller receptive fields, thereby accumulating information across space (Hubel, 1988; DeYoe et al., 1995; Malach et al., 2002; Wandell et al., 2007; Silver & Kastner, 2009). Yet, our understanding of the neuronal basis underlying the gathering and parsing information over time is limited. One way to advance our knowledge is to show modification of neural information processing under deficient cognitive states. A great model for such deficiency is schizophrenia, severe mental disorder, characterized by abnormality in long-range integration of neural information across many brain regions.
Embarking on this new field poses new challenges for researchers and calls for the development of new paradigms and analysis tools. One such challenge is the need to recreate real-life situations in the constrained conditions of the fMRI environment. Another difficulty is the need to come up with an adequate analysis tool for dealing with the inherent complexity of the data. Recently, we developed a novel paradigm (Lerner et al., 2011) and started to explore its potential for studying different aspects of information processing. The empirical protocol involves measuring brain activity during free listening to an engaging real-life story. Stories were chosen as they are capable of simulating aspects of real-life experience, by recreating complex context within a carefully produced and reproducible environment. Since conventional analysis methods are unsuitable for such an open-ended experiment, we have used an advanced inter-subject correlation (inter-SC) approach, which measures the overall similarities in the spatio-temporal response patterns across individuals. The level of inter-subject synchronization provides a quantitative measure of the similarities as well as the differences in cortical response patterns to the shared reality across subjects.
In essence, as will be demonstrated below, this methodological approach opens up new frontiers for cognitive neurosciences. Below is a short description of each of the projects we worked on during the grant period. Each project explores different aspects of high-level cognitive functions such as the temporal scale of processing, the neural basis of inter-group differences, memory and learning.
Completed projects
Up to date two of these works were accepted for publication, and one is under review:
Project I. Levels of information processing in schizophrenia patients
Disturbances in information processing are considered as a hallmark feature of schizophrenia (Nuechterlein et al., 1994). Recently, Krishnan et al. (2009) pointed to impairment in high levels of perception in schizophrenia, which resulted in impairment in use of context of information. Numerous imaging studies have reported disturbed large-scale functional organization in patients with schizophrenia compared to healthy controls. Namely, neural circuits that support sensory, cognitive, and emotional processes are the substrates for cognitive and affective impairments in schizophrenia (Shenton et al., 2001, Davidson and Heinrichs, 2003, Hill et al., 2004). In particular, the 'dysconnection hypothesis', a leading assumption in schizophrenia, claims for a disturbed long-distance cortical connectivity, between frontal and posterior brain regions which underlies the impairments found in the disorder (Friston and Frith, 1995, Friston, 1998, Andreasen et al., 1999).
Recent studies in healthy brains revealed a hierarchical organization of areas involved in process continuously incoming information (Hasson et al., 2008, Lerner et al., 2011). Specifically, it has been shown that capacity to accumulate and process information over different time scales increased gradually in correspondence with the length of coherent temporal structures in the stimulus. To characterize the informational capacity of brain regions involved in processing of information over different time scales, i.e. different levels of information complexity, we parametrically varied the temporal structure of a story by scrambling it at different temporal scales (e.g. words, sentences). Using an inter-SC analysis, we compared neural responses evoked by these conditions in participants with schizophrenia against healthy response benchmark.
The results within the schizophrenia group (N=15) showed that early auditory areas perform their processing over a short temporal scale (e.g. words) similarly to the healthy group. Scrambled sentences (intermediate temporal scale) evoked reliable, although spatially less extended responses in patients. However, the analysis of responses to story condition (long temporal scale) revealed robust and widespread disruption of the inter-SCs. In contrast to healthy group, the response time courses to the story were highly variable within the schizophrenia group, although some inter-SCs in the TPJ and precuneus were found. Importantly, there were no regions exhibited high correlations exclusively within the patients with schizophrenia but not within the healthy individuals.
Project II. Levels of information processing in non-psychotic siblings of patients with schizophrenia
Healthy siblings of schizophrenia patients are known to be impaired, albeit to a lesser degree than patients, on a wide array of cognitive tasks (Grove et al., 1991, Faraone et al., 1995, Cosway et al., 2000, Appels et al., 2003), including working memory tasks (Keshavan et al., 2002, Callicott et al., 2003), attentional processes (Morey et al., 2005), language lateralization (Li et al., 2007, Li et al., 2007) and facial expression (Habel et al., 2004). Moreover, imaging studies present that first degree relatives may share regional brain volume abnormalities with their sick siblings in many brain regions (Gogtay et al., 2007). Which of the brain circuits do subserve predisposition for an illness in unaffected family members? In this study, we adapted the inter-SC protocol in order to seek out and identify the brain regions in which the correlation across siblings is synchronized during different time scales. This analysis revealed that the temporal topography observed in healthy participants and healthy siblings was similar. However, a direct comparison of responses to story condition in both groups revealed overlapping regions in A1+, along the STG up to TPJ, in the IFS and DLPFC, but not in the angular gyrus, superior and middle frontal gyri, lateral and medial PFC that were found in healthy group only.
Projects I & II are the first to report a hierarchy of information processing in patients with schizophrenia and their non-psychotic siblings. It extends traditional findings in schizophrenia to a rich, dynamic and more ecologically valid situation. I believe that these findings and the experimental protocol will open the way for studying complex processes unfolding over time. In the upcoming years, I plan to extend this research- using the correlation-based analysis I plan to explore neuro-typicality in other groups with psychiatric and neurological problems.
Project III. The architecture of neural circuits involving in processing of non-verbal stimuli
Processing of non-verbal meaningful auditory stimuli might be very different from processing of regular verbal information, such as a story. Music processing is a prime example of our ability to decipher other’s beliefs and desires in a non-verbal situation. On the one hand, music contains structures at multiple time scales, and as such is a potentially powerful entry point into the question of how the brain integrates complex streams of information. On the other hand, these designative meanings are often less precise and specific than those arising in linguistic communication and people may be pretty indifferent to the shuffling of music structures. In this study, we adopted the inter-SC to characterize the time scale of music processing at different stages. To that end, we manipulated the way participants (performing pianists) interpret the music content by rendering the temporal structure of Brahms Piano Concerto No1. The temporal order of the intact piece was scrambled at three time scales: bars, phrases, and sections. We found that response reliability depended systematically on musical structure coherence, revealing a topographically organized hierarchy of processing time scales. Early auditory areas (at the bottom of the hierarchy) responded reliably in all conditions. For brain areas at the top of the hierarchy, the original (unscrambled) excerpt evoked more reliable responses than any of the scrambled excerpts, indicating that these brain areas process long time scale musical structures, on the order of minutes. The topography of processing time scales was analogous with that reported for speech, but the time scale gradients for music and speech overlapped with one another only partially, suggesting that temporally analogous structures – words/measures, sentences/musical phrases, paragraph/sections – are processed separately. These results provided additional evidence that the time scale of processing is a functional property that may be a general organizing principle for the human cortex.
Project IV. Hierarchy in information processing as a unique signature for patients with mild cognitive impairment
The brain undergoes plastic changes constantly with age. However, our knowledge of cortical plasticity involving aging-related processes and understanding the types of processing, when the breakdown may lead to cognitive impairments, are still limited. In this study we characterized the processes by which the brain accumulates and integrates temporally extended information in patients suffering from the amnestic mild cognitive impairment (a-MCI) and healthy older adults. The study mirrored our previous procedure, which used a narrated story and its scrambled versions as stimuli. An inter-SC analysis applied to the data set revealed prominent difference between groups at the level of long time scale processing. In healthy older adults, parietal and frontal regions responded reliably to paragraphs and intact story, reproducing hierarchy as previously reported in young individuals. Participants with a-MCI exhibited a robust anatomical shift in the neural mechanisms involved in long time scale processing to the pre- and post-central sulci. These results suggest reorganization of the central nervous system as a compensation for the dysfunction of other high cognitive areas. The present observation might serve as an early marker of neurodegeneration, reflecting functional disturbance prior to actual neuronal or synaptic loss.
The MCI project is the first study that adopted the inter-SC analysis for investigating cortical differences in neurodegeneration. In the upcoming years I plan to develop this approach for tracing systematic differences in this clinical population. Moreover, I hope to use our findings to reveal the underlying neuronal substrates that correlate with physical and social training in patients with MCI.
Ongoing projects
We are currently working on the following manuscript:
Project V. Shared feelings: investigating neural and emotional attunement to co-present participant
In addition to the study of high-level cognitive functions, inter-SC analysis can provide a new quantitative measure of the similarities as well as the differences in response patterns to a shared emotional experience across participants. Emotional experiences are frequently formed within social context. Co-attending to external events in social presence, people rapidly monitor and adapt to the incoming emotional signals from co-present others. Yet, the neural processes underlying emotion propagation between people have not been directly studied. As an initial step, we investigated how the brain processes emotional cues, arriving from another, co-attending individual. Participants viewed an engaging movie in the MRI scanner while simultaneously received continuous emotional feedback. Participants in the social group (but not in the control group) believed that the feedback is coming from another individual who co-views the same movie. The results demonstrated that social-emotional feedback affected the neural dynamics both in the core affect regions and in the evaluative medial prefrontal regions. Strikingly, the neural response patterns in these regions exhibited robust temporal alignment with the emotional timeline of the feedback. Moreover, the neural effects of the social-emotional feedback in the dorsomedial prefrontal region and in the right amygdala were modulated by the subjective experience of social presence during the experiment. Taken in conjunction with previous research, our findings suggest that emotional cues from others dynamically shape the activity across the whole neural continuum of emotional processing in the brain.
The ability to store, integrate and manipulate information is crucial to our functioning in a ever-changing world. In the most real-life situations (e.g. watching a movie, engaging in a conversation), the information gathered at each particular moment becomes meaningful only in the context of previous events. Thus, process a continuous flow of information is required for comprehension of complex situations, as processing occurs at different time scales and relies on accumulation of information over time.
Hierarchies of information processing have been long expected to reflect organization levels in brain processing (Gazzaniga, Ivry & Mangun, 2008). It is a well-established organizing principle of the visual system that neurons along visual cortical pathways have increasingly larger spatial receptive fields, such as neurons in the higher-level visual areas receive input from the low-level neurons with smaller receptive fields, thereby accumulating information across space (Hubel, 1988; DeYoe et al., 1995; Malach et al., 2002; Wandell et al., 2007; Silver & Kastner, 2009). Yet, our understanding of the neuronal basis underlying the gathering and parsing information over time is limited. One way to advance our knowledge is to show modification of neural information processing under deficient cognitive states. A great model for such deficiency is schizophrenia, severe mental disorder, characterized by abnormality in long-range integration of neural information across many brain regions.
Embarking on this new field poses new challenges for researchers and calls for the development of new paradigms and analysis tools. One such challenge is the need to recreate real-life situations in the constrained conditions of the fMRI environment. Another difficulty is the need to come up with an adequate analysis tool for dealing with the inherent complexity of the data. Recently, we developed a novel paradigm (Lerner et al., 2011) and started to explore its potential for studying different aspects of information processing. The empirical protocol involves measuring brain activity during free listening to an engaging real-life story. Stories were chosen as they are capable of simulating aspects of real-life experience, by recreating complex context within a carefully produced and reproducible environment. Since conventional analysis methods are unsuitable for such an open-ended experiment, we have used an advanced inter-subject correlation (inter-SC) approach, which measures the overall similarities in the spatio-temporal response patterns across individuals. The level of inter-subject synchronization provides a quantitative measure of the similarities as well as the differences in cortical response patterns to the shared reality across subjects.
In essence, as will be demonstrated below, this methodological approach opens up new frontiers for cognitive neurosciences. Below is a short description of each of the projects we worked on during the grant period. Each project explores different aspects of high-level cognitive functions such as the temporal scale of processing, the neural basis of inter-group differences, memory and learning.
Completed projects
Up to date two of these works were accepted for publication, and one is under review:
Project I. Levels of information processing in schizophrenia patients
Disturbances in information processing are considered as a hallmark feature of schizophrenia (Nuechterlein et al., 1994). Recently, Krishnan et al. (2009) pointed to impairment in high levels of perception in schizophrenia, which resulted in impairment in use of context of information. Numerous imaging studies have reported disturbed large-scale functional organization in patients with schizophrenia compared to healthy controls. Namely, neural circuits that support sensory, cognitive, and emotional processes are the substrates for cognitive and affective impairments in schizophrenia (Shenton et al., 2001, Davidson and Heinrichs, 2003, Hill et al., 2004). In particular, the 'dysconnection hypothesis', a leading assumption in schizophrenia, claims for a disturbed long-distance cortical connectivity, between frontal and posterior brain regions which underlies the impairments found in the disorder (Friston and Frith, 1995, Friston, 1998, Andreasen et al., 1999).
Recent studies in healthy brains revealed a hierarchical organization of areas involved in process continuously incoming information (Hasson et al., 2008, Lerner et al., 2011). Specifically, it has been shown that capacity to accumulate and process information over different time scales increased gradually in correspondence with the length of coherent temporal structures in the stimulus. To characterize the informational capacity of brain regions involved in processing of information over different time scales, i.e. different levels of information complexity, we parametrically varied the temporal structure of a story by scrambling it at different temporal scales (e.g. words, sentences). Using an inter-SC analysis, we compared neural responses evoked by these conditions in participants with schizophrenia against healthy response benchmark.
The results within the schizophrenia group (N=15) showed that early auditory areas perform their processing over a short temporal scale (e.g. words) similarly to the healthy group. Scrambled sentences (intermediate temporal scale) evoked reliable, although spatially less extended responses in patients. However, the analysis of responses to story condition (long temporal scale) revealed robust and widespread disruption of the inter-SCs. In contrast to healthy group, the response time courses to the story were highly variable within the schizophrenia group, although some inter-SCs in the TPJ and precuneus were found. Importantly, there were no regions exhibited high correlations exclusively within the patients with schizophrenia but not within the healthy individuals.
Project II. Levels of information processing in non-psychotic siblings of patients with schizophrenia
Healthy siblings of schizophrenia patients are known to be impaired, albeit to a lesser degree than patients, on a wide array of cognitive tasks (Grove et al., 1991, Faraone et al., 1995, Cosway et al., 2000, Appels et al., 2003), including working memory tasks (Keshavan et al., 2002, Callicott et al., 2003), attentional processes (Morey et al., 2005), language lateralization (Li et al., 2007, Li et al., 2007) and facial expression (Habel et al., 2004). Moreover, imaging studies present that first degree relatives may share regional brain volume abnormalities with their sick siblings in many brain regions (Gogtay et al., 2007). Which of the brain circuits do subserve predisposition for an illness in unaffected family members? In this study, we adapted the inter-SC protocol in order to seek out and identify the brain regions in which the correlation across siblings is synchronized during different time scales. This analysis revealed that the temporal topography observed in healthy participants and healthy siblings was similar. However, a direct comparison of responses to story condition in both groups revealed overlapping regions in A1+, along the STG up to TPJ, in the IFS and DLPFC, but not in the angular gyrus, superior and middle frontal gyri, lateral and medial PFC that were found in healthy group only.
Projects I & II are the first to report a hierarchy of information processing in patients with schizophrenia and their non-psychotic siblings. It extends traditional findings in schizophrenia to a rich, dynamic and more ecologically valid situation. I believe that these findings and the experimental protocol will open the way for studying complex processes unfolding over time. In the upcoming years, I plan to extend this research- using the correlation-based analysis I plan to explore neuro-typicality in other groups with psychiatric and neurological problems.
Project III. The architecture of neural circuits involving in processing of non-verbal stimuli
Processing of non-verbal meaningful auditory stimuli might be very different from processing of regular verbal information, such as a story. Music processing is a prime example of our ability to decipher other’s beliefs and desires in a non-verbal situation. On the one hand, music contains structures at multiple time scales, and as such is a potentially powerful entry point into the question of how the brain integrates complex streams of information. On the other hand, these designative meanings are often less precise and specific than those arising in linguistic communication and people may be pretty indifferent to the shuffling of music structures. In this study, we adopted the inter-SC to characterize the time scale of music processing at different stages. To that end, we manipulated the way participants (performing pianists) interpret the music content by rendering the temporal structure of Brahms Piano Concerto No1. The temporal order of the intact piece was scrambled at three time scales: bars, phrases, and sections. We found that response reliability depended systematically on musical structure coherence, revealing a topographically organized hierarchy of processing time scales. Early auditory areas (at the bottom of the hierarchy) responded reliably in all conditions. For brain areas at the top of the hierarchy, the original (unscrambled) excerpt evoked more reliable responses than any of the scrambled excerpts, indicating that these brain areas process long time scale musical structures, on the order of minutes. The topography of processing time scales was analogous with that reported for speech, but the time scale gradients for music and speech overlapped with one another only partially, suggesting that temporally analogous structures – words/measures, sentences/musical phrases, paragraph/sections – are processed separately. These results provided additional evidence that the time scale of processing is a functional property that may be a general organizing principle for the human cortex.
Project IV. Hierarchy in information processing as a unique signature for patients with mild cognitive impairment
The brain undergoes plastic changes constantly with age. However, our knowledge of cortical plasticity involving aging-related processes and understanding the types of processing, when the breakdown may lead to cognitive impairments, are still limited. In this study we characterized the processes by which the brain accumulates and integrates temporally extended information in patients suffering from the amnestic mild cognitive impairment (a-MCI) and healthy older adults. The study mirrored our previous procedure, which used a narrated story and its scrambled versions as stimuli. An inter-SC analysis applied to the data set revealed prominent difference between groups at the level of long time scale processing. In healthy older adults, parietal and frontal regions responded reliably to paragraphs and intact story, reproducing hierarchy as previously reported in young individuals. Participants with a-MCI exhibited a robust anatomical shift in the neural mechanisms involved in long time scale processing to the pre- and post-central sulci. These results suggest reorganization of the central nervous system as a compensation for the dysfunction of other high cognitive areas. The present observation might serve as an early marker of neurodegeneration, reflecting functional disturbance prior to actual neuronal or synaptic loss.
The MCI project is the first study that adopted the inter-SC analysis for investigating cortical differences in neurodegeneration. In the upcoming years I plan to develop this approach for tracing systematic differences in this clinical population. Moreover, I hope to use our findings to reveal the underlying neuronal substrates that correlate with physical and social training in patients with MCI.
Ongoing projects
We are currently working on the following manuscript:
Project V. Shared feelings: investigating neural and emotional attunement to co-present participant
In addition to the study of high-level cognitive functions, inter-SC analysis can provide a new quantitative measure of the similarities as well as the differences in response patterns to a shared emotional experience across participants. Emotional experiences are frequently formed within social context. Co-attending to external events in social presence, people rapidly monitor and adapt to the incoming emotional signals from co-present others. Yet, the neural processes underlying emotion propagation between people have not been directly studied. As an initial step, we investigated how the brain processes emotional cues, arriving from another, co-attending individual. Participants viewed an engaging movie in the MRI scanner while simultaneously received continuous emotional feedback. Participants in the social group (but not in the control group) believed that the feedback is coming from another individual who co-views the same movie. The results demonstrated that social-emotional feedback affected the neural dynamics both in the core affect regions and in the evaluative medial prefrontal regions. Strikingly, the neural response patterns in these regions exhibited robust temporal alignment with the emotional timeline of the feedback. Moreover, the neural effects of the social-emotional feedback in the dorsomedial prefrontal region and in the right amygdala were modulated by the subjective experience of social presence during the experiment. Taken in conjunction with previous research, our findings suggest that emotional cues from others dynamically shape the activity across the whole neural continuum of emotional processing in the brain.
