Social Cognition, Autism, and Religiosity
Thursday, May 18, 2017
Pivotal to our ability to manage the complexities of the social world is our capacity to draw inferences readily about the states of other people’s minds, i.e., about the contents of their mental representations, their attitudes towards those contents, and their influences on behavior. Variously described as “theory of mind,” “mentalizing,” and “mindreading,” that capacity seems central to human social cognition. Some theorists also accord theory of mind an important role in making sense of many features of autism, proposing that people with autism are “mindblind,” i.e., that for them theory of mind abilities do not emerge unconsciously and comparatively effortlessly in the typical developmental time course (if they emerge at all). Cognitive scientists of religion have proposed that social cognition and theory of mind capacities, in particular, play a prominent role in shaping many recurring features of religious representations and conduct. This workshop examines the confluence of the three topics comprising its title and the consequence of the proposals in the cognitive science of religion that if people with autism have impaired or atypical theory of mind capacities, then they might exhibit impaired or atypical understandings of some aspects of religion.
Department of Psychology
University of British Columbia, Canada
Director, Spoken Communication Laboratory
Marcus Autism Center
Social Cognition, Theory of Mind, and Belief in Gods - Dr. Ara Norenzayan
For a given person to believe in a deity or deities, she must (a) be able to form intuitive mental representations of supernatural agents; (b) be motivated to commit to supernatural agents (and related rituals) as real and relevantsources of meaning and control; and (c) have received specific cultural inputs that, of all the supernatural agents or forces one could possibly think of, one or more specific deities should be believed in and committed to. In this talk, I present these interrelated hypotheses from the new cognitive science of religion and the science of cultural evolution in light of the growing evidence from diverse fields. I also present new research about belief in karma in relation to cognitive theories. Throughout the talk I explore the current controversies and debates about the social cognitive and cultural learning capacities that make human beings a believing species.
A Tale of Intertwining Spectrums: Is There a Link Between Autistic Tendencies and Disbelief in Gods? - Dr. Ara Norenzayan
Are non-clinical populations high on the autistic spectrum less likely to “get” religion? Building on the first talk, I ask whether autism increases the odds of disbelief, as has been predicted by some cognitive theories of religious belief. Probing further, I ask whether this link is statistically explained by the selective deficits in theory of mind associated with the autistic spectrum. Next I explore whether gender differences in autism and theory of mind offer a novel, if partial, explanation for the well-documented gender gap in religious belief. Further, I present new research on links between the schizotypal spectrum in non-clinical populations – a cluster of traits partly characterized by a hyperactive theory of mind – and hyper-religiosity. This link in turn may offer insights into the psychological profile of the “spiritual but not religious“ phenomenon.
Social Neuroscience and the Nature and Origin of Religious Experience: Lessons and Non-Lessons from Autism and Related Neurodevelopmental Disorders - Dr. Gordon Ramsay
Recent attempts to use findings in neuroscience to inform our understanding of religious experience have focused on explaining the origins of religious activity and belief as potential byproducts of neural structures that evolved for, and were exapted from, other biological functions. Brain mechanisms implicated in attributing agency, detecting intentions, social reward, pro-social adaptation, and other aspects of social cognition have variously been proposed as potential pathways leading to the emergence of commonalities in religion and ritual across cultures. Conversely, conditions where those mechanisms are perturbed or impaired are potentially useful in testing new theories in neurotheology. Most proposals in this area have neglected the role of development and early experience in shaping neural function throughout the lifespan. This presentation will provide an overview of recent research in developmental social neuroscience, in the context of autism, in order to explore the extent to which social cognition in general and neurodevelopmental disorders in particular may or may not be able to shed light on religiosity.
LEARNING ABOUT THE VOCAL WORLD:
Wednesday, May 20th, 2015
Deciphering the Statistics of Communication
Learning about the Vocal World: Deciphering the Statistics of Communication
This interdsciplinary symposium brought together Emory faculty and internationally-recognized scholars from around the country, spanning fields as diverse as psychology, neuroscience, physics, and medicine, to share with post-doctoral fellows and students how they make use of computational and quantitative methods to study the production and perception of vocal communication signals. The conference consisted of the main conference discussions followed by lunch time round-tables and concluded with a poster session where Emory graduate students (and any undergraduates on campus) presented posters, judged by the invited speakers. The all-day event was held on at the Emory Conference Center in Atlanta, GA.
The event was sponsored by Emory Conference Subvention Fund, the Institute for Quantitative Theory and Methods, and the Center for Mind, Brain and Culture. Co-organizers for this event were Samuel Sober and Robert Liu.
VOCAL STATISTICS AND LEARNING I
Comments by session chair: Lynne Nygaard (Psychology, Emory University)
Sequence learning and the cultural evolution of language -- Morten Christiansen (Psychology, Princeton University)
How a song is learned: mechanisms of template matching -- Ofer Tchenichovski (Psychology, Hunter College, City University of New York)
VOCAL STATISTICS AND LEARNING II
Comments by session chair: Gordon Ramsay (Pediatrics, Emory University)
Topography of Human Vocal Development -- Eugene Buder (Communication Sciences & Disorders, University of Memphis)
Statistical constraints on vocal learning in songbirds -- Sam Sober (Biology, Emory University)
NEURAL MECHANISMS OF PERCEPTION
Comments by session chair: Robert Liu (Biology, Emory University)
Functional organization of human auditory cortex in speech perception -- Edward Chang (Neurosurgery, University of California at San Francisco)
Neural mechanisms of auditory-vocal communication: Mapping receiver tuning to sender behavior -- Sarah N. M. Woolley (Psychology, Columbia University)
The role of auditory cortex in perceptual acuity and emotional learning of sounds -- Maria Geffen (Otolaryngology, University of Pennsylvania)
VOCALIZATIONS AND EMOTIONS
Comments by session chair: Donald Rainnie (Psychiatry, Emory University)
Evaluating the communicative value of mouse vocalizations -- Katrin Schenk (Physics and Astronomy, Randolph College)
Emotional content in acoustic communication: Messages sent, messages received -- Jeffrey J. Wenstrup (Neurobiology, Northeast Ohio Medical University)
Vocal Alignment toward Accented Speech
Eva M. Lewandowski *, Lynne C. Nygaard *
*Department of Psychology, Emory University, Atlanta, GA
Individual differences in statistical learning and language: A psychometric study
Ethan Jost ‐ Cornell University; Memorial Sloan Kettering Cancer Center
Jennifer Misyak ‐ University of Warwick
Morten Christiansen ‐ Cornell University
Statistical Learning Ability Can Overcome the Negative Impact of Low Socioeconomic Status on Language Development
Joanne A. Deocampo, Leyla Eghbalzad, & Christopher M. Conway
Department of Psychology, Georgia State University
What just happened? Exploring the neural mechanisms underlying structured sequence processing and language and their role in detection of statistical/sequential violations
Gretchen N.L. Smith, Sanjay D. Pardasani, Gerardo E. Valdez, Gwen A. Frishkoff, & Christopher M. Conway
Georgia State University, Department of Psychology
Early vocal development in infants at risk of autism: prosody and social interaction
Gordon Ramsay (1,2), Kristin Muench (1,2), Ami Klin (1,2)
(1) Department of Pediatrics, Emory University School of Medicine, Atlanta, GA
(2) Spoken Communication Laboratory, Marcus Autism Center, Children's Healthcare of Atlanta, Atlanta, GA
Vocal plasticity and coordinated singing behavior in adult zebra finches
Julia Hyland Bruno & Ofer Tchernichovski
Projections of the cerebellar nuclei in a songbird
David Nicholson, Biology Department, Neuroscience Graduate Program, Emory
Sam Sober, Biology Department, Emory University
Focal intraoperative cooling modifies speech production in a location‐specific manner
Michael A. Long1, Kalman A. Katlowitz1, Rachel C. Clary1, Mario Svirsky1, Hiroyuki Oya2, Matthew A. Howard III2, Jeremy D. W. Greenlee2
1 ‐ NYU School of Medicine
2 ‐ University of Iowa Neurosurgery
Development of a speech prosthesis from recordings of single units in a human
Phil Kennedy, MD, PhD, Neural Signals Inc. Duluth Ga 30096
Modulation of core auditory cortex single unit responses by vocalization categories depends on
Kathryn N. Shepard (1), Frank G. Lin (2), Kelly K. Chong (1,2); Charles Zhao (2); Robert C. Liu (1)
1. Emory University Department of Biology, Atlanta, GA
2. Georgia Institute of Technology/Emory University Department of Biomedical Engineering, Atlanta, GA
Familiarity with a vocal category revealed through the expression of a synaptic plasticity gene in
Tamara Ivanova, Robert C. Liu
Emory Biology, Atlanta, GA USA
Human auditory cortex correlates of perceptual spoken word boundaries
Tang, Claire 1 and Chang, Edward F. 2
1. Neuroscience Graduate Program, University of California, San Francisco, 675 Nelson Rising Lane, San
Francisco, CA 94158, USA.
2. Department of Neurological Surgery and Department of Physiology, University of California, San Francisco,
675 Nelson Rising Lane, San Francisco, CA 94158, USA.
State‐dependent changes in auditory responses determined from single trial analysis of local field
Emily G. Hazlett (1,2), Jasmine M. S. Grimsley (1), Jeffrey J. Wenstrup (1)
1‐ Northeast Ohio Medical University
2‐ Kent State University
Affective Priming Effect of Music on Emotional Faces and Prosody in Williams Syndrome
Miriam Lense, Emory University, Marcus Autism Center, and Children's Healthcare of Atlanta; Cyrille Magne, Middle Tennessee State University; Michael Pridmore, Middle Tennessee State University, Reyna Gordon,Vanderbilt University; Sasha Key, Vanderbilt University; Elisabeth Dykens, Vanderbilt University
Song‐induced dopamine release in the reward pathway of a seasonally‐breeding songbird
Carlos Rodríguez‐Saltos (1), Susan Lyons (2), Keith Sockman (2), Donna Maney (1)
1 Emory University, Atlanta, GA; 2 University of North Carolina, Chapel Hill, NC
Dopaminergic contributions to vocal learning
Lukas Hoffmann, Emory University Neuroscience
Samuel Sober, PhD, Emory University Biology
Varun Saravanan, Emory University Neuroscience
Estrogen receptor alpha expression in the auditory cortex changes across motherhood
Amielle Moreno (1), Kelly Chong (1,2), Tamara Ivanova (1), Robert C. Liu (1)
1. Emory University Department of Biology, Atlanta, GA
2. Georgia Institute of Technology/Emory University Department of Biomedical Engineering, Atlanta, GA
NEUROSCIENCE WORKSHOP: Dimensionality Reduction Methods
October 30 - 31, 2015
Compressing Animal Behavior.
Animals perform a complex array of behaviors, from changes in body posture to vocalizations to other dynamic outputs. Far from being a disordered collection of actions, however, there is thought to be an intrinsic structure to the set of behaviors and their temporal and functional organization. In this talk, I will introduce a novel method for mapping the behavioral space of organisms. This method relies only upon the underlying structure of postural movement data to organize and classify behavior, eschewing ad hoc behavioral definitions entirely and effectively compressing the vast amounts of data being collected. Applying this method to videos of freely-behaving fruit flies (D. melanogaster), I will show that the organisms’ behavioral repertoire consists of a hierarchically-organized set of stereotyped behaviors. This hierarchical patterning results in the emergence of long time scales of memory in the system, providing insight into the mechanisms of behavioral control over that occur over seconds, minutes, hours, days, and the entire lifetime of the fly. Lastly, I will show the generality of this approach to behavioral analysis — specifically its applicability to other species, alternative behavioral modalities, and high-throughput screens investigating the underlying neurobiology and genetics of behavior.
Data Signal Processing
Georgia Institute of Technology
Dimensionality Reduction as a Model of Efficient Coding in the Visual Pathway
The engineering and applied math communities often exploit the fact that natural stimuli have significant structure that lends itself well to dimensionality reduction. The efficient coding hypothesis for sensory neural coding postulates that stages of neural processing should sequentially make the representations more efficient by removing stimulus redundancies, and this is often expressed in the language of information theory. In this talk I will present our work exploring efficient coding models of vision based on dimensionality reduction, including sparsity, low-rank matrix factorizations and random projections. I will show that such approaches are able to account for many observed properties in visual cortex, including classical receptive fields, response properties based on nonclassical or nonlinear receptive fields, and properties of the inhibitory interneurons.
Electrical & Computer Engineering
Carnegie Mellon University
Dimensionality Reduction of Large-Scale Neural Recordings during Sensorimotor Control
Most sensory, cognitive, and motor functions rely on the interaction among many neurons. To analyze the activity of many neurons together, many groups are now adopting advanced statistical methods, such as dimensionality reduction. In this talk, I will first describe how dimensionality reduction can be used in a closed-loop experimental setting to understand how learning is shaped by the underlying neural circuitry. Then, I will describe a novel latent variable model that extracts a subject's internal model during sensorimotor control.
The Large-Scale Structure of the Mental Dictionary: A Data Mining Approach Using Word2Vec, t-SNE, and GMeans
Advancements in machine learning and data mining have already led to amazing breakthroughs in the natural sciences, including the unlocking of the human genome and the detection of subatomic particles. Such techniques promise to wield a similar impact on the study of mind. In my talk I will discuss how the large-scale structure of the human mental lexicon, roughly 50,000 words, can be recovered from billions of words at a level of resolution that includes the differentiation of word senses. Central to this effort are several machine learning and dimensionality reduction techniques, including deep learning, t-Distributed Stochastic Neighbor Embedding (t-SNE), and the clustering technique called GMeans. In addition to the extraction of the mental lexicon, I will discuss how an approach to topic modeling, based on neural networks, might be used to partially automate the process of theory generation. I also raise implications for research on physical and mental wellbeing.
Georgia Institute of Technology
Modularity in Neural Control of Movement
Neuromechanical principles define the properties and problems that shape neural solutions for movement. Although the theoretical and experimental evidence is debated, I will present arguments for consistent modular structures in motor patterns that are neuromechanical solutions for movement particular to an individual and shaped by evolutionary, developmental, and learning processes.
Jessica L. Allen (Post-Doctoral Fellow; Biomedical Engineering, Emory University)
Using non-negative matrix factorization to compare muscle coordination patterns across walking and balance post-stroke
Muscle coordination for walking and balance is often severely impaired post-stroke. Understanding how muscles are coordinated across tasks could have important implications for rehabilitation. However, simply recording muscle activity (electromyography, EMG) from multiple muscles results in large and variable datasets that are difficult to interpret. Motor module analysis, which uses non-negative matrix factorization, can reveal the underlying coordination patterns within a large dataset of EMG that would be hard to identify using individual muscle patterns. This technique has previously revealed that a common set of motor modules is recruited during both walking and reactive balance. We do not know if this persists after stroke. Therefore, the objective of this preliminary study was to examine if common motor modules are recruited across walking and balance post-stroke and how this relates to locomotor performance. We predicted that the number of common motor modules is reduced on the paretic leg and that this reduction is related to impaired locomotor performance. We collected data from five stroke survivors and two healthy older adults. EMG data from 13 muscles per leg were collected during standing reactive balance and walking at self-selected speed. Motor modules were identified from EMG using non-negative matrix factorization. The number of common motor modules between reactive balance and walking was found using Pearson’s correlations and we correlated walking speed to the number of motor modules on the paretic leg that were common across both tasks. The number of motor modules on the paretic leg common across both balance and walking was reduced compared to the nonparetic leg and HOA. We found a moderate positive relationship between walking speed and the number of motor modules common across balance and walking, providing evidence that recruiting common motor modules across reactive balance and walking is impaired post-stroke and is related to impaired mobility.
Elizabeth A. Amadei (Graduate Student; Biomedical Engineering, Emory University)
Measuring and manipulating corticostriatal functional neural circuitry in the socially monogamous prairie vole
The ability to form positive social relationships is key to mental health, and yet the underlying functional neural circuitry remains poorly understood. The socially monogamous prairie vole is a canonical animal model for social bonding. Previous anatomical, genetic, and pharmacological studies have implicated two corticostriatal nodes – the medial prefrontal cortex (mPFC) and nucleus accumbens (NAcc) – in vole social bond formation. However, these approaches do not provide a dynamic view of the functional neural activity and connectivity of these regions during social interactions leading to a bond. To address this, we measured and manipulated neural activity within this circuit in socially-behaving female voles. We found an enhancement in low-frequency mPFC-to-NAcc connectivity during mating, a behavior that accelerates vole bond formation, compared to the control, non-social behavior of self-grooming. Further, optogenetically stimulating mPFC afferents to the NAcc at low frequencies in the absence of mating shifts later behavioral preference towards a partner, suggesting that low-frequency activation of this circuit is functionally relevant for bond formation. Finally, phase-amplitude coupling from mPFC to NAcc is enhanced during mating, suggesting that mPFC activation during social bonding drives NAcc by rhythmically modulating its excitability. Together, these results reveal a dynamic picture of corticostriatal activation during bond formation, with exciting implications for how affiliative social interactions can recruit reward and reinforcement systems to drive changes in behavior. A key ongoing direction is to determine the role of neurochemicals (e.g. oxytocin) in modulating this system.
Jacob Billings (Graduate Student; Neuroscience, Emory University)
Multiscale functional connectivity: Fractionation and recomposition in space and frequency
Recent advances in functional connectivity (FC) analysis of functional magnetic resonance imaging (fMRI) data facilitate the characterization of the brain’s intrinsic functional networks (FC-fMRI). Because the fMRI signal does not provides a perfect representation of neuronal activity, the potential for FC-fMRI to identify functionally relevant networks critically depends upon separating overlapping signals from one another and from external noise. As a step in data preconditioning, researchers often band-pass filter fMRI signals to the range from 0.01 Hz to 0.1 Hz. However, coordinated network oscillations operate across multiple frequencies. Thus, it is not clear that the view of FC-fMRI networks within a single spectral range produces the fullest characterization of brain’s multiple and overlapping systems. The following study addresses this limitation by advancing a multiscale fractionation of FC-fMRI networks, as well methods for quantifying cross-spectral network similarity. These methods clearly and consistently represent group-level brains as composed of well-known functional networks.
Alexander W. Calhoun (Graduate Student; Biomedical Engineering, Georgia Institute of Technology)
Lasting effects of estradiol on network activity of rat cortical neurons in vitro
In rats, sexual differentiation of the developing brain is largely driven by 17β-estradiol (E2), locally produced by conversion of gonadal testosterone. Exposure to E2 during a critical period of embryonic development masculinizes and defeminizes the brain, producing behavioral sex differences in adults and priming the brain for further changes at puberty. At the cellular level, E2 promotes synaptogenesis and increases the excitatory drive of neural circuits. While it is widely believed that such changes in connectivity underlie observable sex differences in behavior, it is unknown how E2 affects the function of neural networks, a critical missing link between morphology and behavior. We used dissociated cultures of cortical neurons harvested from E18 rats and grown on 59-channel multielectrode arrays to investigate how early exposure to estradiol alters population-level activity. Cultures were treated with 10 nM E2 or vehicle for the first week in vitro, and we recorded spontaneous activity every two days from 10-26 days in vitro. E2-treated cultures showed similar patterns of activity to vehicle-treated controls throughout their development, including the emergence of network-wide bursts. However, even weeks after exposure, E2-treated cultures showed greater intra-network variance in the strength, latency, and temporal span of functional connections between their constituent units. This suggests that early E2 exposure has a lasting organizational effect on the complexity of synaptic networks and potentially their ability to represent and process information, even when stripped of the structure of the brain in vivo.
Xiangxi Gao (Undergraduate Student; Emory University)
Crohn’s Disease classifier using gut microbiome composition data
The human gut microbiome, consisting of all the microorganisms in the gastrointestinal (GI) tract, has become recognized for its diverse roles in shaping our health and development. However, the insufficiently characterized microbial networks and interactions occurring in the GI tract pose challenges toward fully understanding the gut microbiome’s role in human health. Alteration of the gut microbiome’s composition, along with genetic risk factors, have been implicated in Crohn’s disease (CD). Compositional differences in the gut microbiomes of CD and control subjects have been well documented, but changes in microbial interactions associated with CD remain unknown. Using composition data of bacterial taxa from RISK, a large cohort study of the gut microbiome in newly diagnosed and untreated pediatric CD subjects, significant compositionally different bacterial taxa between CD and control subjects and groups of bacterial taxa that tend to co-occur together were identified at the Operational Taxonomical Unit (OTU), genus, and family levels. A multivariate Gaussian classifier, which incorporated interactions between significant bacterial taxa in the form of an inverse covariance (precision) matrix, performed equally well as an independent Gaussian classifier and other conventional classifiers, indicating that interactions were not informative for predicting CD diagnosis. However, differences in co-occurrence among bacterial taxa were observed when a group of weakly correlated taxa in the control subjects became strongly correlated in CD subjects at both the genus and family levels. In particular, the Ruminococcaceae and Lachnospiraceae families exhibited a strong correlated decrease in relative abundance in CD subjects and are known to be involved in butyrate production which affects homeostasis of the gut immune system. Deciphering such changes, though not informative for classification purposes, may still provide insights into CD pathogenesis.
Sara M. List (Graduate Student; Neuroscience, Emory University)
Representational similarity analysis examining body-metaphor fMRI data
Representational similarity analysis (RSA) is a form of multivariate pattern analysis (MVPA), which can be used to compare voxel-wise differences in activity across various conditions, allowing more in-depth information to be extracted from neural responses than can be seen from standard univariate activation differences. RSA can foster the comparison of response patterns for a set of stimuli through representational distance matrices, such that the distinction among stimuli that are functionally relevant can be determined. This method has proven to be an effective complement to standard univariate analysis. The data that was analyzed for the current project consists of blood oxygenation level dependent signals obtained when participants were visually presented with images of body parts, such as the face, arm, and leg, as well as when participants were presented with metaphors referring to body parts (i.e. “give a hand”). The goal of this project was to compare voxel-wise activity across all combinations of body parts (i.e. visual face to visual arm, visual face to visual leg, etc) and then to compare the visual to the metaphor condition. This analysis indicated no additional voxel-wise patterns between region of interest and visual or metaphoric presentation of body parts that survived significance tests. However, searchlight MVPA using support vector machine classifiers revealed a topographic mapping that somatosensory and motor areas share, which suggested that the extrastriate body area is sensitive to motor as well as visual stimulation. These findings confirm what the standard univariate analyses indicated for this experiment and are empirically backed by previous studies. RSA and other multivariate approaches are recommended for current neuroscience projects comparing fine-grained activity across multiple conditions.
David A. Nicholson (Graduate Student; Biology, Emory University)
Features required for support vector classification of birdsong elements
Songbirds provide a tractable model system for the study of vocalization and other sequenced motor skills. In the same way that infants learn to speak from their parents, songbirds learn their song by listening to tutor birds and then practicing this behavior millions of times. Each bird’s song consists of repeated elements, often separated by brief silent gaps, referred to as “syllables”. To analyze how different experiments affect production and learning of song, experimenters must label these elements by hand. However, the songbird species that typically are used for laboratory studies will sing without training or extrinsic reward from experimenters, meaning the birds can produce hundreds or thousands of songs a day, many more than can be labeled. Hence some labs have contributed automated analyses of birdsong to the community, including the use of various machine learning algorithms to label elements (cf., Tchernichovski et al 2000, Kogan Margoliash 1998). However, no study has applied standard methods to demonstrate the effectiveness of such machine learning algorithms, such as n-fold cross validation, or to determine which features provide the best discrimination of elements, such as rating the features with f-scores. One recent study in particular made use of support vector machines (SVMs) to label the songs of Bengalese Finches (Tachibana et al. 2014). Using the same SVM method as Tachibana et al., I have reproduced their result on Bengalese Finch song collected in our lab. I go on to show with 5-fold cross validation on typically-sized data sets that this is a viable technique for automatic classification of Bengalese Finch song elements. Discriminability scores demonstrate that power spectra of song elements averaged across time are the best features to use for classification. Preliminary attempts to reduce dimensionality of the feature vectors with the “Best n features” method (Sanders et al 2013) show that most (18/24) of the features are required to maximize discriminability. Studies in progress will test whether other imensionality reduction methods can improve on this result, and whether SVMs can classify the song of a related species, the zebra finch, with a comparable level of accuracy.
Aiden M. Payne (Graduate Student; Biomedical Engineering, Emory University and Georgia Institute of Technology)
Redundant levels of modular motor control revealed by peripheral sensory loss
Motor modules are coordinated groups of simultaneously activated muscles thought to encode a wide range of motor behaviors within the central nervous system (CNS). Alternatively, motor modules may not require coordination from the CNS, but may reflect biomechanically constrained proprioceptive feedback from the peripheral nervous system (PNS). The role of proprioception in the structure and activation of motor modules was investigated in a reactive balance task before and after pyridoxine-induced loss of large-diameter peripheral afferents (n = 5 cats, Stapley 2002, Lockhart 2007). We hypothesize that centrally encoded modules would be unchanged by sensory loss, whereas modules shaped by sensory feedback should be lost or altered. Histological assessment of axon diameters in mixed and cutaneous nerves revealed a range of loss from severely affected (loss of group I and partial loss of group II fibers >5-6um), moderately affected (loss of group I fibers >7-8um), and mildly affected (partial loss of group I fibers >15-16um). Motor modules were identified by non-negative matrix factorization in responses to postural perturbations at three time points: intact (day 0), at the onset of sensory loss (days 4-5, determined by absence of tendon tap response), and at the final time point before histological collection (days 9-18). Consistent with the idea that motor modules are centrally encoded, a moderately affected cat had similar motor structure across time points. In contrast, another moderately affected cat showed a loss of module structure at the onset of sensory loss, followed by the appearance of new modules, suggesting initial reliance on feedback and the availability of other means of coordinating motor modules. These results suggest motor modules are not exclusively organized by the CNS but are not entirely the result of biomechanically constrained proprioceptive feedback. Redundant levels of modular motor control may have important implications for rehabilitation of movement disorders.
Krishna Pusuluri (Graduate Student; Neuroscience, Georgia State University)
Dynamical analysis of connected neuronal motifs with OpenAcc and OpenMPI
Large scale analysis of the dynamical behavior of Central Pattern Generators (CPGs) formed by neuronal networks of even small sizes is computationally intensive and grows exponentially with network size. We have developed a suite of tools to exhaustively study the behavior of such networks on modern GPGPU accelerators using the directive based approach of OpenAcc. We also achieve parallelization across clusters of such machines using OpenMPI. Directive based approaches simplify the task of porting serial code onto GPUs, without the necessity for expertise in lower level approaches to GPU programming, such as CUDA and OpenCL. 3-cell neuronal CPGs have been explored previously using various GPGPU tools . As motifs form the building blocks of larger networks, we have employed our framework to study 4-cell CPGS and two connected 3-cell motifs. We discuss the performance improvements achieved using this framework and present some of our results.
Clarissa J. Whitmire (Graduate Student; Biomedical Engineering, Georgia Institute of Technology and Emory University)
Information coding through adaptive control of synchronized thalamic bursting
It has been posited that the regulation of burst/tonic firing in the thalamus could function as a mechanism for controlling not only how much, but what kind of information is conveyed to downstream cortical targets. Yet how this gating mechanism is adaptively modulated on fast time scales by ongoing sensory inputs in rich sensory environments remains unknown. Using single unit recordings in the rat vibrissa thalamus (VPm), we found that the degree of bottom-up adaptation modulated thalamic burst/tonic firing as well as the synchronization of bursting across the thalamic population along a continuum for which the extremes facilitate detection or discrimination of sensory inputs. Optogenetic control of baseline membrane potential in thalamus further suggests that this regulation may result from an interplay between adaptive changes in thalamic membrane potential and synaptic drive from inputs to thalamus, setting the stage for an intricate control strategy upon which cortical computation is built.
Adam A. Willats (Graduate Student; Biomedical Engineering, Georgia Institute of Technology and Emory University)
Closed loop optogenetic control: Closing the loop around neural circuits in vivo
Since the advent of optogenetics, there has been ever¬ growing interest in applying this technology to the control of neural activity. However, factors such as variable opsin expression, differential distribution of light, and spontaneous neural states make robust control of neural activity intractable with open loop stimulation. To overcome these limitations, we have engineered a system for closed ¬loop control of neural activity which combines single ¬unit thalamic recording and optogenetic stimulation through real¬ time interfacing in the rat vibrissa system in ¬vivo. Here we present an overview of this technique and its success in controlling step changes in firing rate as well as technical refinements to the process of estimating firing rate from spikes which allow much improved control of dynamic firing rate targets.
Riley T. Zeller-Townson (Graduate Student; Biomedical Engineering, Georgia Institute of Technology)
In vitro simultaneous recording of multiple axonal compartments identified using entirely electrical measurements on a novel micro electrode array
Conduction velocity dynamics of action potentials in cortex impact the transmission of spike timing information, and may shed light on the neural code. Measurement of changes in action potential conduction velocity along axons in cortex is hampered by several factors, including: low signal amplitude of the axon compared to measurement noise, presence of many competing signals from other neurites and spatial constraints on measuring from multiple electrical compartments of the same axon. The HiDens CMOS Micro Electrode Array system is a novel platform for the investigation of electrophysiological phenomena in the axon in vitro. This platform provides 126 channels of acquisition, which can be reconfigured in seconds to record from a nearly arbitrary subset of the 11015 electrodes on the array. I describe an experiment protocol for the identification and stimulation of individual axons grown in vitro on the array, using entirely electrical methods. This protocol uses a combination of scanning experiments and targeted experiments. Scanning experiments combine data recorded over many configurations and trials to increase spatial sampling and SNR. These experiments were used to gather information on axon location, and responsiveness to electrical stimulation. This then informed the design of targeted experiments, where a single configuration (with subsequently reduced spatial sampling) was used to perform single trial recordings from multiple compartments of interest of single axons. The dimensionality of the targeted recording was further reduced down to a single recording per axonal compartment through the use of matched filters, which were based on prototypical waveforms recorded during the scanning experiment. The data generated through this protocol are used to measure conduction reliability and velocity changes of evoked spikes traveling antidromically and orthodromically, allowing direct comparison between the antidromic conduction that is more accessible for in vivo experimentation, and the orthodromic conduction which is believed to be more computationally relevant.
Charles L. Zhao (Graduate Student; Biomedical Engineering, Georgia Institute of Technology)
Examining genetic background and synaptic morphology with heterozygotes
The model organism Caenorhabditis elegans is prized for ease of handling and genetic manipulation. This encourages use for genetic studies, which has yielded ground-breaking results. However, the vast majority of genetic studies in C. elegans have been done on the strain N2, in which decades of cultivation has resulted in behavioral, physiological, and genetic divergence from wild populations. The exact significance of this is unknown, but evaluation of this has been bottlenecked by difficulties in experimental procedure, as well as the subtlety of background effects, whose observation requires much larger sampler sizes and fluorescent markers in every background, rendering examination of more than a handful of genetic backgrounds impracticable. To address this, we introduce a methodology tying together several innovations made by this lab. These include microfluidics for high-throughput imaging, computer vision for rapid and accurate quantitative phenotyping, and the use of heterozygotes for comparison of genetic backgrounds without burdensome, repeated outcrossing. We use a combination of the two recently introduced dominant synaptic mutants jkk-1 (km2) and unc-104 (wy673), along with the synaptic marker Pmig-13:snb-1::yfp. By crossing these strains with wildtype strains of C. elegans, producing heterozygous F1 progeny, we show that background effects on synaptic morphology can be discerned, using the N2-cross as a control. In particular, we show that the background of the Hawaiian strain CB4856 exerts an effect on synaptic morphology similar to km2 and wy673, without reinforcing these mutations when present. Since CB4856 does not carry mutations known to affect synaptic morphology, the CB4856 wildtype background exerts a novel effect on these phenotypes.
We thus demonstrate that our method is capable of detecting subtle genetic background effects, and also that these effects are an important confound to genetic discovery.
Embedding High-Dimensional Clusters with t-SNE
Almost all dimensionality-reduction schemes have the property that data points initially far away in a high dimensional space will remain approximately the same distance away in the lower-dimensional representation that is created. A consequence of this is that local ordering in the high-dimensional space often is warped or destroyed, thus preventing study of clustering and other meso-scale patterns in the data set. A recently developed technique for dimensionality reduction, t-Distributed Stochastic Neighbor Embedding (t-SNE), is able to perform the precise opposite — maintaining local structure but creating long length-scale distortions. The price to be paid here, however, is that this algorithm is non-convex and has very poor scaling properties, going like N^2 in both time and memory. In this tutorial, I will introduce methods for implementing this algorithm and its variants, showing the relative strengths and weaknesses of this type of embedding. We will apply t-SNE to both conjured and real-world data using Matlab scripts and compare the results to more traditional non-linear embedding techniques. In addition, we will see how, under certain conditions and approximations, it is possible to scale this approach up to include hundreds of millions of data points.
Electrical & Computer Engineering
Carnegie Mellon University
Dimensionality Reduction in Neural Modeling
In this tutorial, we will cover dimensionality reduction methods that are commonly used to interpret neural population activity, including principal components analysis, factor analysis, and latent dynamical systems. We will implement these methods in MATLAB and apply them to population recordings from the motor cortex.
Techniques for Recovering Conceptual Structure from Text
Neural network approaches to the representation of words and images are now the state of the art. In this tutorial I will discuss two instantiations of this approach, word2vec and sent2vec, including how these techniques can be implemented in cython, a C++ extension to Python. Time allowing, we will also discuss t-SNE, a dimensionality reduction technique (like PCA) that is particularly well suited for visualizing high-dimensionality datasets. The example datasets will involve both words and images.
The Mathematics of Dimensionality Reduction Methods
Sponsored by CMBC with additional support from the Emory College Neuroscience Fund, the Institute for Quantitative Methods (QuanTM), and the Cognitive Neuroscience Training Grant (CNTG).
Cognitive Science and Religion
Over the last twenty-five years, scholars have brought the theories and findings as well as the tools and methods of the various cognitive and brain sciences to bear on religious thought and behavior. From its beginnings the cognitive science of religion has been a thoroughly inter-disciplinary undertaking, seeking to integrate formal modeling, experimental psychology, ethnographic research, and evolutionary insights. More recently, general trends in cognitive science, from deploying the tools of brain imaging to incorporating insights about embodiment, have swept across the cognitive science of religion as well, shaping accounts of religious beliefs and representations (both mental and public), ritual and other forms of religious conduct, religious attitudes and values, and types of religious experience. This workshop explores some of the most prominent directions such research has taken over these twenty-five years, focusing specifically on some of the most exciting recent developments.
May 15-16, 2014
Speakers and Topics:
Cristine Legare (Department of Psychology, University of Texas at Austin): The Cognitive Foundations of Cultural Learning
Imitation is multifunctional; it is crucial not only for the transmission of instrumental skills but also for learning cultural conventions such as rituals (Herrmann, Legare, Harris, & Whitehouse, 2013; Legare & Herrmann, 2013). Despite the fact that imitation is a pervasive feature of children’s behavior, little is known about the kinds of information children use to determine when an event provides an opportunity for learning instrumental skills versus cultural conventions. In my talk I will discuss a program of research aimed at developing an integrated theoretical account of how children use imitation flexibly as a tool for cultural learning. I propose that the cognitive systems supporting flexible imitation are facilitated by the differential activation of an instrumental stance (i.e., rationale based on physical causation) and a ritual stance (i.e., rationale based on cultural convention). I will present evidence that the instrumental stance increases innovation and the ritual stance increases imitative fidelity, the dual engines of cultural learning.
Cristine Legare (Department of Psychology, University of Texas at Austin): The Coexistence of Natural and Supernatural Explanations across Cultures and Development
In both lay and scientific writing, natural explanations (potentially knowable and empirically verifiable phenomena of the physical world) and supernatural explanations (phenomena that violate or operate outside of, or distinct from, the natural world) are often conceptualized in contradictory or incompatible terms. My research has demonstrated that this common assumption is psychologically inaccurate. I propose instead that the same individuals frequently use both natural and supernatural explanations to interpret the very same events. To support this hypothesis, my colleagues and I reviewed converging developmental data on the coexistence of natural and supernatural explanations from diverse cultural contexts in three areas of biological thought: the origin of species, the acquisition of illnesses, and the causes of death (Legare, Evans, Rosengren, & Harris, 2012; Legare & Visala, 2011; Legare & Gelman, 2008). We identified multiple predictable and universal ways in which both kinds of explanations coexist in individual minds at proximate and ultimate levels of analysis. For example, synthetic thinking (i.e., combining two kinds of explanations without integration), integrative thinking (i.e., integrating two kinds of explanations by distinguishing proximate and ultimate causes), and target-dependent thinking (i.e., two kinds of explanations remain distinct and are used to explain different aspects of an event, depending on contextual information) all illustrate different kinds of explanatory coexistence. We also discovered that supernatural explanations often increase, rather than decrease, with age. Reasoning about supernatural phenomena, in short, seems to be an integral and enduring aspect of human cognition, not a transient or ephemeral element of childhood cognition.
Cristine Legare (Department of Psychology, University of Texas at Austin): Ritual and the Rationality Problem: Old Wine in a New Bottle
As a group, we will examine the kinds of ritualistic remedies used to treat a great variety of problems across highly diverse cultural contexts and vast stretches of historical time. Our objective will be to identify the kinds of information people may use to evaluate the efficacy of these pervasive cultural practices.
Cristine Legare (Department of Psychology, University of Texas at Austin): Evidence from the Supernatural: Evaluating Ritual Efficacy
Rituals pose a cognitive paradox: although widely used to treat problems, they are cultural conventions and lack causal explanations for their effects. How do people evaluate the efficacy of rituals in the absence of causal information? To address this question, I have examined the kinds of information that influence perceptions of ritual efficacy experimentally (Legare & Souza, 2012; 2013). I conducted three studies (N = 162) in Brazil, a cultural context in which rituals called simpatias are used to treat a great variety of problems ranging from asthma to infidelity. Using ecologically-valid content, I designed experimental simpatias to manipulate the kinds of information that influence perceptions of efficacy (e.g., repetition, number of procedural steps). The results provide evidence that information reflecting intuitive causal principles affects how people evaluate ritual efficacy. I propose that the structure of ritual is the product of an evolved cognitive system of intuitive causality.
E. Thomas Lawson (Queens University and Western Michigan University (emeritus)): Obstacles and Opportunities: Reflections on the Origins of the Cognitive Science of Religion
A focus on interpretation at the expense of explanation in the humanities, particularly religious studies, an insistence on the autonomy of the social sciences at the cost of underestimating the value of psychology, and an overemphasis on cultural differences while being blind to human commonalities in anthropology all presented obstacles to developing a theoretically sophisticated, empirically tractable science of religion until the cognitive revolution provide the means and methods to do so.
Greg Berns (Facility for Education and Research in Neuroscience (FERN), Emory University):Brain Imaging Studies of Sacred Values and Social Norms
We hypothesize that when people engage sacred values that underpin many political conflicts, they behave differently than when operating with the more mundane values of the marketplace and normal social interactions. Given the importance of sacred values, and their potential for triggering violent conflict, it is important to understand how sacred values become intertwined in decision making. Traditionally, this type of investigation has been the purview of anthropology and sociology. However, recent advances in functional brain imaging make it possible to use this technology to uncover biological signatures in the brain for sacred values and the neural systems that come online when they are violated.
John Dunne (Graduate Division of Religion, Emory University): Scientific Research on Meditation and the Cognitive Science of Religion: Anything Shared?
Scientific research on meditation has grown exponentially in the last two decades, yet that research often remains disconnected from the academic study of religion. Likewise, cognitive scientific approaches to religion often seem irrelevant to the scientific study of meditation. Why do these fields of research largely fail to interact, and what does it tell us about our notion of religion?
Robert N. McCauley (Center for Mind, Brain, and Culture (CMBC), Emory University): The Cognitive Science of Religion: Seminal Findings and New Trends
Theorists in the cognitive science of religion have proposed that many religious proclivities are by-products of garden-variety cognitive systems that humans share. This general theoretical proposal has generated a variety of notable experimental findings pertaining to such matters as the character and memorability of religious representations, the failure of religious participants to deploy orthodox beliefs in on-line cognitive processing, and the human penchant for “promiscuous teleology.” Subsequent influences on the cognitive science of religion over the past fifteen years do not differ from those affecting cognitive science more broadly. Perhaps the three most prominent of those influences concern evolutionary considerations, the growing availability of brain imaging tools, and an interest in religious experience and embodiment. Each has inspired experimental studies that have produced comparably significant findings concerning such topics as developmental regularities in reasoning about the afterlife, the impact of public ritual participation and other forms of costly signaling on commitment to religious groups (in particular), neural evidence implicating theory of mind in prayer, the impact of synchronous bodily movements on pain thresholds, and more.
Vernon K. Robbins (Graduate Division of Religion, Emory University):Conceptual Blending and Interactive Emergence in Early Christian Writings
In the context of three major Mediterranean modes of religious thought and practice—mythical, philosophical, and ritual—early Christians produced writings during the first century CE that exhibit six discursive-religious forms of life. The conceptual blending of time, space, and body in this discursive-religious environment created interactive emergence identifiable as prophetic, apocalyptic, wisdom, precreation, miracle, and priestly thought and practice. Rhetography, which is rhetoric that evokes graphic images and pictures in the mind, working interactively with rhetology, which is rhetoric that produces verbal argumentation, nurtured such energetic cognitive-conceptual blends that their effects are still observable in Christianity today. This presentation will feature a combination of past results and recent insights from Emory Sawyer Seminars on Visual Exegesis and Hermeneutics
Bradd Shore (Department of Anthropology, Emory University): Religion and Ritual: A Marriage Made in Heaven
While ritualized behavior is not exclusively associated with religious experience, there is clearly a powerful affinity between religion and ritual. A look at the evolutionary roots of human ritual and several of its cognitive and experiential characteristics sheds interesting light on some of the underlying reasons for this affinity.
Functional Magnetic Resonance Imaging (fMRI): What It Is, How It Works, and How to Use It in Your Research
May 17, 2013
Led by Kate Pirog Revill (Facility for Education and Research in Neuroscience [FERN], Emory University) with additional lectures by Emory faculty.
MORNING SESSION: THE BASICSKate Pirog Revill (Facility for Education and Research in Neuroscience [FERN], Emory University): Introduction to the Basics, Current Issues and Controversies
Faculty Neuroimaging Showcase:Greg Berns (Facility for Education and Research in Neuroscience [FERN], Emory University)
Lynne Nygaard (Psychology, Emory University)
Shella Keilholz (Biomedical Engineering, Emory University)
Jim Rilling (Anthropology, Emory University)
AFTERNOON SESSION: GETTING DOWN TO BRASS TACKS
Kate Pirog Revill (Facility for Education and Research in Neuroscience [FERN], Emory University):Safety Presentaion, FERN Tour and Scanner Demonstration, Data Processing Tutorial and DemonstrationThis one-day workshop focused on what fMRI is, what it can and cannot do, and how to use this technology in research. The morning session was an overview of what fMRI is and how to interpret fMRI findings, including a discussion of current methodological controversies and short demonstrations by facutly members currently using this technology in their research. The afternoon session focused on the steps and skills involved in conducting fMRI research. Following a brief discussion of fMRI safety, participants had the opportunity to tour the new FERN Center and observe a live scanner demonstration. The workshop concluded with a demonstration and discussion of fMRI data analysis.
What's Human About the Human Brain? Exploring Evolutionary Specializations of the Human Brain
May 25-27, 2011
Led by Todd Preuss (Yerkes National Primate Research Center and Emory University) including additional lectures by Jim Rilling, Dietrich Stout, and Craig Hadley (Emory Anthropology), and Krish Sathian (Emory Neurology), and included a visit to the Biomedical Imaging Technology Center at Emory Hospital.
How is the human brain distinct from that of other primates? Discover the evidence from a variety of methodologies and levels of analysis that reveal the evolutionary changes that contributed to the distinct structural and functional characteristics of the human brain.
Todd Preuss (Yerkes National Primate Research Center and Emory University): Introduction and History of the Study of Brain Evolution
Todd Preuss (Yerkes National Primate Research Center and Emory University): Fundamentals of Evolutionary Neuroscience
Jim Rilling (Department of Anthropology, Emory University): Structural Brain Imaging Methods and Applications
Krish Sathian (Department of Neurology, Emory University School of Medicine): Functional Brain Imaging
Dietrich Stout (Department of Anthropology, Emory University): The Archaeological and Paleontological Record of Human Cognitive Evolution
Dietrich Stout (Department of Anthropology, Emory University): Technology and Cognitive Evolution (with stone tool-making demonstration)
Jim Rilling and Craig Hadley (Department of Anthropology, Emory University): Evolution of Human Life History
Craig Hadley (Department of Anthropology, Emory University): Social and Cultural Evolution
All Speakers: Theories of Human Brain Evolution: Implications and Next Steps
May 25-27, 2010
Discourse analysis is a cover term for a loosely connected set of methods for doing research on texts and records of talk. Each method has been used mainly for certain topics rather than others and has a somewhat different intellectual and disciplinary lineage and somewhat different theoretical underpinnings. All, however, require close attention to the linguistic and (in most cases) non-linguistic details of human interaction: text structure, sentence structure, verb tense, mood, and modality, word choice, intonation, gesture, gaze, and so on. In this workshop we began with an overview of key concepts in the study of discourse, asking how "texts," "conversations," and the like get plucked out of the flow of experience and how "texts" shape "contexts" and vice versa. We then turned to two sets of topics of particular interest to workshop participants: methods for studying talk in interaction, and narrative and identity. Along the way, three Emory faculty members presented their research and other participants shared the data they are working with.
Barbara Johnstone (Department of Rhetoric and Linguistics, Carnegie Mellon University): Key Concepts: Text, Context, and Entextualization
Barbara Johnstone (Department of Rhetoric and Linguistics, Carnegie Mellon University): Face to Face Interaction: Conversation Analysis and Interactional Sociolinguistics
Debra Spitulnik (Department of Anthropology, Emory University): Interdiscursivity, Stance, and Voicing
Barbara Johnstone (Department of Rhetoric and Linguistics, Carnegie Mellon University): Narrative and Identity: Sociolinguistic Approaches
Robyn Fivush (Department of Psychology, Emory University): Narrative and Identity: Theory and Method across Disciplines -- Narrative Theory for Psychological Research
Roberto Franzosi (Department of Sociology, Emory University): Narrative and Identity: Theory and Method across Disciplines -- Quantitative Narrative Analysis for Socio-historical Research
Barbara Johnstone (Department of Rhetoric and Linguistics, Carnegie Mellon University): Workshop on Your Data
Computational Modeling of Complex Human Systems
Sue Becker, McMaster University,
Dieter Jaeger (Biology, Emory University)