Brian Uzzi presents two studies on innovation in science. Study one examines conditions favoring the generation of innovative, high impact ideas and study two examines the conditions that effectively communicate innovative ideas to others. In study one, he examines the role of gender diverse teams in the creation of high impact and novel research relative to same gender teams. The study analyzes of over 26 million scientific papers. They find that mixed gender teams across in medical science produce more highly cited and more novel research papers than same gender teams, a relationship that increases with the team’s gender balance and regardless of whether the team’s leader is a woman or a man. Further, despite the impact of gender diverse teams, and their broad generalizability across science, gender diverse teams remain underrepresented in science – potentially slowing innovations. In study 2, they investigate how the merits of innovative ideas communicated in science. Here we conduct semantic analyses of grant application success with a focus on scientific promotional language, which purportedly helps to convey an innovative idea’s originality. Our analysis attempts to surmount limitations of prior studies by examining the full text of tens of thousands of both funded and unfunded grants from three leading public and private funding agencies, the NIH, NSF, and Novo Nordisk Foundation, one of the world’s largest private science foundations. We find a robust association between promotional language and the support and adoption of innovative ideas. They find that the percentage of promotional language in a grant proposal is associated with up to a doubling in its probability of being funded, with a grant’s inherent level of innovativeness, and with its predicted citation impact and productivity. Lastly, a computer-assisted experiment that manipulated the promotional language in our data helps to uncover how promotional language may communicate the merits of ideas through cognitive activation.

Learning Objectives: 

1. Review two-step process of innovation in science.

2. Demonstrate the link between gender diverse teams and the generation of innovative and high impact ideas.

3. Demonstrate the link between promotional language successful grant applications and high impact publications.

The rapid growth of technologies capable of comprehensively collecting, managing, and allowing us to gain an in-depth picture of the brain, nervous system, and behavior across diverse spatial and temporal dimensions, as well as across different scales of biological organization and data modalities has opened unprecedented avenues for understanding human experiences within their social and physical contexts. In this talk I will discuss some of the key ethical and societal considerations that are raised with these new capabilities. While these cutting-edge capabilities hold immense potential for advancing our knowledge, critical questions must be addressed. This talk will explore inquiries such as what can and should be measured, as well as the ethical considerations surrounding the access and utilization of this wealth of data. From safeguarding user privacy to navigating the potential privacy risks associated with capturing multimodal and multiscale data for individuals and communities. In addition, I will discuss biases, encompassing considerations of which data is deemed superior and the biases embedded in algorithms used to interpret these data. Addressing the potential dissonance between digital phenotypes and self-reported measures further complicates the ethical terrain. Recognizing that a one-size-fit-all solution is elusive, the presentation emphasizes the importance of establishing responsible governance data frameworks. Such governance is essential to ensure the quality, availability, findability, accessibility, usability and utility of these multidimensional datasets. Moreover, a responsible research and innovation can help us navigate the challenges posed by these transformative technologies, ensuring that their societal impact aligns with shared values and principles.

Learning objectives:

1. Identify key ethical and societal considerations raised by the comprehensive, continuous, real-time collection of behavioral data and the use of that data.

2. Recognize the importance of establishing and using responsible governance data frameworks.

3. Assessing facilitators and barriers for ensuring that the societal impact of these data collection and use aligns with shared values and principles.

The diagnosis, treatment and study of psychiatric illnesses relies primarily on subjective impressions and clinical ratings of behavior. Unlike other fields of medicine, psychiatry does not have objective tests that can be used for diagnosis and treatment. In psychiatry, therefore, there is a clear need to operationalize and quantify behavior, in particular behavior related to social communication. Human communication entails a rich repertoire of expression, including spoken language, face expression and gaze. Language has evolved in humans within a social context, beginning with the parent-infant dyad, and mirroring of sounds and facial expressions. Its natural ecology is face-to-face interpersonal interactions, both in-person and increasingly remotely via platforms such as Zoom, including telepsychiatry. Human dyadic social communication entails a rich repertoire of expression, including not only language, but also face expression, acoustics (prosody), pauses and turn-taking, gaze (shared attention), and gestures, highly coordinated between speakers, with regular turn-taking and alignment of facial expression, gesture, semantic content, and speech rates. Overall, given the synchronization and speed needed to successfully communicate with others, it is remarkable how well this process normally works. However, in schizophrenia and its risk states, communication frequently breaks down, leading to frustration, isolation and significant difficulties in social and role function.  This presentation will describe the development of objective measures of dyadic social communication behavior using natural language processing and computer vision, with schizophrenia and its risk states as a use case.

Learning Objectives:

1. Describe language disturbances in schizophrenia and its risk states.

2. Describe face expression disturbances in schizophrenia and its risk states.

3. Describe ethical and technical issues related to computational phenotyping.

The nervous system of social species has evolved to perceive and evaluate signals within a social context. Social information therefore must impact how the brain processes information, yet little is still known about how the brain integrates social information to produce actions in a social context. This lack of knowledge exists in part because social context is difficult to quantify and because most studies are performed in species that do not have a particularly rich social structure. Here Marc Schmidt will discuss ongoing work with the brown-headed cowbird (Molothrus ater), a highly gregarious songbird species whose social behavior has been well studied and where vocal and non-vocal communication signals form a central and critical component of its social system.

Marc's team has created a “smart aviary” equipped with an array of cameras, microphones, and other sensors with the aim to generate a fully automated pipeline to evaluate moment-to-moment behavioral interactions in a group of birds over the entire breeding season using machine learning approaches that track each bird’s position, orientation, pose, and shape directly from images. Supplied with their rich dataset, they aim to develop mathematical tools necessary to generate social network models that will allow us to quantify the specific state of the bird social network associated with neural activation patterns associated with individual behavioral events. To evaluate brain dynamics, they are developing a miniature non-invasive wirelessly powered and transmitting recording device optimized for long duration recording that critically does not impact bird individual or social behavior.

In Marc's talk, he will use our proposed work to track social behavior, and its neural correlates, in our “smart aviary” to highlight the new emerging field of “neuroscience in a naturalistic context”, highlighting both the possibilities and challenges that this new direction presents.

Learning Objectives:

1. Review changing opportunities and challenges in the study of the neural bases of behavior.

2. Recognize advantages and disadvantages of using birds as a model system for studying the neural bases of complex behavior within a social context.

3. Demonstrate knowledge of the different technical elements and expertise required for studying brain function in a naturalistic context.

Parkinson’s disease (PD) belongs the group of synucleinopathies characterized by progressive aggregation of the neuronal protein alpha-synuclein (asyn). Diagnosis and evidence of disease pathology in tissue rely on the demonstration of loss of dopaminergic neurons in substantia nigra, pars compacta and the presence of Lewy bodies in some of the remaining neurons. Lewy bodies are intraneuronal inclusions rich in aggregated asyn. Lewy-like inclusions (Lewy bodies in cell bodies, and Lewy neurites in axons) represents the appreciated pathology in PD and dementia with Lewy bodies. Such inclusions are also used as a marker of asyn pathology in cell and animal models. The patients typically suffer from a range of non-motor symptoms the become worse during the disease. Recent data demonstrates that wider areas of the nervous system is affected by the presence of asyn aggregates than what can be appreciated by immunohistochemical staining of Lewy bodies, and such aggregates can also be detected in biofluids. This less appreciated asyn aggregates tells us that PD initiates long before diagnosis and impact the nervous tissue on a larger anatomical scale than anticipated from the presence of LB. The nature of these non-LB aggregates is still unknown, like their pathophysiological impact but they have contributed to the newly proposed biological staging systems for PD and Lewy body dementia. Their presence in apparently unaffected neurons suggests they are strongly toxic but may contribute to symptomatology by making neuronal circuitries dysfunctional. This opens for new potential symptomatic treatments and disease modifying strategies.

Learning Objectives: 

1. Discuss different hypotheses for the development and progression of Parkinson’s disease.

2. Explain what neuropathology represents the established criteria for diagnosing Parkinson’s disease and their limitation.

3. Discuss the basis for novel analyses that demonstrated alpha-synuclein aggregates in tissue and biofluids that cannot be detected by the standard techniques used to demonstrate the established alpha-synuclein pathology.

Speech neuroprostheses have the potential to restore communication to people living with paralysis, but naturalistic speed and expressivity are elusive. Here we use high-density surface recordings of the speech cortex in a clinical-trial participant with severe limb and vocal paralysis to achieve high-performance real-time decoding across three complementary speech-related output modalities: text, speech audio and facial-avatar animation. We trained and evaluated deep-learning models using neural data collected as the participant attempted to silently speak sentences. For text, we demonstrate accurate and rapid large-vocabulary decoding with a median rate of 78 words per minute and median word error rate of 25%. For speech audio, we demonstrate intelligible and rapid speech synthesis and personalization to the participant’s pre-injury voice. For facial-avatar animation, we demonstrate the control of virtual orofacial movements for speech and non-speech communicative gestures. The decoders reached high performance with less than two weeks of training. Our findings introduce a multimodal speech-neuroprosthetic approach that has substantial promise to restore full, embodied communication to people living with severe paralysis.

Learning Objectives: 

1. Explain the need for reliable speech-neuroprostheses for restoring communication

2. Summarize the current most accurate approach to decoding speech from the brain.

3. Identify speech brain-computer interfaces as a positive use case for AI for benefiting people in need.

The study of psychopathology is changing with the advent and pervasive use of smartphones, wearables, and connected devices together with significant advances in brain imaging. These new avenues offer a unique opportunity to understand phenotypic variation in a large population, to track individuals longitudinally, and relate these data to life and clinical outcomes. Despite modern sensors’ ability to capture real-world behavioral data at high temporal frequency and over long periods of time, access to and availability of longitudinal data about individuals is severely lacking. In this talk, a few of these approaches connecting brain imaging, voice, and language to psychopathology are presented through the lens of analytical approaches spanning deep learning, signal processing, and natural language processing. The integration of these sensors and newer analytic techniques will enable a change from the current symptomatic and subjective assessment and treatment of mental health to direct quantitative measurement of behavior and brain function and subsequent adjustments to alter mental health.

Learning objectives:

1. Summarize how different data modalities provide different views of brain disorders.

2. Review the use of ML/AI tools in psychopathology research.

3. Explain the role of voice, speech and language as a window into the mind.

The ability to record every spike from every neuron in a behaving animal is one of the holy grails of neuroscience. Here, Alison Hanson discusses a collaborative effort aimed at coming one step closer towards this goal with the development of an end-to-end pipeline that automatically tracks and extracts calcium signals from individual neurons in the cnidarian Hydra vulgaris. Together, her team imaged dually labeled (nuclear tdTomato and cytoplasmic GCaMP7s) transgenic Hydra and developed an open-source Python platform (TraSE-IN) for the Tracking and Spike Estimation of Individual Neurons in the animal during behavior. The TraSE-IN platform described, here, comprises a series of modules that segments and tracks each nucleus over time and extracts the corresponding calcium activity in the GCaMP channel. Another series of signal processing modules allows robust prediction of individual spikes from each neuron’s calcium signal. This complete pipeline will facilitate the automatic generation and analysis of large-scale datasets of single-cell resolution neural activity in Hydra, and potentially other model organisms, paving the way towards deciphering the neural code of an entire animal.

Learning Objectives: 

1. Explain the advantages of using Hydra as a simple system in neuroscience.

2. Review the advantages of using deep learning for segmenting individual nuclei.

3. Summarize how and why the TraSE-IN pipeline was generated and its potential uses for dually labeled organisms.

To make movement and foraging decisions in a naturalistic environment, multiple neural populations must work synergistically to produce successful actions. These decisions span multiple scales, from kinematic control of vigor, to broader decision motifs associated with a given context. We have recently developed a suite of tools for rapid, accurate behavioral segmentation that now allows for the identification and interrogation of spontaneous, naturalistic behaviors. This suite includes both unsupervised (B-SOiD) as well as supervised algorithms (A-SOiD) which learn behaviors and add interpretability to pose estimation data. We simultaneously recorded from neurons across motor cortex and striatum as a mouse freely explored an open-field over the course of several days. We then these techniques to identify distinct movements, behaviors, and action motifs. B-SOiD identified significant patterns of co-activation across neurons in these areas. Surprisingly, all of the identified behaviors elicited a robust neural signature, though we note several differences in the sparsity and dimensionality of the representation in different areas. A simple decoder based upon individual single-neuron spike rates was able to predict the behavior performed at a level many times greater than chance, demonstrating the strength of the neurobehavioral relationships. This strength also enables the responses of single cells to be well-explained by a small set of underlying ‘latent’ dimensions. Traditional methods involving linear tools (e.g. PCA, GPFA) have been successful in explaining shared variability in neural populations during artificial, overtrained tasks; however, how populations covary in naturalistic settings absent reductionist tasks and rigid restrictions on movement has yet to be evaluated. We show that the naturalistic behaviors of mice in the open field require an appreciation of more complex dimensionality. Specifically, we observed that many latent dimensions were nonlinearly related to recorded neural firing rates, particularly in dorsal striatum. These results, including the differing levels of complexity underlying the representation of behavioral features, shed new light into how the coordinated translation of information between areas orchestrate complex, ethological behaviors.

Learning Objectives: 

1. Understand the difference in substance and application between supervised and unsupervised machine learning algorithms.

2. Determine a baseline understanding of how to apply the B-SOiD behavioral segmentation platform to study a variety of behaviors and datasets.

3. Identify basic differences in how motor cortex and basal ganglia represent spontaneous, naturalistic behavior.

The ultimate goal of neuroscience is to understand and explain real-world behavior in terms of brain activity, and to use these insights to develop therapeutic approaches for neural disorders. By using mobile recording devices synchronized with intracranial EEG recordings in epilepsy patients with an implanted deep brain recording system, we can study the neural basis of human activities such as navigation and real-world memory encoding in a way that captures the complexity, scale, and functional characteristics of real-world experiences. In this talk, I’ll describe an experiment in which we asked five participants to learn a 0.75-mile route around campus while hippocampal electrophysiology was recorded. In order to gain measurement control over this rich, real-world, and complex experiment, we synchronized a variety of wearable sensors (eye-tracking, GPS, accelerometry, psychophysiology, etc.) with the participant’s hippocampal recordings and 1st person experience with millisecond precision. Using this synchronized system, subjects walked the route 7-8 times across two days, with the 1st walk guided (encoding) and 6-7 of the walks navigated by the participants themselves (navigation retrieval; 28.5 total miles). Findings across all participants suggest that theta band power significantly increases when participants are navigating outdoors relative to indoor navigation. We also find evidence that temporal lobe theta band power changes immediately around spatial event boundaries. Taken together, these initial neural findings support our hypothesis that medial and lateral temporal lobe activity changes around real-world event boundaries. Future studies will investigate the potential of direct brain recordings to capture the processes of autobiographical memory formation and direct brain stimulation to enhance episodic memory for real-world experiences.

Learning Objectives: 

1. Describe the importance of understanding cognitive neuroscience in the wild, complex, real-world.

2. Review the new opportunities for doing cognitive neuroscience using naturalistic paradigms.

3. Demonstrate the insights that can be gained by recording neural data during real-world behaviors.

Transdisciplinary knowledge producing teams (TDKPTs) dedicate their efforts to solving wicked problems. Successful TDKPTs embody characteristics of complex adaptive systems (CAS) and exemplify behaviors associated with human adaptation. Individuals within these types of work groups are highly interactive and perform within complex environments where structural and interactive dynamics are constantly at play. We use a complex adaptive systems (CAS) lens to identify and define the features of TDKPTs and suggest how team members might develop skills that are more congruent with complex adaptive conditions. Acknowledging these dynamics and mapping the skills required to effectively work within these teams is at the heart of ensuring that teams meet their required outcomes. Once identified, behavioral measurement techniques can be devised that link behaviors to team effectiveness allowing for further research to be conducted using these behavioral competencies as indicators of team member readiness and abilities to work in complex adaptive environments. Science-of-Team-Science (SciTS) and complexity and systems theory contribute to construct a typology of the features of TDKPTs based on systems and interactive complexities. A list of features and corresponding skills foci are developed from a diverse body of literature useful for considering complexity within TDKPTs.

Learning Objectives:

1.  Evaluate a cross-disciplinarity team approach using complex adaptive systems in Transdisciplinary Knowledge Producing Teams (TKPTS). 

2. Identify complexity challenges for TKPTS.

3. Generate Complexity focused competencies for TKPTS.

Virginia will discuss work on single-trial analysis of EEG signals during episodic memory tasks. They develop single-trial classifiers to predict whether people will remember pictures from EEG recorded prior to and during study presentation. The single-trial classification analysis provides discriminative dimensionality reduction which can be used to project untrained data, including untrained classes, to reveal novel electrophysiological distinctions between behavioral states. They also analyze EEG signals recorded during the test phase and show that they can model all 13 different classes of trials from the combination of participants' objective memory performance (e.g. correct rejects, false alarms,...) as well as their perceived performance (e.g. sure new, maybe new ...) as a function of only three different basis classifiers. Together these results show how projecting untrained classes onto classifiers trained on other classes can help decompose EEG signals and reveal novel electrophysiological correlates of behavior.

Learning Objectives:

1. Review the basic terms and concepts in recognition memory research.

2. Summarize the basic limitations of EEG signals as well as some notable features that relate to memory processing.

3. Recognize the benefits of single trial classification for EEG analysis.

Robots are useful tools for studying the nervous system and its control of body mechanics because they can be used to perform experiments that would be difficult to perform with an animal. For example, in a robot, data can be recorded from every sensor on the body simultaneously as the robot performs a task. Such data may approximate what sensory feedback the animal would experience while performing the same task. Unanticipated results may inspire future experiments to describe the function of the animal’s nervous system in greater detail, and the robot can be used to generate concrete hypotheses for such experiments. The utility of a robotic model is greatest when it mimics key characteristics of the model organism, e.g., its kinematics, dynamic scale, sensing modalities, and computation and control.

To study the sensorimotor control of legged locomotion, my group constructs robots modeled after walking insects. When modeling a particular species, the kinematics of the robot are designed to mimic those of the animal. To ensure that the same forces (e.g., gravity, inertia, elastic) dominate the motion of the robot and the animal, the speed of the robot’s motion is dynamically scaled to that of the animal. To provide the control system with sensory feedback similar to that which the animal experiences, the robot has special sensors whose signals are processed through biomimetic filters. These filters can provide necessary adjustments for inherent differences in the mass, material, and other properties between the robot and the animal. To create control systems that are constrained to the types of computations the nervous system performs, the robot’s control system is a real-time simulation of neural dynamics. The resulting robot is a substrate onto which to accumulate experimental discoveries and test that particular mechanisms function as understood in vivo.

Learning Objectives: 

1. Describe some ways in which body mechanics affect the neural control of movement.

2. Explain some interdisciplinary methods for investigating nervous system function using biomimetic robots.

3. Evaluate the applicability of a biomimetic robot to the organism it is modeling.

1. Review background on deep brain stimulation for OCD.

2. Discuss recording capable deep brain stimulation devices.

3. Demonstrate knowledge on how recording-capable DBS devices have been used to identify neural biomarkers of OCD symptoms.

With the internet, society has been exposed to puzzling and exquisite animal behavior, while neuroscience has vastly concentrated on a few inbred animal models studied in trained unnatural settings. Changing this trend is imperative. The onset of wireless recording technologies has allowed the development of more naturalistic inquiries into the brain and behavior of animals. By developing a novel behavioral paradigm in neuroscience, playing ‘Hide and Seek’ with rats, we were able to study mammalian play and its neural bases. We played ‘Hide and Seek’ with rats in a 30 square meter room, and found that they acquired the game easily and played by the rules. Rats played strategically and, without being conditioned, developed remarkable game specific vocalizations patterns. The freedom afforded to our subject animals allowed us to probe new forms of linking neural and behavioral phenomena. We reversed the arrow of inquiry by doing unsupervised clustering of neural population data allowing us to blindly refer back to the behavior of the rat during the game and uncover neural-behavioral associations unsuspected by the experimenters. My work in large scale wireless rat play laid the foundation for taking novel neurotechnologies out of the lab to study the brains and behavior of animals in the wild.

Learning Objectives: 

1. Review play in rats and its capabilities for enhancing animal welfare.

2. Explain the reverse physiology approach and wireless technologies.

3. Summarize Hidden Markov Models.

Infant motor development is measured clinically using short, subjective, observational scales. While these scales are useful, they have limitations. They provide us with a “snapshot” of the infant’s performance at a certain point in time, when the infant may be tired or fussy, and do not capture the entire repertoire of an infant’s motor abilities. Further, due to their subjective nature and short observation window, these assessments lack sensitivity and specificity. Wearable sensors provide the ability to capture measurements across full days, providing quantitative data that can capture an infant’s true repertoire of movement skills. Wearable sensors have the potential to provide much more sensitive and specific assessment of motor development than current scales. This would support more accurate early identification of atypical development and identify more subtle responses to intervention. I will discuss the potential of using wearable sensors to measure infant movement patterns across full days in the natural environment, as well as limitations and challenges. In addition, I will discuss the potential of integrating wearable sensor measures of behavior with electroencephalography measures of brain function, as well as limitations and challenges.

Learning Objectives: 

1. Discuss 2 potential applications of wearable sensors to improve infant assessment and intervention.

2. Identify 2 challenges to application of wearable sensors to improve infant assessment and intervention.

3.Identify 1 challenge to the integration of wearable sensor measures of behavior with electroencephalography measures of brain function.

How do the cognitive and physiological traits of individuals translate into real-world group dynamics? There is an empirical and theoretical gap between understanding individual-level processes driving behavior, and the group-level outcomes of these processes. In this talk, I will present an overview of my group’s work in African ape sanctuaries aimed at bridging this divide. African sanctuaries have become a new focus of research on primate behavior, cognition, and health. Primates living in African sanctuaries are typically wild-born and semi-free-range in species-appropriate habitats as part of large social groups with variation in age and sex, but are also familiar with humans to a degree not possible in wild populations. This combination of naturalistic ecological and social contexts, where controlled interventions are also possible, makes these primate populations well-positioned to bridge traditional research approaches that focus either on experimental manipulations with captive animals, or observational approaches of behavior in the wild. I will present an overview of our work using an integrative approach to examine the cognition, behavior, and health of chimpanzees, one of human’s closest living relatives and a key model system for understanding the biological origins of human behavior. This work hinges on strong reciprocal interactions with wild field work focused on long-term observational research, to synthesize the strengths of both approaches. Finally, I will discuss how collaborations between sanctuary researchers and field researchers can build a new research program to understand how complex behavior emerges across different contexts.

Learning Objectives:

1. Explain why research with chimpanzees is relevant for understanding human cognition, behavior, and health.

2. Review the experimental methods to characterize individual and developmental differences in decision-making and cognitive control in non-human animals.

3. Infer why lab-living chimpanzees might have different patterns of health compared to wild or semi-free-ranging sanctuary chimpanzees.

Studying motor behaviors can reveal fundamental principles of the brain and mind. This presentation focuses on our work towards developing the hardware and computational tools that will allow scientist to measure movement of an individual in a dynamic environment. The octopus has many features that makes it advantageous for pursuing a holistic understanding of motor control and behavior and mapping central and peripheral neural circuits. We will show novel technologies to record electrophysiology signals from octopus’s nervous system and the potential of new computer vision platforms to measure octopus movement with high resolution. Using these approaches, we can start modeling octopus arm movement with high precision. These methods could improve brain-machine interface neuroprostheses and help develop new assistive technologies that will allow compensation and augmentation of specific circuits. We will also demonstrate how understanding motor behavior is critical to measure normal development in a translational animal model, and how neurotrauma interferes with normal motor behaviors.

Learning Objectives:

1. Review concepts of motion analytics.

2. Discuss new computational technologies to analyze movement.

3. Summarize how to analyze motor behaviors in animal models.

Recent advances in gene editing has allowed the generation of advanced rodent research models, mimicking some aspects of human conditions. For neurodegenerative disorders such as Alzheimer’s Disease, a long list of models exist with variable pathologies and not always convincingly linking pathology with behavioural phenotypes. As the field is moving more towards the use of in vitro models, it is nevertheless recognised that validation of cellular research and drug discovery must ultimately include systems endpoints, so that e.g. benefits and detrimental impacts of treatments in terms of health, behaviour and most importantly cognition are observed.

In order to develop a paradigm that allows accelerated insights into multiple behavioural aspects, we have designed and validated a novel behavioural paradigm, OF-NO-SI, which comprises activity measures in a circular open field (OF) arena, followed by a novel object (NO) stage and lastly presentation of a sex and age-matched conspecific for social interaction (SI) in one session. Each segment consisted of two repeated trials to measure within- and between-trial recognition memory. Data were correlated with post-mortem molecular outcomes.

This presentation will discuss results from different mouse lines (rTg4510 tau over expression model, 5X APP overexpression model, and in house generated PLB lines expressing low levels of tau and/or APP). During OF-NO-SI, transgenic mice displayed deficits in object and conspecific exploration and recognition memory. The paradigm robustly detected differences between models, genotypes and sexes for activity and cognitive measures.

Both overexpression models, 5X and rTg4510, displayed hyperactivity and spent less time interacting with object/stranger compared to WTs. Compared to low expression models (PLBAPP/PLBTau), the overexpression models differed only in their sustained hyperactivity. Despite being on the same genetic background, ‘control’ animals from different lines did not behave the same. Strongest correlations between behavioural and molecular phenotypes were detected for inflammatory markers.

Overall, disease-relevant behavioural phenotypes were identified, but our results identified over-expression per se as a potential confounder, and emphasise essential issues with control littermates.

Learning Objectives: 

1. Discuss and compare genetically modified mouse models of dementia.

2. Explain a new stream-lined behavioural tests paradigm.

3. Describe behavioural and cognitive phenotypes detected in the paradigm.

Social interactions are essential for the survival of social animals. Across species the medial prefrontal cortex (mPFC) has been identified as a hub of social cognition. In mouse models we can dissect the neural circuits and dynamics underlying the role of mPFC in specific social behaviors. Combining novel behavioral assays, optogenetics, machine learning and wireless electrophysiology in mice, we demonstrate a role for the mPFC-hypothalamic pathway in driving social competition. We also discuss recent unpublished work regarding prefrontal and subcortical network interactions that change with competitive success. Finally, we explore the potential relationships between social memory and social rank encoding in the mPFC.

Learning Objectives:

1. Learn how social rank is measured in mouse models.

2. Demonstrate knowledge about the role of the prefrontal cortex in social competition.

3. Demonstrate knowledge about prefrontal subcortical interactions during social competition.

Many behaviors are learned through trial and error by matching performance to internal goals, yet neural mechanisms of performance evaluation remain poorly understood. We recorded basal ganglia–projecting dopamine neurons in singing zebra finches as we controlled perceived song quality with distorted auditory feedback. Dopamine activity was suppressed after distorted syllables, consistent with worse-than-predicted performance, and activated when a predicted distortion did not occur, consistent with better-than-predicted performance. Thus, dopaminergic error signals can evaluate behaviors that are learned, not for reward, but by matching performance to internal goals. We then developed new computational methods to show that spontaneous dopamine activity correlated with natural song variations, demonstrating that dopamine can evaluate natural behavior unperturbed by experimental events such as cues, distortions, or rewards. Attending to mistakes during practicing alone provides opportunities for learning, but self-evaluation during audience-directed performance could distract from ongoing execution. It remains unknown how animals switch between, and process errors during, practice and performance modes. When male zebra finches transitioned from singing alone to singing female-directed courtship song, singing-related error signals were reduced or gated off and dopamine neurons were instead activated by female calls. Dopamine neurons can thus dynamically alter their tuning with changes in social context.

Learning Objectives:

1. Discuss three advantages of the songbird as a model system in neuroscience.

2. Describe the nature of the signal provided by dopamine for song learning.

3. Compare the dopamine signal during solo practice to female-directed performance in zebra finches.

This talk will present the methodology of using the guided ultrasonic waves generated by the rodent to track its physiology, behavior, and psychological state. When behaving in an environment, muscle contractions (voluntary and involuntary) apply subtle forces to the surfaces the animal contacts. These forces generate elastic waves, which propagate through the material (i.e., waveguide) at ultrasonic frequencies and contain information about the mouse's physiology, behavior, and underlying mental states. To date, this stream of data has been ignored. If the information in this data were mined, it would lead to the development of a novel, non-invasive way to study rodent behavior.

In the experiments, an aluminum plate was used as the floor of an open field for the animal to stand and move on. As the rodent behaved, moved, or reacted in this modified open field, the subtle voluntary and involuntary forces applied to the plate's surface resulted in the generation of guided ultrasonic waves (GUWs) in the plate. Highly sensitive ultrasonic sensors were bonded to the bottom of the plate to detect and acquire the generated waves.

Two types of GUWs exist in the plate: the Lamb waves and shear horizontal (SH) waves. Lamb waves are characterized by in-plane and out-of-plane particle vibrations, while in the SH waves, particles vibrate in-plane perpendicular to the direction of propagation. Both these waves are dispersive (the wave’s velocity varies with frequency) and multimodal (wherein multiple modes can propagate simultaneously). These properties enable (1) differentiating GUWs resulting from physiological functions from the ones associated with ambulation and behavior and (2) conducting an intricate analysis of rodent behavior. Owing to the dispersive and multimodal nature, wherein a minuscule change in the applied energy (i.e., the frequency) leads to changes in the mode profiles (change in velocity, wavelength, and types of modes), the differences between involuntary forces (arising from heartbeat and breathing) and voluntary forces will cause a noticeable change in the generated GUWs, i.e., their wave characteristics. Further, as slight changes in the behavior and ambulatory movements will cause significant changes in the characteristics of the generated waves, variations between and within behaviors can also be identified with GUWs. Two types of experiments were conducted with male and female mice from two strains, C57BL/6J and 129sv/evTac mice. In these experiments, each mouse was treated as an acoustic source, and the generated GUWs were acquired as acoustic emissions (AE) based on the amplitude threshold-based data acquisition. This converts the continuous data stream into AE hits, such that each AE hit is a consequence of a behavior/movement of the animal. Each AE hit is analyzed in terms of the features of the AE hits, like Duration, Amplitude, Absolute Energy, Rise Time, etc. Effectively, this analysis converts the information in the physiological process/behavior into values of various AE hit features and thus enables a simpler mode of analysis. In the first set of studies, the mice were placed one at a time into the open field arena and allowed to explore freely for five minutes without any external stimulus. It was observed that every movement/behavior of the animal resulted in the generation of the Lamb wave packets in the frequency range of the R6a sensor (30-100 kHz). This demonstrated the capability of the R6a ultrasonic sensor to detect even the subtlest movements of the animal—which were imperceptible even in the visual feed. Further, plots also showed the technology's ability to identify variations in wave signatures in accordance with different behaviors and variations within the behaviors. In contrast to the R6a sensors, the sensors used to measure SH waves (in the range 75 to 230 kHz) respond to selective behaviors like sniffing and rearing. Simply put, SH modes in the 75 to 230 kHz range were generated when the animal reared (or sometimes sniffed), and this was a consistent observation across all the tested C57BL/6J mice. In addition to this, we were able to spot variations within the supported and unsupported rearing. Thus, with the AE data, we could note a variability within the behavior, which was not discernible visually. These results showed that GUWs contained valuable information on behaviors that can be used not only to classify them but also to identify variations within them. We believe that with the information in all the AE features across the entire frequency spectrum (20 to 600 kHz), all the known behaviors can be classified through a deep neural network model.

In the second type of experiment, while the animal was in the open field, the tweeters attached to the open field played an auditory startle stimulus (4 kHz, 110 dB, and 10 msec) to evoke a startle response from the animal. The free-moving animal’s response was quantified from the GUW data. Certain features of the AE hits were analyzed to interpret the animal’s startle response. Particularly, features like the absolute energy, duration, and amplitude revealed physical details of the animal’s reaction. The absolute energy of the AE hit quantified the “intensity” of the startle. Reaction time, calculated from the Time-of-hit, gives the time the animal takes to respond to the tone (from the start of the tone). Duration specifies the length of time the animal applied the “high-intensity” force, and Amplitude is an indirect indication of the maximum amplitude of the applied force. We hypothesize that the intensity of the startle (represented by the absolute energy of the AE hit) reflects the animal's psychological state, specifically regarding stress and anxiety. With just a video-based analysis, such detailed quantification of startle is imprecise. On the other hand, the evaluation from AE data offers a reliable and multi-dimensional description of the acoustic startle response. In our tests, owing to the animal being in an unconstrained environment, a wider variety of startle responses were observed, including retraction, jump, head/body jerk, and halt/freezing. Thus, we hypothesize the AE-based open-field startle analysis will yield richer, more informative insights about psychological state and the influence of behavioral, pharmacological, and neural manipulations. The results of these studies showcase sensors’ ability to detect and measure the subtlest motions and demonstrate the presence of valuable information in GUWs, which we believe can be used to undertake physiological measurements and perform enhanced behavioral analysis. Physiological measurements (like heart and breathing rates) require the detection of waves originating from involuntary forces—much like the acoustic startle response. The success of performing detailed acoustic startle response studies attests to the GUWs-based technology's capability to undertake other parameter (heart rate and respiratory) measurements. The techniques used in the startle response studies will serve as the starting point of the research aiming to undertake physiological measurements. Similarly, the results of the open-field behavioral studies have demonstrated the AE technology's potential to enhance rodent behavior analysis. They will serve as a platform to develop deep neural network models to assess freely moving rodents' behavioral and psychological parameters.

Learning Objectives: 

1. Summarize the methodology to analyze free-moving rodents' behavior from the generated guided-ultrasonic waves (GUWs).

2. Identify the wave signatures associated with common behaviors (freezing and sniffing) to establish the uniqueness of generated waves for behaviors.

3. Review acoustic startle response tests in the GUW-enabled open field to obtain multidimensional data of the response.

Brain organoids are self-assembled three-dimensional cellular aggregates with cell types and tissue architectures that resemble the embryonic human brain. As they recapitulate many key features of early human brain, they have emerged as novel model systems to investigate human brain development and disorders in vitro. More recently, region-specific brain organoids have come forward as tools to study diseases and processes affecting specific parts of the brain. In this regard, midbrain organoids are essential to study one of the most affected regions in Parkinson’s disease, the midbrain. We make use of Parkinson’s disease patient specific midbrain organoids for the discovery and development of effective drug candidates, which target neurodegeneration. These models are complex and recapitulate many features of the human midbrain, including the presence of dopaminergic, glutamatergic, GABAergic and serotonergic neurons, astrocytes, oligodendrocytes, microglia and other cell types. We focus on disease relevant phenotypes including the loss of dopaminergic neurons, the appearance of alpha-Synuclein aggregates and neuroinflammation. Our phenotypic approach allows for the detection of disease associated conditions as well as their amelioration after drug treatment. Importantly, this is not limited to screening for efficacy of novel small molecule drugs, but can also cover the testing of anti-sense oligonucleotides, peptides or other therapeutic approaches. We are committed to the further development of our models and the improvement of assays and computational tools to perform deep phenotyping of our models to better detect cellular defects and their rescue following compound treatments.

Learning Objectives: 

1. Describe the concept of organoids and midbrain organoids.

2. Certify the use of 3D in vitro culture models for modeling complex neurodegenerative diseases.

3. Illustrate the application of the organoid technology for drug screening.

Individuals with substance use disorders often show impaired sensitivity to reward value. However, it is not clear whether such impairment impacts reward-related decision making and whether it can be quantified objectively via brain imaging modalities. This talk presents results from two studies: the first study employed reinforcement learning modeling to show that individuals with frequent alcohol use show impaired value-based modulation between computationally cheaper model-free (habitual) control and costly model-based (goal-directed) control in reward-related decision making, which in turn is associated with risk-taking behavior and anhedonia. The second study used electroencephalography (EEG) to objectively quantify reduced sensitivity to monetary value in individuals with chronic cocaine use, such that the EEG-derived signal(i.e., P300 amplitude) that is sensitive to reward value did not differentiate between high and low money reward in cocaine users, as it did in non-cocaine using controls.

Learning Objectives: 

1. Review reward sensitivity in substance use disorders.

2. Examine deficits in reward related decision making in frequent alcohol users, via computational modeling of task behavior.

3. Discuss impaired sensitivity to reward value using EEG in chronic cocaine users.

A core goal of neuroscience is to understand how the brain adaptively orchestrates movements to execute complex behaviors. Quantifying behavioral dynamics, however, has historically been prohibitively laborious or technically intractable, particularly for unconstrained and naturalistic behaviors which the brain evolved to produce. Driven by advances in computer vision and deep learning, new methods are being developed to overcome these limitations and enable precise and automated quantification of behavior from conventional across species and experimental settings. In this talk we will: introduce the problem of pose tracking for behavioral quantification; show how deep learning can be employed to achieve markerless motion capture; and highlight examples of how our work on making this technology accessible through open-source tools like SLEAP (sleap.ai) is enabling studies across domains and application areas ranging from social and motor neuroscience in flies, rodents, and primates, to ecology, digital humanities, and even plant biology to tackle climate change. We will conclude with preliminary results from a large-scale home cage phenotyping project designed to detect prodromal markers of disease.

Learning objectives:

1. Summarize deep learning technologies for markerless motion capture in animals.

2. Review open-source software tools that can be used for behavioral analysis (SLEAP and Keypoint-MoSeq).

3. Identify how behavior quantification technologies can be used to enable the identification of behavioral biomarkers of neurodegenerative disease progression.

A central goal for neuroscience is to understand how our brains process information in real life, such as faces during natural social interactions. While we have gained important knowledge using simplified controlled experiments in laboratory settings, they cannot capture all aspects of how cognition works in the real world. Understanding the neurocognitive basis of natural behavior in humans requires both innovative data collection paradigms and statistical-computational models to understand real world brain-behavior relationships. We harnessed multi-electrode intracranial recordings from hours of unscripted interactions participants had with friends, family, experimenters, and others. We demonstrate that we can reconstruct both what participants saw on a fixation-by-fixation basis during natural social interactions based on the neural activity and accurately reconstruct the neural activity patterns based on the visual input using bidirectional, multivariate machine learning models. The results highlighted the social-vision pathway as particularly important to natural face perception. Sharper tuning was revealed for the type of facial expression over its intensity; and for subtler expressions and movements around a person’s resting expression versus strong expressions – a Weber’s law for facial expressions. These results suggest that oval-shaped neural tuning for the kind and intensity of dynamic facial expressions reflects the neural code for real-world face processing. This work illustrates a paradigm that allows us to both robustly model the neural underpinnings of real-world social perception and test and develop novel hypotheses about how we encode natural faces.

Learning Objectives:

1. Review the paradigms for studying real-world neuroscience.

2. Summarize the computational tools needed to model real-world variability to build robust statistical models.

3. Demonstrate knowledge about the neural code for real-world face perception.

Many nonhuman primate species show sex differences in behavior, which suggests that there may also be sex differences in brain organization. In order to better understand the evolution of sex differences in the human brain, it is important to investigate neuroanatomical sex differences in other primate species. The capuchin monkey (Sapajus [Cebus] apella) is a neotropical primate species that shares several behavioral characteristics with humans that are relevant to social organization. Using magnetic resonance imaging, we conducted a whole-brain volumetric analysis of gray matter using voxel-based morphometry on T1-weighted scans from 20 capuchin monkeys (15 female, 5 male). While males showed significant expansion in areas of the hypothalamus, we found that females showed expansion in the early visual cortex, the cerebellum, and higher-order areas across the occipital and temporal cortex. Several of these regions showed non-overlapping ranges between males and females. We then investigated sex differences in white matter tracts using tract-based spatial statistical analysis on fractional anisotropy (FA) images from the T1-weighted scans. We found that females showed significantly higher FA than males in the right cerebral hemisphere, specifically in regions of frontal-parietal white matter, and these values were also non-overlapping. We suggest that distinct socioecological niches which male and female capuchins occupy could be associated with the sex differences found in cortical gray and white matter. The sex differences in neuroanatomy that we report here appear more pronounced than what is typically observed in humans. These differences could be associated with human adaptations for prolonged neurodevelopmental trajectories and increased plasticity.

Learning Objectives:

1. Identify differences that exist in the capuchin monkey male and female social environment.

2. Understand which type of brain tissue is analyzed with voxel-based morphometry and tract-based spatial statistical analysis.

3. Examine whether selection pressures on social behavior drive sexual differentiation in primate neuroanatomy.

Mental health is a multidimensional challenge, marked by rapid brain changes and essential interpersonal relationships. This complicates the application of traditional statistical methods and neuropsychiatric approaches classically focused on isolated individuals. The combination of artificial intelligence and multi-brain neuroscience offers an innovative path towards a holistic understanding that encompasses neurobiological, behavioral and social scales. These advanced technologies facilitate the analysis of dynamics and interactions, whether between genes, neurons, or even individuals. This more integrative perspective paves the way for inter-personalized computational psychiatry. Combining mathematical tools and models with an ecosocial approach promises to reinvent the detection and treatment of psychiatric disorders, but also opens the way to medicine beyond the individual.

Learning Objectives: 

1. Explain the basic principles of computational psychiatry.

2. Review how multi-brain neuroscience can integrate biological, behavioral, and social factors in mental health.

3. Summarize the potential of inter-personalized computational psychiatry in improving psychiatric care.

Social animals congregate in groups and communicate with vocalizations. To study the dynamics of natural vocal communication and their neural basis, one must characterize signals used for communication and determine the sender and receiver of the signal [1]. To this end, we established two complementary approaches for (1) quantifying vocal repertoire using a variational autoencoder (VAE) with longitudinal audio recordings in a naturalistic social environment, and (2) vocal call attribution using a deep neural network. We pursued this research by establishing a unique and favorable model organism — the Mongolian gerbil — which has a sophisticated vocal repertoire and complex social hierarchy, including pair bond formation [2]. Here, we made continuous acoustic recordings of three separate gerbil families for 20 days each, and used a VAE for unsupervised representation learning of acoustic features to show that gerbil families have family specific vocal repertoires. Although this result positions the gerbil as an intriguing model of social vocal interactions, the inability to attribute vocalizations to individuals in a group limits interpretability of these family vocal differences, and remains a persistent problem for others in the field. We have therefore developed (1) a supervised deep learning framework with calibrated uncertainty estimates that achieves state-of-the-art sound source localization performance, (2) novel hardware solutions to generate benchmark datasets for training/evaluating sound source localization models across labs, and (3) curated and released the first large-scale benchmark datasets for vocal call localization in social rodents.

Learning Objectives: 

1. Demonstrate a model for natural vocal communication.

2. Identify vocalizations to individuals.

3. Generate scalable computational analyses.

The Brain Behavior Quantification and Synchronization (BBQS) program is a basic research effort in the NIH BRAIN Initiative to develop new tools and approaches in support of a more comprehensive mechanistic understanding of the neural basis of behavior. The goals of the BBQS program are to 1) Develop high-resolution tools, platforms, and theories to precisely quantify behaviors as a multi-dimensional response with high resolution and synchronize these with simultaneously recorded brain activity; 2) Build new conceptual and computational models of behavioral systems, with which to establish causal brain/behavior relationships; 3) Establish a cross-disciplinary consortium to develop and disseminate new tools, ontologies, research designs, and ethical frameworks that will transform how mechanistic brain-behavioral research is conducted.

We are developing the three-banded panther worm Hofstenia miamia as a new model system for neuroscience. Hofstenia is a rare, lab-tractable member of a lineage of early-branching marine invertebrates called acoel worms. Hofstenia is capable of complex mating, foraging, and egg-laying behavior. The worm has a diffuse brain that is intermediate - phylogenetically and morphologically - between the decentralized nerve nets of cnidarians and the fully centralized brains of vertebrates and typical model organisms. In addition, Hofstenia has a remarkable ability to regenerate any and all tissues from virtually any initial configuration. Thus, it is a promising new system for the study of distributed neural computation, brain regeneration, and the evolution of the first centralized brains. In this talk, I will discuss the approaches we are using to study the brain and behavior of Hofstenia, particularly in the context of its environment.

Learning Objectives: 

1. Summarize the value of diversifying the model systems for neuroscience.

2. Review the acoel Hofstenia miamia as a new research organism.

3. Explain approaches for studying brain, behavior, and the environment in this new system.

Behaving animals continually reconcile the internal states of their neural circuits brain-wide with incoming sensory and environmental evidence to evaluate when and how to act. The brains of animals including humans exploit many evolutionary innovations, chiefly modularity—observable at the level of anatomically-defined brain regions, cortical layers, and cell types among others—that can be repurposed in a compositional manner to endow the animal with a highly flexible behavioral repertoire. Accordingly, behavioral output shows its own modularity, yet these behavioral modules seldom correspond directly to modularity in the brain. It remains unclear how to link neural and behavioral modularity compositionally. Here, we propose a comprehensive framework—compositional modes—to identify overarching compositional structure across specialized submodules such as brain regions. Our framework directly links the behavioral repertoire with distributed patterns of population activity brain-wide at multiple concurrent spatial and temporal scales. Using data from whole-brain recordings of larval zebrafish, we introduce an unsupervised pipeline based on neural network models to reveal highly conserved compositional modes across individuals despite the spontaneous nature of the behavior. These modes provided scaffolding for other modes that account for the idiosyncratic behavior of each fish. We then demonstrate experimentally that behavioral and pharmacological manipulation can consistently manipulate compositional modes. Our results demonstrate that even spontaneous behavior in different individuals can be decomposed and understood using a relatively small number of neurobehavioral modules—the compositional modes—and elucidate a compositional neural basis of behavior.

Learning Objectives: 

1. Explain the concept of modularity in the brain and its significance for understanding behavioral flexibility.

2. Describe the framework of compositional modes to link brain-wide neural activity with behavior.

3. Demonstrate how data-constrained modeling can reveal the neural basis of behavior in models like the larval zebrafish.

Drug discovery programs require a robust panel of models and assays to identify and validate drug targets, support compound optimisation, and run mechanistic studies. Fully human in vitro model systems comprising iPSC-derived neurons and/or glia represent robust and flexible systems to provide translationally relevant data and support decision making. Complexity can be built using multi-culture systems comprising multiple cell types and/or by working in 3D.

Use of patient-derived iPSCs harbouring autosomal dominant mutations (e.g. APP and PSEN1 mutations causal for monogenic Alzheimer’s disease, AD) enable cellular dysfunction underlying disease pathogenesis to be studied and exploited. Cellular phenotypes including altered b-amyloid ratios and endolysosomal dysfunction can be studied in human monogenic AD iPSC-derived neurons.

In addition to neuronal dysfunction, glia are increasingly being recognised as important players in neurological disease. Genetic associations between multiple immune genes and risk of developing AD highlight the role of neuroinflammation in disease pathogenesis. We have generated a fully human iPSC-derived cortical neuron-microglia coculture system to enable mechanistic studies of neuroinflammation, suitable for neuroinflammation target identification, validation and compound profiling. Microglia display both homeostatic and damage response behaviours in coculture, are phagocytically active and release cytokines in response to proinflammatory stimulation.

Human iPSC-derived neuronal and glial models comprising multiple cell types and/or patient-derived cells capture relevant neurodegenerative and neuroinflammatory outcomes, enabling generation of translationally relevant data to advance drug discovery.

Learning Objectives: 

1. Review how stem cell models of neurological disease can be used to support drug discovery.

2. Demonstrate knowledge on how cellular phenotypes can be studied in monogenic Alzheimer’s disease neurons.

3. Demonstrate knowledge on how neuron-microglia cocultures can be used to study neuroinflammation.

Vestibular sensation anchors our perception of the environment and is crucial for maintaining balance during locomotion. Vertebrate animals use the evolutionarily conserved vestibular system to sense linear and angular acceleration. However, tetrapods on land exhibit complex balance behaviors that are difficult to capture and quantify. Zebrafish (Danio rerio) is a tractable small vertebrate exhibiting innate balance behavior that is straightforward to measure under water. Larval zebrafish are inherently unstable due to their front-heavy body plane. Larval zebrafish rely on swimming to prevent nose-down rotation and maintain balance. To visualize balance behavior, we developed a scalable apparatus to measure posture and locomotion (SAMPL). SAMPL employs a simple marker-less pose estimation algorithm to extract posture and locomotion data and can be adapted for various small animals. Through a combination of high-throughput behavior assays, 2-photon laser ablation, and functional imaging techniques, we have identified the neuronal substrates underlying vestibular control of balance. These studies establish a foundation for comprehending the neural mechanisms of vestibular perturbations, compensation, and rehabilitation under pathological conditions.

Learning Objectives:

1. Explain how the vestibular system sense acceleration.

2. Understand a basic algorithm for pose estimation.

3. Describe vestibular contribution to locomotion in zebrafish.

Protein misfolding and aggregation are key features observed in numerous neurodegenerative conditions, such as Alzheimer's and Parkinson's disease. The oligomers formed during the aggregation process likely play a significant role in the development of these diseases and represent promising candidates for biomarkers. Nevertheless, due to their small size and low concentrations, there is a current lack of specific tools for quantifying and characterizing these aggregates in complex biological samples. In this presentation, I will introduce our recently devised single-molecule techniques, designed to overcome this obstacle by investigating immobilized proteins using detection antibodies labeled in an orthogonal manner. Through the analysis of colocalized signals, we can eliminate monomeric protein, and accurately quantify aggregated proteins. Using the aggregation-prone alpha-synuclein protein as a model, we illustrate that this approach can selectively identify aggregates with a detection limit of 5 picomolar. Furthermore, we demonstrate its applicability across various samples, including human biofluids. Our method is versatile and can be applied to protein aggregates associated with diverse disorders, contributing to the identification of crucial biomarkers for improved diagnostic efforts in these diseases.

Learning Objectives: 

1. Review the relevance of protein aggregates in neurodegenerative disorders.

2. Summarize how they can be visualized specifically at the single-molecule level.

3. Explain the concepts of super-resolution microscopy, and how it can be utilized for protein aggregate detection.

Our ability to record large-scale neural and behavioral data has substantially improved in the last decade. However, the inference of quantitative dynamical models for cognition and motor control remains challenging due to their unconstrained nature. Here, we incorporate constraints from anatomy and physiology to tame machine learning models of neural activity and behavior. I will show that these constraints-based modeling approaches allow us to predictively understand the relationship between neural activity and behavior. How does the motor cortex achieve generalizable and purposeful movements from the complex, nonlinear musculoskeletal system? I will present a deep reinforcement learning framework for training recurrent neural network controllers that act on anatomically accurate limb models such that they achieve desired movements. We apply this framework to kinematic and neural recordings made in macaques as they perform movements at different speeds. This framework for the control of the musculoskeletal system mimics biologically observed neural strategies and enables hypothesis generation for prediction and analysis of novel movements and neural strategies.

Modeling neural activity and behavior across different subjects and in a naturalistic setting remains a significant challenge. Here, we develop novel explainable AI methods for modeling continuously varying differences in behavior, which successfully represent distinct features of multi-subject and social behavior in an unsupervised manner. These methods are also successful at uncovering the relationships between recorded neural data and the ensuing behavior. We end with future work on understanding the interactions between two or more subjects during social behavior.

Learning Objectives:

1. Recognize the role of behavior in constrained machine learning models of the brain.

2. Discuss the different levels at which we can model behavior.

3. Summarize the downstream applications on motor control, modeling inter-subject variability, and social behavior.