Biology subjects

Yttri, E. A.

Publications and source records attributed to Yttri, E. A..

8 recordsLinked to original sources

bioRxiv2026

B-REAL: A keypoint-less behavior platform for offline classification and real-time detection

Two obstacles limit automated behavior analysis in neuroethology: the burden of first training pose estimation models, and the inability to identify behavior quickly enough for real-time classification that could trigger stimuli, drug delivery, or environmental manipulations. Here we present B-REAL, a pose-free classifier that maps raw video to behavior faster than video frames are recorded, even on a CPU. Given substantially less training data, B-REAL matched or surpassed established methods trained on full benchmark rodent or fly datasets, particularly on rare and even single-frame behaviors. Pretraining on unlabeled video further improves its accuracy without slowing inference or training time. In closed loop, rewarding mice for B-REAL-detected grooming more than doubled their performance of grooming bouts. B-REAL thus makes closed-loop behavioral experiments practical with minimal annotation and standard hardware.

neuroscience↗

bioRxiv2024

Striatal indirect pathway mediates hesitation

Determining the best possible action in an uncertain situation is often challenging, and organisms frequently need extra time to deliberate. This pause in behavior in response to uncertainty - also known as hesitation - commonly occurs in many aspects of daily life, yet its neural circuits are poorly understood. Here we present the first experimental paradigm that reliably evokes hesitation in mice. Using cell-type specific electrophysiology and optogenetics, we show that indirect, but not direct, pathway spiny projection neurons specifically in the dorsomedial striatum mediate hesitation. These data indicate that the basal ganglia circuits controlling the pausing involved in cognitive processes like hesitation are distinct from those that control other types of behavioral inhibition, such as cue-induced stopping.

neuroscience↗

bioRxiv2024

Mimicking opioid analgesia in cortical pain circuits

The anterior cingulate cortex is a key brain region involved in the affective and motivational dimensions of pain, yet how opioid analgesics modulate this cortical circuit remains unclear. Uncovering how opioids alter nociceptive neural dynamics to produce pain relief is essential for developing safer and more targeted treatments for chronic pain. Here we show that a population of cingulate neurons encodes spontaneous pain-related behaviors and is selectively modulated by morphine. Using deep-learning behavioral analyses combined with longitudinal neural recordings in mice, we identified a persistent shift in cortical activity patterns following nerve injury that reflects the emergence of an unpleasant, affective chronic pain state. Morphine reversed these neuropathic neural dynamics and reduced affective-motivational behaviors without altering sensory detection or reflexive responses, mirroring how opioids alleviate pain unpleasantness in humans. Leveraging these findings, we built a biologically inspired gene therapy that targets opioid-sensitive neurons in the cingulate using a synthetic mu-opioid receptor promoter to drive chemogenetic inhibition. This opioid-mimetic gene therapy recapitulated the analgesic effects of morphine during chronic neuropathic pain, thereby offering a new strategy for precision pain management targeting a key nociceptive cortical opioid circuit with safe, on-demand analgesia. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=40 SRC="FIGDIR/small/591113v2_ufig1.gif" ALT="Figure 1"> View larger version (28K): org.highwire.dtl.DTLVardef@d294c0org.highwire.dtl.DTLVardef@123d4f4org.highwire.dtl.DTLVardef@1720deborg.highwire.dtl.DTLVardef@efc632_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2022

A-SOiD, an active learning platform for expert-guided, data efficient discovery of behavior

To identify and extract naturalistic behavior, two schools of methods have become popular: supervised and unsupervised. Each approach carries its own strengths and weaknesses, which the user must weigh in on their decision. Here, a new active learning platform, A-SOiD, blends these strengths and, in doing so, overcomes several of their inherent drawbacks. A-SOiD iteratively learns user-defined groups and can considerably reduce the necessary training data while attaining expansive classification through directed unsupervised classification. In socially-interacting mice, A-SOiD outperformed other methods and required 85% less training data than was available. Additionally, it isolated two additional ethologically-distinct mouse interactions via unsupervised classification. Similar performance and efficiency were observed using non-human primate 3D pose data. In both cases, the transparency in A-SOiDs cluster definitions revealed the defining features of the supervised classification through a game-theoretic approach. Lastly, we show the potential of A-SOiD to segment a large and rich variety of human social and single-person behaviors with 3D position keypoints. To facilitate use, A-SOiD comes as an intuitive, open-source interface for efficient segmentation of user-defined behaviors and discovered sub-actions.

animal behavior and cognition↗

bioRxiv2021

Behavioral and nociceptor states of inflammatory painacross timescales in 2D and 3D

Ongoing pain is often driven by direct activation of pain-sensing neurons and neuroimmune mediated sensitization. These heightened states of pain alter physiology, reduce motor function, and alter motivation to engage in normal behaviors. The complexity of the pain state has evaded a comprehensive definition, especially in nonverbal animals. Here in mice, we capture the physiological state of sensitized pain neurons at different time points post-inflammation and used computational tools to automatically map behavioral signatures of evoked and spontaneous displays of pain. First, retrograde labeling coupled with electrophysiology of neurons innervating the site of localized inflammation defined critical time points of pain sensitization. Next, we used high-speed videography combined with supervised and unsupervised machine learning tools and uncovered sensory-evoked defensive coping postures to pain. Using 3D pose analytics inspired by natural language processing, we identify movement sequences that correspond to robust representations of ongoing pain states. Surprisingly, with this analytical framework, we find that a commonly used anti-inflammatory painkiller does not return an animals behavior back to a pre-injury state. Together, these findings reveal the previously unidentified signatures of pain and analgesia at timescales when inflammation induces heightened pain states.

neuroscience↗

bioRxiv2019

B-SOiD: An Open Source Unsupervised Algorithm for Discovery of Spontaneous Behaviors

Studying naturalistic behavior remains a prohibitively difficult objective. Recent machine learning advances have enabled limb localization. Extracting behaviors, however, requires ascertaining the spatiotemporal patterns of these positions. To provide the missing bridge from poses to actions and their kinematics, we developed B-SOiD - an open-source, unsupervised algorithm that identifies behavior without user bias. By training a machine classifier on pose pattern statistics clustered using new methods, our approach achieves greatly improved processing speed and the ability to generalize across subjects or labs. Using a frameshift alignment paradigm, B-SOiD overcomes previous temporal resolution barriers that prevent the use of other algorithms with electrophysiological recordings. Using only a single, off-the-shelf camera, B-SOiD provides categories of sub-action for trained behaviors and kinematic measures of individual limb trajectories in an animal model. These behavioral and kinematic measures are difficult but critical to obtain, particularly in the study of pain, OCD, and movement disorders.

neuroscience↗

bioRxiv2019

The CMU Array. A 3D Nano-Printed, Fully Customizable Ultra-High-Density Microelectrode Array

Microelectrode arrays (MEAs) provide the means to record electrophysiological activity fundamental to both basic and clinical neuroscience (e.g. brain-computer interfaces). Despite recent advances, current MEAs have significant limitations - including low recording density, fragility, expense, and the inability to optimize the probe to individualized study or patient needs. Here we address the technological limitations through the utilization of the newest developments in 3D nanoparticle printing.1 Our CMU Arrays possess previously impossible electrode densities (> 6000 channels/cm2) with tip diameters as small as 10 m. Most importantly, the probes are entirely customizable owing to the adaptive manufacturing process. Any combination of individual shank lengths, impedances, and layouts are possible. This is achieved in part via our new multi-layer, multi material, custom 3D-printed circuit boards, a fabrication advancement in itself. This device design enables new experimental avenues of targeted, large-scale recording of electrical signals from a variety of biological tissues.

neuroscience↗

bioRxiv2019

Build a better mouse task: can an open-source rodent joystick enhance reaching behavior outcomes through improved monitoring of real-time spatiotemporal kinematics?

For decades, advanced behavioral tasks have only been used in human and non-human primates. However, with improved analytical and genetic techniques, there has been a growing drive to implement complex reaching, decision-making, and reaction time tasks - not in primates - but in rodents. Here, we assess the hypothesis that a mouse can learn a cued reaction time task. Moreover, we tested multiple training regimens and found that introducing elements of the reaction time task serially hindered, rather than helped task acquisition. Additionally, we include a step-by-step manual for inexpensive implementation and use of a rodent joystick for behavioral analysis. Task and analysis code for the evaluated behaviors are included such that they may be replicated and tested further. With these, we also include code for a probabilistic reward two-arm bandit task. These various tasks, and the method to construct and implement them, will enable greatly improved study of the neural correlates of behavior in the powerful mouse model organism. In summary, we have tested and demonstrated that mice can learn sophisticated tasks with A joystick, and that targeted task design provides a significant advantage. These results of this study stand to inform the implementation of other sophisticated tasks using the mouse model.

animal behavior and cognition↗