A5006832724- Neural dynamics and brain function
- Advanced Memory and Neural Computing
- Memory and Neural Mechanisms
- Visual perception and processing mechanisms
- Face Recognition and Perception
- Neuroscience and Neuropharmacology Research
- Retinal Development and Disorders
- Neural Networks and Applications
- Photoreceptor and optogenetics research
- Visual Attention and Saliency Detection
- Neurotransmitter Receptor Influence on Behavior
- Motor Control and Adaptation
- Domain Adaptation and Few-Shot Learning
- Machine Learning and ELM
- Neurological disorders and treatments
- Neurobiology and Insect Physiology Research
- Insect and Arachnid Ecology and Behavior
- Zebrafish Biomedical Research Applications
- Muscle activation and electromyography studies
- Receptor Mechanisms and Signaling
- Neural and Behavioral Psychology Studies
- EEG and Brain-Computer Interfaces
- Complex Systems and Time Series Analysis
- Advanced Computational Techniques and Applications
- Ferroelectric and Negative Capacitance Devices
Columbia University
2018-2025
Brain (Germany)
2023-2025
Bureau of Meteorology
2023
Manaaki Whenua – Landcare Research
2023
Environment and Climate Change Canada
2023
Commonwealth Scientific and Industrial Research Organisation
2023
Météo-France
2023
Stanford University
2021
Stanford Medicine
2019
Google (United States)
2018
Making inferences about the computations performed by neuronal circuits from synapse-level connectivity maps is an emerging opportunity in neuroscience. The mushroom body (MB) well positioned for developing and testing such approach due to its conserved architecture, recently completed dense connectome, extensive prior experimental studies of roles learning, memory, activity regulation. Here, we identify new components MB circuit Drosophila, including visual input output neurons (MBONs) with...
Mechanistic interpretability aims to understand the computational mechanisms underlying neural networks' capabilities in order accomplish concrete scientific and engineering goals. Progress this field thus promises provide greater assurance over AI system behavior shed light on exciting questions about nature of intelligence. Despite recent progress toward these goals, there are many open problems that require solutions before practical benefits can be realized: Our methods both conceptual...
A bstract The vertebrate visual system is hierarchically organized to process information in successive stages. Neural representations vary drastically across the first stages of processing: at output retina, ganglion cell receptive fields (RFs) exhibit a clear antagonistic center-surround structure, whereas primary cortex (V1), typical RFs are sharply tuned precise orientation. There currently no unified theory explaining these differences layers. Here, using deep convolutional neural...
The utility of a learned neural representation depends on how well its geometry supports performance in downstream tasks. This the structure inputs, target outputs, and architecture network. By studying learning dynamics networks with one hidden layer, we discovered that network's activation function has an unexpectedly strong impact representational geometry: Tanh tend to learn representations reflect while ReLU retain more information about raw inputs. difference is consistently observed...
Abstract Spiny projection neurons (SPNs) in dorsal striatum are often proposed as a locus of reinforcement learning the basal ganglia. Here, we identify and resolve fundamental inconsistency between striatal models known SPN synaptic plasticity rules. Direct-pathway (dSPN) indirect-pathway (iSPN) neurons, which promote suppress actions, respectively, exhibit that reinforces activity associated with elevated or suppressed dopamine release. We show iSPN prevents successful learning, it...
Abstract One of the most striking aspects early visual processing in retina is immediate parcellation information into multiple parallel pathways, formed by different retinal ganglion cell types each tiling entire field. Existing theories efficient coding have been unable to account for functional advantages such cell-type diversity encoding natural scenes. Here we go beyond previous analyze how a simple linear model with convolutional efficiently encodes naturalistic spatiotemporal movies...
Forming an episodic memory requires binding together disparate elements that co-occur in a single experience. One model of this process is neurons representing different components bind to “index” — subset unique memory. Evidence for has recently been found chickadees, which use hippocampal store and recall locations cached food. Chickadee hippocampus produces sparse, high-dimensional patterns (“barcodes”) uniquely specify each caching event. Unexpectedly, the same participate barcodes also...
Forming an episodic memory requires binding together disparate elements that co-occur in a single experience. One model of this process is neurons representing different components bind to “index” — subset unique memory. Evidence for has recently been found chickadees, which use hippocampal store and recall locations cached food. Chickadee hippocampus produces sparse, high-dimensional patterns (“barcodes”) uniquely specify each caching event. Unexpectedly, the same participate barcodes also...
Spiny projection neurons (SPNs) in dorsal striatum are often proposed as a locus of reinforcement learning the basal ganglia. Here, we identify and resolve fundamental inconsistency between striatal models known SPN synaptic plasticity rules. Direct-pathway (dSPN) indirect-pathway (iSPN) neurons, which promote suppress actions, respectively, exhibit that reinforces activity associated with elevated or suppressed dopamine release. We show iSPN prevents successful learning, it patterns...
Spiny projection neurons (SPNs) in dorsal striatum are often proposed as a locus of reinforcement learning the basal ganglia. Here, we identify and resolve fundamental inconsistency between striatal models known SPN synaptic plasticity rules. Direct-pathway (dSPN) indirect-pathway (iSPN) neurons, which promote suppress actions, respectively, exhibit that reinforces activity associated with elevated or suppressed dopamine release. We show iSPN prevents successful learning, it patterns...
Abstract Making inferences about the computations performed by neuronal circuits from synapse-level connectivity maps is an emerging opportunity in neuroscience. The mushroom body (MB) well positioned for developing and testing such approach due to its conserved architecture, recently completed dense connectome, extensive prior experimental studies of roles learning, memory activity regulation. Here we identify new components MB circuit Drosophila , including visual input output neurons...
How motor cortex contributes to sequence execution is much debated, with studies supporting disparate views. Here we probe the degree which cortex's engagement depends on task demands, specifically whether its role differs for highly practiced, or 'automatic', sequences versus flexible informed by external events. To test this, trained rats generate three-element either overtraining them a single having follow instructive visual cues. Lesioning revealed that it necessary cue-driven but...
Object classification has been proposed as a principal objective of the primate ventral visual stream and used an optimization target for deep neural network models (DNNs) system. However, brain areas represent many different types information, optimizing object identity alone does not constrain how other information may be encoded in representations. Information about scene parameters discarded altogether ('invariance'), represented non-interfering subspaces population activity...
Object classification has been proposed as a principal objective of the primate ventral visual stream and used an optimization target for deep neural network models (DNNs) system. However, brain areas represent many different types information, optimizing object identity alone does not constrain how other information may be encoded in representations. Information about scene parameters discarded altogether (‘invariance’), represented non-interfering subspaces population activity...
Abstract Forming an episodic memory requires binding together disparate elements that co-occur in a single experience. One model of this process is neurons representing different components bind to “index” — subset unique memory. Evidence for has recently been found chickadees, which use hippocampal store and recall locations cached food. Chickadee hippocampus produces sparse, high-dimensional patterns (“barcodes”) uniquely specify each caching event. Unexpectedly, the same participate...
Animal behavior is driven by multiple brain regions working in parallel with distinct control policies. We present a biologically plausible model of off-policy reinforcement learning the basal ganglia, which enables such an architecture. The accounts for action-related modulation dopamine activity that not captured previous models implement on-policy algorithms. In particular, predicts signals combination reward prediction error (as classic models) and "action surprise," measure how...
In a variety of species and behavioral contexts, learning memory formation recruits two neural systems, with initial plasticity in one system being consolidated into the other over time. Moreover, consolidation is known to be selective; that is, some experiences are more likely long-term than others. Here, we propose analyze model captures common computational principles underlying such phenomena. The key component this mechanism by which prioritizes storage synaptic changes consistent prior...