Shengxuan Jerry Ye

Shengxuan Jerry Ye

Singapore, Singapore
1K followers 500+ connections

Join to follow

About

Established in December 2017, Whale helps retail companies and brands innovate in…

Activity

Join now to see all activity

Experience

  • Facebook Graphic

    Facebook

    Greater New York City Area

  • -

    Greater Boston Area

  • -

    Greater Los Angeles Area

  • -

    Pasadena, California

  • -

    Charlottesville, Virginia Area

Education

  • Caltech Graphic

    California Institute of Technology

    -

    Activities and Societies: Caltech Chinese Student Association (Caltech C)

    Study focused on a combination of single-neuron and LFP data analysis, machine learning and behavioral prediction. Major projects include building a online, real-time and parallel system for predicting action content in humans using intracranial EEG with epilepsy patients, as well as human single-neuron analysis of memory strength model in a memory retrieval task.

  • -

    Energy sharing strategies for multi-energy systems with multiple flexible resources.

  • -

    Activities and Societies: Enactus at the University of Virginia (Founder)

    Graduated with Highest Distinction. Study in computer science focused on security and privacy-preserving computation, and their applications in biology, as well as on extreme computing (MPI & CUDA). Study in economics focused on behavioral and experimental economics with its application to marketing.

Publications

  • Representation of retrieval confidence by single neurons in the human medial temporal lobe

    Nature Neuroscience

    Memory-based decisions are often accompanied by an assessment of choice certainty, but the mechanisms of such confidence judgments remain unknown. We studied the response of 1,065 individual neurons in the human hippocampus and amygdala while neurosurgical patients made memory retrieval decisions together with a confidence judgment. Combining behavioral, neuronal and computational analysis, we identified a population of memory-selective (MS) neurons whose activity signaled stimulus familiarity…

    Memory-based decisions are often accompanied by an assessment of choice certainty, but the mechanisms of such confidence judgments remain unknown. We studied the response of 1,065 individual neurons in the human hippocampus and amygdala while neurosurgical patients made memory retrieval decisions together with a confidence judgment. Combining behavioral, neuronal and computational analysis, we identified a population of memory-selective (MS) neurons whose activity signaled stimulus familiarity and confidence, as assessed by subjective report. In contrast, the activity of visually selective (VS) neurons was not sensitive to memory strength. The groups further differed in response latency, tuning and extracellular waveforms. The information provided by MS neurons was sufficient for a race model to decide stimulus familiarity and retrieval confidence. Together, our results indicate a trial-by-trial relationship between a specific group of neurons and declared memory strength in humans. We suggest that VS and MS neurons are a substrate for declarative memories.

    Other authors
    See publication

  • The relationship between single human hippocampal and amygdala neurons and subjective memory strength

    Society of Neuroscience (SfN) 2013

    Episodic memories allow us to remember not only that we have seen an item before (familiarity) but also provide a subjective impression of how sure we are that we have seen an item before (confidence). The neuronal mechanisms of confidence judgment about one's own memories, a decision about one's own internal state, are poorly understood. The medial temporal lobe (MTL) plays an important role in the estimation of memory strength, but its role in confidence judgments is unclear. We recorded…

    Episodic memories allow us to remember not only that we have seen an item before (familiarity) but also provide a subjective impression of how sure we are that we have seen an item before (confidence). The neuronal mechanisms of confidence judgment about one's own memories, a decision about one's own internal state, are poorly understood. The medial temporal lobe (MTL) plays an important role in the estimation of memory strength, but its role in confidence judgments is unclear. We recorded single neurons in the MTL of patients implanted with depth electrodes for the purpose of localizing epileptic seizures. Patients first viewed a sequence of 100 novel images, followed by another sequence of the same images intermixed with novel images. Patients rated each image as new/old on a 1-6 confidence scale.

    We recorded 381 neurons from the hippocampus and amygdala from 14 patients and quantified the relationship of neuronal spiking activity with behavior using Receiver Operating Characteristic (ROC) analysis. 56 (15%) of units significantly differentiated new from old stimuli as expressed by an area under the curve (AUC) of the ROC significantly different from chance. 66% of the significant neurons predicted behavior better than ground truth. The average AUC was significantly larger when calculated with respect to behavioral choice as opposed to ground truth (p<0.001). Thus, neurons predicted behavior rather than ground truth. AUCs were 0.64±0.05 (all errors are ±s.d.), indicating that single neurons predicted behavior choices with about 65% accuracy. In contrast, behavioral performance had an AUC of 0.75±0.05. For neurons that increased their firing rate for familiar stimuli relative to novel stimuli, the neuronal AUC was significantly larger for trials that resulted in a memory with high compared to low subjective confidence (0.62±0.05 vs. 0.55±0.12, p<0.05). Together, this indicates that neuronal firing was related to subjective memory strength.

    Other authors
    • Ian Ross
    • Jeffery Chung
    • Adam Mamelak
    • Ueli Rutishauser

  • An Intracortical Study of Online Realtime Action-Content Prediction in Patients

    Neural Interfaces Conference (NIC) 2012

    Brain-computer interface (BCI) systems ideally rely on robust brain signals highly correlated to the mental process or state that the system is designed to decode. Therefore, movement-related BCI typically depends on the subject imagining one of several neurally separable movements (e.g., left- versus right-hand movement) for a few seconds, while the system attempts to classify it. But movement is preceded by preparatory neural activity, as part of the intention to move, which may be decoded…

    Brain-computer interface (BCI) systems ideally rely on robust brain signals highly correlated to the mental process or state that the system is designed to decode. Therefore, movement-related BCI typically depends on the subject imagining one of several neurally separable movements (e.g., left- versus right-hand movement) for a few seconds, while the system attempts to classify it. But movement is preceded by preparatory neural activity, as part of the intention to move, which may be decoded earlier. This activity was the focus of our investigation.
    We recorded from microwires at the tips of depth electrodes that were implanted in the brains of consenting intractable-epilepsy patients as part of their presurgical clinical evaluation. The electrodes were implanted in the left and right supplementary motor cortex (SMA), the anterior cingulate cortex (ACC), amygdala and hippocampus. The patients were engaged in a matching-pennies game against the computer or the experimenter, where in each round they had to raise their left or right hand at the go signal, winning and gaining monetary reward if they raised the same hand as their opponent and losing otherwise. We gathered local-field potential data from the electrodes’ microwires. Focusing on slow waves in the 0.1-5Hz range, we found that brain signals from a combination of channels – especially ACC and SMA – tended to differentiate well between the intended left- or right-hand movements before go-signal onset. We constructed an analysis system and tested it in simulated online-realtime conditions on 6 patients over 20 sessions and in actual online-realtime on 2 patients over 6 sessions, so far. Our results suggest that the accuracy gradually rises from 65-75% correct 3-4s before the go-signal to 80-90% correct shortly before go-signal onset.

    Other authors
    • Uri Maoz
    • Ian Ross
    • Adam Mamelak
    • Christof Koch
    See publication

  • A System for Predicting Action Content On-line and in Real Time before Action Onset in Humans – an Intracranial Study

    Neural Information Processing Systems (NIPS)

    The ability to predict action content from neural signals in real time before action onset has been long sought in the neuroscientific study of decision-making, agency and volition. On-line real-time (ORT) prediction is important for understanding the relation between neural correlates of decision-making and conscious, voluntary action. Here, epilepsy patients, implanted with intracranial depth microelectrodes or subdural grid electrodes for clinical purposes, participated in a…

    The ability to predict action content from neural signals in real time before action onset has been long sought in the neuroscientific study of decision-making, agency and volition. On-line real-time (ORT) prediction is important for understanding the relation between neural correlates of decision-making and conscious, voluntary action. Here, epilepsy patients, implanted with intracranial depth microelectrodes or subdural grid electrodes for clinical purposes, participated in a “matching-pennies” game against either the experimenter or a computer. In each trial, subjects were given a 5s countdown, after which they had to raise their left or right hand immediately as the “go” signal appeared on a computer screen. They won a fixed amount of money if they raised a different hand than their opponent and lost that amount otherwise. The working hypothesis of this experiment was that neural precursors of the subjects’ decisions precede action onset and potentially also the awareness of the decision to move, and that these signals could be detected in intracranial local field potentials (LFP). We found that low-frequency LFP signals from a combination of 10 channels, especially bilateral anterior cingulate cortex and supplementary motor area, were predictive of the intended left-/right-hand movements before the onset of the go signal. Our ORT system predicted which hand the patient would raise 0.5s before the go signal with 68±3% accuracy in two patients. Based on these results, we constructed an ORT system that tracked up to 30 channels simultaneously, and tested it on retrospective data from 6 patients. On average, we could predict the correct hand choice in 80% of the trials, which rose to 90% correct if we let the system drop about 1/3 of the trials on which it was less confident. Our system demonstrates – for the first time – the feasibility of accurately predicting a binary action in real time for patients with intracranial recordings, well before the action occurs.

    Other authors
    • Uri Maoz
    • Ian Ross
    • Adam Mamelak
    • Christof Koch
    See publication

Languages

  • Chinese

    Native or bilingual proficiency

  • English

    Native or bilingual proficiency

More activity by Shengxuan Jerry

View Shengxuan Jerry’s full profile

  • See who you know in common
  • Get introduced
  • Contact Shengxuan Jerry directly
Join to view full profile

Other similar profiles

Explore more posts

Explore collaborative articles

We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.

Explore More

Add new skills with these courses

See all courses