scholarly journals Correcting the Hebbian Mistake: Toward a Fully Error-Driven Hippocampus

2021 ◽  
Author(s):  
Yicong Zheng ◽  
Xiaonan L. Liu ◽  
Satoru Nishiyama ◽  
Charan Ranganath ◽  
Randall C. O'Reilly
Keyword(s):  
Computational Models ◽  
Hebbian Learning ◽  
Critical Role ◽  
Neural Data ◽  
Learning Capacity ◽  
Learning Speed ◽  
Hippocampal Ca1 ◽  
Sparse Autoencoder ◽  
Episodic Memories ◽  
Area Ca3

The hippocampus plays a critical role in the rapid learning of new episodic memories. Many computational models propose that the hippocampus is an autoassociator that relies on Hebbian learning (i.e., "cells that fire together, wire together"). However, Hebbian learning is computationally suboptimal as it modifies weights unnecessarily beyond what is actually needed to achieve effective retrieval, causing more interference and resulting in a lower learning capacity. Our previous computational models have utilized a powerful, biologically plausible form of error-driven learning in hippocampal CA1 and entorhinal cortex (EC) (functioning as a sparse autoencoder) by contrasting local activity states at different phases in the theta cycle. Based on specific neural data and a recent abstract computational model, we propose a new model called Theremin (Total Hippocampal ERror MINimization) that extends error-driven learning to area CA3 --- the mnemonic heart of the hippocampal system. In the model, CA3 responds to the EC monosynaptic input prior to the EC disynaptic input through dentate gyrus (DG), giving rise to a temporal difference between these two activation states, which drives error-driven learning in the EC->CA3 and CA3<->CA3 projections. In effect, DG serves as a teacher to CA3, correcting its patterns into more pattern-separated ones, thereby reducing interference. Results showed that Theremin, compared with our original model, has significantly increased capacity and learning speed. The model makes several novel predictions that can be tested in future studies.

scholarly journals Neural Components of Reading Revealed by Distributed and Symbolic Computational Models

2020 ◽  
Vol 1 (4) ◽  
pp. 381-401
Author(s):  
Ryan Staples ◽  
William W. Graves
Keyword(s):  
Neural Activity ◽  
Neural Coding ◽  
Computational Models ◽  
Ann Model ◽  
Distributed Representations ◽  
Neural Data ◽  
Cognitive Components ◽  
Components Of Reading ◽  
Artificial Neural Network Ann ◽  
Models Of Reading

Determining how the cognitive components of reading—orthographic, phonological, and semantic representations—are instantiated in the brain has been a long-standing goal of psychology and human cognitive neuroscience. The two most prominent computational models of reading instantiate different cognitive processes, implying different neural processes. Artificial neural network (ANN) models of reading posit nonsymbolic, distributed representations. The dual-route cascaded (DRC) model instead suggests two routes of processing, one representing symbolic rules of spelling–to–sound correspondence, the other representing orthographic and phonological lexicons. These models are not adjudicated by behavioral data and have never before been directly compared in terms of neural plausibility. We used representational similarity analysis to compare the predictions of these models to neural data from participants reading aloud. Both the ANN and DRC model representations corresponded to neural activity. However, the ANN model representations correlated to more reading-relevant areas of cortex. When contributions from the DRC model were statistically controlled, partial correlations revealed that the ANN model accounted for significant variance in the neural data. The opposite analysis, examining the variance explained by the DRC model with contributions from the ANN model factored out, revealed no correspondence to neural activity. Our results suggest that ANNs trained using distributed representations provide a better correspondence between cognitive and neural coding. Additionally, this framework provides a principled approach for comparing computational models of cognitive function to gain insight into neural representations.

Distance-Dependent Ni2+-Sensitivity of Synaptic Plasticity in Apical Dendrites of Hippocampal CA1 Pyramidal Cells

2002 ◽  
Vol 87 (2) ◽  
pp. 1169-1174 ◽  
Author(s):  
Yoshikazu Isomura ◽  
Yoko Fujiwara-Tsukamoto ◽  
Michiko Imanishi ◽  
Atsushi Nambu ◽  
Masahiko Takada
Keyword(s):  
Calcium Influx ◽  
Critical Role ◽  
Field Potential ◽  
Pyramidal Cells ◽  
Long Term Potentiation ◽  
Apical Dendrite ◽  
Excitatory Postsynaptic Potential ◽  
Distal Dendrite ◽  
Ca1 Pyramidal Cells ◽  
Hippocampal Ca1

Low concentration of Ni2+, a T- and R-type voltage-dependent calcium channel (VDCC) blocker, is known to inhibit the induction of long-term potentiation (LTP) in the hippocampal CA1 pyramidal cells. These VDCCs are distributed more abundantly at the distal area of the apical dendrite than at the proximal dendritic area or soma. Therefore we investigated the relationship between the Ni2+-sensitivity of LTP induction and the synaptic location along the apical dendrite. Field potential recordings revealed that 25 μM Ni2+ hardly influenced LTP at the proximal dendritic area (50 μm distant from the somata). In contrast, the same concentration of Ni2+ inhibited the LTP induction mildly at the middle dendritic area (150 μm) and strongly at the distal dendritic area (250 μm). Ni2+ did not significantly affect either the synaptic transmission at the distal dendrite or the burst-firing ability at the soma. However, synaptically evoked population spikes recorded near the somata were slightly reduced by Ni2+ application, probably owing to occlusion of dendritic excitatory postsynaptic potential (EPSP) amplification. Even when the stimulating intensity was strengthened sufficiently to overcome such a reduction in spike generation during LTP induction, the magnitude of distal LTP was not significantly recovered from the Ni2+-dependent inhibition. These results suggest that Ni2+ may inhibit the induction of distal LTP directly by blocking calcium influx through T- and/or R-type VDCCs. The differentially distributed calcium channels may play a critical role in the induction of LTP at dendritic synapses of the hippocampal pyramidal cells.

scholarly journals Animacy and the prediction of behaviour

2020 ◽  
Author(s):  
Johannes Schultz ◽  
Chris D. Frith
Keyword(s):  
Visual Cues ◽  
Computational Models ◽  
Action Observation ◽  
Critical Role ◽  
Mirror Neuron ◽  
Neuron Network ◽  
Short Term ◽  
Prior Beliefs ◽  
Physical Constraints ◽  
Intentional Agents

To survive, all animals need to predict what other agents are going to do next. The first step is to detect that an object is an agent and, if so, how sophisticated it is. To this end, visual cues are especially important: the form of the agent and the nature of its movements. Once identified, the movements of an agent, however sophisticated, can be anticipated in the short term on the basis of purely physical constraints, but, in the longer term, it is useful to take account of the agent’s goals and intentions. Goal directed agents are marked by the rationality of their movements, reaching their goals by the shortest or least effortful path. Observing goal directed behaviour activates the brain’s action observation/mirror neuron network. The observer’s own action generating mechanism has an important role in predicting future movements of goal directed agents.Intentions have a critical role in determining actions when agents interact with other agents. In such interactions, movements can become communicative rather than directed to immediate goals. Also, each agent can be trying to predict the behaviour of the other, leading to a recursive arms race. It is difficult to infer intentional behaviour from movement kinematics and interpretation is much more dependent upon prior beliefs about the agent. When people believe that they are interacting with an intentional agent, the brain’s mentalising system is activated as the person tries to assess the degree of sophistication of the agent. Several biologically-constrained computational models of action recognition are available, but equivalent models for understanding intentional agents remain to be developed.

MDAPlatform: a Component-based Platform for Constructing and Assessing miRNA-disease association Prediction Methods

2021 ◽  
Vol 16 ◽  
Author(s):  
Yayan Zhang ◽  
Guihua Duan ◽  
Cheng Yan ◽  
Haolun Yi ◽  
Fang-Xiang Wu ◽  
...  
Keyword(s):  
Computational Models ◽  
Critical Role ◽  
Prediction Method ◽  
Disease Association ◽  
Prediction Methods ◽  
Comparison Results ◽  
Disease Associations ◽  
Pros And Cons ◽  
Clinical Drugs ◽  
Validation Experiments

Background: Increasing evidence has indicated that miRNA-disease association prediction plays a critical role in the study of clinical drugs. Researchers have proposed many computational models for miRNA-disease prediction. However, there is no unified platform to compare and analyze the pros and cons or share the code and data of these models. Objective: In this study, we develop an easy-to-use platform (MDAPlatform) to construct and assess miRNA-disease association prediction method. Methods: MDAPlatform integrates the relevant data of miRNA, disease and miRNA-disease associations that are used in previous miRNA-disease association prediction studies. Based on the componentized model, it develops differet components of previous computational methods. Results: Users can conduct cross validation experiments and compare their methods with other methods, and the visualized comparison results are also provided. Conclusion: Based on the componentized model, MDAPlatform provides easy-to-operate interfaces to construct the miRNA-disease association method, which is beneficial to develop new miRNA-disease association prediction methods in the future.

Peri-personal space as an interface for self-environment interaction

2021 ◽  
pp. 17-46
Author(s):  
Jean-Paul Noel ◽  
Tommaso Bertoni ◽  
Andrea Serino
Keyword(s):  
Motor System ◽  
Computational Models ◽  
Multisensory Processing ◽  
Critical Role ◽  
Personal Space ◽  
Environment Interaction ◽  
Specific System ◽  
Physical Interactions ◽  
Psychophysical Studies ◽  
The Brain

The brain has developed a specific system to encode the space closely surrounding our body, our peri-personal space (PPS). This space is the theatre where all physical interactions with objects in the environment occur, and thus is postulated to play a critical role in both approaching and defensive behaviour. Here, we first describe the classic neurophysiological findings that have led researchers to conceive of PPS as a multisensory-motor interface. This historical perspective is given to clarify what properties are strictly related to PPS encoding, and what characteristics bear out or are related to PPS. Then, in an effort to uncover gaps in knowledge that often go unnoticed, we critically examine the association between PPS and i) multisensory processing, and ii) the motor system—its strongest allies. We do not mean to say that PPS isn’t multisensory-motor, simply to pinpoint current research shortcomings. Subsequently, we detail more recent psychophysical studies, highlighting the extreme plasticity of PPS, and its putative role in bodily self-consciousness and social cognition. Lastly, we briefly discuss computational models of PPS. Throughout the chapter, we particularly attempt to emphasize open areas of investigation. By critically evaluating past findings, many of them our own, we hope to provide a forward-looking perspective on the study of PPS.

scholarly journals Intrinsic motivation and episodic memories for robot exploration of high-dimensional sensory spaces

2020 ◽  
pp. 105971232092291
Author(s):  
Guido Schillaci ◽  
Antonio Pico Villalpando ◽  
Verena V Hafner ◽  
Peter Hanappe ◽  
David Colliaux ◽  
...  
Keyword(s):  
Neural Networks ◽  
Episodic Memory ◽  
Intrinsic Motivation ◽  
Computational Models ◽  
Deep Neural Networks ◽  
Image Sensor ◽  
Forward Kinematics ◽  
High Dimensional ◽  
Episodic Memories ◽  
Low Dimensional

This work presents an architecture that generates curiosity-driven goal-directed exploration behaviours for an image sensor of a microfarming robot. A combination of deep neural networks for offline unsupervised learning of low-dimensional features from images and of online learning of shallow neural networks representing the inverse and forward kinematics of the system have been used. The artificial curiosity system assigns interest values to a set of pre-defined goals and drives the exploration towards those that are expected to maximise the learning progress. We propose the integration of an episodic memory in intrinsic motivation systems to face catastrophic forgetting issues, typically experienced when performing online updates of artificial neural networks. Our results show that adopting an episodic memory system not only prevents the computational models from quickly forgetting knowledge that has been previously acquired but also provides new avenues for modulating the balance between plasticity and stability of the models.

The School Principal and Student Learning Capacity

2017 ◽  
Vol 53 (4) ◽  
pp. 556-584 ◽  
Author(s):  
Curt M. Adams ◽  
Jentre J. Olsen ◽  
Jordan K. Ware
Keyword(s):  
Student Learning ◽  
Instructional Practices ◽  
Critical Role ◽  
Psychological Needs ◽  
School Principal ◽  
Principal Support ◽  
Learning Capacity ◽  
Perceived Need ◽  
Need Support ◽  
School Differences

Purpose: The purpose of this study was to define student learning capacity and to examine the role of the school principal in nurturing it. Method: The study used cross-sectional data from 3,175 students in 70 schools located in a metropolitan area of a Southwestern city. We tested three hypotheses by following a conventional modeling building process in HLM 7.0: Hypothesis 1—Principal Support for Student Psychological Needs (PSSPN) is related to school differences in student-perceived autonomy-support; Hypothesis 2—PSSPN is related to school differences in student-perceived competence-support; Hypothesis 3—Student-perceived need-support mediates the relationship between PSSPN and grit. Results: Evidence aligns with our initial theorizing about student learning capacity and principal support for student psychological needs. Student-perceived need-support, as a social characteristic of capacity, manifests itself through teacher–student interactions in the learning process. Need-supporting interactions varied significantly across schools, and principals played a critical role in developing an instructional environment that students experienced as nurturing autonomy and competence. Implications: PSSPN highlights the transformative effects that regular principal–teacher social exchanges can have on instructional practices. School principals who interacted with teachers about student psychological needs and need-supporting instructional practices contributed to a learning environment that students experienced as autonomy-supporting and competence-supporting.

scholarly journals Entorhinal cortical Island cells regulate temporal association learning with long trace period

2021 ◽  
Vol 28 (9) ◽  
pp. 319-328
Author(s):  
Jun Yokose ◽  
William D. Marks ◽  
Naoki Yamamoto ◽  
Sachie K. Ogawa ◽  
Takashi Kitamura
Keyword(s):  
Fear Conditioning ◽  
Critical Role ◽  
Pyramidal Cells ◽  
Temporal Association ◽  
Association Learning ◽  
Trace Fear Conditioning ◽  
Dorsal Hippocampal ◽  
Neural Input ◽  
Hippocampal Ca1 ◽  
Trace Fear

Temporal association learning (TAL) allows for the linkage of distinct, nonsynchronous events across a period of time. This function is driven by neural interactions in the entorhinal cortical–hippocampal network, especially the neural input from the pyramidal cells in layer III of medial entorhinal cortex (MECIII) to hippocampal CA1 is crucial for TAL. Successful TAL depends on the strength of event stimuli and the duration of the temporal gap between events. Whereas it has been demonstrated that the neural input from pyramidal cells in layer II of MEC, referred to as Island cells, to inhibitory neurons in dorsal hippocampal CA1 controls TAL when the strength of event stimuli is weak, it remains unknown whether Island cells regulate TAL with long trace periods as well. To understand the role of Island cells in regulating the duration of the learnable trace period in TAL, we used Pavlovian trace fear conditioning (TFC) with a 60-sec long trace period (long trace fear conditioning [L-TFC]) coupled with optogenetic and chemogenetic neural activity manipulations as well as cell type-specific neural ablation. We found that ablation of Island cells in MECII partially increases L-TFC performance. Chemogenetic manipulation of Island cells causes differential effectiveness in Island cell activity and leads to a circuit imbalance that disrupts L-TFC. However, optogenetic terminal inhibition of Island cell input to dorsal hippocampal CA1 during the temporal association period allows for long trace intervals to be learned in TFC. These results demonstrate that Island cells have a critical role in regulating the duration of time bridgeable between associated events in TAL.

scholarly journals A Universal Ankle–Foot Prosthesis Emulator for Human Locomotion Experiments

2014 ◽  
Vol 136 (3) ◽  
Author(s):  
Joshua M. Caputo ◽  
Steven H. Collins
Keyword(s):  
Degrees Of Freedom ◽  
Computational Models ◽  
Early Stage ◽  
Human Subjects ◽  
Critical Role ◽  
Tracking Error ◽  
Peak Torque ◽  
Human Robot Interaction ◽  
Limb Amputation ◽  
Wide Range

Robotic prostheses have the potential to significantly improve mobility for people with lower-limb amputation. Humans exhibit complex responses to mechanical interactions with these devices, however, and computational models are not yet able to predict such responses meaningfully. Experiments therefore play a critical role in development, but have been limited by the use of product-like prototypes, each requiring years of development and specialized for a narrow range of functions. Here we describe a robotic ankle–foot prosthesis system that enables rapid exploration of a wide range of dynamical behaviors in experiments with human subjects. This emulator comprises powerful off-board motor and control hardware, a flexible Bowden cable tether, and a lightweight instrumented prosthesis, resulting in a combination of low mass worn by the human (0.96 kg) and high mechatronic performance compared to prior platforms. Benchtop tests demonstrated closed-loop torque bandwidth of 17 Hz, peak torque of 175 Nm, and peak power of 1.0 kW. Tests with an anthropomorphic pendulum “leg” demonstrated low interference from the tether, less than 1 Nm about the hip. This combination of low worn mass, high bandwidth, high torque, and unrestricted movement makes the platform exceptionally versatile. To demonstrate suitability for human experiments, we performed preliminary tests in which a subject with unilateral transtibial amputation walked on a treadmill at 1.25 ms-1 while the prosthesis behaved in various ways. These tests revealed low torque tracking error (RMS error of 2.8 Nm) and the capacity to systematically vary work production or absorption across a broad range (from −5 to 21 J per step). These results support the use of robotic emulators during early stage assessment of proposed device functionalities and for scientific study of fundamental aspects of human–robot interaction. The design of simple, alternate end-effectors would enable studies at other joints or with additional degrees of freedom.

scholarly journals Critical role of trkB receptors in reactive axonal sprouting and hyperexcitability after axonal injury

2013 ◽  
Vol 109 (3) ◽  
pp. 813-824 ◽  
Author(s):  
Stephanie Aungst ◽  
Pamela M. England ◽  
Scott M. Thompson
Keyword(s):  
Critical Role ◽  
Neurological Complications ◽  
Receptor Activation ◽  
Axonal Sprouting ◽  
Posttraumatic Epilepsy ◽  
Trkb Receptor ◽  
Stratum Pyramidale ◽  
Trkb Receptors ◽  
Collateral Pathway ◽  
Area Ca3

Traumatic brain injury (TBI) causes many long-term neurological complications. Some of these conditions, such as posttraumatic epilepsy, are characterized by increased excitability that typically arises after a latent period lasting from months to years, suggesting that slow injury-induced processes are critical. We tested the hypothesis that trkB activation promotes delayed injury-induced hyperexcitability in part by promoting reactive axonal sprouting. We modeled penetrative TBI with transection of the Schaffer collateral pathway in knock-in mice having an introduced mutation in the trkB receptor (trkB F616A) that renders it susceptible to inhibition by the novel small molecule 1NMPP1. We observed that trkB activation was increased in area CA3 1 day after injury and that expression of a marker of axonal growth, GAP43, was increased 7 days after lesion. Extracellular field potentials in stratum pyramidale of area CA3 in acute slices from sham-operated and lesioned mice were normal in control saline. Abnormal bursts of population spikes were observed under conditions that were mildly proconvulsive but only in slices taken from mice lesioned 7–21 days earlier and not in slices from control mice. trkB activation, GAP43 upregulation, and hyperexcitability were diminished by systemic administration of 1NMPP1 for 7 days after the lesion. Synaptic transmission from area CA3 to area CA1 recovered 7 days after lesion in untreated mice but not in mice treated with 1NMPP1. We conclude that trkB receptor activation and reactive axonal sprouting are critical factors in injury-induced hyperexcitability and may contribute to the neurological complications of TBI.

Sign in / Sign up

Export Citation Format

Share Document