scholarly journals Context-dependent extinction learning emerging from raw sensory inputs: A reinforcement learning approach

2020 ◽  
Author(s):  
Thomas Walther ◽  
Nicolas Diekmann ◽  
Sandhiya Vijayabaskaran ◽  
José R. Donoso ◽  
Denise Manahan-Vaughan ◽  
...  
Keyword(s):  
Reinforcement Learning ◽  
Deep Neural Networks ◽  
Neural Mechanisms ◽  
Context Dependence ◽  
Learning Approach ◽  
Context Change ◽  
Extinction Learning ◽  
Sensory Inputs ◽  
Visual Inputs

AbstractThe context-dependence of extinction learning has been well studied and requires the hippocampus. However, the underlying neural mechanisms are still poorly understood. Using memory-driven reinforcement learning and deep neural networks, we developed a model that learns to navigate autonomously in biologically realistic VR environments based on raw camera inputs alone. Neither is context represented explicitly in our model, nor is context change signaled. We find that memory-intact agents learn distinct context representations, and develop ABA renewal, whereas memory-impaired agents do not. These findings reproduce the behavior of control and hippocampal animals, respectively. We therefore propose that the role of the hippocampus in the context-dependence of extinction learning might stem from its function in episodic-like memory and not in context-representation per se. We conclude that context-dependence can emerge from raw visual inputs.

scholarly journals Context-dependent extinction learning emerging from raw sensory inputs: a reinforcement learning approach

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Thomas Walther ◽  
Nicolas Diekmann ◽  
Sandhiya Vijayabaskaran ◽  
José R. Donoso ◽  
Denise Manahan-Vaughan ◽  
...  
Keyword(s):  
Reinforcement Learning ◽  
Deep Neural Networks ◽  
Neural Mechanisms ◽  
Context Dependence ◽  
Learning Approach ◽  
Context Change ◽  
Extinction Learning ◽  
Sensory Inputs ◽  
Visual Inputs

AbstractThe context-dependence of extinction learning has been well studied and requires the hippocampus. However, the underlying neural mechanisms are still poorly understood. Using memory-driven reinforcement learning and deep neural networks, we developed a model that learns to navigate autonomously in biologically realistic virtual reality environments based on raw camera inputs alone. Neither is context represented explicitly in our model, nor is context change signaled. We find that memory-intact agents learn distinct context representations, and develop ABA renewal, whereas memory-impaired agents do not. These findings reproduce the behavior of control and hippocampal animals, respectively. We therefore propose that the role of the hippocampus in the context-dependence of extinction learning might stem from its function in episodic-like memory and not in context-representation per se. We conclude that context-dependence can emerge from raw visual inputs.

scholarly journals The Role of Temporal Contingency and Integrity of Visual Inputs in the Sense of Agency: A Psychophysical Study

2021 ◽  
Vol 12 ◽  
Author(s):  
Hiroaki Mizuhara ◽  
Peter Uhlhaas
Keyword(s):  
Likert Scale ◽  
Sense Of Agency ◽  
Subjective Feeling ◽  
Finger Movements ◽  
Computer Screen ◽  
Sensory Inputs ◽  
Visual Inputs ◽  
Psychophysical Study ◽  
Psychophysical Task

The sense of agency is a subjective feeling that one's own actions drive action outcomes. Previous studies have focused primarily on the temporal contingency between actions and sensory inputs as a possible mechanism for the sense of agency. However, the contribution of the integrity of visual inputs has not been systematically addressed. In the current study, we developed a psychophysical task to examine the role of visual inputs as well as temporal contingencies toward the sense of agency. Specifically, participants were required to track a target on a sinusoidal curve on a computer screen. Visual integrity of sensory inputs was manipulated by gradually occluding a computer cursor, and participants were asked to report the sense of agency on a nine-point Likert scale. Temporal contingency was manipulated by varying the delay between finger movements on a touchpad and cursor movements. The results showed that the sense of agency was influenced by both visual integrity and temporal contingency. These results are discussed in the context of current models that have proposed that the sense of agency emerges from the comparison of visual inputs with motor commands.

Audio-Visual Multisensory Integration in Superior Parietal Lobule Revealed by Human Intracranial Recordings

2006 ◽  
Vol 96 (2) ◽  
pp. 721-729 ◽  
Author(s):  
Sophie Molholm ◽  
Pejman Sehatpour ◽  
Ashesh D. Mehta ◽  
Marina Shpaner ◽  
Manuel Gomez-Ramirez ◽  
...  
Keyword(s):  
Multisensory Integration ◽  
Visual Stimulation ◽  
Multisensory Processing ◽  
Human Subjects ◽  
Superior Parietal Lobule ◽  
Sensory Inputs ◽  
Visual Inputs ◽  
Parietal Lobule ◽  
Intracranial Recordings

Intracranial recordings from three human subjects provide the first direct electrophysiological evidence for audio-visual multisensory processing in the human superior parietal lobule (SPL). Auditory and visual sensory inputs project to the same highly localized region of the parietal cortex with auditory inputs arriving considerably earlier (30 ms) than visual inputs (75 ms). Multisensory integration processes in this region were assessed by comparing the response to simultaneous audio-visual stimulation with the algebraic sum of responses to the constituent auditory and visual unisensory stimulus conditions. Significant integration effects were seen with almost identical morphology across the three subjects, beginning between 120 and 160 ms. These results are discussed in the context of the role of SPL in supramodal spatial attention and sensory-motor transformations.

scholarly journals The role of the optic tectum for visually evoked orienting and evasive movements

2019 ◽  
Vol 116 (30) ◽  
pp. 15272-15281 ◽  
Author(s):  
Daichi G. Suzuki ◽  
Juan Pérez-Fernández ◽  
Tobias Wibble ◽  
Andreas A. Kardamakis ◽  
Sten Grillner
Keyword(s):  
Optic Tectum ◽  
Neural Mechanisms ◽  
Visual Features ◽  
Motor Patterns ◽  
Behavioral Choice ◽  
Visual Inputs ◽  
Pharmacological Blockade ◽  
The Common ◽  
Motor Actions

As animals forage for food and water or evade predators, they must rapidly decide what visual features in the environment deserve attention. In vertebrates, this visuomotor computation is implemented within the neural circuits of the optic tectum (superior colliculus in mammals). However, the mechanisms by which tectum decides whether to approach or evade remain unclear, and also which neural mechanisms underlie this behavioral choice. To address this problem, we used an eye–brain–spinal cord preparation to evaluate how the lamprey responds to visual inputs with distinct stimulus-dependent motor patterns. Using ventral root activity as a behavioral readout, we classified 2 main types of fictive motor responses: (i) a unilateral burst response corresponding to orientation of the head toward slowly expanding or moving stimuli, particularly within the anterior visual field, and (ii) a unilateral or bilateral burst response triggering fictive avoidance in response to rapidly expanding looming stimuli or moving bars. A selective pharmacological blockade revealed that the brainstem-projecting neurons in the deep layer of the tectum in interaction with local inhibitory interneurons are responsible for selecting between these 2 visually triggered motor actions conveyed through downstream reticulospinal circuits. We suggest that these visual decision-making circuits had evolved in the common ancestor of vertebrates and have been conserved throughout vertebrate phylogeny.

Energy-efficient UAV trajectory design for backscatter communication: A deep reinforcement learning approach

2020 ◽  
Vol 17 (10) ◽  
pp. 129-141
Author(s):  
Yiwen Nie ◽  
Junhui Zhao ◽  
Jun Liu ◽  
Jing Jiang ◽  
Ruijin Ding
Keyword(s):  
Reinforcement Learning ◽  
Energy Efficient ◽  
Learning Approach ◽  
Trajectory Design ◽  
Backscatter Communication

The Role of Tahfidzul Quran Learning to Build the Students’ Character at Madrasah Ibtidaiyah Istiqomah Sambas Purbalingga

2018 ◽  
Vol 2 ◽  
pp. 235-245
Author(s):  
Ikhwandi Arifin
Keyword(s):  
Elementary School ◽  
Primary School ◽  
Quality Of Education ◽  
Learning Approach ◽  
The Future ◽  
Character Building

This paper discusses the importance of the students ‘character building in the level of primary school/Islamic elementary school to face the globalization and information era. Education is the process of determining the nation’s character. Good or bad character of the nation in the future will be determined by the present quality of education. Building the character through Tahfidzul Quran learning approach is expected to be the main foundation to improve the degree and prestige of learners as the asset of the nation. This study aimed to describe the process of Tahfidzul Quran learning which included planning, organizing, doing action and monitoring the important role of learning itself to build the learners’ character, especially in Madrasah Ibtidaiyah Istiqomah Sambas Purbalingga.

Sign in / Sign up

Export Citation Format

Share Document