Review
Neurosciences
Yujun Deng, Da Song, Junjun Ni, Hong Qing, Zhenzhen Quan
Summary: Learning is a complex process where our opinions and decisions can be easily influenced by unexpected information. The neural mechanism underlying revision and correction during learning is still unclear.
FRONTIERS IN NEUROSCIENCE
(2023)
Article
Psychology, Clinical
Paola Fuentes-Claramonte, Maria Angeles Garcia-Leon, Pilar Salgado-Pineda, Nuria Ramiro, Joan Soler-Vidal, Maria Llanos Torres, Ramon Cano, Isabel Argila-Plaza, Francesco Panicali, Carmen Sarri, Nuria Jaurrieta, Manel Sanchez, Ester Boix-Quintana, Auria Albacete, Teresa Maristany, Salvador Sarro, Joaquim Radua, Peter. J. McKenna, Raymond Salvador, Edith Pomarol-Clotet
Summary: The negative symptoms of schizophrenia may be due to reduced responsiveness to rewarding stimuli, which is associated with abnormal dopamine function in the disorder. However, few imaging studies have examined whether patients with negative symptoms show reduced activation related to reward prediction error (RPE). The findings suggest that negative symptoms are not caused by a generalized reduction in RPE signaling, but rather by specific dysfunction in the lateral frontal and possibly the orbitofrontal cortex.
PSYCHOLOGICAL MEDICINE
(2023)
Article
Neurosciences
Huw Jarvis, Isabelle Stevenson, Amy Q. Huynh, Emily Babbage, James Coxon, Trevor T. -J. Chong
Summary: Recent research suggests that the act of investing effort may influence learning. This study tested whether effort modulates teaching signals in a reinforcement learning paradigm. The results showed that effort resulted in more efficient learning from positive outcomes and less efficient learning from negative outcomes. Interestingly, this effect varied across individuals and was more pronounced in those who were more averse to investing effort in the first place. These findings highlight the importance of motivational factors in a common framework of reward-based learning, integrating the computational principles of reinforcement learning with those of value-based decision-making.
JOURNAL OF NEUROSCIENCE
(2022)
Article
Psychology, Clinical
Wei Lei, Kezhi Liu, Guangxiang Chen, Serenella Tolomeo, Cuizhen Liu, Zhenlei Peng, Boya Liu, Xuemei Liang, Chaohua Huang, Bo Xiang, Jia Zhou, Fulin Zhao, Rongjun Yu, Jing Chen
Summary: This study found that patients with Internet gaming disorder (IGD) have impaired reinforcement learning and blunted reward prediction error (RPE) signals in the brain reward system, as well as hyperconnectivity between regions of the reward system. These results suggest that reinforcement learning deficits may be crucial characteristics of IGD pathophysiology.
PSYCHOLOGICAL MEDICINE
(2022)
Article
Psychology, Mathematical
Kate Ergo, Luna De Vilder, Esther De Loof, Tom Verguts
Summary: Research shows that any type of reward prediction error (RPE), whether from the participant's own response or other sources, can drive declarative learning. This finding has important implications for declarative learning theory.
PSYCHONOMIC BULLETIN & REVIEW
(2021)
Article
Neurosciences
Shuyuan Xu, Wei Ren
Summary: This study used electroencephalogram to investigate the neural correlates of state prediction errors (SPEs) in goal-directed reinforcement learning. The results suggest that the parietal correlate is responsible for explicit learning of state transition structure, while the frontal and central correlates may be involved in cognitive control.
Article
Neurosciences
Cristian B. Calderon, Esther De Loof, Kate Ergo, Anna Snoeck, Carsten N. Boehler, Tom Verguts
Summary: Behavioral evidence suggests that reward prediction errors play a key role in episodic memory acquisition. In a novel task where RPEs were manipulated, fMRI results confirmed that signed RPEs are encoded in the ventral striatum and mediate their effects on episodic memory accuracy. Connectivity between processing areas and the hippocampus and ventral striatum increased with RPE value, supporting their central role in episodic memory formation.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Computer Science, Artificial Intelligence
Xiaoshu Zhou, Fei Zhu, Peiyao Zhao
Summary: The method of prediction based on uncertainty exploration (SPE) improves the quality of exploration and reduces noise interference in deep reinforcement learning, leading to significant improvements in simulated environments.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Psychology, Clinical
Qiang Shen, Shiguang Fu, Xiaoying Jiang, Xiaoyu Huang, Doudou Lin, Qingyan Xiao, Sitti Khadijah, Yaping Yan, Xiaoxing Xiong, Jia Jin, Richard P. Ebstein, Ting Xu, Yiquan Wang, Jun Feng
Summary: This study investigates the differences in learning behavior between adolescent depressive patients and healthy controls using an instrumental learning task. The results show that depressive patients perform worse, have slower learning rates, and exhibit pessimistic biases and counterfactual outcome biases. These biases are also linked with the severity of depressive symptoms.
PSYCHOLOGICAL MEDICINE
(2023)
Article
Behavioral Sciences
Eric Rawls, Connie Lamm
Summary: The study demonstrates that mediofrontal PE signals are the mechanism underlying negative reinforcement learning, and that the increase in central delta power may contribute to the aversion positivity. PEs systematically influence future behavior in both positive and negative reinforcement conditions. In negative reinforcement conditions, cortical PE signals vary in different time windows.
Review
Psychology, Biological
Laura Fassbender, Daniel Krause, Matthias Weigelt
Summary: This study highlights the significance of feedback processing in motor learning and compares it to the cognitive domain. The findings indicate that the FRN amplitude is higher and the latency is shorter in motor tasks, possibly due to higher task complexity and feedback ambiguity.
Article
Computer Science, Artificial Intelligence
Guofeng Zhu, Fei Zhu
Summary: The article proposes a policy framework called PGMP, which utilizes multi-step prediction to guide exploration in reinforcement learning. The framework includes a curiosity mechanism and a safety bonus model to encourage exploration in safe and task-relevant areas. Additionally, a looking-ahead model is introduced to predict future states, actions, and rewards, allowing the agent to optimize its policy for predicted future states.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Computer Science, Artificial Intelligence
Shivam Kalhan, Marta I. Garrido, Robert Hester, A. David Redish
Summary: Dysfunction in learning and motivational systems is believed to contribute to addictive behaviors. Previous models have limitations in explaining the key features of addictive behaviors, but this study proposes a novel mathematical model that combines dopamine's role in learning and motivation to simulate addictive behaviors accurately. The model also explains some key characteristics of addictive behaviors.
Article
Computer Science, Artificial Intelligence
Yongming Wu, Zijun Fu, Xiaoxuan Liu, Yuan Bing
Summary: This paper proposes a prediction model that combines unsupervised learning with reinforcement learning to address the difficulty of predicting the stock market accurately. The model captures the stock trend from historical data and constructs the trading environment state using unsupervised learning algorithm, then uses a novel trading agent algorithm, Triple Q-learning, to execute trading behaviors and make comprehensive predictions. Experimental results show that the proposed model outperforms other comparative models.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Haitao Xu, Lech Szymanski, Brendan McCane
Summary: Exploration in environments with continuous control and sparse rewards is a challenging task in reinforcement learning. VASE, a surprise-based exploration method, introduces intrinsic rewards to encourage more systematic and efficient exploration. Experimental results demonstrate that VASE outperforms other surprise-based exploration techniques in such environments.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)