Pub Date : 2025-05-18DOI: 10.1016/j.cogsys.2025.101361
Jacquelyn H. Berry
What is the best way to give feedback to improve task performance? Informing someone of their success after the fact, which they can often plainly see, is effective for simple tasks. However, for complex, ecologically-based tasks with multiple subskills such as piloting a helicopter, remotely operating a robot arm, or playing Tetris, this type of feedback may be less effective. Some research suggests that certain types of feedback given during task performance maybe preferred for complex tasks rather than feedback given after the fact. This question was addressed by this pilot study which compared performance across sessions in the video game Tetris. Novice Tetris players were provided Reinforcement-based feedback, Instructive feedback, or a combination of the two. Results suggest that Instructive feedback, followed by combining the two, was most effective for improving performance over time.
{"title":"Reactive AI feedback Improves task performance over time","authors":"Jacquelyn H. Berry","doi":"10.1016/j.cogsys.2025.101361","DOIUrl":"10.1016/j.cogsys.2025.101361","url":null,"abstract":"<div><div>What is the best way to give feedback to improve task performance? Informing someone of their success after the fact, which they can often plainly see, is effective for simple tasks. However, for complex, ecologically-based tasks with multiple subskills such as piloting a helicopter, remotely operating a robot arm, or playing Tetris, this type of feedback may be less effective. Some research suggests that certain types of feedback given <em>during</em> task performance maybe preferred for complex tasks rather than feedback given after the fact. This question was addressed by this pilot study which compared performance across sessions in the video game Tetris. Novice Tetris players were provided Reinforcement-based feedback, Instructive feedback, or a combination of the two. Results suggest that Instructive feedback, followed by combining the two, was most effective for improving performance over time.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"92 ","pages":"Article 101361"},"PeriodicalIF":2.1,"publicationDate":"2025-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144115884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-05-17DOI: 10.1016/j.cogsys.2025.101362
Mengya Yin , Xilin Zhou , Qunfeng Ji , Huairen Peng , Shize Yang , Chuancheng Li
Many studies have been conducted on the effects of color, light, and signage location on the visual saliency of underground signage. However, few studies have investigated the influence of indoor visual environments on the saliency of pedestrian signage. To explore the factors that influence the visual saliency of signage in metro stations, we developed a novel analysis method using a combination of saliency and focus maps. Then, questionnaires were utilized to unify the various formats of results from the saliency and focus maps. The factors that influence the visual saliency of signage were explored using the proposed method at selected sites and validated through virtual reality experiments. Additionally, this study proposes an image-analysis-based method that reveals the multilevel factors affecting pedestrian attention to signage in underground metro stations, including spatial interfaces, crowd flow, and ambient light. The results indicate that crowd flow has the greatest impact on pedestrian attention to signage. The study’s findings underscore the significance of considering pedestrian dynamics in the design of railway stations, which is crucial for delivering a high-quality subway experience.
{"title":"Image-analysis-based method for exploring factors influencing the visual saliency of signage in metro stations","authors":"Mengya Yin , Xilin Zhou , Qunfeng Ji , Huairen Peng , Shize Yang , Chuancheng Li","doi":"10.1016/j.cogsys.2025.101362","DOIUrl":"10.1016/j.cogsys.2025.101362","url":null,"abstract":"<div><div>Many studies have been conducted on the effects of color, light, and signage location on the visual saliency of underground signage. However, few studies have investigated the influence of indoor visual environments on the saliency of pedestrian signage. To explore the factors that influence the visual saliency of signage in metro stations, we developed a novel analysis method using a combination of saliency and focus maps. Then, questionnaires were utilized to unify the various formats of results from the saliency and focus maps. The factors that influence the visual saliency of signage were explored using the proposed method at selected sites and validated through virtual reality experiments. Additionally, this study proposes an image-analysis-based method that reveals the multilevel factors affecting pedestrian attention to signage in underground metro stations, including spatial interfaces, crowd flow, and ambient light. The results indicate that crowd flow has the greatest impact on pedestrian attention to signage. The study’s findings underscore the significance of considering pedestrian dynamics in the design of railway stations, which is crucial for delivering a high-quality subway experience.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"92 ","pages":"Article 101362"},"PeriodicalIF":2.1,"publicationDate":"2025-05-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144106247","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Foodstragramming refers to how people share images of food through social media in order to have an impact on potential consumers. Hence, foodstragramming images is a way for small and medium enterprises (SMEs) to create loyal customers and promote gastronomic tourism.
An approach for analyzing foodstragramming images is presented in this paper and also their related comments published by the Getcookingcanada Instagram account, which belongs to a cooking school. The S-O-R (Stimulus-Organism-Response) model is used to study the emotions and impact on viewers of these Instagram images. Our approach evaluates the user’s preferences according to the gastronomic images and their comments and number of likes. The approach suggests possible moods or emotions that a user can have when looking at these gastronomic images based in the colors of the images, and also it studies the sentiment elicited from the comments. The analysis was performed using a variance-based structural equation modeling method called Partial Least Squares (PLS).
The obtained results show a structural model between the sentiment associated to the comments, the number of likes, and moods or emotions that can be extracted from each image. Images that evoke a positive sentiment also have a higher number of likes and comments. Also, it is shown that gastronomic images evoke the adjectives Romantic and Healthy due to the food color of the images. These adjectives produce a positive sentiment, which drive a positive behavioral intention.
{"title":"Analyzing aesthetics, attractiveness and color of gastronomic images for user engagement","authors":"Vicente Casales-Garcia , Lledó Museros , Ismael Sanz , Luis Gonzalez-Abril","doi":"10.1016/j.cogsys.2025.101358","DOIUrl":"10.1016/j.cogsys.2025.101358","url":null,"abstract":"<div><div><em>Foodstragramming</em> refers to how people share images of food through social media in order to have an impact on potential consumers. Hence, <em>foodstragramming</em> images is a way for small and medium enterprises (SMEs) to create loyal customers and promote gastronomic tourism.</div><div>An approach for analyzing <em>foodstragramming</em> images is presented in this paper and also their related comments published by the <em>Getcookingcanada</em> Instagram account, which belongs to a cooking school. The S-O-R (Stimulus-Organism-Response) model is used to study the emotions and impact on viewers of these Instagram images. Our approach evaluates the user’s preferences according to the gastronomic images and their comments and number of <em>likes</em>. The approach suggests possible moods or emotions that a user can have when looking at these gastronomic images based in the colors of the images, and also it studies the sentiment elicited from the comments. The analysis was performed using a variance-based structural equation modeling method called Partial Least Squares (PLS).</div><div>The obtained results show a structural model between the sentiment associated to the comments, the number of <em>likes</em>, and moods or emotions that can be extracted from each image. Images that evoke a positive sentiment also have a higher number of <em>likes</em> and comments. Also, it is shown that gastronomic images evoke the adjectives <em>Romantic</em> and <em>Healthy</em> due to the food color of the images. These adjectives produce a positive sentiment, which drive a positive behavioral intention.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101358"},"PeriodicalIF":2.1,"publicationDate":"2025-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143895687","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-27DOI: 10.1016/j.cogsys.2025.101357
Roel Visser , Tobias M. Peters , Ingrid Scharlau , Barbara Hammer
A current concern in the field of Artificial Intelligence (AI) is to ensure the trustworthiness of AI systems. The development of explainability methods is one prominent way to address this, which has often resulted in the assumption that the use of explainability will lead to an increase in the trust of users and wider society. However, the dynamics between explainability and trust are not well established and empirical investigations of their relation remain mixed or inconclusive.
In this paper we provide a detailed description of the concepts of user trust and distrust in AI and their relation to appropriate reliance. For that we draw from the fields of machine learning, human–computer interaction, and the social sciences. Based on these insights, we have created a focused study of empirical literature of existing empirical studies that investigate the effects of AI systems and XAI methods on user (dis)trust, in order to substantiate our conceptualization of trust, distrust, and reliance. With respect to our conceptual understanding we identify gaps in existing empirical work. With clarifying the concepts and summarizing the empirical studies, we aim to provide researchers, who examine user trust in AI, with an improved starting point for developing user studies to measure and evaluate the user’s attitude towards and reliance on AI systems.
{"title":"Trust, distrust, and appropriate reliance in (X)AI: A conceptual clarification of user trust and survey of its empirical evaluation","authors":"Roel Visser , Tobias M. Peters , Ingrid Scharlau , Barbara Hammer","doi":"10.1016/j.cogsys.2025.101357","DOIUrl":"10.1016/j.cogsys.2025.101357","url":null,"abstract":"<div><div>A current concern in the field of Artificial Intelligence (AI) is to ensure the trustworthiness of AI systems. The development of explainability methods is one prominent way to address this, which has often resulted in the assumption that the use of explainability will lead to an increase in the trust of users and wider society. However, the dynamics between explainability and trust are not well established and empirical investigations of their relation remain mixed or inconclusive.</div><div>In this paper we provide a detailed description of the concepts of user trust and distrust in AI and their relation to appropriate reliance. For that we draw from the fields of machine learning, human–computer interaction, and the social sciences. Based on these insights, we have created a focused study of empirical literature of existing empirical studies that investigate the effects of AI systems and XAI methods on user (dis)trust, in order to substantiate our conceptualization of trust, distrust, and reliance. With respect to our conceptual understanding we identify gaps in existing empirical work. With clarifying the concepts and summarizing the empirical studies, we aim to provide researchers, who examine user trust in AI, with an improved starting point for developing user studies to measure and evaluate the user’s attitude towards and reliance on AI systems.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101357"},"PeriodicalIF":2.1,"publicationDate":"2025-04-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143928469","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-21DOI: 10.1016/j.cogsys.2025.101360
Stjepan Sambol, Emra Suleyman, Michelle Ball
Executive functions (EFs) are integral to emotion regulation, yet current models often overlook the interactions between hot and cool EFs. This study aimed to develop a new framework by incorporating both hot and cool EF constructs. A sample of 150 participants (18–58 years, M = 25.87; SD = 7.48) completed assessments of hot EF (Iowa Gambling Task, Columbia Card Task, and the Understanding Emotions branch of the MSCEIT) and cool EF (Digit Span, N-Back, Keep Track, Modified Card Sorting Task, Colour-Shape Task, and Stop-Signal Task). A confirmatory factor analysis (CFA) tested a three-factor model: hot EF, working memory, and cognitive flexibility. Results partially supported this structure: while a robust working memory factor emerged and the hot EF construct was upheld, albeit with weaker loadings, item loadings for the cognitive flexibility factor were non-significant, indicating inadequate measurement. The hot EF and working memory latent factors were shown to be significantly related, supporting notions that hot and cool EFs interact. Subsequent regression analyses revealed that only the cognitive flexibility factor significantly predicted planning performance on later, more difficult Tower of Hanoi trials. This pattern suggests that demanding planning tasks appear to rely particularly on the cognitive flexibility aspect of avoiding perseverative errors—a relationship largely driven by the MCST. However, because the cognitive flexibility factor was generally weakly measured, these findings should be interpreted with caution. Ultimately, integrating hot EF into EF assessments remains essential, yet existing hot EF tasks must be further refined to more accurately reflect real-world executive demands.
{"title":"The interplay of hot and cool executive functions: Implications for a unified executive framework","authors":"Stjepan Sambol, Emra Suleyman, Michelle Ball","doi":"10.1016/j.cogsys.2025.101360","DOIUrl":"10.1016/j.cogsys.2025.101360","url":null,"abstract":"<div><div>Executive functions (EFs) are integral to emotion regulation, yet current models often overlook the interactions between hot and cool EFs. This study aimed to develop a new framework by incorporating both hot and cool EF constructs. A sample of 150 participants (18–58 years, <em>M</em> = 25.87; <em>SD</em> = 7.48) completed assessments of hot EF (Iowa Gambling Task, Columbia Card Task, and the Understanding Emotions branch of the MSCEIT) and cool EF (Digit Span, N-Back, Keep Track, Modified Card Sorting Task, Colour-Shape Task, and Stop-Signal Task). A confirmatory factor analysis (CFA) tested a three-factor model: hot EF, working memory, and cognitive flexibility. Results partially supported this structure: while a robust working memory factor emerged and the hot EF construct was upheld, albeit with weaker loadings, item loadings for the cognitive flexibility factor were non-significant, indicating inadequate measurement. The hot EF and working memory latent factors were shown to be significantly related, supporting notions that hot and cool EFs interact. Subsequent regression analyses revealed that only the cognitive flexibility factor significantly predicted planning performance on later, more difficult Tower of Hanoi trials. This pattern suggests that demanding planning tasks appear to rely particularly on the cognitive flexibility aspect of avoiding perseverative errors—a relationship largely driven by the MCST. However, because the cognitive flexibility factor was generally weakly measured, these findings should be interpreted with caution. Ultimately, integrating hot EF into EF assessments remains essential, yet existing hot EF tasks must be further refined to more accurately reflect real-world executive demands.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101360"},"PeriodicalIF":2.1,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143873641","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-20DOI: 10.1016/j.cogsys.2025.101359
Stanislav Matousek , Pavel Skobrtal , Radim Badosek
Despite the ongoing controversy, the Rorschach remains one of the most popular tests among professionals conducting personality assessments. The present study focuses on using eye tracking to obtain gaze density maps of 20 healthy participants on standardized visual material, offering insight into the patterns of visual attention distribution during the examination of standard Rorschach inkblots. Specifically, our analysis shows that gaze density confirms our empirical experience and exhibits considerable symmetry around the central axis of symmetry of the inkblots. In addition, inkblots commonly interpreted as depicting human or animal figures show increased human attention to the ‘head’ and upper ‘torso’ regions. These hypotheses are supported by the significance testing results, except the symmetry hypothesis, which suggests considerable but incomplete symmetry. Interpreting these findings in the context of current cognitive research, we consider them evidence for an evolutionarily rooted system governing automatic perception and attention. These findings could enhance psychologists’ understanding of their responses to the Rorschach test. They may also help in the development of more efficient systems, particularly in the field of artificial intelligence learning. This approximation of knowledge regarding human behavior in visuomotor preferences could also benefit other disciplines.
{"title":"Hidden rules of visual perception in the Rorschach test","authors":"Stanislav Matousek , Pavel Skobrtal , Radim Badosek","doi":"10.1016/j.cogsys.2025.101359","DOIUrl":"10.1016/j.cogsys.2025.101359","url":null,"abstract":"<div><div>Despite the ongoing controversy, the Rorschach remains one of the most popular tests among professionals conducting personality assessments. The present study focuses on using eye tracking to obtain gaze density maps of 20 healthy participants on standardized visual material, offering insight into the patterns of visual attention distribution during the examination of standard Rorschach inkblots. Specifically, our analysis shows that gaze density confirms our empirical experience and exhibits considerable symmetry around the central axis of symmetry of the inkblots. In addition, inkblots commonly interpreted as depicting human or animal figures show increased human attention to the ‘head’ and upper ‘torso’ regions. These hypotheses are supported by the significance testing results, except the symmetry hypothesis, which suggests considerable but incomplete symmetry. Interpreting these findings in the context of current cognitive research, we consider them evidence for an evolutionarily rooted system governing automatic perception and attention. These findings could enhance psychologists’ understanding of their responses to the Rorschach test. They may also help in the development of more efficient systems, particularly in the field of artificial intelligence learning. This approximation of knowledge regarding human behavior in visuomotor preferences could also benefit other disciplines.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101359"},"PeriodicalIF":2.1,"publicationDate":"2025-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143869719","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-04DOI: 10.1016/j.cogsys.2025.101356
Dongyu Zhang , Xingyuan Lu , Mulin Zhuang , Senqi Yang , Hongjun Chen
Metaphor is a way of thinking and cognition prevalent in human language. With the development of social media and multimodal data, metaphor recognition research has expanded from the traditional unimodal scope (such as text or images) to the multimodality. However, current multimodal metaphor processing methods mainly focus on fusion techniques for multiple modalities such as text and image, but neglect the cognitive mechanism of metaphor as a way of thinking, and are deficient in utilizing pre-trained information from large language models. Therefore, this paper proposes a chain-of-cognition prompting (CoC) method to address multimodal metaphor recognition task, which makes full use of the pre-training information of the large model in order to better recognize metaphors. The method utilizes prompting words to construct inputs that guide the large language model to reason about potential metaphorical source and target domain related entities and associations between entities in the sample. At the same time, visual information is obtained through image caption extraction and a visual encoder to enable the model to reason and produce metaphor recognition results. The experimental results show that the method performs well on the metaphor recognition task, which is better than the existing baseline model, verifying the effectiveness of the method on the metaphor recognition task.
{"title":"Multimodal metaphor recognition based on chain-of-cognition prompting","authors":"Dongyu Zhang , Xingyuan Lu , Mulin Zhuang , Senqi Yang , Hongjun Chen","doi":"10.1016/j.cogsys.2025.101356","DOIUrl":"10.1016/j.cogsys.2025.101356","url":null,"abstract":"<div><div>Metaphor is a way of thinking and cognition prevalent in human language. With the development of social media and multimodal data, metaphor recognition research has expanded from the traditional unimodal scope (such as text or images) to the multimodality. However, current multimodal metaphor processing methods mainly focus on fusion techniques for multiple modalities such as text and image, but neglect the cognitive mechanism of metaphor as a way of thinking, and are deficient in utilizing pre-trained information from large language models. Therefore, this paper proposes a chain-of-cognition prompting (CoC) method to address multimodal metaphor recognition task, which makes full use of the pre-training information of the large model in order to better recognize metaphors. The method utilizes prompting words to construct inputs that guide the large language model to reason about potential metaphorical source and target domain related entities and associations between entities in the sample. At the same time, visual information is obtained through image caption extraction and a visual encoder to enable the model to reason and produce metaphor recognition results. The experimental results show that the method performs well on the metaphor recognition task, which is better than the existing baseline model, verifying the effectiveness of the method on the metaphor recognition task.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101356"},"PeriodicalIF":2.1,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143807020","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-01DOI: 10.1016/j.cogsys.2025.101352
Carmelo Fabio Longo , Misael Mongiovì , Luana Bulla , Antonio Lieto
The introduction of Large Language Models (LLMs) able to exhibit a number of linguistic and extra-linguistic capabilities has represented, in the last years, one of the main frontiers in Artificial Intelligence (AI) research. Researcher from various disciplines debate about whether or not, among the capabilities of LLMs, there is the one of using knowledge about knowledge – usually considered one of the antechambers of meta-cognition in cognitive agents – about a particular task in order to improve or self-correct previous errors. In this work we propose a novel fine-tuning approach for LLMs, named exar, based on a multi-stage process leveraging past predictions from an early version of the same, and aimed at injecting metacognitive features for the task of Question-Answering. The conducted experiments on Llama-2-7B-chat showed promising improvements on the quality of the outcomes, due to the fact that the LLM acquired the ability to detect its own wrong predictions forcing itself to repeat submissions, thorough a prompt designed to fix inadmissible predictions, whenever detected. Such detection is achieved by enquiring the same LLM acting as meta-validator, through another prompt specifically designed for such purpose.
{"title":"Eliciting metaknowledge in Large Language Models","authors":"Carmelo Fabio Longo , Misael Mongiovì , Luana Bulla , Antonio Lieto","doi":"10.1016/j.cogsys.2025.101352","DOIUrl":"10.1016/j.cogsys.2025.101352","url":null,"abstract":"<div><div>The introduction of Large Language Models (LLMs) able to exhibit a number of linguistic and extra-linguistic capabilities has represented, in the last years, one of the main frontiers in Artificial Intelligence (AI) research. Researcher from various disciplines debate about whether or not, among the capabilities of LLMs, there is the one of using <em>knowledge about knowledge</em> – usually considered one of the antechambers of <em>meta-cognition</em> in cognitive agents – about a particular task in order to improve or self-correct previous errors. In this work we propose a novel fine-tuning approach for LLMs, named <span>exar</span>, based on a multi-stage process leveraging past predictions from an early version of the same, and aimed at <em>injecting</em> metacognitive features for the task of Question-Answering. The conducted experiments on <span>Llama-2-7B-chat</span> showed promising improvements on the quality of the outcomes, due to the fact that the LLM acquired the ability to detect its own wrong predictions forcing itself to repeat submissions, thorough a prompt designed to fix inadmissible predictions, whenever detected. Such detection is achieved by enquiring the same LLM acting as meta-validator, through another prompt specifically designed for such purpose.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101352"},"PeriodicalIF":2.1,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143768761","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-01DOI: 10.1016/j.cogsys.2025.101354
Gustavo Morais , Eduardo Yuji , Paula Costa , Alexandre Simões , Ricardo Gudwin , Esther Colombini
Recent advancements in reinforcement learning (RL), particularly deep RL, show the capacity of this paradigm to perform varied and complex tasks. However, a series of exploration, generalization, and adaptation challenges hold RL back from operating in more general contexts. In this paper, we explore integrating techniques originating from cognitive research into existing RL algorithms by defining a general framework to standardize interoperation between arbitrary cognitive modules and arbitrary RL techniques. We show the potential of hybrid approaches through a comparative experiment that integrates an episodic memory encoder with a well-known deep RL algorithm. Furthermore, we show that built-in RL algorithms with different cognitive modules can fit our framework, as well as remotely run algorithms. Hence, we propose a way forward for RL in the form of innovative solutions that integrate research in cognitive systems with recent RL techniques.
{"title":"A general framework for reinforcement learning in cognitive architectures","authors":"Gustavo Morais , Eduardo Yuji , Paula Costa , Alexandre Simões , Ricardo Gudwin , Esther Colombini","doi":"10.1016/j.cogsys.2025.101354","DOIUrl":"10.1016/j.cogsys.2025.101354","url":null,"abstract":"<div><div>Recent advancements in reinforcement learning (RL), particularly deep RL, show the capacity of this paradigm to perform varied and complex tasks. However, a series of exploration, generalization, and adaptation challenges hold RL back from operating in more general contexts. In this paper, we explore integrating techniques originating from cognitive research into existing RL algorithms by defining a general framework to standardize interoperation between arbitrary cognitive modules and arbitrary RL techniques. We show the potential of hybrid approaches through a comparative experiment that integrates an episodic memory encoder with a well-known deep RL algorithm. Furthermore, we show that built-in RL algorithms with different cognitive modules can fit our framework, as well as remotely run algorithms. Hence, we propose a way forward for RL in the form of innovative solutions that integrate research in cognitive systems with recent RL techniques.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101354"},"PeriodicalIF":2.1,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143777264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Data processing capability of lower power networks can be improved by Mobile Edge Computing (MEC) extending to the wireless sensor networks and IoT. Creating a replication of MEC network with an offloading policy where a choice is made in the Wireless devices (WDs) for each computation task is the focus of this study. Deciding whether the task execution proceeds locally in the same environment or can be handed over to a remote MEC server, an optimized algorithm is needed which adopts task offloading decisions and wireless resource allocation in real time. But adopting this is a challenging solution to the real time fast combinatorial optimization problems, and impossible with the available traditional approaches. As a solution, heuristic algorithms encompassing Deep reinforcement learning (DRL) are emerging; however, it doesn’t make fair use of connection data like device-to-device interaction in MEC network. Moreover, heuristic algorithms rely on precise mathematical models for MEC systems which brought a new theory to the stage. This study revolves around this emerging technique relying on Graph neural networks (GNNs) learns from graph data while forwarding messages in the network. Utilizing GNN benefits, a Graph reinforcement learning-based online offloading framework (GROO) is proposed in this research, where the offloading policy is visualized as a graph state migration and MEC as an acyclic graph. The GROO achieves the lowest weighted task response latency (0.96 s) as compared to the existing DRL method (1.32 s) whereas on unseen circumstances and complex network topologies, GROO achieved lowest average latency up to 25 %.
{"title":"A graph reinforcement LearningPowered Online-Computational task offloading and latency minimization framework for wireless mobile edge computing networks","authors":"Akshat Agrawal , Aayush Agrawal , Nilesh Kumar Verma , Arepalli Peda Gopi , K. Jairam Naik","doi":"10.1016/j.cogsys.2025.101355","DOIUrl":"10.1016/j.cogsys.2025.101355","url":null,"abstract":"<div><div>Data processing capability of lower power networks can be improved by Mobile Edge Computing (MEC) extending to the wireless sensor networks and IoT. Creating a replication of MEC network with an offloading policy where a choice is made in the Wireless devices (WDs) for each computation task is the focus of this study. Deciding whether the task execution proceeds locally in the same environment or can be handed over to a remote MEC server, an optimized algorithm is needed which adopts task offloading decisions and wireless resource allocation in real time. But adopting this is a challenging solution to the real time fast combinatorial optimization problems, and impossible with the available traditional approaches. As a solution, heuristic algorithms encompassing Deep reinforcement learning (DRL) are emerging; however, it doesn’t make fair use of connection data like device-to-device interaction in MEC network. Moreover, heuristic algorithms rely on precise mathematical models for MEC systems which brought a new theory to the stage. This study revolves around this emerging technique relying on Graph neural networks (GNNs) learns from graph data while forwarding messages in the network. Utilizing GNN benefits, a Graph reinforcement learning-based online offloading framework (GROO) is proposed in this research, where the offloading policy is visualized as a graph state migration and MEC as an acyclic graph. The GROO achieves the lowest weighted task response latency (0.96 s) as compared to the existing DRL method (1.32 s) whereas on unseen circumstances and complex network topologies, GROO achieved lowest average latency up to 25 %.</div></div>","PeriodicalId":55242,"journal":{"name":"Cognitive Systems Research","volume":"91 ","pages":"Article 101355"},"PeriodicalIF":2.1,"publicationDate":"2025-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687970","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号