Visuospatial Research Articles

The Influence of Space Standards on Housing Typologies: The Evolution of the Nuclear Family Dwelling in England

This article explores the evolution of dwelling typologies in London, examining how regulations and standards have transformed housing layouts and indirectly informed personal and social interactions at home. Housing policy of the past century, as reflected through space standards, type plans, housing manuals, and reports, reveal a socio-political agenda of promoting nuclear family dwellings, traditionally a household of parents and their children. This article contributes to this discussion by exploring the contextual drivers, spatial reasoning, and evidence underpinning the decisions of housing reports such as the <em>Tudor Walters Report</em> (1918), the <em>Dudley Report</em> (1944), and the <em>Parker Morris Report</em> (1961). Changes in household structures can be seen through regulations: from multiple families sharing a house, to the separation of individual families into single homes, to prioritising the individual over the family unit. This article analyses how five historical moments in which typological shifts promoted nuclear family dwellings that have determined spatial hierarchies and family dynamics around cooking, eating, and socialising. Similarly, societal shifts in housing expectations, such as the changing perceptions of social status symbols, privacy, gender roles, and household dynamics have contributed to the spatial arrangement and layout of homes. By shedding light on the socio-technical transformations, this research highlights the need for innovative design solutions and evidence-based space standards to meet contemporary needs.

Advancing Spatial Reasoning in Large Language Models: An In-Depth Evaluation and Enhancement Using the StepGame Benchmark

Artificial intelligence (AI) has made remarkable progress across various domains, with large language models like ChatGPT gaining substantial attention for their human-like text-generation capabilities. Despite these achievements, improving spatial reasoning remains a significant challenge for these models. Benchmarks like StepGame evaluate AI spatial reasoning, where ChatGPT has shown unsatisfactory performance. However, the presence of template errors in the benchmark has an impact on the evaluation results. Thus there is potential for ChatGPT to perform better if these template errors are addressed, leading to more accurate assessments of its spatial reasoning capabilities. In this study, we refine the StepGame benchmark, providing a more accurate dataset for model evaluation. We analyze GPT’s spatial reasoning performance on the rectified benchmark, identifying proficiency in mapping natural language text to spatial relations but limitations in multi-hop reasoning. We provide a flawless solution to the benchmark by combining template-to-relation mapping with logic-based reasoning. This combination demonstrates proficiency in performing qualitative reasoning on StepGame without encountering any errors. We then address the limitations of GPT models in spatial reasoning. To improve spatial reasoning, we deploy Chain-of-Thought and Tree-of-thoughts prompting strategies, offering insights into GPT’s cognitive process. Our investigation not only sheds light on model deficiencies but also proposes enhancements, contributing to the advancement of AI with more robust spatial reasoning capabilities.

STAIR: Spatial-Temporal Reasoning with Auditable Intermediate Results for Video Question Answering

Recently we have witnessed the rapid development of video question answering models. However, most models can only handle simple videos in terms of temporal reasoning, and their performance tends to drop when answering temporal-reasoning questions on long and informative videos. To tackle this problem we propose STAIR, a Spatial-Temporal Reasoning model with Auditable Intermediate Results for video question answering. STAIR is a neural module network, which contains a program generator to decompose a given question into a hierarchical combination of several sub-tasks, and a set of lightweight neural modules to complete each of these sub-tasks. Though neural module networks are already widely studied on image-text tasks, applying them to videos is a non-trivial task, as reasoning on videos requires different abilities. In this paper, we define a set of basic video-text sub-tasks for video question answering and design a set of lightweight modules to complete them. Different from most prior works, modules of STAIR return intermediate outputs specific to their intentions instead of always returning attention maps, which makes it easier to interpret and collaborate with pre-trained models. We also introduce intermediate supervision to make these intermediate outputs more accurate. We conduct extensive experiments on several video question answering datasets under various settings to show STAIR's performance, explainability, compatibility with pre-trained models, and applicability when program annotations are not available. Code: https://github.com/yellow-binary-tree/STAIR

VELMA: Verbalization Embodiment of LLM Agents for Vision and Language Navigation in Street View

Incremental decision making in real-world environments is one of the most challenging tasks in embodied artificial intelligence. One particularly demanding scenario is Vision and Language Navigation (VLN) which requires visual and natural language understanding as well as spatial and temporal reasoning capabilities. The embodied agent needs to ground its understanding of navigation instructions in observations of a real-world environment like Street View. Despite the impressive results of LLMs in other research areas, it is an ongoing problem of how to best connect them with an interactive visual environment. In this work, we propose VELMA, an embodied LLM agent that uses a verbalization of the trajectory and of visual environment observations as contextual prompt for the next action. Visual information is verbalized by a pipeline that extracts landmarks from the human written navigation instructions and uses CLIP to determine their visibility in the current panorama view. We show that VELMA is able to successfully follow navigation instructions in Street View with only two in-context examples. We further finetune the LLM agent on a few thousand examples and achieve around 25% relative improvement in task completion over the previous state-of-the-art for two datasets.

Spatial-Logic-Aware Weakly Supervised Learning for Flood Mapping on Earth Imagery

Flood mapping on Earth imagery is crucial for disaster management, but its efficacy is hampered by the lack of high-quality training labels. Given high-resolution Earth imagery with coarse and noisy training labels, a base deep neural network model, and a spatial knowledge base with label constraints, our problem is to infer the true high-resolution labels while training neural network parameters. Traditional methods are largely based on specific physical properties and thus fall short of capturing the rich domain constraints expressed by symbolic logic. Neural-symbolic models can capture rich domain knowledge, but existing methods do not address the unique spatial challenges inherent in flood mapping on high-resolution imagery. To fill this gap, we propose a spatial-logic-aware weakly supervised learning framework. Our framework integrates symbolic spatial logic inference into probabilistic learning in a weakly supervised setting. To reduce the time costs of logic inference on vast high-resolution pixels, we propose a multi-resolution spatial reasoning algorithm to infer true labels while training neural network parameters. Evaluations of real-world flood datasets show that our model outperforms several baselines in prediction accuracy. The code is available at https://github.com/spatialdatasciencegroup/SLWSL.

BLIVA: A Simple Multimodal LLM for Better Handling of Text-Rich Visual Questions

Vision Language Models (VLMs), which extend Large Language Models (LLM) by incorporating visual understanding capability, have demonstrated significant advancements in addressing open-ended visual question-answering (VQA) tasks. However, these models cannot accurately interpret images infused with text, a common occurrence in real-world scenarios. Standard procedures for extracting information from images often involve learning a fixed set of query embeddings. These embeddings are designed to encapsulate image contexts and are later used as soft prompt inputs in LLMs. Yet, this process is limited to the token count, potentially curtailing the recognition of scenes with text-rich context. To improve upon them, the present study introduces BLIVA: an augmented version of InstructBLIP with Visual Assistant. BLIVA incorporates the query embeddings from InstructBLIP and also directly projects encoded patch embeddings into the LLM, a technique inspired by LLaVA. This approach assists the model to capture intricate details potentially missed during the query decoding process. Empirical evidence demonstrates that our model, BLIVA, significantly enhances performance in processing text-rich VQA benchmarks (up to 17.76% in OCR-VQA benchmark) and in undertaking general (not particularly text-rich) VQA benchmarks (up to 7.9% in Visual Spatial Reasoning benchmark), and achieved 17.72% overall improvement in a comprehensive multimodal LLM benchmark (MME), comparing to our baseline InstructBLIP. BLIVA demonstrates significant capability in decoding real-world images, irrespective of text presence. To demonstrate the broad industry applications enabled by BLIVA, we evaluate the model using a new dataset comprising YouTube thumbnails paired with question-answer sets across 11 diverse categories. For researchers interested in further exploration, our code and models are freely accessible at https://github.com/mlpc-ucsd/BLIVA.

Visuospatial cognition predicts performance on an obstructed vision obstacle walking task in older adults

Walking performance and cognitive function demonstrate strong associations in older adults, with both declining with advancing age. Walking requires the use of cognitive resources, particularly in complex environments like stepping over obstacles. A commonly implemented approach for measuring the cognitive control of walking is a dual-task walking assessment, in which walking is combined with a second task. However, dual-task assessments have shortcomings, including issues with scaling the task difficulty and controlling for task prioritization. Here we present a new assessment designed to be less susceptible to these shortcomings while still challenging cognitive control of walking: the Obstructed Vision Obstacle (OBVIO) task. During the task, participants hold a lightweight tray at waist level obstructing their view of upcoming foam blocks, which are intermittently spaced along a 10 m walkway. This forces the participants to use cognitive resources (e.g., attention and working memory) to remember the exact placement of upcoming obstacles to facilitate successful crossing. The results demonstrate that adding the obstructed vision board significantly slowed walking speed by an average of 0.26 m/s and increased the number of obstacle strikes by 8-fold in healthy older adults (n = 74). Additionally, OBVIO walking performance (a score based on both speed and number of obstacle strikes) significantly correlated with computer-based assessments of visuospatial working memory, attention, and verbal working memory. These results provide initial support that the OBVIO task is a feasible walking test that demands cognitive resources. This study lays the groundwork for using the OBVIO task in future assessment and intervention studies.

A Qualitative Description of Biochemistry Students’ Visuospatial Reasoning Difficulties Associated with Amino Acid Models

The adoption of multimedia learning tools, including visual models in biochemistry, has increased considerably over the last few decades. Adopting these tools necessitates the development of visual literacy competencies, such as visuospatial reasoning, for students to learn effectively. The extent to which biochemistry students have the necessary visuospatial reasoning skills to learn from visual models effectively is a subject of continuing research. The current qualitative research sought to describe biochemistry students’ visuospatial reasoning difficulties associated with amino acids. Seven purposively selected students were interviewed using a semi-structured protocol to solicit their learning difficulties when asked to perform specific tasks related to visuospatial reasoning. The research question explored in this study is: “What are biochemistry students’ visuospatial reasoning difficulties associated with amino acids models?” Learning difficulties related to perceiving spatial orientation, mental rotation, interpreting colour codes, and generating a visual model depicting spatial depth were identified. Identifying and describing these learning difficulties could help teachers adopt remediation strategies to enhance content understanding. The researcher concludes that students may not always have the visuospatial reasoning skills required for successful learning in biochemistry. Given the complex nature of visual literacy, the researcher recommends explicit remediation strategies to address learning difficulties associated with visuospatial reasoning in biochemistry.

CASPER: Cognitive Architecture for Social Perception and Engagement in Robots

AbstractOur world is being increasingly pervaded by intelligent robots with varying degrees of autonomy. To seamlessly integrate themselves in our society, these machines should possess the ability to navigate the complexities of our daily routines even in the absence of a human’s direct input. In other words, we want these robots to understand the intentions of their partners with the purpose of predicting the best way to help them. In this paper, we present the initial iteration of cognitive architecture for social perception and engagement in robots: a symbolic cognitive architecture that uses qualitative spatial reasoning to anticipate the pursued goal of another agent and to calculate the best collaborative behavior. This is performed through an ensemble of parallel processes that model a low-level action recognition and a high-level goal understanding, both of which are formally verified. We have tested this architecture in a simulated kitchen environment and the results we have collected show that the robot is able to both recognize an ongoing goal and to properly collaborate towards its achievement. This demonstrates a new use of qualitative spatial relations applied to the problem of intention reading in the domain of human–robot interaction.

Projective geometry and spatial reasoning for STEM learning

Projective geometry is a prominent area in many fields including art, design, architecture, and mathematics, but how it can contribute to children’s spatial reasoning as well as a supplementary geometry to that of Euclid’s in school mathematics curricula raises the need for further consideration. With emphasis on the dynamic relationship between 2D and 3D objects, projective geometry not only offers important concepts but a meaningful context in which to develop spatial reasoning for STEM learning. In this article, we overview spatial reasoning as it relates to STEM and in particular, mathematics. Making a case for projective geometry, we present activities for the classroom, demonstrating how this mathematics topic connects to and opens new possibilities to promote spatial reasoning for STEM learning in the elementary grades.

A systematic review of transcranial magnetic and direct current stimulation application on the improvement of reading ability of developmental dyslexia

Transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS), two non-invasive brain stimulation techniques (NIBS), are modulatory tools to probe into a possible causal and interaction relationship between brain structure and function. Developmental dyslexia (DD) is a neurodevelopmental disorder, which is characterized by significant and persistent difficulty in learning and reading-related skills. Previous studies have shown that TMS and tDCS play an important role in reading ability and neurocognitive plasticity, such as visuospatial working memory, attention, speech recognition, and motor perception. However, research on the effectiveness of TMS/tDCS in treating DD is still limited. The current study aimed to systematically review the broadly application of TMS/tDCS in the treatment of dyslexic children and adolescents as well as adults. Twelve studies involving 128 dyslexic subjects (199 children/adolescents and 29 adults) were included in this systematic review. Overall, TMS/tDCS appeared to be an effective technique for treating dyslexia. However, further future studies with larger sample sizes as well as more different languages are essential to demonstrate its potential for successful intervention in DD. This systematic review suggested that the combination of tDCS and cognitive training was effective. Moreover, the treatment with multiple sessions and coupled with behavioral training appeared to result in greater efficacy than stand-alone NIBS treatment.

Sex-specific neuropsychological correlates of apathy and depression across neurodegenerative disorders.

Apathy and depression are common neuropsychiatric symptoms across neurodegenerative disorders and are associated with impairment in several cognitive domains, yet little is known about the influence of sex on these relationships. We examined the relationship between these symptoms with neuropsychological performance across a combined cohort with mild or major neurodegenerative disorders, then evaluated the impact of sex. We conducted a cohort analysis of participants in the COMPASS-ND study with mild cognitive impairment (MCI), vascular MCI, Alzheimer's disease, mixed dementia, Parkinson's disease, frontotemporal dementia, and cognitively unimpaired (CU) controls. Participants with neurodegenerative disease and CU controls were stratified by the presence (severity ≥1 on Neuropsychiatric Inventory Questionnaire) of either depressive symptoms alone, apathy symptoms alone, both symptoms, or neither. A neuropsychological battery evaluated executive function, verbal fluency, verbal learning, working memory, and visuospatial reasoning. Analysis of covariance was used to assess group differences with age, sex, and education as covariates. Groups included depressive symptoms only (n=70), apathy symptoms only (n=52), both (n=68), or neither (n=262). The apathy and depression+apathy groups performed worse than the neither group on tests of working memory (t(312) =-2.4, p=0.02 and t(328) =-3.8, p=0.001, respectively) and visuospatial reasoning (t(301) =-2.3, p=0.02 and t(321) =-2.6, p=0.01, respectively). The depression, apathy, and depression+apathy groups demonstrated a similar degree of impairment on tests of executive function, processing speed, verbal fluency, and verbal learning when compared to participants without apathy or depression. Sex-stratified analyses revealed that compared to the male neither group, the male apathy and depression+apathy groups were impaired broadly across all cognitive domains except for working memory. Females with depression alone showed deficits on tests of executive function (t(166) =2.4, p=0.01) and verbal learning (t(167) =-4.3, p=0.001) compared to the female neither group. This study demonstrated that in neurodegenerative diseases, apathy with or without depression in males was associated with broad cognitive impairments. In females, depression was associated with deficits in executive function and verbal learning. These findings highlight the importance of effectively treating apathy and depression across the spectrum of neurodegenerative disorders with the goal of optimizing neuropsychological outcomes.

The Complexity of Reading Revealed by a Study with Healthy Older Adults.

Aging, even when healthy, involves changes in cognitive functioning that can gradually affect the everyday activities and well-being of older people. Reading, which requires the integrity of several functions and their integration, is important to maintaining high cognitive and emotional stimulation over time. Our study aimed to investigate whether reading ability declines with aging. To explore also why reading would decline, we explored the changes in the performance of visual and attention tasks. A group of 58 neurologically healthy older people aged from 65 to 75 underwent neuropsychological assessment to investigate their global cognitive functioning, reading skills, crowding, and attention components. We found a decline in reading abilities as a function of aging (β = 0.34, p < 0.05). We did not find an increase in crowding or difficulties in visual acuity. Furthermore, we found no decline with age in tasks of simple reaction times, visuospatial attention, and other single components of attention. Interestingly, we instead found a worsening with age in the Symbol Digit Modalities Test (β = -0.26, p < 0.05), involving attention, working memory, and processing speed, which explains part of the reading decline. Our results suggest that task complexity is a fundamental aspect to account for aging changes.

Do-It-Yourself Ultrasonography Phantom Model for Improving Hand Skills: A Technical Report.

This technical report focuses on developing a do-it-yourself (DIY) model of a water phantom for training in ultrasound-guided needle insertion techniques. Ultrasound technology is becoming more widely used in perioperative and intensive care settings. However, accurate needle placement using ultrasound guidance necessitates strong spatial reasoning and hand-eye coordination. To address this, the authors experimented with a water phantom model that is cost-effective, easily accessible, and efficient for training. The DIY water phantom was made using materials such as an examination glove, a used vial rubber cap, water, adhesive tape, sealing glue, and a target object. This technical report discusses the process of assembling the water phantom and the potential benefits it offers for ultrasound training.

Educating on spatial skills using a paper-folding-and-punched-hole videogame: gameplay data analysis

IntroductionPaper folding and punched hole tests are used to measure spatial abilities in humans. These abilities are relevant since they are associated with success in STEM (Science, Technology, Engineering, and Mathematics). This study addresses the challenge of teaching spatial reasoning skills using an educational videogame, the Paper Folding Reasoning Game.MethodsThe Paper Folding Reasoning Game is an interactive game which presents activities intended to help users train and understand how to fold a paper to get a specific shape (Part I) and the consequence of punching a hole on a folded paper (Part II). This educational videogame can automatically generate paper-folding-and-punched-hole questions with varying degrees of difficulty depending on the number of folds and holes made, thus producing additional levels for training due to its embedded reasoning mechanisms (Part III).ResultsThis manuscript presents the results of analyzing the gameplay data gathered by the Paper Folding Reasoning Game in its three parts. For Parts I and II, the data provided by 225 anonymous unique players are analyzed. For Part III (Mastermode), the data obtained from 894 gameplays by 311 anonymous unique players are analyzed. In our analysis, we found out a significant difference in performance regarding the players who trained (i.e., played Parts I and II) before playing the Mastermode (Part III) vs. the group of players who did not train. We also found a significant difference in players' performance who used the visual help (i.e., re-watch the animated sequence of paper folds) vs. the group of players who did not use it, confirming the effectiveness of the Paper Folding Reasoning Game to train paper-folding-and-punched-hole reasoning skills. Statistically significant gender differences in performance were also found.

Enhancing mathematical education with spatial visualization tools

IntroductionThis article highlights the innovative approach of incorporating spatial visualization tools in mathematics education, specifically focusing on the exploration of virtual environments and 3D printing. By integrating these advanced technologies, educators can enhance the learning experience and facilitate a deeper understanding of mathematical concepts. This research delves into the potential benefits and practical applications of these tools, shedding light on their effectiveness in fostering spatial reasoning skills and promoting active student engagement. The study aims to uncover new avenues for leveraging technology in the mathematics classroom and re-imagining the teaching and learning process.MethodsTo assess the progress in spatial visualization ability, the Revised Purdue Spatial Visualization Test (Revised PSVT: R) was administered as both a pre-test and a post-test to a sample of 255 students divided into control and experimental groups.ResultsThe findings indicate that the control group exhibited no significant development (p = 0.163) and had a minimal effect size of change (g = 0.035), whereas the experimental group demonstrated a substantial improvement in spatial visualization ability (p &amp;lt; 0.05) with a noteworthy medium effect size of change (g = 0.325). Furthermore, an analysis of variance was conducted to compare the mean normalized change in spatial visualization ability between the control and experimental groups. The results, with 95% confidence, establish that the normalized spatial visualization change is significantly higher in the experimental group. Specifically, the experimental group witnessed a 25% increase in spatial visualization skills, while the control group only experienced a 5% increase.DiscussionOur study demonstrates the significance and effectiveness of integrating spatial visualization tools in mathematics education to enhance students’ spatial reasoning skills and mathematics understanding.

Classification algorithm for land use in the giant panda habitat of Jiajinshan based on spatial case-based reasoning

Jiajin Mountain, where the giant pandas reside, is an essential nature reserve in China. To comprehend the land use classification of the habitat, this article proposes a remote sensing interpretation algorithm based on spatial case reasoning, known as spatial case-based reasoning (SCBR). The algorithm incorporates specific spatial factors into its framework and does not require an extensive amount of domain knowledge and eliminates the need for a complex model training process, making it capable of completing land use classification in the study area. SCBR comprises a spatial case expression model and a spatial case similarity reasoning model. The paper conducted comparative experiments between the proposed algorithm and support vector machine (SVM), U-Net, vision transformer (ViT), and Trans-Unet, and the results demonstrate that spatial case-based reasoning produces superior classification outcomes. The land use classification experiment based on spatial case-based reasoning at the Jiajinshan giant panda habitat produced satisfactory experimental results. In the comparative experiments, the overall accuracy of SCBR classification reached 95%, and the Kappa coefficient reached 90%. The paper further analyzed the changes in land use classification from 2018 to 2022, and the average accuracy consistently exceeds 80%. We discovered that the ecological environment in the region where the giant pandas reside has experienced significant improvement, particularly in forest protection and restoration. This study provides a theoretical basis for the ecological environment protection of the area.

A transversal and practical education as a business success factor: literature review of learning process of basic design through ICT tools

The ongoing societal transformation propelled by innovation and digitalization is amplifying the demand for technological adeptness among current and upcoming professionals. Consequently, there's a pressing need to revamp the educational system and its methodologies to elevate the requisite skill set. Spatial reasoning, innovative thinking, and creativity stand as pivotal proficiencies essential for enabling future professionals to swiftly craft adaptable prototypes catering to client needs. However, conventional teaching approaches exhibit shortcomings in nurturing spatial reasoning, necessitating deeper exploration within the realm of education. This study delves into novel 3D design tools aimed at facilitating three-dimensional modelling within educational settings. The findings underscore the imperative use of digital tools in crafting 3D objects, fostering the development of spatial visualization skills. Moreover, they emphasize the significance of intertwining ICT knowledge, programming skills, and pertinent soft skills. Such an amalgamation equips future professionals with enhanced work capabilities, a comprehensive grasp of market needs, and refined product commercialization strategies.

Is recalibration more important than realignment in prism adaptation training for visuospatial neglect? A randomized controlled trial

ABSTRACT Prism adaptation training (PAT) as a treatment for visuospatial neglect (VSN) involves two components: recalibration and realignment. We conducted a randomized controlled trial with PAT protocols requiring different degrees of recalibration and realignment, by using a single or multi-step protocol and varying visibility of the pointing movement. Twenty-five VSN patients received an alertness treatment without prisms, followed by four PAT protocols, encompassing a multi- or single-step procedure with terminal exposure, a single-step procedure with concurrent exposure, and sham PAT, presented in random order. The primary outcome parameter was the mean response time (RT) to left-sided targets in an endogenous variant of the Posner task, and we also measured the sensorimotor aftereffect. The two protocols without visibility of most of the movement trajectory produced significant aftereffects. The single-step protocol without movement visibility resulted in shorter RTs to left-sided targets. Hence, aftereffects depended on the partial invisibility of the movement. Moreover, only allowing VSN patients to recalibrate several times and direct feedback from the pointing errors had a beneficial effect on non-motor leftward visuospatial attention. We provide preliminary evidence that maximizing the conscious experience of movement errors may be an important component for remediating VSN. Trial registration: German Clinical Trials Register identifier: DRKS00025938.

Cognitive Effort during Visuospatial Problem Solving in Physical Real World, on Computer Screen, and in Virtual Reality.

Spatial cognition plays a crucial role in academic achievement, particularly in science, technology, engineering, and mathematics (STEM) domains. Immersive virtual environments (VRs) have the growing potential to reduce cognitive load and improve spatial reasoning. However, traditional methods struggle to assess the mental effort required for visuospatial processes due to the difficulty in verbalizing actions and other limitations in self-reported evaluations. In this neuroergonomics study, we aimed to capture the neural activity associated with cognitive workload during visuospatial tasks and evaluate the impact of the visualization medium on visuospatial task performance. We utilized functional near-infrared spectroscopy (fNIRS) wearable neuroimaging to assess cognitive effort during spatial-reasoning-based problem-solving and compared a VR, a computer screen, and a physical real-world task presentation. Our results reveal a higher neural efficiency in the prefrontal cortex (PFC) during 3D geometry puzzles in VR settings compared to the settings in the physical world and on the computer screen. VR appears to reduce the visuospatial task load by facilitating spatial visualization and providing visual cues. This makes it a valuable tool for spatial cognition training, especially for beginners. Additionally, our multimodal approach allows for progressively increasing task complexity, maintaining a challenge throughout training. This study underscores the potential of VR in developing spatial skills and highlights the value of comparing brain data and human interaction across different training settings.