Spatial Bias Research Articles

Spatial Bias for attention-free non-local neural networks

In this paper, we introduce the Spatial Bias to learn global knowledge without self-attention in convolutional neural networks. Owing to the limited receptive field, conventional convolutional neural networks suffer from learning long-range dependencies. Non-local neural networks have struggled to learn global knowledge, but unavoidably have too heavy a network design due to the self-attention operation. Therefore, we propose a fast and lightweight Spatial Bias that efficiently encodes global knowledge without self-attention on convolutional neural networks. Spatial Bias is stacked on the feature map and convolved together to adjust the spatial structure of the convolutional features. Because we only use the convolution operation in this process, ours is lighter and faster than traditional methods based on the heavy self-attention operation. Therefore, we learn the global knowledge on the convolution layer directly with very few additional resources. Our method is very fast and lightweight due to the attention-free non-local method while improving the performance of neural networks considerably. Compared to non-local neural networks, the Spatial Bias use about ×10 times fewer parameters while achieving comparable performance with 1.6∼3.3 times more throughput on a very little budget. Furthermore, the Spatial Bias can be used with conventional non-local neural networks to further improve the performance of the backbone model. We show that the Spatial Bias achieves competitive performance that improves the classification accuracy by +0.79% and +1.5% on ImageNet-1K and CIFAR-100 datasets. Additionally, we validate our method on the MS-COCO and ADE20K datasets for downstream tasks involving object detection and semantic segmentation.

When intentions do not matter: Climate change mitigation and adaptation innovations in the Forest-based sector

Climate change mitigation and adaptation innovations in the forest-based sector can potentially have important environmental benefits but it is not clear what drives such innovations. In this paper we present the results of a 2021 survey on a sample of 293 Austrian and Slovenian enterprises in the sector using the theory of planned behaviour. Consistent with previous studies that applied this theory to forest management and other environmental topics, we have confirmed that attitudes, norms and perceived behavioural control significantly affect intentions to act. However, it was not possible to confirm an association between intentions and actual behaviour. To explain the intention-behaviour gap, we explored two factors: First, the respondent's perception of climate change as an opportunity for the forest-based sector to mitigate climate change or as a threat to operations and a need to do adaptation. Second, the belief that environmental problems are a bigger threat in more geographically distant locations (spatial bias). Whereas data on mitigation innovations were insufficient, for adaptation innovations our data indicate that the more strongly climate change is perceived as a threat or an opportunity, the more likely a company is to adapt. Results have implications both for innovation policies addressing the forest-based sector in Europe and for future research on climate change adaptation and mitigation behaviour.

Eye movements reveal that young school children shift attention when solving additions and subtractions.

Adults shift their attention to the right or to the left along a spatial continuum when solving additions and subtractions, respectively. Studies suggest that these shifts not only support the exact computation of the results but also anticipatively narrow down the range of plausible answers when processing the operands. However, little is known on when and how these attentional shifts arise in childhood during the acquisition of arithmetic. Here, an eye-tracker with high spatio-temporal resolution was used to measure spontaneous eye movements, used as a proxy for attentional shifts, while children of 2nd (8 y-o; N=50) and 4th (10 y-o; N=48) Grade solved simple additions (e.g., 4+3) and subtractions (e.g., 3-2). Gaze patterns revealed horizontal and vertical attentional shifts in both groups. Critically, horizontal eye movements were observed in 4th Graders as soon as the first operand and the operator were presented and thus before the beginning of the exact computation. In 2nd Graders, attentional shifts were only observed after the presentation of the second operand just before the response was made. This demonstrates that spatial attention is recruited when children solve arithmetic problems, even in the early stages of learning mathematics. The time course of these attentional shifts suggests that with practice in arithmetic children start to use spatial attention to anticipatively guide the search for the answer and facilitate the implementation of solving procedures. RESEARCH HIGHLIGHTS: Additions and subtractions are associated to right and left attentional shifts in adults, but it is unknown when these mechanisms arise in childhood. Children of 8-10 years old solved single-digit additions and subtractions while looking at a blank screen. Eye movements showed that children of 8 years old already show spatial biases possibly to represent the response when knowing both operands. Children of 10 years old shift attention before knowing the second operand to anticipatively guide the search for plausible answers.

Modification of Turing patterns through the use of time-varying anisotropic diffusion

The Turing mechanism underpinning pattern formation in reaction–diffusion relies on the interplay between diffusion parameters and reaction kinetics. Diffusion is typically assumed isotropic, however anisotropic diffusion is known to arise in both nature and laboratory conditions. We study how the Turing instability and resulting Turing patterns are modified when the underlying diffusion tensor is both anisotropic and time-dependent, modelling, for instance, chemical species reacting and diffusing through time-varying anisotropic media. We show that the set of unstable wavenumber vectors corresponding to Turing modes evolve in a spatially biased manner under this anisotropy, thereby modifying the spatial scale and structure of any resulting Turing patterns differently along each spatial coordinate. We employ this spatial bias to develop control strategies to modify the shape or even structure of Turing patterns over time. We are able to make minor changes to the aspect ratio of Turing patterns, such as morphing circular Schnakenberg spots into elliptical spots, as well as more major changes to the structure of patterns, for instance converting Gierer–Meinhardt spots into stripes or FitzHugh–Nagumo labyrinthine patterns into target patterns. Our results suggest that time-varying anisotropic media may be used as a tool by which to modify and even control Turing patterns.

Self-Supervised Interactive Embedding for One-shot Organ Segmentation.

One-shot organ segmentation (OS2) aims at segmenting the desired organ regions from the input medi-cal imaging data with only one pre-annotated example as the reference. By using the minimal annotation data to facilitate organ segmentation, OS2 receives great attention in the medical image analysis community due to its weak re-quirement on human annotation. In OS2, one core issue is to explore the mutual information between the support (ref-erence slice) and the query (test slice). Existing methods rely heavily on the similarity between slices, and additional slice allocation mechanisms need to be designed to reduce the impact of the similarity between slices on the segmen-tation performance. To address this issue, we build a novel support-query interactive embedding (SQIE) module, which is equipped with the channel-wise co-attention, spatial-wise co-attention, and spatial bias transformation blocks to identify "what to look", "where to look", and "how to look" in the input test slice. By combining the three mechanisms, we can mine the interactive information of the intersection area and the disputed area between slices, and establish the feature connection between the target in slices with low similarity. We also propose a self-supervised contrastive learning framework, which transforms knowledge from the physical position to the embedding space to facilitate the self-supervised interactive embedding of the query and support slices. Comprehensive experiments on two large benchmarks demonstrate the superior capacity of the pro-posed approach when compared with the current alterna-tives and baseline models.

Evaluation of Experimental High-Resolution Model Forecasts of Tropical Cyclone Precipitation Using Object-Based Metrics

Abstract Tropical cyclone (TC) precipitation poses serious hazards including freshwater flooding. High-resolution hurricane models predict the location and intensity of TC rainfall, which can influence local evacuation and preparedness policies. This study evaluates 0–72-h precipitation forecasts from two experimental models, the Hurricane Analysis and Forecast System (HAFS) model and the basin-scale Hurricane Weather Research and Forecasting (HWRF-B) Model, for 2020 North Atlantic landfalling TCs. We use an object-based method that quantifies the shape and size of the forecast and observed precipitation. Precipitation objects are then compared for light, moderate, and heavy precipitation using spatial metrics (e.g., area, perimeter, elongation). Results show that both models forecast precipitation that is too connected, too close to the TC center, and too enclosed around the TC center. Collectively, these spatial biases suggest that the model forecasts are too intense even though there is a negative intensity bias for both models, indicating there may be an inconsistency between the precipitation configuration and the maximum sustained winds in the model forecasts. The HAFS model struggles with forecasting stratiform versus convective precipitation and with the representation of lighter (stratiform) precipitation during the first 6 h after initialization. No such spinup issues are seen in the HWRF-B forecasts, which instead exhibit systematic biases at all lead times and systematic issues across all rain-rate thresholds. Future work will investigate spinup issues in the HAFS model forecast and how the microphysics parameterization affects the representation of precipitation in both models.

Crowdsourcing social values data: Flickr and public participation GIS provide different perspectives of ecosystem services in a remote coastal region

Spatial planning and environmental management are expected to adopt participatory processes. However, the needed spatial data on social values of ecosystem services are limited, especially for marine spatial planning for large, remote coastal areas, and the collection of such information can be time and resource intensive. Crowdsourcing techniques are cost efficient sources of social values data, but must meet the information needs of planners and managers with sufficient confidence. We evaluated the reliability of crowdsourced social values data by assessing the agreement between geotagged photos posted to the social media platform Flickr and data from an online public participation GIS (PPGIS) survey conducted to support marine spatial planning in the Kimberley region of Western Australia.Flickr could not represent all of the social values mapped in the PPGIS dataset, was notably dominated by scenic values, and under-represented the importance of recreational fishing and Aboriginal cultural values. The Flickr dataset was more affected by accessibility from settlements and transportation infrastructure (roads, boat ramps) and more restricted spatially. Maximum entropy (maxent) distribution models developed to explore the socioecological drivers determining where social values occur also highlighted that Flickr and PPGIS samples provide different perspectives on social values, even after accounting for differences in sampling intensity. The Flickr dataset suggested that the social values occurred under narrower ranges of socioecological conditions, and the PPGIS dataset was more likely to find protected areas to be valued, likely because participants believe protected areas to support those social values, but may not themselves pursue them there.Flickr and PPGIS crowdsourced datasets provide different perspectives on the spatial distribution of social values for several reasons: They differ in their spatial and demographic biases, as well as their measurement of revealed vs. stated preferences. Flickr is most effective at representing social values that are easily photographed and interpretable in photos, and may not capture some information needs of managers. PPGIS is under greater investigator control but, because stated preferences can be disconnected from actual use of the sites mapped, may be best suited for assessing management preferences and social acceptability. With understanding of the nuances of these datasets, crowdsourced social values data can be applied most appropriately to support successful planning outcomes.

Unsupervised Domain Adaptive Dose Prediction Via Cross-Attention Transformer and Target-Specific Knowledge Preservation.

Radiotherapy is one of the leading treatments for cancer. To accelerate the implementation of radiotherapy in clinic, various deep learning-based methods have been developed for automatic dose prediction. However, the effectiveness of these methods heavily relies on the availability of a substantial amount of data with labels, i.e. the dose distribution maps, which cost dosimetrists considerable time and effort to acquire. For cancers of low-incidence, such as cervical cancer, it is often a luxury to collect an adequate amount of labeled data to train a well-performing deep learning (DL) model. To mitigate this problem, in this paper, we resort to the unsupervised domain adaptation (UDA) strategy to achieve accurate dose prediction for cervical cancer (target domain) by leveraging the well-labeled high-incidence rectal cancer (source domain). Specifically, we introduce the cross-attention mechanism to learn the domain-invariant features and develop a cross-attention transformer-based encoder to align the two different cancer domains. Meanwhile, to preserve the target-specific knowledge, we employ multiple domain classifiers to enforce the network to extract more discriminative target features. In addition, we employ two independent convolutional neural network (CNN) decoders to compensate for the lack of spatial inductive bias in the pure transformer and generate accurate dose maps for both domains. Furthermore, to enhance the performance, two additional losses, i.e. a knowledge distillation loss (KDL) and a domain classification loss (DCL), are incorporated to transfer the domain-invariant features while preserving domain-specific information. Experimental results on a rectal cancer dataset and a cervical cancer dataset have demonstrated that our method achieves the best quantitative results with [Formula: see text], [Formula: see text], and HI of 1.446, 1.231, and 0.082, respectively, and outperforms other methods in terms of qualitative assessment.

Visual images of disgusting creatures facilitated attentional orienting and delayed attentional disengagement.

Numerous studies have suggested that threatening stimuli induce a spatial attention bias; however, only a few studies have investigated spatial attention biases for disgusting stimuli. Moreover, past studies generally reported that the spatial attention bias to disgusting images is not robustly in normal individuals. We hypothesized that this was due to the unfamiliar of the images, so we prepared the creature's images that were clearly categorized as disgusting and examined the effects of disgusting images on spatial attention bias. A disgusting or an emotionally neutral image was paired and presented with an (emotionally neutral) filler image. After a temporal interval, a target appeared at either the position where a disgusting or a neutral image was presented (valid condition) or where a filler image was presented (invalid condition). Participants pressed a key corresponding to the target's position as quickly and accurately as possible. We varied the position-response correspondence among three experiments. The results showed that the RTs in the invalid condition was longer for the disgusting images than for the neutral images when the position of a disgusting image was not naturally associated with the left-right hand position. We interpreted the results in that that disgusting images generally slowed down attentional disengagement process but the manual responses were inhibited for the position where a disgusting image appeared when the locations of keys and targets were congruent. The present results suggest that disgusting images affect not only attentional processes but also manual responses related to the selection and initiation of responses.

Paying attention in metaverse: an experiment on spatial attention allocation in extended reality shopping

PurposeMetaverse, that is extended reality (XR)-based technologies such as augmented reality (AR) and virtual reality (VR), are increasingly believed to facilitate fundamental human practice in the future. One of the vanguards of this development has been the consumption domain, where the multi-modal and multi-sensory technology-mediated immersion is expected to enrich consumers' experience. However, it remains unclear whether these expectations have been warranted in reality and whether, rather than enhancing the experience, metaverse technologies inhibit the functioning and experience, such as cognitive functioning and experience.Design/methodology/approachThis study utilizes a 2 (VR: yes vs no) × 2 (AR: yes vs no) between-subjects laboratory experiment. A total of 159 student participants are randomly assigned to one condition — a brick-and-mortar store, a VR store, an AR store and an augmented virtuality (AV) store — to complete a typical shopping task. Four spatial attention indicators — visit shift, duration shift, visit variation and duration variation — are compared based on attention allocation data converted from head movements extracted from recorded videos during the experiments.FindingsThis study identifies three essential effects of XR technologies on consumers' spatial attention allocation: the inattention effect, acceleration effect and imbalance effect. Specifically, the inattention effect (the attentional visit shift from showcased products to the environmental periphery) appears when VR or AR technology is applied to virtualize the store and disappears when AR and VR are used together. The acceleration effect (the attentional duration shift from showcased products to the environmental periphery) exists in the VR store. Additionally, AR causes an imbalance effect (the attentional duration variation increases horizontally among the showcased products).Originality/valueThis study provides valuable empirical evidence of how VR and AR influence consumers' spatial bias in attention allocation, filling the research gap on cognitive function in the metaverse. This study also provides practical guidelines for retailers and XR designers and developers.

DepthFormer: Exploiting Long-range Correlation and Local Information for Accurate Monocular Depth Estimation

This paper aims to address the problem of supervised monocular depth estimation. We start with a meticulous pilot study to demonstrate that the long-range correlation is essential for accurate depth estimation. Moreover, the Transformer and convolution are good at long-range and close-range depth estimation, respectively. Therefore, we propose to adopt a parallel encoder architecture consisting of a Transformer branch and a convolution branch. The former can model global context with the effective attention mechanism and the latter aims to preserve the local information as the Transformer lacks the spatial inductive bias in modeling such contents. However, independent branches lead to a shortage of connections between features. To bridge this gap, we design a hierarchical aggregation and heterogeneous interaction module to enhance the Transformer features and model the affinity between the heterogeneous features in a set-to-set translation manner. Due to the unbearable memory cost introduced by the global attention on high-resolution feature maps, we adopt the deformable scheme to reduce the complexity. Extensive experiments on the KITTI, NYU, and SUN RGB-D datasets demonstrate that our proposed model, termed DepthFormer, surpasses state-of-the-art monocular depth estimation methods with prominent margins. The effectiveness of each proposed module is elaborately evaluated through meticulous and intensive ablation studies.

A common format for representing spatial location in visual and motor working memory.

Does the mind rely on similar systems of spatial representation for both perception and action? Here, we assessed the format of location representations in two simple spatial localization tasks. In one task, participants simply remembered the location of an item based solely on visual input. In another, participants remembered the location of a point in space based solely on kinesthetic input. Participants' recall errors were more consistent with the use of polar coordinates than Cartesian coordinates in both tasks. Moreover, measures of spatial bias and performance were correlated across modalities. In a subsequent study, we tested the flexibility with which people use polar coordinates to represent space; we show that the format in which the information is presented to participants influences how that information is encoded and the errors that are made as a result. We suggest that polar coordinates may be a common means of representing location information across visual and motor modalities, but that these representations are also flexible in form.

A Generalized Area‐Based Framework to Quantify River Mobility From Remotely Sensed Imagery

AbstractRivers are the primary conduits of water and sediment across Earth's surface. In recent decades, rivers have been increasingly impacted by climate change and human activities. The availability of global‐coverage satellite imagery provides a powerful avenue to study river mobility and quantify the impacts of these perturbations on global river behavior. However, we lack remote sensing methods for quantifying river mobility that can be generally applied across the diversity of river planforms (e.g., meandering, braided) and fluvial processes (e.g., channel migration, avulsion). Here, we upscale area‐based methods from laboratory flume experiments to build a generalized remote sensing framework for quantifying river mobility. The framework utilizes binary channel‐mask time series to determine time‐ and area‐integrated rates and scales of river floodplain reworking and channel‐thread reorganization. We apply the framework to numerical models to demonstrate that these rates and scales are sensitive to specific river processes (channel migration, channel‐bend cut‐off, and avulsion). We then apply the framework to natural migrating and avulsing rivers with meandering and braided planforms. Results show that our area‐based framework provides an objective and accurate means to quantify river mobility at reach‐ to floodplain‐scales, which is largely insensitive to spatial and temporal biases that can arise in traditional mobility metrics. Our work provides a framework for investigating global controls on river mobility, testing hypotheses about river response to environmental gradients, and quantifying the timescales of terrestrial organic carbon cycling.

Opportunities and challenges for monitoring a recolonizing large herbivore using citizen science.

Monitoring is a prerequisite for evidence-based wildlife management and conservation planning, yet conventional monitoring approaches are often ineffective for species occurring at low densities. However, some species such as large mammals are often observed by lay people and this information can be leveraged through citizen science monitoring schemes. To ensure that such wildlife monitoring efforts provide robust inferences, assessing the quantity, quality, and potential biases of citizen science data is crucial. For Eurasian moose (Alces alces), a species currently recolonizing north-eastern Germany and occurring in very low numbers, we applied three citizen science tools: a mail/email report system, a smartphone application, and a webpage. Among these monitoring tools, the mail/email report system yielded the greatest number of moose reports in absolute and in standardized (corrected for time effort) terms. The reported moose were predominantly identified as single, adult, male individuals, and reports occurred mostly during late summer. Overlaying citizen science data with independently generated habitat suitability and connectivity maps showed that members of the public detected moose in suitable habitats but not necessarily in movement corridors. Also, moose detections were often recorded near roads, suggestive of spatial bias in the sampling effort. Our results suggest that citizen science-based data collection can be facilitated by brief, intuitive digital reporting systems. However, inference from the resulting data can be limited due to unquantified and possibly biased sampling effort. To overcome these challenges, we offer specific recommendations such as more structured monitoring efforts involving the public in areas likely to be roamed by moose for improving quantity, quality, and analysis of citizen science-based data for making robust inferences.

Altered Allocation of Vertical Attention in Individuals With Autism Spectrum Disorder.

Typical adults most frequently orient their attention to other people's eyes, whereas individuals with autism spectrum disorder (ASD) orient their attention to other people's mouths. Typical adults also reveal visuospatial biases on tasks such as vertical and horizontal line bisections. Therefore, the difference in face viewing might be related to a more general group difference in the allocation of vertical attention. To use vertical line bisection and quadrisection tasks to evaluate whether individuals with ASD have a more downward-oriented vertical attentional bias than do typical individuals. We recruited 20 individuals with ASD and 20 control participants matched for age (6-23 years), IQ, and sex. We asked the individuals to bisect and quadrisect lines on the top and bottom when the vertical lines were placed at the intersection of their right, left, and center egocentric sagittal planes and their coronal plane. The distances from the true midpoint and quadripoint were measured, and between-group performances were compared. No significant difference was found between the ASD and control groups for vertical line bisections or lower line quadrisections. However, when the ASD group was compared with the control group for higher line quadrisections, the ASD group exhibited a greater upward deviation. There is no downward vertical attentional spatial bias associated with ASD that could help to explain these individuals' attentional bias toward the mouth. However, additional studies are required to learn if this atypical upward vertical attentional bias might account for some of the symptoms and signs associated with ASD.

Cicada-MET: an efficient ecological monitoring protocol of cicada populations

Cicadas are a fascinating group of insects that play an essential role in terrestrial ecosystems. Their long-lasting association with plant roots encourages their use as indicators of vegetation and soil integrity. Cicada-MET is a novel, standardized method for monitoring cicada populations by counting cicada exuviae (i.e., the skin of emerged nymphs), providing an effective and efficient means to study their distribution, abundance, and ecology. The method involves annual exuviae counts along transects and fixed plots sampled throughout the emergence season. We validated Cicada-MET using a database of 466 counts from 64 transects over 10 years and the sampling of 60 plots for one season. Methodological aspects tested included sampling speed, exuviae detectability in successive counts, exuviae loss due to weather, and cicada species detection performance using exuviae counts compared to auditory methods. Transects captured approximately 10% of the total number of emerged nymphs across one season, demonstrating the protocol’s reliability in estimating emerging cicada population numbers. However, caution is needed when inferring densities for larger areas, separated from paths where transects are located. The standardized nature of Cicada-MET reduces spatial and temporal biases, allowing for interspecific comparisons and monitoring interannual variations in abundances and emergence timing. This method is well-suited for studying the impact of natural and anthropogenic disturbances. The high-resolution data obtained can be easily combined with environmental variables, enhancing the value of cicada data as bioindicators. In summary, Cicada-MET offers a versatile and efficient tool for monitoring cicada populations, with applications in ecological indication, conservation, and management. The adaptability of Cicada-MET to various research questions, spatial scales, and long-term approaches, along with its quantitative accuracy and ease of use, make it a valuable resource for researchers and practitioners working with cicadas and their associated ecosystems.

Marine inundation history during the last 3000 years at Lake Kogare-ike, a coastal lake on the Pacific coast of central Japan

Sediment cores collected at Lake Kogare-ike, a coastal lake on the Pacific coast of central Japan, record the marine inundation history during the last 3000 years. The sediments consist mainly of organic mud, sand, gravel, inorganic mud, and volcanic ash, and inundation events were recognized as 19 event deposits (E1–E19, from top to bottom) interbedded with the organic mud. Visual observation by naked eyes and X-ray computed tomography (CT) images identified 16 event deposits based on quantitative and qualitative changes in sand contents and changes in the textures and colors of the sediment samples (E1–E3, E5, E6, E8, E9, and E11–E19). The other three event deposits (E4, E7, and E10) were identified only on the CT images as layers with higher radiodensity than the underlying and overlying organic mud layers. The sedimentary features, the spatial bias of the event deposits toward seaward areas, the diatom assemblages, and the frequency of inundation events suggest that 13 (E1–E10 and E12–E14) of the 19 event deposits were formed by tsunamis or extraordinary storms. To constrain the depositional ages of the event deposits, Bayesian age–depth models were constructed based on radiocarbon dating of plant macrofossils and concentrated fossil pollen and the 137Cs profile. The depositional ages of the event deposits indicate that five or possibly six event deposits can be correlated with historical tsunamis along the Nankai Trough: E2, either of E3 or E4, E5, E7, and E9 correspond to the 1707 CE Hoei, the 1605 CE Keicho, the 1498 CE Meio, the 1096 CE Eicho, and the 684 CE Hakuho tsunamis, respectively. E1 was possibly formed by the 1944 CE Showa-Tonankai tsunami, the 1854 Ansei–Tokai tsunami, the 1959 Isewan typhoon, or a combination of two or all three events.

Relating social, ecological, and technological vulnerability to future flood exposure at two spatial scales in four U.S. cities

Flooding occurs at different scales and unevenly affects urban populations based on the broader social, ecological, and technological system (SETS) characteristics particular to cities. As hydrological models improve in spatial scale and account for more mechanisms of flooding, there is a continuous need to examine the relationships between flood exposure and SETS drivers of flood vulnerability. In this study, we related fine-scale measures of future flood exposure—the First Street Foundation's Flood Factor and estimated change in chance of extreme flood exposure—to SETS indicators like building age, poverty, and historical redlining, at the parcel and census block group (CBG) scales in Portland, OR, Phoenix, AZ, Baltimore, MD, and Atlanta, GA. We used standard regression models and accounted for spatial bias in relationships. The results show that flood exposure was more often correlated with SETS variables at the parcel scale than at the CBG scale, indicating scale dependence. However, these relationships were often inconsistent among cities, indicating place-dependence. We found that marginalized populations were significantly more exposed to future flooding at the CBG scale. Combining newly-available, high-resolution future flood risk estimates with SETS data available at multiple scales offers cities a new set of tools to assess the exposure and multi-dimensional vulnerability of populations. These tools will better equip city managers to proactively plan and implement equitable interventions to meet evolving hazard exposure.

Related Variety and Regional Development: A Critique

Evolutionary approaches in economic geography have contributed substantially to the growing body of knowledge of regional development processes and their underlying mechanisms. One key concept in the literature on evolutionary economic geography is that of related variety. Herein, regional industry structure is represented through the level of related variety of technologies, skills, or outputs. The related variety concept proposes that regional economic development is favored when an economy diversifies into products or technologies that are closely related to the stock of existing activities. In this article, we raise substantive questions regarding the internal logic of the concept of related variety, its spatial expressions, measurement specifics, empirical regularities and biases, and its possible short- and long-term effects on regional development. Based on this investigation, we make suggestions for improvements to future research.

Cognitive Path Planning with Spatial Memory Distortion.

Human path-planning operates differently from deterministic AI-based path-planning algorithms due to the decay and distortion in a human's spatial memory and the lack of complete scene knowledge. Here, we present a cognitive model of path-planning that simulates human-like learning of unfamiliar environments, supports systematic degradation in spatial memory, and distorts spatial recall during path-planning. We propose a Dynamic Hierarchical Cognitive Graph (DHCG) representation to encode the environment structure by incorporating two critical spatial memory biases during exploration: categorical adjustment and \sequence order effect. We then extend the '`Fine-To-Coarse'' (FTC), the most prevalent path-planning heuristic, to incorporate spatial uncertainty during recall through the DHCG. We conducted a lab-based Virtual Reality (VR) experiment to validate the proposed cognitive path-planning model and made three observations: (1) a statistically significant impact of sequence order effect on participants' route-choices, (2) approximately three hierarchical levels in the DHCG according to participants' recall data, and (3) similar trajectories and significantly similar wayfinding performances between participants and simulated cognitive agents on identical path-planning tasks. Furthermore, we performed two detailed simulation experiments with different FTC variants on a Manhattan-style grid. Experimental results demonstrate that the proposed cognitive path-planning model successfully produces human-like paths and can capture human wayfinding's complex and dynamic nature, which traditional AI-based path-planning algorithms cannot capture.