Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Slide 13Slide 14Slide 15Slide 16Slide 17Slide 18Slide 19Slide 20Slide 21Slide 22Slide 23Mind and MazeAnn Sloan Devlin, 2001Preetha LakshmiChris Muellerhttp://www.cs.umn.edu/~cmueller/cs8715/CSCI 8715Professor Shashi ShekharOutlineOverview of Mind and MazeCommon themesKey conceptsDevelopment of spatial cognitionNeuropsychologyNatural languageApplications: maps and urban planningCritiqueConclusionWhat is Mind and MazeA summary of research in fields related to spatial cognition:Cognitive developmentGender studiesNeuropsychology Map-makingUrban PlanningA loose framework for classifying research in these disparate fieldsWhat is the problem?There is need for a coherent, comprehensive overview of spatial cognition research.Why is it important?Spatial cognition research guides many industries, including map-making, urban planning, architecture, GIS, etc.Why is it hard?Research in these areas is divided among many journals, books, and disciplines.Related workThe book is a survey of related work across many fields.NoveltyThis is a one-of-a-kind book. It was published in 2003, and presents relatively new research.ValidationThe author claims a “review of the research in detail”. We do not know the author's research methodology, but we trust it was thorough.AssumptionOnly the five fields of spatial development, gender studies, neuropsychology, urban planning, and map-making are relevant to the book.OverviewInput: Five disciplines: cognitive development, gender studies, neuropsychology, map-making, and urban planning. Output: A general-purpose overview of spatial cognition.Objective: Combine the five disciplines separate theories into a substantive whole.Constraints: Research outside of these five disciplines is not considered.Relevance to CS8715: A theoretical/psychological background on human understanding of space.Themes in Spatial Cognitionlandmarks: landmarks and relationships among them form the basis for how we think about spacechunking: people separate spatial information into aggregate chunks to help with processingSpatial understanding is a form of non-Euclidean geometry Gestalt's principles: the whole cannot be understood by just looking at the partsDevelopment of Spatial CognitionSpatial understanding develops over time; children tend to be more egocentricProposed model for spatial understanding:landmarks are learned firsthierarchies indicate importanceroutes are relationships between landmarksroutes are built into networksBAWorkHomeDECMother as LandmarkMotherChild (9 mos. old)Base orientationChildren are cued to look at the left screen. A colored slide appears.Child rotated 180°Mother remains stationaryChild rotated 180°Mother's movement matches infant's rotation80% accuracy in determining which screen should be looked at43% accuracy in determining which screen should be looked atMovement and self-motionMovement through an environment is critical for spatial understanding, especially compared to film or slides of a spaceSelf-motion promotes better spatial understanding than other forms of movementVirtual 3d tours of spaces have proven effective for developing a spatial understandingCognitive MapsCognitive maps are another way to conceptualize human understanding of space. In these decision-based models, we see four components:Recognizing where one is locatedPredicting what will happen nextDetermining the outcome of an action (+/-)Acting out a planAnchor Point TheoryAnchor points are points of “cognitive salience,” similar to landmarks.In a case study involving “choice points” and “plots” within a neighborhood with an 11-yr-old boy, researchers drew three main conclusions:saliency of a plot decreases with distance from major choice pointsthe more alternative actions possible at a choice point, the greater its importancethe choice points along a route (where navigational decisions are made) are similar to the route's storage in long-term memoryNatural Language and Spatial CognitionNatural language plays an important role in communicating spatial informationHemispherical brain division:left brain processes complex spatial information and visual interpretationright brain processes simple spatial tasks e.g. line orientation Language is processed in left brain, where it might be closely related to complex spatial cognitive tasksWhen given a task to describe a place, most people (>80%) provided a high-level description rather than route or way-finding descriptions.Neurophysiology of Spatial Cognition(O'Keefe 1978)(McNaughton, et al 2006)neuron firings indicate grid patternsNeurons support cognition of relative directional motion based on strength of synaptic activity in “head direction cells”Hippocampus neuronal structures also exist that track relative translational distance measurements; stored in repeating orthogonal grids of “place cells” (torus topology) Grids of place cells are uniquely patterned to match unique environments; upon entering a new environment, a new grid is created, but existing patterns are used for previously visited environments.MapsEuclidean information is difficult for many to understandLegibility is crucialSimplification, classification, symbolizationShould not contain too much clutterImportant landmarks should be made clearly visibleUniformity among colors and stylesGestalt's PrinciplesProximityChunkingimages from http://www.atpm.com/9.10/design.shtmlUrban Planning: the GridABABUrban Planning: Vocabularypaths: roads and sidewalksedges: a fence or river bank districts: neighborhoods nodes: major intersections or transit terminals Similar to “landmarks” or “anchor points” in cognitive psychology literaturelandmarks: may be cultural or historicRewriting Mind and MazeStructure content by concept rather than by author/timeline demonstrate similarities among research provide richer chapter summaries and a conclusion to the book as a wholeGive more energy to neuropsychologyPrune outdated and irrelevant researchAddress more topics in Urban Planning, e.g. sustainability and scalabilityDiscuss a wider set of applications: architecture, virtual reality, video games, user interfaces, etc.ConclusionMind and Maze
View Full Document