Mapping the mindNew England Complex Systems InstituteOne-Week Intensive Course: Complex Physical, Biological and Social SystemsMIT, January 6-10, 2003Supervisor: Prof. Yaneer Bar-YamSubmitted by:Fumiaki KatagiriIvelisse LazzariniFlorence SingerShai Shen-OrrCopyright © 2003 New England Complex Systems Institute. All rights reserved.NECSI One Week Intensive CourseJanuary 6-10, 2003Supervisor: Prof. Yaneer Bar-YamMapping the mindFumiaki Katagiri1, Ivelisse Lazzarini2, Florence Singer3, Shai Shen-Orr41 Dept. Plant Health, Torrey Mesa Research Institute, San Diego, CA, 2 Dept. of Pharmacy and Health Professions,Creighton University, Omaha, NE, 3 Graduate School of Education, Harvard University, Cambridge, MA, 4 Dept. ofMolecular and Cellular Biology, Harvard University, Cambridge, MAMainstream social and psychological sciences has traditionally disregardedindividual differences with respect to perception. While studies on perception have beenperformed on individuals, the differences in their perception were discarded and only the“average” perception consensus reported. Only in recent years, have perceptual systemsbeen thought of in dynamical sense (Freeman 1995; Kelso 1999). In this study weattempt to investigate the visual perceptual system of individuals with respect to aspecific object. In particular we are trying to probe what is it that an individual sees whenhe/she is looking at something? For many investigators the quandary of perceptionresides in how properties of the world come to be represented in the mind of theperceiver. In our project, however, we are less concerned with the contents of perceptionand more interested with the dynamics of perceiving.Touching, looking, listening and smelling are obviously dynamical processes. Butwhat is the nature of the dynamics? If we describe our concept spectrum as an everchanging energy landscape were each concept is a basin of attraction, then describing thedynamics of the basin implies mapping its size in concept space and its slopes in everydirection in continuous time. The dimensionality of our concept space is extremely large– the number of different concepts that we know. More concretely, if we examine aspecific concept, i.e. –chair, then how an individual perceives a chair can be described ashaving the consensus chair be the bottom of the basin, and differences in criteria, such ascomfort, utility and aesthetics be dimensions in our infinite concept space, describing theslopes of the basin. Chairs less similar than the ideal consensus chair could be mappedalong the slopes of the chair basin. Obviously, in every individual the basin for eachconcept would be shaped differently.This study is a first attempt to explore individual concept space at a specific pointof time, and contrast it with the concept space of other individuals. To achieve this goal,we constructed a survey which attempts to explore the space for a specific concept.MethodsData Collection: We constructed a questionnaire containing two parts: the first withgeneral questions about respondents’ interests, the second – containing the questionstargeting our analysis, with the task to classify 11 images as how they fit a specificconcept (i.e. chair) according to four criteria (utility/functionality, comfort, aesthetics andprice). Fitness to the classification concepts was done by ranking the different images foreach one of the criteria. For utility/functionality, comfort, aesthetics ranking was done byusing a numbering system between 1 and 11, while for the price criteria, participantswere asked to name a price in US dollars they would be willing to purchase the item inthe image provided. Questionnaires were filled out by all participants at the same timeindependently of one another. The respondents were 43 participants in an intensivecomplex systems course given by the New England Complex Systems Institute. Seeappendix A for questionnaire.Data Analysis: Data from questionnaire was entered into a spreadsheet. Pricing rankingwas converted to sequential discrete number ranking between 1-11. Complete linkageanalysis was used to cluster the participants according to each criteria as well as all thecriteria together. The number of clusters for each criteria was decided upon by manualobservation of branch lengths in the linkage dendogram and the data. Clustering wasperformed by using the public domain Cluster package and visualized using TreeView.Spearman’s rank correlation was used to try and detect if any correlation existed betweeneach of the first three criteria and the price criteria. Any criteria that was missing a datapoint in the form was disregarded and all correlation analysis between that criteria andothers was not done. Participants answers for the general questions section of thequestionnaire was reviewed by the authors. The data in those answers was deemed to betoo diverse to include in the study.ResultsClustering participants by category: Linkage analysis and manual cluster cutoffyielded 4 utility clusters, 3 comfort clusters and four aesthetic clusters. Table 1 displaysthe clusters per criteria category as well as the commonalties in the answers of theparticipants of each clusters.Table 1: Clusters of participants’ rankings by criteria. Bold font in the commonaltiescolumn indicates the top and bottom picks of each clusters.CategoryClusterNumberNumber ofParticipantsCommonalties13Ranked on the highest levels the objects:C, D, E and on the lowest levels theobjects: A, J, H, K.213Ranked on the highest levels the objects:A, C, F, J and on the lowest levels theobjects: G, B, H, I.321Ranked on the highest levels the objects:A, F and on the lowest levels the objects:E, J, H, G.Utility46Ranked on the highest levels the objects:B, K, F and on the lowest levels theobjects: C, H, J.11Ranked on the highest levels the objects:G, F, E and on the lowest levels theobjects: A, J, B.223Ranked on the highest levels the object Dand on the lowest levels the objects: G, J,H.Comfort319ranked on the highest levels the objects: A,F, D and on the lowest levels the objects:J, H113Ranked on the highest levels the object Hand on the lowest levels the objects: C, A.215Ranked on the highest levels the objects: I,D and on the lowest levels the object H.aesthetics36Ranked on the highest levels the objects: Dand on the lowest levels the objects: K, G.49Ranked on the highest levels the objects: I,H and on the lowest levels the objects: A,K.Utility and