- Email: [email protected]
Thesis for environmental psychophysics
Journal of Environmental Psychology (1989) 9, 345-356
THESIS FOR ENVIRONMENTAL PSYCHOPHYSICS J O H N C. B A I R D * and B I R G I T T A B E R G L U N D t
* Dartmouth College, Hanover, N.H.U.S.A. t University of Stockholm and Karolinska Institute, Stockholm, Sweden
Abstract The newly minted 'Handbook of Environmental Psychology' is much stronger inthe area of social psychology than it is in perception and psychophysics. While several chapters acknowledge the influence of these older fields of experimental psychology, important historical roots are scarcely noted and guidelines for future interleaving of theory and application are not explored. We present a thesis for environmental psychophysics that emphasizes the bidirectional interaction between field and laboratory practice, and the advantages accruing from viewing both psychophysics and environmental science as inextricable parts of a larger research picture. The value and implications of our thesis are illustrated through discussion of two recent case studies: cognitive mapping and the sick building syndrome.
Introduction Twenty years before Wilhelm Wundt founded what the textbooks tell us was the first laboratory in experimental psychology, Gustav Fechner was putting the finishing touches on his monumental two-volume work, 'Elemente der Psychophysik' in which he states the thesis that subjective perceptual experience can be quantified and systematically related to physical measures of the external environment (see Boring, 1950). Fechner (1860) did m a n y careful experiments to support this contention, but in all cases he dealt with simple, unidimensional stimuli presented under controlled laboratory conditions. Twenty years ago Kenneth Craik (1970) reviewed a growing literature in social science especially devoted to understanding human-environment interactions, and thereby announced the arrival of environmental psychology as one of the newest branches of applied psychology. Unlike psychophysics, the thesis of environmental psychology was that perceptual experience could best be studied as a response to complex, multidimensional stimuli usually encountered under naturalistic conditions outside the laboratory. The recent two-volume ' H a n d b o o k of Environmental Psychology' has updated much of the progress made so far to establish environmental psychology as a viable experimental discipline, but the treatment is a bit light in the area in which we w o r k - perception and psychophysics. Rather, the chapters are heavily skewed toward issues In Volume 8, 1988, the Journal of Environmental Psychology carried a review of the Handbook of Environmental Psychology,edited by D. Stockols and I, Altman (Wiley, 1987). This review consisted of a number of short responsesto the achievementof the Handbook. Baird and Berglund were originallygoing to contribute to that issue but their response has grown to the present review essay. 0272-4944/89/040345 + 12 $03.00/0
© 1989 Academic Press Limited
346
John C. Baird and Birgitta Berglund
in applied social psychology, and in fact, those sections dealing at all with perception pay scant attention to the laboratory studies that were largely responsible for launching much of the current work in environmental perception. In particular, we would have liked to see at least one chapter tracing the contributions of psychophysics to the emergence of environmental psychology. We have no pretensions about our own ability to remedy this shortcoming here, but rather offer a new thesis to guide the future direction of environmental psychophysics; a thesis illustrated by two case studies with which we are intimately familiar. The long-range goal of psychophysics, though seldom stated, must be to provide a theoretical picture of the perception of the natural environment, and from this perspective, the methods and results of psychophysics should be applicable to environmental issues. In reality, however, the arrow has gone the other way equally often; that is, the complexity of the environment has forced changes in the way we view the methods and theory of pure psychophysics. The two-way street has been well traveled in both directions. The environmental side furnishes many research issues that find justification within society at large, while at the same time the psychophysical side offers a theoretical path for Conducting sound environmental research in field settings. The interplay between these areas is reciprocal and thoroughly dynamic at all levels, including identification of a research problem, determination of procedures, and interpretation of results that eventually increment our knowledge of both psychophysics and the environment. This thesis for environmental psychophysics is depicted by the bubble diagram shown in Figure 1. The traditional psychophysical loop is evident in the bottom half of the diagram, and the traditional, albeit younger, environmental loop is evident in the upper half. Our contention is that one can enter the bubble with either an environmental or a psychophysical problem in hand, apply methodology appropriate to either area, and end up contributing knowledge to either field. For instance, what starts out looking like a clearcut environmental problem may cycle around a few times and eventually conclude with a set of findings of great importance to psychophysics; and vice versa. It is not as though certain parts of the bubble are any more 'basic' or 'applied' than other parts. The direction of movement through the structure is constrained only by the imagination of the researcher, not by the intrinsic character of the subject matter.
Environmentol Knowledge ~ E
Environmento(Prob(em Psychophysico(
Problem
Psychophysicol Methods
J
L...Psycnophysica(Results
PsychophysicolKnow(edge~
FIGURE1.
Bubble diagram of the thesis for environmental psychophysics.
Environmental Psychophysics
347
The implications of this thesis become clearer when seen against the backdrop of specific case studies. We present two such cases; one concerning cognitive mapping, the other concerning the sick building syndrome.
Case 1: cognitive mapping One of the most active areas of environmental psychology is 'cognitive mapping' in which subjects usually sketch or otherwise locate items on a board, piece of paper, or computer screen in order to represent the relative location of objects in the familiar, large-scale environment. Alternatively, they assign numbers or produce intensities to convey their judgment of inter-item distances. The major interest in these techniques comes from two directions: Experimentalists wish to know how accurate people are at representing their everyday environment as a function of such variables as a subject's age and familiarity with the setting, the nature of the setting itself, and the method employed to reveal such judgments; while theorists wish to construct models of the psychological processes involved in the creation of cognitive maps and to incorporate experimental results into broader theories of information processing in natural and laboratory settings. The Handbook provides two excellent reviews of the literature on cognitive mapping, one pertaining to adults (Golledge, 1987), the other to children (Heft & Wohlwill, 1987). What is a bit surprising in both chapters, however, is that so little attention is devoted to the historical source of the method and theory that presently makes this a viable topic for environmental science. This root source lies deep within scaling and psychophysics, and yet, there are virtually no references to key individuals responsible for the growth of this discipline over the past 30 years. For example, although the method of magnitude estimation is noted on numerous occasions, as well as the power law, there is no reference at all to S. S. Stevens. Similarly, the use of non-metric multidimensional scaling is mentioned quite frequently without any reference to its originator, Roger Shepard. On the other hand, both chapters find it necessary to discuss the perceptual theory of J. J. Gibson (1966, 1979) as somehow being crucial to the evolution of the field. In point of fact Gibson's adherence to an epistemological position of naive realism coupled with a reluctance to assign any importance to cognitive factors make a perfect antithesis to the purposes driving the study of 'cognitive' maps, though the same cannot be said about his influence on models of human navigation in naturalistic settings. As fully documented in the Handbook, the original impetus for research on cognitive mapping dates back to an influential book, 'The Image of the City', by Kevin Lynch published in 1960. As is well known, Lynch was a city planner at M.I.T. seeking ways to improve the livability of urban environments through the physical design process. Based on his understanding of the city from a designer's perspective he asked whether some cities were so constituted in their physical characteristics as to lead their inhabitants to agree as to m~jor, distinctive features; whereas other cities were built so as to elicit no such consensus in this regard. Those cities in which citizens demonstrated high agreement Lynch termed 'legible' or 'imageable'. His main point was that a legible, clearly articulated environment was a desirable place to live as compared to one which was low in imageability. Rather than actually test this notion, however, he formulated the environmental problem as one of determining the imageability of different cities, and subsequently, of identifying the physical features associated with high imageability. In this way he hoped to lay the foundation for a theory that would guide the design and construction of new cities. Cognitive mapping was merely a tool to
348
John C. Baird and Birgitta Berglund
achieve this end. In terms of the research loop in Figure 1 the idea was to address an environmental problem by applying environmental methods. The results would supposedly enhance environmental knowledge which in turn would be used to improve physical conditions. Lynch was successful in reaching his first two goals, but apparently never felt his work was influential in achieving the third and most important goal of changing the way designers actually went about building cities (Stea, 1982). On the other hand, his work unintentionally raised psychophysical issues, the solution to which have led some of us to re-examine a fundamental premise of psychophysics; namely, that psychophysical methods and theory can be generalized to explain perception of molar, large-scale environments. The cause for this re-examination stems from the early environmental research in which standard questionnaires were used to determine city images as perceived by residents of Boston, Los Angeles, and Jersey City. As part of this test battery, Lynch also asked people to sketch a map of their city. His map-sketching technique was subsequently refined by experimental psychologists. The main outcome of Lynch's study was that of the three cities, Boston was the most imageable in the sense that residents agreed on major landmarks, avenues, districts and the like. Their sketches seemed to resemble each other both in the number and type of items noted and in their mapped locations. No attempt was made to quantify Lynch's observations on these matters, and indeed, it is difficult to see how this could be done in view of the freedom allowed subjects in sketching their images. In later applications of the technique these inadequacies became more apparent. For example, if not all subjects draw the same items then how can one come up with a composite, average map for a group of individuals? A second difficulty is that some subjects may be good at producing a sketch of their home town from memory and others may be poor at this task. Are these observed differences due to differences in subjects' internal 'cognitive maps', or rather are they due to differences in their artistic skills? Another troublesome problem is that of scale: subjects often fail to plan far enough ahead in sketching an environment from memory, and consequently, find themselves running out of room on the paper to complete their sketch. Unless the size scale is appropriate and constant, as increasingly more items are placed on the map, distortions of distance will be evident in the final product. These and other drawbacks associated with the original map-sketching technique are noted in the Handbook and have been reviewed earlier by Evans (1980) and Siegel (1982). It soon became clear, then, that the methodological issues arising in map-sketching are approached more profitably through the disciplines of scaling and psychophysics than by city planning. Over the course of a decade or so what was initially considered an environmental problem gradually was transformed into a psychophysical one, the resolution of which required the application of well-known methods and data analyses. In terms of Figure 1 attention was diverted away from the environmental loop onto the psychophysical loop. Nowadays within psychophysics two primary methods are used to investigate the memory representation of familiar stimuli arranged spatially--the situation of interest in cognitive mapping. In the first, the recalled distances between pairs of items are estimated by assigning numbers or marking offdistances to reflect the ratios among the distances separating the objects under consideration; for example, all pairs of buildings and landmarks in a city or on a campus. An early application was reported by Ekman
Environmental Psychophysics
349
and Bratisch (1965) who had subjects judge the relative distances between well-known cities of the world. Since then, numerous studies of this sort have been done on a variety of environmental scales and with varying degrees of methodological sophistication (e.g., Canter & Tagg, 1975; Foley & Cohen, 1984; Kirasic, Allen, & Siegel, 1984; Golledge & Zannaras, 1973; Sherman, Croxton, & Giovanatto, 1979). This method avoids the most glaring deficiencies of map-sketching, but does not in itself yield a two-dimensional representation of the perceived location of items in the environment. If the pair-wise distance estimates are treated by standard techniques of multidimensional scaling, however, a two-dimensional map can be obtained and compared for accuracy with the actual spatial arrangement of items (Baird, Merrill, & Tannenbaum, 1979; Magnana, Evans, & Romney, 1981). The second approach is a more obvious extension of Lynch's ideas. Here a grid of fixed size and containing a set number of empty cells is provided to the subject, along with a set of items to be mapped. The task is to locate the items in the grid to represent their relative location in the familiar environment under investigation. This method has been simply called direct mapping. Because of the commonality of items and the fixed nature of the space within which mapping occurs, it is possible to aggregate data from different subjects. Potential differences in drawing ability can be obviated by using a computer to present the grid and stimuli, and by allowing subjects to place items in the space and edit their positions until they are satisfied with the map (Baird et aL, 1979). The final result is treated as a subjective analogue of the perceived environment that can be compared against the physical environment or against a map created by multidimensional scaling of pair-wise distance estimates--that is, secured by the first psychophysical method described above. Either of these methods is potentially effective in revealing the general structure of cognitive maps, and as such, either would suit the needs of those environmental psychologists who seek to uncover an individual's overview of the perceived environment, as represented in memory. On the other hand, the same cannot be said about the two methods as they apply to more detailed psychophysical problems. The maps obtained by direct mapping and magnitude estimation are not the same, and in fact, subjects definitely see them as different. In our experiments designed to compare the two methods (Baird et al., 1979), subjects preferred the direct map they created over the one arrived at by multidimensional scaling (based on pair-wise distance estimates). From the standpoint of psychophysics this is a rather unsettling development, since it brings into question the belief that all methods are equally good for expressing an individual's subjective views of the environment. In addition, it raises the broader point concerning the validity of multidimensional scaling when applied in many other domains of behavioral science, especially in cases where the nature of the dimensions revealed by the method are not readily interpretable. Fortunately, the psychophysical problem can be quantified and made explicit (Baird, Wagner, & Noma, 1982). Magnitude estimates (J) of unidimensional distances between points in the recalled environment relate to actual physical distances (D) by a power function (e.g., Cadwallader, 1979). J = 2D ~
(1)
Equation 1 is well-known as Stevens's law; it applies generally to the relationship between magnitude estimates and physical intensities presented to the various sensory systems (Baird & Noma, 1978; Stevens, 1975), where the exponent depends on the
350
John C. Baird and Birgitta Berglund
particular stimulus attribute and the multiplicative constant depends on the response format. Since unidimensional distances are being evaluated, there are only two constraints on the judgments, which themselves can be treated as perceived distances: (i)
J~,y = 0
if x = y
Jx,y > 0 if x :~ y
(ii)
On the other hand, drawn maps in two dimensions must satisfy two additional axioms: (iii)
J~,r = Jr,x
(iv)
Jx.y + Jr,~ 1> L,~
These four axioms define a metric (Baird & Noma, 1978; Golledge, 1987). Minor violations of (iii) and (iv) have been reported for distance estimates of items in twodimensional space by Cadwallader (1979) and by Burroughs and Sadalla (1979). MoreJcritically, the exponent (7) of Equation 1, based on magnitude estimates of distances, is not always the same as the exponent based on the distances between items produced in direct mapping (Baird et al., 1979; Sherman et aL, 1979). This indicates that the cognitive representation of space depends on the method used to reveal it. Conflicts arise between results of these two methods because the distortions (based on perception or memory) of distance that occur in magnitude estimation of one-dimensional distances are not closely linked to the distortions that occur in direct mapping of twodimensional distances. The two approaches yield different, sometimes incompatible, cognitive maps. This invites the query as to whether both methods are measuring the same thing. By way of example, suppose a subject marks on a sheet of paper his or her memory of the relative positions of three buildings located along the same street. x
y
z
All four axioms for a metric are obviously satisfied. In particular,
~x,y+ Jy,z=Jx,z Now suppose the subject estimates the distances between all possible pairs of the three buildings by assigning a numerical rating to each inter-building distance. Assume a power function (Equation 1) describes the relationship between physical and judged distance. It almost certainly will; that is, the judged distances for each pair are
Jx,y = 2(Ox,y Jy,z = 2(Dy,z)~ Jx,~ = 2(D~,y So long as the subject is accurate, by which we mean that judgments are linearly related to physical distance, the exponent (7) will be l, and hence,
J~,y+4,~=Jx,z On the other hand, suppose the relative distances are overestimated such that ? = 2, in this case,
~,y + ~,z < J.,z
Environmental Psychophysics
351
and axiom (iv) is violated. The problem is that the magnitude estimates only satisfy a Euclidean metric if they are linearly related to the physical distances (exponent equals 1). Otherwise, contradictions occur in trying to press, squash, or expand the distances to fit into a Euclidean space that must by definition hold sway in direct mapping, simply because of the inherent constraints of the two-dimensional surface on which the items are placed. Individual differences abound when a group of subjects are required to give magnitude estimates of the same set of stimuli, and although the mean exponent may be one, there surely will be great dispersion of individual values around the mean. Several specific examples of how sets of magnitude estimates imply serious distortions of Euclidean space are given in Baird et al. (1982). Direct mapping and magnitude estimation are both used to uncover cognitive representations of familiar environments. Under a potentially wide range of experimental conditions, however, these two methods produce results that can be shown to be mathematically contradictory. Such a situation requires a reappraisal of the role of psychophysics in the broader field of perception. It has often been assumed that psychophysics would someday provide the foundation for a theory of perception, and yet, it is difficult to see how this can happen if the results from different methods fail to dovetail into a unified picture. What began as a city planner's interest in map sketching as a means to reveal residents' views of the everyday environment has wound its way into a theoretical issue with deep implications for all of perception and cognition. This latter development would never have occurred if researchers had confined their efforts to the environmental loop in Figure 1 and not ventured forth to explore the wider implications of their findings. Similarly, the theoretical results obtained in psychophysics may now feed back to enrich discussions of cognitive mapping within environmental psychology. Case 2: the sick building syndrome Our second example comes from quite a different area of environmental psychology, and one not considered at any great length in the Handbook, though it is briefly noted in the chapter by Kiiller (1987). The problem arises within the environment proper and its origin is more recent than that of cognitive mapping. Over the last decade there has been a marked increase in reports of ill health associated with modern, energy-efficient buildings. When a sufficient number of individuals occupying such structures report a specific constellation of medical symptoms, the building may be classified as "sick." Once the cause of the symptoms is attributed to the built environment it becomes an issue addressed by environmental science, including the disciplines of health psychology and psychophysics. Following the energy crisis of 1972 new buildings were constructed with more efficlent insulation, and older structures were tightened in order to conserve energy. Shortly thereafter people throughout the industrialized world began reporting health problems from stuffy, strange-smelling new work settings. In Sweden, for example some 125 of the 691 preschools built from 1977 to 1982 in greater Stockholm became the subject of health concerns, and several of the buildings were actually closed down. Psychophysics originally became involved in the issue of sick buildings because most of the symptoms reported by occupants were sensory in nature: irritation of the eyes, nose, and throat; dry skin and mucosa, erythema, mental fatigue, and the perception of weak but persistent odors (Berglund & Lindvall, 1986; WHO, 1983). Such a pattern of symptoms is quite similar to that found after exposure to formaldehyde, a chemical
352
John C. Baird and Birgitta Berglund
prevalent in many modern building materials. Formaldehyde was also implicated as a causative agent by the fact that some of the buildings classified as 'sick' showed improvement after efforts were made to sanitize them against formaldehyde sources (bring air concentrations below 100#g m-3). On the other hand, some buildings showed no improvement after treatment. Starting with the assumption of a chemical (airborne) origin to the sick building problem, the psychophysical loop in Figure 1 was entered and the problem was temporarily removed from the environmental arena and brought into the laboratory, where a kind of psycho-chemical approach was instituted to determine more carefully human sensitivity to formaldehyde. One of the initial problems addressed was the great variability between studies reported in the literature on the absolute odor threshold, the value of which is important in setting limit values for formaldehyde exposure in non-industrial indoor environments. In the process of finding out that this variability was probably not due to intra or inter-individual differences, nor due to the method of making formaldehyde, a basic psychophysical result was obtained as well (Berglund, Berglund, H6gman, Johansson & Lindvall, 1987). Both a signal detection and free magnitude estimation paradigm were employed and it was found that the psychophysical function (between judgments and stimulus intensity) was consistent with the percent detection values obtained from the yes-no task. Such a result tends to validate alternative psychophysical methods and to demonstrate the internal consistency of the psychophysical paradigm, quite apart from its role in unraveling environmental problems. In related studies (Ahlstr6m, Berglund, Berglund, & Lindvall, 1984) it was found that mixing formaldehyde with indoor air had a large relative effect on perceived odor strength when the concentrations of formaldehyde were low, but that no substantial effects of such mixing were evident when the concentrations were high. This again is a result with implications both of a practical and theoretical nature, since chemical substances are seldom found alone in the natural environment, but rather, are usually present as part of a chemical mixture. This line of research did not, however, implicate formaldehyde in the sick building syndrome, since most of the concentrations found in the buildings were well below the obtained detection thresholds. The return to the environmental domain meant facing the same issue in a new light. Rather than look for a single chemical or chemically defined group of substances that might be responsible for the sick building syndrome we began thinking about the matter in terms of a pattern of many substances, and that possibly, this pattern is different in healthy and sick buildings. This raised a host of new environmental considerations. For example, how does one diagnose a sick building? What is the proper control case? And most importantly, how does one conduct pattern recognition on chemical patterns obtained from analysis of the air in buildings? Air samples were taken from a healthy and sick preschool and from a healthy office building on a number of occasions and then analysed by gas chromatography for volatile organic compounds, which were subsequently identified by mass spectrometry. In addition, subjects sniffed a fraction of the output eluate from the chemical analysis over time and judged its intensity by magnitude estimation. In this way it is possible to relate the magnitude estimate to the concentration of particular substances being judged. Results indicated that the judgement of chemical intensity showed no significant correlation with the chemical concentration (pbb) actually present. Some chemicals of high concentration are barely noticed, and others of low concentration are
Environmental Psychophysics
353
judged to be quite strong smelling (Berglund, Berglund, Lindvall & NicanderBredberg, 1982), a finding with importance for both psychophysics and human health. More recently, we performed a number of analyses to isolate particular patterns in both the chemical and judgement data. These included a number of techniques developed within the tradition of scaling and psychophysics, such as factor (component), cluster, and correspondence analysis (Berglund, Berglund, Lindvall & Nicander-Bredberg, 1982; Baird, Berglund, Berglund, Nicander-Bredberg & Noma, 1987; Noma, Berglund, Berglund, Johansson, & Baird, 1988). The indications from these analyses are that distinctive patterns of chemicals can be found in sick and healthy buildings, although the exact identification of chemical constituents has yet to be completed (approximately 40 have been positively identified as relevant so far). Beside distinguishing between different types of buildings these analysis techniques can find patterns of chemical substances that are characteristic of locations within buildings; for example, the pattern present for supply air entering a building, air after it goes through a ventilation system, exhaust air, etc. The impact of these studies is that a sort ofpsycho-chemics has emerged, combining psychophysical methods and analysis techniques (e.g., cluster analysis, factor analysis) with standard tests applied in analytic chemistry to identify the constituents of air (McLafferty, 1980; McNair & Bonelli, 1968). It is interesting to note that many of the scaling methods familiar to psychologists for over 20 years have only recently been discovered by chemists working on problems of air pollution (cf. Massart & Kaufman, 1983; Varmuza, 1980). When we consider the environmental loop in Figure ! in the broadest sense, it is now possible through an interdisciplinary effort to make some headway toward evaluating the impact of the built environment on human health. It is not a t all clear how this could be accomplished without the experimental and quantitative models of scaling and psychophysics. The implications for theoretical psychophysics are also powerful in that we are not dealing here with a simple dose-response relationship, similar to what are referred to as "psychophysical laws' (Bair & Noma, 1978; Stevens, 1975). Pattern psychophysics is not unidimensional, and as such, it suggests a re-evaluation of the multidimensional nature of the stimulus in psychophysics. It seems especially true, as with the case of cognitive mapping, that general perceptual theories will require a richer paradigm than the unidimensional approach so religiously followed throughout the history of psychophysics. Discussion Our thesis for environmental psychophysics does not plead for more applied work or for a psychophysics of complex stimuli. Nor do we advocate that environmental perception be reduced in some way to pure psychophysics. Rather it is that the growth of knowledge in either field;is inextricably tied to developments in the other. Although we feel that all phases of the bubble diagram of Figure 1 arc impacted to some extent by the two cases we have presented, the impact will be greater in some places than in others and the locus of such impact will depend on the specific examples being considered. Nonetheless, a few general principles can in fact be taken from the case studies of cognitive mapping and the sick building syndrome. (1) The problem you start with may not be the problem you solve. Kevin Lynch wanted to influence the design of new cities to make them more
354
John C. Baird and Birgitta Berglund
imageable for their residents. The map-sketching technique he created to document levels of imageability was streamlined by experimental psychologists to the point where subsequent tests of two-dimensional and one-dimensional knowledge of spatial relations led to theoretical contradictions at the very heart of psychophysics. (2) The problem you start with may not be the problem you end with. The original intent of introducing psychophysics into the picture for the study of sick buildings was that the perception of odorous substances might be used to distinguish between sick and healthy locations. The scaling methods adapted to do this turned out to be used primarily for the separation of locations based on the chemical constituents of air. What started out to be a psychophysical problem ended up being one of pattern recognition among chemicals in different environments--though the psychophysical work-continues as well. (3) The point of impact of new knowledge cannot be predicted with much certainty. One major implication of our thesis is that an individual or a research team must possess different kinds of training and expertise in order to function comfortably at different places in the bubble diagram. The impact of one field on another can only be seen by individuals (or a team) who have access-to information in both realms of knowledge and who can make the necessary connections across disciplines. This interdisciplinary promise of environmental psychology remains its greatest virtue, despite the difficulties to be overcome in securing its rightful position among the environmental sciences. The early textbooks on environmental psychology had two broad themes: perception of complex stimulus contexts, and the social interactions that occur within them (e.g., Bell, Fisher, & Loomis, 1978; Ittelson, Proshansky, Rivlin, & Winkel, 1974). The recently issued Handbook thoroughly updates the contributions of social psychology to the field, but in our opinion takes little note of the equally-significant progress in environmental perception and psychophysics.
References
Ahlstr6m, R., Berglund, B., Berglund, U. & Lindvall, T. (1986). Formaldehyde odor and its interaction with the air of a sick building. Environment International, 12, 289-295. Baird, J. C., Berglund, B., Berglund, U., Nicander-Bredberg, H. & Noma, E. (1987). Distinguishing between healthy and sick preschools by chemical classification. Environment International, 13, 167-174. Baird, J. C., Merill, A. A. & Tannenbaum, J. (1979). Cognitive representation of spatial relations. II. A familiar environment. Journal of Experimental Psychology: General, 108, 92-98. Baird, J. C. & Noma, E. (1978). Fundamentals of Scaling and Psychophysics. New York: WileyInterscience. Baird, J. C., Wagner, M. & Noma, E. (1982). Impossible cognitive spaces. GeographiealAnalysis, 14, 204-216. Bell, P. A., Fisher, J. D. & Loomis, R. J. (1978). Environmental Psychology. Philadelphia: Saunders. Berglund, B., Berglund, U., H6gman, L., Johansson, I. & Lindvall, T. (1985). Measurement of formaldehyde odor indoors. In CLIMA 2000, P. O. Fanger, Ed., vol. 4: Indoor climate, pp. 251-257. VVS Kongres-VVS Messe, Copenhagen. Berglund, B., Berglund, U., Lindvall, T. & Nicander-Bredberg, H. (1982). Olfactory and chemical
Environmental Psychophysies
355
characterization of indoor air. Towards a psychophysical model for air quality. Environment International, 8, 327-332. Berglund, B. & Lindvall, T. (1986). Sensory reactions to 'sick buildings'. Environment International, 12, 147-159. Boring, E. G. (1950). A History of Experimental Psychology. 2nd ed., Appleton-Century-Crofts, New York. Burroughs, W. J. & Sadalla, E. K. (1979). Asymmetries in distance cognition. Geographical Analysis, 11,414-421. Cadwallader, M. (1979). Problems in cognitive distance: implications for cognitive mapping. Environment and Behavior, 11, 559-576. Canter, D. & Tagg, S. K. (1975). Distance estimation in cities. Environment and Behavior, 7, 59-80. Craik, K. H. (1970). Environmental psychology. In K. H. Craik, B. Kleinmuntz, R. L. Rosnow, R. Rosenthal, J. A. Cheyne & R. H. Walters, Eds., New Directions in Psychology, 4, 1-122. Holt, Rhinehart & Winston. Ekman, G. & Bratfisch, O. (1965). Subjective distance and emotional involvement--a psychological mechanism. Acta Psychologica, 24, 446-453. Evans, G. E. Environmental cognition. (1980). Psychological Bulletin, 88, 259-287. Fechner, G. T. (1860). Elemente der Psychophysik. Breitkopf & Hdrtel, Leipzig. Foley, J. E. & Cohen, A. J. (1984). Working mental representations of the environment. Environment and Behavior, 16, 713-729. Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J. J. (1979). The Ecological Approach to Visual Perception. Boston: Houghton Mifflin. Golledge, R. G. (1987). Environmental cognition. In D. Stokols & I. Altman, Eds., Handbook of Environmental Psychology, vol. 1, 131-174. New York: Wiley. Golledge, R. G. & Zannaras, G. (1973). Cognitive approaches to the analysis of human spatial behavior. In W. H. Ittelson Ed., Environment and Cognition. New York: Seminar Press. Heft, H. & Wohlwill, J. F. (1987). Environmental cognition in children. In D. Stokols & I. AItman, Eds., Handbook of Environmental Psychology, vol. 1, 175-204. New York: Wiley. Ittelson, W., Proshansky, H., Rivlin & Winkel, G. (1974). An Introduction to Environmental Psychology. New York: Holt, Rinehart & Winston. Kirasic, K, C., Allen, G. L. & Siegel, A. W. (1984). Expression of configurational knowledge of large-scale environments: students' performance of cognitive tasks. Environment and Behavior, 16, 687 712. Kfiller, R. (1987). Environmental psychology from a Swedish perspective. In D. Stokols & I. Altman, Eds., Handbook of Environmental Psychology, voL 2, 1243-1279. Lynch, K. (1960). The Image of the City. Cambridge, Mass.: MIT Press. Magana, J. R., Evans, G. W. & Romney, A. K. (1981). Scaling techniques in the analysis of environmental cognition data. Professional Geographer, 33, 294-301. Massart, D. L. & Kaufman, L. (1983). The Interpretation of Analytic Chemical Data by the Use of Cluster Analysis. Wiley Interscience, New York. McLafferty, F. W. (1980). Interpretation of Mass Spectra. 3rd ed., Mill Valley, California, University Science Books. McNair, H. M. & Bonelli, E. J. (1968). Basic Gas Chromatography. Palo Alto, California, Varian. Noma, E., Berglund, B., Berglund, U., Johansson, I. & Baird, J. C. (1988). Joint representation of physical locations and volatile organic compounds in indoor air from a healthy and a sick building. Atmospheric Environment, 22, 451-460. Sherman, R. C., Croxton, J. & Giovanatto, J. (1979). Investigating cognitive representations of spatial relationships. Environment and Behavior, 11, 209-226. Siegel, A. W. (1982). Methodological and metatheoretical issues in the development of cognitive mapping of large-scale environments. In J. C. Baird & A. D. Lutkus, Eds., Mind Child Architecture. Hanover, N. H.: University Press of New England. Stea, D. (1982). Cross-cultural environmental modeling. In J. C. Baird & A. D. Lutkus, Eds., Mind Child Architecture. Hanover, N. H.: University Press of New England. Stevens, S. S. (1975). Psychophysics: Introduction to its Perceptual, Neural, and Social Prospects. New York: Wiley-Interscience. Tell, W. (1983). Hiilsofarliga ByggnadsmateriaL VVS Special No. 2, 43-46.
356
John C. Baird and Birgitta Berglund
Varmuza, K. (1980). Pattern Recognition in Chemistry. Berlin: Springer-Verlag. WHO (1983). Indoor Air Pollutants: Exposure and Health Effects. Copenhagen,Denmark: World Health Organization. EURO Reports and and Studies, 78. Acknowledgments This research was supported by the Swedish Council for Building Research and the Swedish Council for Research in the Humanities and Social Sciences. The paper was written while John C. Baird was a visiting professor at the Department of Psychology, University of Stockholm, and the Department of Hygiene of the National Institute of Environmental Medicine. Reprint requests should be addressed to Professor Birgitta Berglund, Department of Psychology, University of Stockholm, S-106 91 Stockholm, Sweden.
THESIS FOR ENVIRONMENTAL PSYCHOPHYSICS J O H N C. B A I R D * and B I R G I T T A B E R G L U N D t
* Dartmouth College, Hanover, N.H.U.S.A. t University of Stockholm and Karolinska Institute, Stockholm, Sweden
Abstract The newly minted 'Handbook of Environmental Psychology' is much stronger inthe area of social psychology than it is in perception and psychophysics. While several chapters acknowledge the influence of these older fields of experimental psychology, important historical roots are scarcely noted and guidelines for future interleaving of theory and application are not explored. We present a thesis for environmental psychophysics that emphasizes the bidirectional interaction between field and laboratory practice, and the advantages accruing from viewing both psychophysics and environmental science as inextricable parts of a larger research picture. The value and implications of our thesis are illustrated through discussion of two recent case studies: cognitive mapping and the sick building syndrome.
Introduction Twenty years before Wilhelm Wundt founded what the textbooks tell us was the first laboratory in experimental psychology, Gustav Fechner was putting the finishing touches on his monumental two-volume work, 'Elemente der Psychophysik' in which he states the thesis that subjective perceptual experience can be quantified and systematically related to physical measures of the external environment (see Boring, 1950). Fechner (1860) did m a n y careful experiments to support this contention, but in all cases he dealt with simple, unidimensional stimuli presented under controlled laboratory conditions. Twenty years ago Kenneth Craik (1970) reviewed a growing literature in social science especially devoted to understanding human-environment interactions, and thereby announced the arrival of environmental psychology as one of the newest branches of applied psychology. Unlike psychophysics, the thesis of environmental psychology was that perceptual experience could best be studied as a response to complex, multidimensional stimuli usually encountered under naturalistic conditions outside the laboratory. The recent two-volume ' H a n d b o o k of Environmental Psychology' has updated much of the progress made so far to establish environmental psychology as a viable experimental discipline, but the treatment is a bit light in the area in which we w o r k - perception and psychophysics. Rather, the chapters are heavily skewed toward issues In Volume 8, 1988, the Journal of Environmental Psychology carried a review of the Handbook of Environmental Psychology,edited by D. Stockols and I, Altman (Wiley, 1987). This review consisted of a number of short responsesto the achievementof the Handbook. Baird and Berglund were originallygoing to contribute to that issue but their response has grown to the present review essay. 0272-4944/89/040345 + 12 $03.00/0
© 1989 Academic Press Limited
346
John C. Baird and Birgitta Berglund
in applied social psychology, and in fact, those sections dealing at all with perception pay scant attention to the laboratory studies that were largely responsible for launching much of the current work in environmental perception. In particular, we would have liked to see at least one chapter tracing the contributions of psychophysics to the emergence of environmental psychology. We have no pretensions about our own ability to remedy this shortcoming here, but rather offer a new thesis to guide the future direction of environmental psychophysics; a thesis illustrated by two case studies with which we are intimately familiar. The long-range goal of psychophysics, though seldom stated, must be to provide a theoretical picture of the perception of the natural environment, and from this perspective, the methods and results of psychophysics should be applicable to environmental issues. In reality, however, the arrow has gone the other way equally often; that is, the complexity of the environment has forced changes in the way we view the methods and theory of pure psychophysics. The two-way street has been well traveled in both directions. The environmental side furnishes many research issues that find justification within society at large, while at the same time the psychophysical side offers a theoretical path for Conducting sound environmental research in field settings. The interplay between these areas is reciprocal and thoroughly dynamic at all levels, including identification of a research problem, determination of procedures, and interpretation of results that eventually increment our knowledge of both psychophysics and the environment. This thesis for environmental psychophysics is depicted by the bubble diagram shown in Figure 1. The traditional psychophysical loop is evident in the bottom half of the diagram, and the traditional, albeit younger, environmental loop is evident in the upper half. Our contention is that one can enter the bubble with either an environmental or a psychophysical problem in hand, apply methodology appropriate to either area, and end up contributing knowledge to either field. For instance, what starts out looking like a clearcut environmental problem may cycle around a few times and eventually conclude with a set of findings of great importance to psychophysics; and vice versa. It is not as though certain parts of the bubble are any more 'basic' or 'applied' than other parts. The direction of movement through the structure is constrained only by the imagination of the researcher, not by the intrinsic character of the subject matter.
Environmentol Knowledge ~ E
Environmento(Prob(em Psychophysico(
Problem
Psychophysicol Methods
J
L...Psycnophysica(Results
PsychophysicolKnow(edge~
FIGURE1.
Bubble diagram of the thesis for environmental psychophysics.
Environmental Psychophysics
347
The implications of this thesis become clearer when seen against the backdrop of specific case studies. We present two such cases; one concerning cognitive mapping, the other concerning the sick building syndrome.
Case 1: cognitive mapping One of the most active areas of environmental psychology is 'cognitive mapping' in which subjects usually sketch or otherwise locate items on a board, piece of paper, or computer screen in order to represent the relative location of objects in the familiar, large-scale environment. Alternatively, they assign numbers or produce intensities to convey their judgment of inter-item distances. The major interest in these techniques comes from two directions: Experimentalists wish to know how accurate people are at representing their everyday environment as a function of such variables as a subject's age and familiarity with the setting, the nature of the setting itself, and the method employed to reveal such judgments; while theorists wish to construct models of the psychological processes involved in the creation of cognitive maps and to incorporate experimental results into broader theories of information processing in natural and laboratory settings. The Handbook provides two excellent reviews of the literature on cognitive mapping, one pertaining to adults (Golledge, 1987), the other to children (Heft & Wohlwill, 1987). What is a bit surprising in both chapters, however, is that so little attention is devoted to the historical source of the method and theory that presently makes this a viable topic for environmental science. This root source lies deep within scaling and psychophysics, and yet, there are virtually no references to key individuals responsible for the growth of this discipline over the past 30 years. For example, although the method of magnitude estimation is noted on numerous occasions, as well as the power law, there is no reference at all to S. S. Stevens. Similarly, the use of non-metric multidimensional scaling is mentioned quite frequently without any reference to its originator, Roger Shepard. On the other hand, both chapters find it necessary to discuss the perceptual theory of J. J. Gibson (1966, 1979) as somehow being crucial to the evolution of the field. In point of fact Gibson's adherence to an epistemological position of naive realism coupled with a reluctance to assign any importance to cognitive factors make a perfect antithesis to the purposes driving the study of 'cognitive' maps, though the same cannot be said about his influence on models of human navigation in naturalistic settings. As fully documented in the Handbook, the original impetus for research on cognitive mapping dates back to an influential book, 'The Image of the City', by Kevin Lynch published in 1960. As is well known, Lynch was a city planner at M.I.T. seeking ways to improve the livability of urban environments through the physical design process. Based on his understanding of the city from a designer's perspective he asked whether some cities were so constituted in their physical characteristics as to lead their inhabitants to agree as to m~jor, distinctive features; whereas other cities were built so as to elicit no such consensus in this regard. Those cities in which citizens demonstrated high agreement Lynch termed 'legible' or 'imageable'. His main point was that a legible, clearly articulated environment was a desirable place to live as compared to one which was low in imageability. Rather than actually test this notion, however, he formulated the environmental problem as one of determining the imageability of different cities, and subsequently, of identifying the physical features associated with high imageability. In this way he hoped to lay the foundation for a theory that would guide the design and construction of new cities. Cognitive mapping was merely a tool to
348
John C. Baird and Birgitta Berglund
achieve this end. In terms of the research loop in Figure 1 the idea was to address an environmental problem by applying environmental methods. The results would supposedly enhance environmental knowledge which in turn would be used to improve physical conditions. Lynch was successful in reaching his first two goals, but apparently never felt his work was influential in achieving the third and most important goal of changing the way designers actually went about building cities (Stea, 1982). On the other hand, his work unintentionally raised psychophysical issues, the solution to which have led some of us to re-examine a fundamental premise of psychophysics; namely, that psychophysical methods and theory can be generalized to explain perception of molar, large-scale environments. The cause for this re-examination stems from the early environmental research in which standard questionnaires were used to determine city images as perceived by residents of Boston, Los Angeles, and Jersey City. As part of this test battery, Lynch also asked people to sketch a map of their city. His map-sketching technique was subsequently refined by experimental psychologists. The main outcome of Lynch's study was that of the three cities, Boston was the most imageable in the sense that residents agreed on major landmarks, avenues, districts and the like. Their sketches seemed to resemble each other both in the number and type of items noted and in their mapped locations. No attempt was made to quantify Lynch's observations on these matters, and indeed, it is difficult to see how this could be done in view of the freedom allowed subjects in sketching their images. In later applications of the technique these inadequacies became more apparent. For example, if not all subjects draw the same items then how can one come up with a composite, average map for a group of individuals? A second difficulty is that some subjects may be good at producing a sketch of their home town from memory and others may be poor at this task. Are these observed differences due to differences in subjects' internal 'cognitive maps', or rather are they due to differences in their artistic skills? Another troublesome problem is that of scale: subjects often fail to plan far enough ahead in sketching an environment from memory, and consequently, find themselves running out of room on the paper to complete their sketch. Unless the size scale is appropriate and constant, as increasingly more items are placed on the map, distortions of distance will be evident in the final product. These and other drawbacks associated with the original map-sketching technique are noted in the Handbook and have been reviewed earlier by Evans (1980) and Siegel (1982). It soon became clear, then, that the methodological issues arising in map-sketching are approached more profitably through the disciplines of scaling and psychophysics than by city planning. Over the course of a decade or so what was initially considered an environmental problem gradually was transformed into a psychophysical one, the resolution of which required the application of well-known methods and data analyses. In terms of Figure 1 attention was diverted away from the environmental loop onto the psychophysical loop. Nowadays within psychophysics two primary methods are used to investigate the memory representation of familiar stimuli arranged spatially--the situation of interest in cognitive mapping. In the first, the recalled distances between pairs of items are estimated by assigning numbers or marking offdistances to reflect the ratios among the distances separating the objects under consideration; for example, all pairs of buildings and landmarks in a city or on a campus. An early application was reported by Ekman
Environmental Psychophysics
349
and Bratisch (1965) who had subjects judge the relative distances between well-known cities of the world. Since then, numerous studies of this sort have been done on a variety of environmental scales and with varying degrees of methodological sophistication (e.g., Canter & Tagg, 1975; Foley & Cohen, 1984; Kirasic, Allen, & Siegel, 1984; Golledge & Zannaras, 1973; Sherman, Croxton, & Giovanatto, 1979). This method avoids the most glaring deficiencies of map-sketching, but does not in itself yield a two-dimensional representation of the perceived location of items in the environment. If the pair-wise distance estimates are treated by standard techniques of multidimensional scaling, however, a two-dimensional map can be obtained and compared for accuracy with the actual spatial arrangement of items (Baird, Merrill, & Tannenbaum, 1979; Magnana, Evans, & Romney, 1981). The second approach is a more obvious extension of Lynch's ideas. Here a grid of fixed size and containing a set number of empty cells is provided to the subject, along with a set of items to be mapped. The task is to locate the items in the grid to represent their relative location in the familiar environment under investigation. This method has been simply called direct mapping. Because of the commonality of items and the fixed nature of the space within which mapping occurs, it is possible to aggregate data from different subjects. Potential differences in drawing ability can be obviated by using a computer to present the grid and stimuli, and by allowing subjects to place items in the space and edit their positions until they are satisfied with the map (Baird et aL, 1979). The final result is treated as a subjective analogue of the perceived environment that can be compared against the physical environment or against a map created by multidimensional scaling of pair-wise distance estimates--that is, secured by the first psychophysical method described above. Either of these methods is potentially effective in revealing the general structure of cognitive maps, and as such, either would suit the needs of those environmental psychologists who seek to uncover an individual's overview of the perceived environment, as represented in memory. On the other hand, the same cannot be said about the two methods as they apply to more detailed psychophysical problems. The maps obtained by direct mapping and magnitude estimation are not the same, and in fact, subjects definitely see them as different. In our experiments designed to compare the two methods (Baird et al., 1979), subjects preferred the direct map they created over the one arrived at by multidimensional scaling (based on pair-wise distance estimates). From the standpoint of psychophysics this is a rather unsettling development, since it brings into question the belief that all methods are equally good for expressing an individual's subjective views of the environment. In addition, it raises the broader point concerning the validity of multidimensional scaling when applied in many other domains of behavioral science, especially in cases where the nature of the dimensions revealed by the method are not readily interpretable. Fortunately, the psychophysical problem can be quantified and made explicit (Baird, Wagner, & Noma, 1982). Magnitude estimates (J) of unidimensional distances between points in the recalled environment relate to actual physical distances (D) by a power function (e.g., Cadwallader, 1979). J = 2D ~
(1)
Equation 1 is well-known as Stevens's law; it applies generally to the relationship between magnitude estimates and physical intensities presented to the various sensory systems (Baird & Noma, 1978; Stevens, 1975), where the exponent depends on the
350
John C. Baird and Birgitta Berglund
particular stimulus attribute and the multiplicative constant depends on the response format. Since unidimensional distances are being evaluated, there are only two constraints on the judgments, which themselves can be treated as perceived distances: (i)
J~,y = 0
if x = y
Jx,y > 0 if x :~ y
(ii)
On the other hand, drawn maps in two dimensions must satisfy two additional axioms: (iii)
J~,r = Jr,x
(iv)
Jx.y + Jr,~ 1> L,~
These four axioms define a metric (Baird & Noma, 1978; Golledge, 1987). Minor violations of (iii) and (iv) have been reported for distance estimates of items in twodimensional space by Cadwallader (1979) and by Burroughs and Sadalla (1979). MoreJcritically, the exponent (7) of Equation 1, based on magnitude estimates of distances, is not always the same as the exponent based on the distances between items produced in direct mapping (Baird et al., 1979; Sherman et aL, 1979). This indicates that the cognitive representation of space depends on the method used to reveal it. Conflicts arise between results of these two methods because the distortions (based on perception or memory) of distance that occur in magnitude estimation of one-dimensional distances are not closely linked to the distortions that occur in direct mapping of twodimensional distances. The two approaches yield different, sometimes incompatible, cognitive maps. This invites the query as to whether both methods are measuring the same thing. By way of example, suppose a subject marks on a sheet of paper his or her memory of the relative positions of three buildings located along the same street. x
y
z
All four axioms for a metric are obviously satisfied. In particular,
~x,y+ Jy,z=Jx,z Now suppose the subject estimates the distances between all possible pairs of the three buildings by assigning a numerical rating to each inter-building distance. Assume a power function (Equation 1) describes the relationship between physical and judged distance. It almost certainly will; that is, the judged distances for each pair are
Jx,y = 2(Ox,y Jy,z = 2(Dy,z)~ Jx,~ = 2(D~,y So long as the subject is accurate, by which we mean that judgments are linearly related to physical distance, the exponent (7) will be l, and hence,
J~,y+4,~=Jx,z On the other hand, suppose the relative distances are overestimated such that ? = 2, in this case,
~,y + ~,z < J.,z
Environmental Psychophysics
351
and axiom (iv) is violated. The problem is that the magnitude estimates only satisfy a Euclidean metric if they are linearly related to the physical distances (exponent equals 1). Otherwise, contradictions occur in trying to press, squash, or expand the distances to fit into a Euclidean space that must by definition hold sway in direct mapping, simply because of the inherent constraints of the two-dimensional surface on which the items are placed. Individual differences abound when a group of subjects are required to give magnitude estimates of the same set of stimuli, and although the mean exponent may be one, there surely will be great dispersion of individual values around the mean. Several specific examples of how sets of magnitude estimates imply serious distortions of Euclidean space are given in Baird et al. (1982). Direct mapping and magnitude estimation are both used to uncover cognitive representations of familiar environments. Under a potentially wide range of experimental conditions, however, these two methods produce results that can be shown to be mathematically contradictory. Such a situation requires a reappraisal of the role of psychophysics in the broader field of perception. It has often been assumed that psychophysics would someday provide the foundation for a theory of perception, and yet, it is difficult to see how this can happen if the results from different methods fail to dovetail into a unified picture. What began as a city planner's interest in map sketching as a means to reveal residents' views of the everyday environment has wound its way into a theoretical issue with deep implications for all of perception and cognition. This latter development would never have occurred if researchers had confined their efforts to the environmental loop in Figure 1 and not ventured forth to explore the wider implications of their findings. Similarly, the theoretical results obtained in psychophysics may now feed back to enrich discussions of cognitive mapping within environmental psychology. Case 2: the sick building syndrome Our second example comes from quite a different area of environmental psychology, and one not considered at any great length in the Handbook, though it is briefly noted in the chapter by Kiiller (1987). The problem arises within the environment proper and its origin is more recent than that of cognitive mapping. Over the last decade there has been a marked increase in reports of ill health associated with modern, energy-efficient buildings. When a sufficient number of individuals occupying such structures report a specific constellation of medical symptoms, the building may be classified as "sick." Once the cause of the symptoms is attributed to the built environment it becomes an issue addressed by environmental science, including the disciplines of health psychology and psychophysics. Following the energy crisis of 1972 new buildings were constructed with more efficlent insulation, and older structures were tightened in order to conserve energy. Shortly thereafter people throughout the industrialized world began reporting health problems from stuffy, strange-smelling new work settings. In Sweden, for example some 125 of the 691 preschools built from 1977 to 1982 in greater Stockholm became the subject of health concerns, and several of the buildings were actually closed down. Psychophysics originally became involved in the issue of sick buildings because most of the symptoms reported by occupants were sensory in nature: irritation of the eyes, nose, and throat; dry skin and mucosa, erythema, mental fatigue, and the perception of weak but persistent odors (Berglund & Lindvall, 1986; WHO, 1983). Such a pattern of symptoms is quite similar to that found after exposure to formaldehyde, a chemical
352
John C. Baird and Birgitta Berglund
prevalent in many modern building materials. Formaldehyde was also implicated as a causative agent by the fact that some of the buildings classified as 'sick' showed improvement after efforts were made to sanitize them against formaldehyde sources (bring air concentrations below 100#g m-3). On the other hand, some buildings showed no improvement after treatment. Starting with the assumption of a chemical (airborne) origin to the sick building problem, the psychophysical loop in Figure 1 was entered and the problem was temporarily removed from the environmental arena and brought into the laboratory, where a kind of psycho-chemical approach was instituted to determine more carefully human sensitivity to formaldehyde. One of the initial problems addressed was the great variability between studies reported in the literature on the absolute odor threshold, the value of which is important in setting limit values for formaldehyde exposure in non-industrial indoor environments. In the process of finding out that this variability was probably not due to intra or inter-individual differences, nor due to the method of making formaldehyde, a basic psychophysical result was obtained as well (Berglund, Berglund, H6gman, Johansson & Lindvall, 1987). Both a signal detection and free magnitude estimation paradigm were employed and it was found that the psychophysical function (between judgments and stimulus intensity) was consistent with the percent detection values obtained from the yes-no task. Such a result tends to validate alternative psychophysical methods and to demonstrate the internal consistency of the psychophysical paradigm, quite apart from its role in unraveling environmental problems. In related studies (Ahlstr6m, Berglund, Berglund, & Lindvall, 1984) it was found that mixing formaldehyde with indoor air had a large relative effect on perceived odor strength when the concentrations of formaldehyde were low, but that no substantial effects of such mixing were evident when the concentrations were high. This again is a result with implications both of a practical and theoretical nature, since chemical substances are seldom found alone in the natural environment, but rather, are usually present as part of a chemical mixture. This line of research did not, however, implicate formaldehyde in the sick building syndrome, since most of the concentrations found in the buildings were well below the obtained detection thresholds. The return to the environmental domain meant facing the same issue in a new light. Rather than look for a single chemical or chemically defined group of substances that might be responsible for the sick building syndrome we began thinking about the matter in terms of a pattern of many substances, and that possibly, this pattern is different in healthy and sick buildings. This raised a host of new environmental considerations. For example, how does one diagnose a sick building? What is the proper control case? And most importantly, how does one conduct pattern recognition on chemical patterns obtained from analysis of the air in buildings? Air samples were taken from a healthy and sick preschool and from a healthy office building on a number of occasions and then analysed by gas chromatography for volatile organic compounds, which were subsequently identified by mass spectrometry. In addition, subjects sniffed a fraction of the output eluate from the chemical analysis over time and judged its intensity by magnitude estimation. In this way it is possible to relate the magnitude estimate to the concentration of particular substances being judged. Results indicated that the judgement of chemical intensity showed no significant correlation with the chemical concentration (pbb) actually present. Some chemicals of high concentration are barely noticed, and others of low concentration are
Environmental Psychophysics
353
judged to be quite strong smelling (Berglund, Berglund, Lindvall & NicanderBredberg, 1982), a finding with importance for both psychophysics and human health. More recently, we performed a number of analyses to isolate particular patterns in both the chemical and judgement data. These included a number of techniques developed within the tradition of scaling and psychophysics, such as factor (component), cluster, and correspondence analysis (Berglund, Berglund, Lindvall & Nicander-Bredberg, 1982; Baird, Berglund, Berglund, Nicander-Bredberg & Noma, 1987; Noma, Berglund, Berglund, Johansson, & Baird, 1988). The indications from these analyses are that distinctive patterns of chemicals can be found in sick and healthy buildings, although the exact identification of chemical constituents has yet to be completed (approximately 40 have been positively identified as relevant so far). Beside distinguishing between different types of buildings these analysis techniques can find patterns of chemical substances that are characteristic of locations within buildings; for example, the pattern present for supply air entering a building, air after it goes through a ventilation system, exhaust air, etc. The impact of these studies is that a sort ofpsycho-chemics has emerged, combining psychophysical methods and analysis techniques (e.g., cluster analysis, factor analysis) with standard tests applied in analytic chemistry to identify the constituents of air (McLafferty, 1980; McNair & Bonelli, 1968). It is interesting to note that many of the scaling methods familiar to psychologists for over 20 years have only recently been discovered by chemists working on problems of air pollution (cf. Massart & Kaufman, 1983; Varmuza, 1980). When we consider the environmental loop in Figure ! in the broadest sense, it is now possible through an interdisciplinary effort to make some headway toward evaluating the impact of the built environment on human health. It is not a t all clear how this could be accomplished without the experimental and quantitative models of scaling and psychophysics. The implications for theoretical psychophysics are also powerful in that we are not dealing here with a simple dose-response relationship, similar to what are referred to as "psychophysical laws' (Bair & Noma, 1978; Stevens, 1975). Pattern psychophysics is not unidimensional, and as such, it suggests a re-evaluation of the multidimensional nature of the stimulus in psychophysics. It seems especially true, as with the case of cognitive mapping, that general perceptual theories will require a richer paradigm than the unidimensional approach so religiously followed throughout the history of psychophysics. Discussion Our thesis for environmental psychophysics does not plead for more applied work or for a psychophysics of complex stimuli. Nor do we advocate that environmental perception be reduced in some way to pure psychophysics. Rather it is that the growth of knowledge in either field;is inextricably tied to developments in the other. Although we feel that all phases of the bubble diagram of Figure 1 arc impacted to some extent by the two cases we have presented, the impact will be greater in some places than in others and the locus of such impact will depend on the specific examples being considered. Nonetheless, a few general principles can in fact be taken from the case studies of cognitive mapping and the sick building syndrome. (1) The problem you start with may not be the problem you solve. Kevin Lynch wanted to influence the design of new cities to make them more
354
John C. Baird and Birgitta Berglund
imageable for their residents. The map-sketching technique he created to document levels of imageability was streamlined by experimental psychologists to the point where subsequent tests of two-dimensional and one-dimensional knowledge of spatial relations led to theoretical contradictions at the very heart of psychophysics. (2) The problem you start with may not be the problem you end with. The original intent of introducing psychophysics into the picture for the study of sick buildings was that the perception of odorous substances might be used to distinguish between sick and healthy locations. The scaling methods adapted to do this turned out to be used primarily for the separation of locations based on the chemical constituents of air. What started out to be a psychophysical problem ended up being one of pattern recognition among chemicals in different environments--though the psychophysical work-continues as well. (3) The point of impact of new knowledge cannot be predicted with much certainty. One major implication of our thesis is that an individual or a research team must possess different kinds of training and expertise in order to function comfortably at different places in the bubble diagram. The impact of one field on another can only be seen by individuals (or a team) who have access-to information in both realms of knowledge and who can make the necessary connections across disciplines. This interdisciplinary promise of environmental psychology remains its greatest virtue, despite the difficulties to be overcome in securing its rightful position among the environmental sciences. The early textbooks on environmental psychology had two broad themes: perception of complex stimulus contexts, and the social interactions that occur within them (e.g., Bell, Fisher, & Loomis, 1978; Ittelson, Proshansky, Rivlin, & Winkel, 1974). The recently issued Handbook thoroughly updates the contributions of social psychology to the field, but in our opinion takes little note of the equally-significant progress in environmental perception and psychophysics.
References
Ahlstr6m, R., Berglund, B., Berglund, U. & Lindvall, T. (1986). Formaldehyde odor and its interaction with the air of a sick building. Environment International, 12, 289-295. Baird, J. C., Berglund, B., Berglund, U., Nicander-Bredberg, H. & Noma, E. (1987). Distinguishing between healthy and sick preschools by chemical classification. Environment International, 13, 167-174. Baird, J. C., Merill, A. A. & Tannenbaum, J. (1979). Cognitive representation of spatial relations. II. A familiar environment. Journal of Experimental Psychology: General, 108, 92-98. Baird, J. C. & Noma, E. (1978). Fundamentals of Scaling and Psychophysics. New York: WileyInterscience. Baird, J. C., Wagner, M. & Noma, E. (1982). Impossible cognitive spaces. GeographiealAnalysis, 14, 204-216. Bell, P. A., Fisher, J. D. & Loomis, R. J. (1978). Environmental Psychology. Philadelphia: Saunders. Berglund, B., Berglund, U., H6gman, L., Johansson, I. & Lindvall, T. (1985). Measurement of formaldehyde odor indoors. In CLIMA 2000, P. O. Fanger, Ed., vol. 4: Indoor climate, pp. 251-257. VVS Kongres-VVS Messe, Copenhagen. Berglund, B., Berglund, U., Lindvall, T. & Nicander-Bredberg, H. (1982). Olfactory and chemical
Environmental Psychophysies
355
characterization of indoor air. Towards a psychophysical model for air quality. Environment International, 8, 327-332. Berglund, B. & Lindvall, T. (1986). Sensory reactions to 'sick buildings'. Environment International, 12, 147-159. Boring, E. G. (1950). A History of Experimental Psychology. 2nd ed., Appleton-Century-Crofts, New York. Burroughs, W. J. & Sadalla, E. K. (1979). Asymmetries in distance cognition. Geographical Analysis, 11,414-421. Cadwallader, M. (1979). Problems in cognitive distance: implications for cognitive mapping. Environment and Behavior, 11, 559-576. Canter, D. & Tagg, S. K. (1975). Distance estimation in cities. Environment and Behavior, 7, 59-80. Craik, K. H. (1970). Environmental psychology. In K. H. Craik, B. Kleinmuntz, R. L. Rosnow, R. Rosenthal, J. A. Cheyne & R. H. Walters, Eds., New Directions in Psychology, 4, 1-122. Holt, Rhinehart & Winston. Ekman, G. & Bratfisch, O. (1965). Subjective distance and emotional involvement--a psychological mechanism. Acta Psychologica, 24, 446-453. Evans, G. E. Environmental cognition. (1980). Psychological Bulletin, 88, 259-287. Fechner, G. T. (1860). Elemente der Psychophysik. Breitkopf & Hdrtel, Leipzig. Foley, J. E. & Cohen, A. J. (1984). Working mental representations of the environment. Environment and Behavior, 16, 713-729. Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J. J. (1979). The Ecological Approach to Visual Perception. Boston: Houghton Mifflin. Golledge, R. G. (1987). Environmental cognition. In D. Stokols & I. Altman, Eds., Handbook of Environmental Psychology, vol. 1, 131-174. New York: Wiley. Golledge, R. G. & Zannaras, G. (1973). Cognitive approaches to the analysis of human spatial behavior. In W. H. Ittelson Ed., Environment and Cognition. New York: Seminar Press. Heft, H. & Wohlwill, J. F. (1987). Environmental cognition in children. In D. Stokols & I. AItman, Eds., Handbook of Environmental Psychology, vol. 1, 175-204. New York: Wiley. Ittelson, W., Proshansky, H., Rivlin & Winkel, G. (1974). An Introduction to Environmental Psychology. New York: Holt, Rinehart & Winston. Kirasic, K, C., Allen, G. L. & Siegel, A. W. (1984). Expression of configurational knowledge of large-scale environments: students' performance of cognitive tasks. Environment and Behavior, 16, 687 712. Kfiller, R. (1987). Environmental psychology from a Swedish perspective. In D. Stokols & I. Altman, Eds., Handbook of Environmental Psychology, voL 2, 1243-1279. Lynch, K. (1960). The Image of the City. Cambridge, Mass.: MIT Press. Magana, J. R., Evans, G. W. & Romney, A. K. (1981). Scaling techniques in the analysis of environmental cognition data. Professional Geographer, 33, 294-301. Massart, D. L. & Kaufman, L. (1983). The Interpretation of Analytic Chemical Data by the Use of Cluster Analysis. Wiley Interscience, New York. McLafferty, F. W. (1980). Interpretation of Mass Spectra. 3rd ed., Mill Valley, California, University Science Books. McNair, H. M. & Bonelli, E. J. (1968). Basic Gas Chromatography. Palo Alto, California, Varian. Noma, E., Berglund, B., Berglund, U., Johansson, I. & Baird, J. C. (1988). Joint representation of physical locations and volatile organic compounds in indoor air from a healthy and a sick building. Atmospheric Environment, 22, 451-460. Sherman, R. C., Croxton, J. & Giovanatto, J. (1979). Investigating cognitive representations of spatial relationships. Environment and Behavior, 11, 209-226. Siegel, A. W. (1982). Methodological and metatheoretical issues in the development of cognitive mapping of large-scale environments. In J. C. Baird & A. D. Lutkus, Eds., Mind Child Architecture. Hanover, N. H.: University Press of New England. Stea, D. (1982). Cross-cultural environmental modeling. In J. C. Baird & A. D. Lutkus, Eds., Mind Child Architecture. Hanover, N. H.: University Press of New England. Stevens, S. S. (1975). Psychophysics: Introduction to its Perceptual, Neural, and Social Prospects. New York: Wiley-Interscience. Tell, W. (1983). Hiilsofarliga ByggnadsmateriaL VVS Special No. 2, 43-46.
356
John C. Baird and Birgitta Berglund
Varmuza, K. (1980). Pattern Recognition in Chemistry. Berlin: Springer-Verlag. WHO (1983). Indoor Air Pollutants: Exposure and Health Effects. Copenhagen,Denmark: World Health Organization. EURO Reports and and Studies, 78. Acknowledgments This research was supported by the Swedish Council for Building Research and the Swedish Council for Research in the Humanities and Social Sciences. The paper was written while John C. Baird was a visiting professor at the Department of Psychology, University of Stockholm, and the Department of Hygiene of the National Institute of Environmental Medicine. Reprint requests should be addressed to Professor Birgitta Berglund, Department of Psychology, University of Stockholm, S-106 91 Stockholm, Sweden.