Effects of Noise, Heatand Indoor Lighting on Cognitive Performance and Self-reported Affect

Theoretical and practical concerns guided the design of an experiment on how ventilation noise (38 and 58 dBA), air temperature (21 and 278C), and illuminance (300 and 1500 lx) combine or interact in their e¡ects on cognitive performance. Self-reports of a¡ective states were taken with an a¡ect circumplex measure (Larsen & Diener, 1992; Knez & Hygge, in press) to study the mediation from the environmental variables over a¡ect to cognitive performance. Arousal models (e.g., Broadbent, 1971) would predict that increased levels of noise and illuminance increase activation and/or a¡ect levels and that mild heat decreases it. The inverted U-hypothesis would further predict that intermediate levels of perceived arousal improve attention, memory and problem solving performance. A distinction was made between synergetic and antagonistic interactions in order to di¡erentiate arousal and nonarousal mediated e¡ects on cognitive performance. The results showed that attention worked faster in noise but at the cost of lesser accuracy, which supports the Speed-Accuracy-Trade-O¡ hypothesis (Hockey, 1984). Interactions were found between noise and heat on the long-term recall of a text, and between noise and light on the free recall of emotionally toned words. These e¡ects on cognitive performance could not be explained as mediated by the a¡ective states, and were not consistent with an arousal model and the inverted-U hypothesis. # 2001 Academic Press E¡ects of Noise, Heat and Indoor Lighting on A¡ect and Cognitive Performance There is both a theoretical and a practical value in knowing how the physical parameters of the indoor environment may combine or interact in producing e¡ects on a¡ect and a cognition. Broadbent (1971) phrased the basic idea behind a synergetic (ordinal) interaction, where the combined e¡ect is more than the sum of its parts, in the following way: If condition A gives a l0 per cent increase in errors, and condition B a 10 per cent increase, then the two together should give 20 per cent. If on the other hand both stresses are a¡ecting the same mechanism, the more drastic impairment may well appear: If each condition on its own lowers performance from a perfect level to one of 10 per cent error, then the addition of the second stress to the ¢rst may produce far more than 20 per cent of error. This is to be expected because the e¡ect of one stress alone will be partly taken up in overcoming the safety margin of the mechanism concerned. When the second stress is applied, there is no longer any margin left within the mechanism, which is being effected. (p. 405 f.) The theoretical signi¢cance of the argument can be probed also for situations where the combination of conditions A and B cancel each other to produce no net change in result (antagonistic interaction, cross-over interaction). Such a result would indicate that the two conditions antagonistically counteract each other at some level of analysis, which is di¡erent from a synergetic interaction indicating a depletion of available processing resources. An arousal model in combination with the inverted-U hypothesis is the major tool to explain a counteracting mediation between environmental in£uences and performance (Broadbent, 1971; Easterbrook, 1959; Hebb, 1972; Lindsley, 1951; Malmo, 1959). The empirical support for the two kinds of interactions is scarce since there are very few studies 292 Sta¡an Hygge and Igor Knez devoted to interactive e¡ects of noise, temperature and indoor lighting on intellectual work of the kind that takes place in schools or o⁄ces. There is no study that varied all the three parameters noise, heat and lighting simultaneously. The most common combination of variables has been that between noise and heat, and the second most common between noise and lighting. Viteles and Smith (1946) reported no interaction between noise (72, 80 and 90 dB fan noise) and heat (corresponding to a range of 22 7^34 48C, corrected for humidity), on psychomotor and attention tasks. However, in a re-analysis of the same data Wilkinson (1969) stated that there was an antagonistic interaction between noise and heat. Wyon et al. (1978) reported an antagonistic interaction between noise (85 dBA industrial noise, 50 dBA quiet) and heat (22 and 308C) on a ¢ve-choice serial reactions task, and on a creativity test (noise from children playing, temperatures 20 0, 23 5, 27 08C). Hygge (1991) also reported an antagonistic interaction between continuous fan noise (37 and 58 dBA) and temperature (20 and 278C) on a problem-solving task (embedded ¢gures), but there were no interactions or main e¡ects for other tasks. Hancock and Pierce (1985) reviewed 13 studies where both noise and heat were independent variables and concluded that the majority of studies neither showed synergetic nor antagonistic interactions. Veitch (1990) studied the e¡ects of illuminance (200, 400, and 600 lx) and intermittent o⁄ce noise (model levels of noise 50 and 70 dBA) on reading comprehension (recognition), but found no interactions or main e¡ects. L ̨fberg et al. (1975) varied illuminance (60, 250 and 1000 lx) and temperature (corresponding to 22 and 278C) for school children. One of their tasks was an addition test written on paper sheets with di¡erent contrasts. For that test there was an interaction between illuminance level, temperature and time-of-day. The high illuminance level improved performance in the warmer condition in the afternoon. In the morning at the lower temperature, performance on the addition test improved with increased illuminance. Nelson et al., (1984) varied air temperatures between 13, 23 and 308C and illuminance between 100 and 300 lx, and reported productivity increases in cool air but no interaction with illuminance. Thus, there is some support for an antagonistic interaction between noise and mild heat, meaning that on attentional and problem solving tasks an increased noise level can be counteracted by a slight increase in temperature. There is no support for an antagonistic interactions between noise and pronounced heat, or illuminance and noise, or illuminance and heat. To account for antagonistic interactions, the arousal model needs to be supplemented with the inverted-U relationship (Broadbent, 1971; Easterbrook, 1959; Hebb, 1972; Lindsley, 1951; Malmo, 1959) between arousal and performance. Performance is assumed to be at its peak in the region of intermediate arousal. For easy tasks the optimum level of performance is in the region of high arousal; for di⁄cult tasks the optimum is on the lower arousal regions. Too low arousal impedes performance by drowsiness and too high arousal produces impaired performance by over-activation. Increases in noise and illumination levels have been assumed to increase arousal, and mild heat (up to *278C) to decrease it. Mere depletion of cognitive processing resources in the mediation of performance would show up as a synergetic interaction, and counteracted arousal as an antagonistic (statistical) interaction between environmental variables in their e¡ects on performance. If an interaction is present, concurrent measures of arousal, a¡ect or activation would provide information about their plausibility as mediators of behavior and cognition. A synergetic interaction between environmental variables in their e¡ects on cognitive performance does not necessitate an inverted-U arousal model. An antagonistic interaction, however, would. Arousal models have been severely criticized. The uni-dimensional and general nature of the arousal concept has been questioned. The hopes to ¢nd physiological correlates of arousal have so far not been successful. Already three decades ago Broadbent (1971, p. 413) explicitly suggested that ‘in the present state of knowledge we are de¢ning (the arousal concept) on the basis purely of behaviour and not of physiology’. Other theorists denounce activation theories and suggest that there are no interactions between e.g., noise and heat (Hockey & Hamilton, 1983, pp. 383 ¡.). Still others e.g., Cohen et al. (1986, pp. 161 ¡.) propose that so called arousal e¡ects rather are decisional or perceptual in nature. The present experiment was designed to study the interaction e¡ects between noise, heat and illuminance levels on attention, memory and problem solving. To evaluate a¡ective states as mediators between environmental impact and performance, a self-report a¡ect circumplex measure was employed (Larsen & Diener, 1992; Knez & Hygge, in press). E¡ects of Noise, Heat and Indoor Lighting 293 It was predicted that increased levels of noise and illuminance lighting would increase perceived activation level and that mild heat would decrease it. In line with the inverted U-hypothesis, intermediate levels of perceived activation were predicted to improve attention, memory and problem solving. In accordance with the Speed-Accuracy-Trade-O¡ (SATO) hypothesis (Hockey, 1984), noise was expected to increase speed at the expense of more errors in attention and working memory.