Seasonal Cognitive Rhythms within the Arctic Circle: an Individual Differences Approach

At 698N the annual swings in the physical environment are considerable. For instance, for two months in winter there is no direct sunlight and for two months in summer the sun does not set. Brennen et al. (1999) tested the cognitive performance of 100 participants living at 698N in summer and in winter. Overall there were very few seasonal e¡ects, and most of these were, contrary to expectation, winter advantages. In order to determine whether particular subgroups of the sample may have shown winter de¢cits, in this paper Brennen et al. ’s database is analysed to investigate whether the age, gender or the proportion of life spent in northern Norway predicts the di¡erence between performance in summer and winter on each test. The multiple regressions showed that age was a signi¢cant predictor on ¢ve tasks, gender on one task, and the proportion of life lived in northern Norway on none. Overall this analysis is in line with Brennen et al. ’s conclusion: there are only very modest seasonal e¡ects in cognition. # 2001 Academic Press In this paper an individual di¡erences approach is used to investigate whether the extreme annual variation in the physical environment of polar regions in£uences cognitive peformance. Studies of mood variations over the annual cycle report a sizeable minority of people who have impaired social and affective functioning in winter compared to summer, the syndrome referred to as Seasonal A¡ective Disorder, or SAD (see for example Rosenthal et al. 1984). On the other hand, until recently there have been almost no studies of cognitive performance over the annual cycle. There are several reasons for which it is timely to investigate possible cognitive rhythms over the annual cycle in a nonclinical population living at a very high latitude. Firstly, it is widely claimed that there is a cognitive component in SAD and the cognitive performance of SAD patients has recently begun to be investigated, e.g. Drake et al. (1997), Hodges and Marks (1999). Secondly, currently available studies of the psychological correlates of the seasonal rhythm have largely used self-report techniques, in contrast to objective measures of performance. Thirdly, there are intuitive and theoretical reasons for expecting psychological seasonal swings to increase with latitude. It seems obvious that places with big di¡erences in amount of daily light over a year should have strong psychological e¡ects compared to places on the equator where length of night and day vary minimally. Furthermore, current accounts of the aetiology of SAD claim that the reduction in amount of daylight triggers the mood disorder. Thus this hypothesis predicts that people at high latitude should be particularly vulnerable to the manifestations of SAD. Since the golden age of polar exploration the effects of ‘wintering-over’ on Antarctica have been documented. In 1898^9 the crew of the Belgica became the ¢rst people to winter over in Antarctica. Their boat became trapped in the ice for 13 months during a scienti¢c expedition, and the doctor reported that homesickness and mental disturbances were widespread symptoms amongst the crew. In the century since, many more expeditions have wintered over on the Antarctic continent, and the symptoms of depression, irritability, insomnia and cognitive impairment have been frequently reported (Edholm & Gundersen, 1973). Palinkas, Houseal and Rosenthal (1996) reported that while the rate of clinical depression and SAD amongst ‘winter-over’ sta¡ are low, subsyndromal symptoms of SAD increase across the winter. FIGURE 2 The University of TromsÖ campus in winter. 192 TIM BRENNEN Taylor and Duncum (1987) investigated how performance of a mental imagery task varied in people on Antarctica, and reported no objective di¡erences over the year, despite the participants’ own intuitions that they were performing worse in winter. Palinkas and Houseal (2000) tested ‘winter-over’ sta¡ monthly throughout the winter at three Antarctic research stations, using a questionnaire that measured anxiety, depression, anger, confusion, fatigue and vigor. The scores got better up to and beyond the winter solstice, before getting worse in the spring, and this e¡ect was larger overall for participants at the South Pole. The present paper reports the extension and reanalysis of a database collected by Brennen et al. (1999). In the initial study the cognitive performance of a population living in TromsÖ, northern Norway was tested. The city of TromsÖ is at 698N, 300 km north of the Arctic Circle. At this latitude, the sun remains below the horizon between mid-November and mid-January and doesn’t set between mid-May and mid-July. TromsÖ provides a useful natural laboratory for the study of circannual rhythms because, despite its very northerly location, it is a lively city, in contrast to the impoverished social environment and restricted daily routine of polar research stations, for instance. TromsÖ’s population is 60,000 which, while small by international standards, puts it among Norway’s ten biggest cities. Figures 1 and 2 show the university campus in summer and winter. Each subject in the original study was tested twice, once in winter and once in summer and about half of the subjects was tested ¢rst in winter while the rest were tested ¢rst in summer. Brennen et al. (1999) used the classic experimental analysis, collapsing data across subjects and running a two-way FIGURE 1 The University of TromsÖ campus in summer. analysis of variance with the factors of season and order of testing, separately for each test. On a battery of tests including explicit and implicit memory for words, verbal £uency, face recognition, and short-term digit memory, the predicted global winter de¢cit did not emerge. On the contrary, of the ¢ve signi¢cant results, four indicated summer de¢cits. In this paper an individual di¡erences approach is taken to determine whether some subgroups of subjects do in fact show the expected winter disadvantage. This reanalysis thus provides another opportunity for seasonal cognitive rhythms to be detected. Another possibility is that even if the independent variables do not reveal a subset of the sample with winter de¢cits, they will demonstrate how the summer de¢cits vary. The independent variables to be used in the regressions are gender, age and the proportion of one’s life lived in northern Norway. Several epidemiological studies have shown that women are more susceptible than men to SAD, (e.g. Thompson & Isaacs, 1988; Kasper et al., 1989, though see Kane & Lowis, 1999 for a contrary view). It might be argued that since Brennen et al. ’s study included over 40 women it was well placed to detect any seasonal cognitive e¡ects even if they should only be observed for women. In addition, Brennen et al., (1999) argued that with a mean age of 31 years old (S.D. = 10) the sample should be highly sensitive to seasonal changes, as determined by the typical age range of SAD patients. So, on the one hand, Brennen et al. ’s sample was of appropriate age and gender to pick up any seasonal e¡ects, but on the other hand, it is possible that any seasonal e¡ect in particularly seasonal subgroups were swamped by null e¡ects in the rest TABLE 1 Characteristics Of The Sample Of Participants