You are on page 1of 18

Organizational Behavior and Human Decision Processes Vol. 78, No. 2, May, pp. 128145, 1999 Article ID obhd.1999.

2827, available online at http://www.idealibrary.com on

One Night of Sleep Loss Impairs Innovative Thinking and Flexible Decision Making
Y. Harrison and J. A. Horne
Sleep Research Laboratory, Loughborough University, Loughborough, United Kingdom

Recent findings with clinically oriented neuropsychological tests suggest that one night without sleep causes particular impairment to tasks requiring flexible thinking and the updating of plans in the light of new information. This relatively little investigated field of sleep deprivation research has real-world implications for decision makers having lost a nights sleep. To explore this latter perspective further, we adapted a dynamic and realistic marketing decision making game embodying the need for these skills, and whereby such performance could be measured. As the task relied on the comprehension of a large amount of written information, a critical reasoning test was also administered to ascertain whether any failure at the marketing game might lie with information acquisition rather than with failures in decision making. Ten healthy highly motivated and trained participants underwent two counterbalanced 36 h trials, sleep vs no sleep. The critical reasoning task was unaffected by sleep loss, whereas performance at the game significantly deteriorated after 3236 h of sleep loss, when sleep deprivation led to more rigid thinking, increased perseverative errors, and marked difficulty in appreciating an updated situation. At this point, and despite the sleep-deprived participants best efforts to do well, their play collapsed, unlike that of the nonsleep-deprived partici1999 Academic Press pants.

Many people have to work through the night and often into the next day, with little or no sleep. This is common among personnel in the emergency services, industrial control rooms, hospitals, and the military. If the individuals are employed in a management or other decision-making capacity, then on
This project was funded by the Wellcome Trust, UK. Address correspondence and reprint requests to Dr. Harrisons new address: Centre for Applied Psychology, Liverpool John Moores University, Henry Cotton Campus, 15-21 Webster Street, Liverpool L3 2ET, UK. Fax: 44(0) 151 2314033. E-mail: humyharr@livjm.ac.uk. 128
0749-5978/99 $30.00 Copyright 1999 by Academic Press All rights of reproduction in any form reserved.

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

129

occasions they will be required to demonstrate flexible, innovative(divergent) thinking and the ability to revise plans in the light of new information. This may be particularly necessary during crises and unforeseen events. While the adverse effects of sleep loss on low-level skills that are dull and monotonous, such as vigilance tests, simple or complex reaction time, are well known (c.f. Horne, 1988; Dinges & Kribbs, 1992; Wilkinson, 1992; Pilcher & Huffcutt, 1996), very little is known about how these higher level behaviors would be affected. Some complex behaviors are relatively impervious to the effects of up to 48 h of sleep loss, possibly because they are stimulating, interesting, and motivate people to apply compensatory effort. This seems to be the case with complex tasks of logical, concrete reasoning (e.g., Haslam, 1981; Percival, Horne & Tilley, 1982; Angus & Heslegrave, 1985; Horne, 1988b; Blagrove, Alexander & Horne, 1995) including IQ-type tests (Wilkinson, 1965). Increased motivation also counteracts detriments in the simple, monotonous vigilance and reaction time tasks, well into a second night without sleep (Horne & Pettitt, 1985). In contrast, there is accumulating evidence that normal subjects are impaired on tests of divergent, innovative, flexible thinking and strategic planning after just one night without sleep (Herscovitch, Stuss, & Broughton, 1980; Horne, 1988b; Wimmer, Hoffman, Bonato, & Moffitt, 1992; Harrison & Horne, 1997). Subjects also perseverate and fail to revise plans in the light of new information (Herscovitch et al., 1980). Such tests are short and stimulating and in these respects might not be expected to be sensitive to sleep loss. Another factor underlying many sleep loss studies is that the conventional, laboratorybased approach to sleep loss usually has no bearing on real-world tasks. As one of the few exceptions, Banderet, Stokes, Francesconi, Kowal, and Naitoh (1981) reported an extended battle simulation involving experienced military personnel in highly motivated five-man cooperative teams. Subjects had difficulty in coping with this task from around 36 h and were forced to give up after about 45 h without sleep. Termination of the game was due to participants not being able to cope with unforeseen, rapid changes in the task and a failure to update and revise information in the light of new events. A second study, by May and Kline (1987), monitored performance across a range of cognitive tasks in military personnel during 2 nights of sleep loss. Participants were able to cope well when required to make logical responses to familiar tasks, whereas innovative thinking and the generation of spontaneous ideas deteriorated markedly. Neither of the two studies focused on these important cognitive failures, but merely described them in a qualitative and operational way and never followed up the findings. Another area of interest, ostensibly having real-world implications, concerns sleep deprivation among junior hospital doctors, who experience sleep loss on a routine basis. Although investigations often focus on cognitive processes that have little to do with the true nature of the job, medical-oriented performance tasks relying on high level cognitive skills have been found to deteriorate after 36 h without sleep (Friedman, Bigger, & Kornfeld, 1971, 1973; Pinnock,

130

HARRISON AND HORNE

Derbyshire, Johnstone, & Smith, 1985; Deary & Tait, 1987; Denisco, Drummond, & Gravenstein, 1987; Leung & Becker, 1992). Of particular interest, Goldman, McDonough, and Rosemond (1972) found that sleep-deprived junior doctors were more hesitant and showed less focused planning ability during an unfamiliar surgical procedure. More recently, Nelson et al. (1995) found decrements in innovative (divergent) thinking for on-call anesthetists following a night of acute sleep loss (less than 30 min sleep in total). This form of thinking is essential for the effective handling of many complex situations. In contrast, skills requiring the assimilation of large amounts of complex information that is more familiar and routine, and less reliant on divergent thinking, are not so sensitive to one night of sleep deprivation (SD), as was shown by Brown et al. (1994) who tested sleep-deprived hospital doctors and medical students for comprehension of detailed and lengthy surgical journal articles. The main aim of the present study was to examine the effects of one night of sleep loss on complex, innovative decision making within the real-word context of a business game, where performance was carefully qualified and quantified. We utilized a competitive, professional game designed to train postgraduate students of business studies and management personnel (MasterplannerSaunders, 1989). The game is both realistic and intrinsically interesting for the participants (i.e., it motivates them to perform at their best). It is a multistage, dynamic planning task requiring an interaction between learned skills necessary for the routine playing of the game and a more innovative capacity to respond to unfamiliar and unusual scenarios. The aim of the game is to make a large profit by manipulating various factors; the more innovative the play, the greater the potential gain. Integral to this game is a continuing need to update oneself on and remember changing events. Participants play for 1-h sessions, but the game is continuous, as each player takes up from where he/she last left off. Progressively, the game becomes more complex as there is more to gain and to lose. We wished to see how performance was affected by sleep loss and, in particular, to ascertain if players were able to avoid serious losses resulting in insolvency and the termination of the game. Before players can plan their responses, Masterplanner involves the reading and assimilation of much factual material (see below). As a benchmark to whether such abilities are impaired by one night of sleep loss, we also administered a variant of the critical reasoning component of the Graduate Management Admission Test (GMATJaffe, 1989), which is based on concrete, logical thinking. It required participants to make subtle distinctions between statements on the basis of the reading and understanding of detailed and extended sections of text. We appreciate that one cannot equate Masterplanner and GMAT in terms of difficulty, and direct comparisons between the two tasks are not being made. GMAT is used here as a guide to whether our participants could assimilate complex written material. The literature indicates that this would not be impaired by one night of sleep loss. On the other hand Masterplanner might well be affected even in highly motivated people. In particular, we were interested in those components requiring innovative and flexible thinking,

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

131

which could be ascertained through measuring profitability, rigidity of response style, and perseverative errors (see below).
METHOD

Participants Ten graduate volunteers (six males, four females, average age 21 years) were selected from respondents to posters around the university. Screening excluded those having sleep difficulties, who napped regularly in the daytime, and were on medication that affected sleep/sleepiness. All were healthy nonsmokers, who normally slept 78 h at night, with no history of epilepsy or alcohol abuse. They were randomly allocated to two groups of five persons, with one group undergoing the order SD then nonsleep deprivation (NSD) and the other vice versa. Volunteers were paid generously to participate and were exhorted to put as much effort as possible into game playing. An additional financial incentive was given to the best performer for each of the four sessions (see below). Design and Procedure The volunteers participated twice in a counterbalanced design, with and without a night of sleep. Each condition commenced 0900 h on Day 1 and was intended to run until 23:00 h the following night. The interim night was either with sleep (NSD) or without sleep (SD). Actimeters and sleep diaries were used for 3 nights prior to each testing session and on the interim night. These devices help to ensure that participants slept normally during these nights (Harrison & Horne, 1996). Lights out was at 23:30 h and rise time at 07:30 h (i.e., beginning of the SD period). During SD, participants were confined to the laboratory and were supervised throughout the whole period. Between testing sessions they remained in a large lounge, watched TV, read, played board games, and talked. Noncaffeinated beverages were available ad lib, and meals were provided at 08:00, 13:00, and 19:00 h. Snacks (e.g., fruit and toasted sandwiches) were available at other times. At the end of SD they were escorted home. During NSD, participants followed the same routine, with the exception that they were escorted home at 23:00 h on Day 1 and instructed to retire to bed and remain there until 07:30 h the following morning (Day 2). Actimeters were again used to ensure compliance. Participants arrived back at the laboratory at 08:30 h and continued with the experimental schedule until termination that night. Masterplanner was scheduled for six periods at 13:30, 19:30, and 22:00 h on both days. Trial times were identical on both days to control for the influence of potential confounds related to circadian changes in mood and subjective arousal. With this in mind, testing was scheduled during afternoon and evening periods, rather than during the morning, in order to target monitoring toward the end of the 36 h period of sleep loss on Day 2. However, during SD, participants were unable to complete the final session (see below). There were no early morning sessions (between midnight and 08:00 h) as these could not be undertaken during NSD without disrupting normal sleep.

132

HARRISON AND HORNE

GMAT: Critical reasoning. Participants were required to make subtle distinctions between written statements on the basis of the participants understanding of a detailed and extended section of text (Jaffe, 1989). Participants had to evaluate the compatibility between statements on the basis of their veracity. The test comprised 20 questions and lasted for 30 min. Tests were performed once per day, at 17:00 h. Four versions were available, presented in a counterbalanced order across Day 1 and Day 2 in both the SD and the NSD conditions. Participants sat in a quiet room throughout the test. Masterplanner. This is a computer-based simulation of a competitive marketing environment (Saunders, 1989). Players act as individual directors for a hypothetical company. The aim is to promote sales for a single, newly developed product and ultimately achieve market dominance and a substantive profit. Alternatively, one can go bankrupt. During each experimental run, Masterplanner was played within the two groups of five individuals (SD and NSD), with each group being independent. Players were in direct competition with the four other players for their share of a finite number of potential sales. Debriefing sessions and pilot studies indicated that the game was both stimulating and enjoyable for the players. The key features of this task are described as follows: Players responsibilities. The simulation takes place over a series of five to six related sessions in which the consequences of decisions taken during the previous session represent the starting point for the next. Players responsibilities and decisions cover 11 cost centers aimed at enhancing company effectiveness (i.e., optimizing sales while minimizing error such as overproduction or overstaffingsee Appendix 1). While most decisions are geared toward the direct marketing of the product, the remainder cover essential financial or production activities. Play sessions and feedback. Although the game is run on a computer, players do not interact directly with it, but receive at the start of each session substantive amounts of information via paper output in the form of individual Company Reports. Previous work has shown (Saunders, 1989) that players find it easier to assimilate this and other market information in this manner. It allows them to make notes and do calculations on the paper, rather then keep scrolling the computer screen. The Company Report covers: (a) a breakdown of the most recent spending decisions for all companies; (b) a breakdown of all sales for the preceding trading period shown as percentage sales per company; (c) Individual company accounts. Note that while Sections a and b provide information covering all companies, Section c provides exclusive access to accounts information relevant to that particular player. Hence, it is possible to dominate the market in terms of sales and for this to be public knowledge (Section b), although, unbeknown to other players, a player can be bordering on bankruptcy due to reckless overspending. Having taken in the latest information the player makes decisions and writes these on preprinted forms which are returned to an independent assistant who enters this routinely into the computer. The assistant does not interact

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

133

with the players other than by acting as a post-box/passive interface between the player and computer. At the first session each player starts with an identical product and a fixed amount of capital. Each session comprises three stages (the time limits for each stage were determined by pilot studies): Stage 1. This stage comprised the performance feedback (except for session 1). Twenty-five minutes were allowed for players to evaluate Company Reports, whereby they could assess the effects of previous decisions and the current company status in relation to competitors. Stage 2. Decisions had to be made about 11 parameters covering, for example, production levels, advertising expenditure, research costs, shareholders dividends (Appendix 1). A total of 15 min was allowed for this stage. Stage 3. Participants had 20 min to provide a confidential written report covering an evaluation of their previous performance, current status, and future play strategy. The report style was open-ended, although, there were guidelines. Throughout all trials players ware seated at fixed distances around a large table to ensure that individual papers remained confidential. They were instructed to try their best and informed about the incentives on offer at the start of the exercise. Discussion between players was permitted and monitored. However, players were under no obligation to disclose details of their own performance to competitors, and we stressed that any information volunteered by players could not be relied on to be truthful. This approach helped to ensure that players didnt simply give up if they felt their own performance to be deteriorating, as others might be in a worse position. In the event, no blatant attempts to mislead other players by offering false information were observed. An environment of cross-talk and discussion is typical of many modern work environments, and we observed no obvious difficulties in concentration with this level of background distraction, and all sections of the task were completed within anticipated time frames, without prompting or intervention from the experimenter. The Product Life Cycle (PLC). The Masterplanner simulation is influenced to a great extent by the concept of the Product Life Cycle (PLCsee Appendix 2), which emphasizes the dynamic nature of product appeal whereby, at any particular point in the cycle, potential buyers are described in terms of their sensitivity to a range of factors such as price, quality or advertising strategy. The PLC covers a simulated period of 18 months and is modeled from product launch to market saturation. Three types of potential customers are identified in terms of buying behaviours: (1) Innovatorsdrawn to new products or technology; (2) Imitatorsinfluenced by trends; (3) Repeatersmaking second or third purchase of same product. To enable players to see these issues in more realistic terms they were given as an example a breakthrough development in home TV/video recording. A key factor in successful play at Masterplanner is to be able to predict and target the current dominant category of potential buyer by making timely adjustments to marketing strategies. For example,

134

HARRISON AND HORNE

sales to innovators are greatly influenced by high initial investment in advertising, but as the game progresses sales to innovators are reduced, and hence the payoff for advertising declines. At this stage, and in order to attract sales to imitators, it is more worthwhile to offer deals such as sales promotions which, in effect, lower prices. Players are able to predict the changing influence of customer categories, and players successes or otherwise at targeting the appropriate category is provided by the individual Company Reports, distributed at the start of each session. A general introduction to the PLC model is provided during training and is incorporated into a series of guidelines appropriate to each session (see Appendix 2). Training. Players were prepared for Masterplanner through a series of intensive and extensive training sessions lasting around 45 h in total; these allowed players to understand the key concepts, rules and guidelines. During training, an experienced instructor was on hand to deal with questions, or the need for further clarification. Throughout the structured sessions, all players were introduced to the same ongoing changes to the task. Guidelines. The complexity, dynamics and range of information presented to players during each session was considerable. Success relied on each players ability to monitor performance relative to three influential factors: (i) the consequences of their own previous decisions; (ii) the actions of competitors; and (iii) the properties of the computer simulation, incorporating basic marketing guidelines such as the PLC. These guidelines, which were utilised throughout the extensive training, predicted the likely impact of increasing or reducing spending for each of the 11 spending areas, for each play sessions (see Appendix 2). If the guidelines were strictly adhered to, then the player would survive, but to do well the player had to be innovative and apply the guidelines more flexibly. Outcome measures. Success or failure depends on the normal rigours of changing market forces generated by the computer. The game becomes more involved as play continues, and several parameters can be measured (see below). Masterplanner is a high level real-world simulation that can not provide simple measures of psychological performance (e.g., speed, accuracy, working memory). Players adopt their own style of playing, which may change with the game. Individual decisions and responses are not timed, as the measurement criteria are pragmatic, being based on what the player achieves. Masterplanner is an unorthodox approach to the measurement of the psychological effects of sleep loss. Nevertheless, we considered it to be a valuable tool owing to its realism and being highly motivating for the player. Its increasing difficulty as play progressed placed participants under pressure to apply more effort, particularly during SD. With successive sessions there is a predictable decline in profitability for all players due to increasing difficulty in securing a place for their product in a saturated market and in an increasingly critical situation. We structured our analyses of the main findings by addressing the following questions:

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

135

1. What are the overall changes in performance across the game, given the anticipated pattern of decline toward the final sessions? How was this affected by SD? 2. Can differences in performance between the two conditions be related to differences in response style? In particular, were participants more rigid in their decision making during SD? Were they able to update their responses appropriately in the light of fresh information? Alternatively, were they more likely to perseverate in their responses, i.e., inappropriately repeat a previous response? The two key indices for measuring performance at Masterplanner, which we shall be reporting, are: 1. Profitability: The percentage of all generated income that is retained as profit after all expenditure has been accounted for. It indicates the ability to balance anticipated gains against appropriate, rather than reckless spending. 2. Production errors: Generated by the Masterplanner simulation, indicating the ability to anticipate fluctuations in product demand on the basis of readily available information. Errors are measured as misjudgments in terms of production accuracy. The scale is arbitrary and includes both under- and overestimations of product demand. We also assessed participants innovation and flexibility by measuring repetitiveness in overall spending behavior, response shifting in accordance with established pretrained guidelines, and subjective reports. For both profitability and production errors, data over Sessions 2 to 5 in NSD and SD conditions were analyzed by repeated-measures ANOVAs using two factors: condition (NSD vs SD) and trials. Significance levels were corrected by adjusting the degrees of freedom using the GreenhouseGeisser epsilon. A two-factor ANOVA (condition day) was utilized for their critical reasoning test.
RESULTS

GMAT: Critical Reasoning This was given just before the fifth Masterplanner session, but was not affected by SD, with scores remaining similar for both days and under both conditions (Fig. 1). The ANOVA revealed no significant main effects or interaction. This suggests that SD participants were able to assimilate the factual material generated by Masterplanner but, as will be seen, they were unable to use this material in an innovative and constructive way. Masterplanner By the fifth session of SD (20:00 h) most (n 8) participants became insolvent or were very near to insolvency, and it was decided to terminate play for all participants at this point, although NSD participants could have continued successfully to the sixth and final session. This was a clear finding during the

136

HARRISON AND HORNE

FIG. 1. Mean (SD) scores on the Critical Reasoning task across SD and NSD conditions (n 10). There were no significant differences between Days 1 and 2 for either SD or NSD conditions.

first run of the study and the sixth session was aborted for the second run. There were no discernible differences within conditions for order effects. That is, the following findings were consistent regardless of whether the participants underwent SD before NSD or vice versa. Throughout all sessions, including the fifth, participants took an active interest in the game, paying particular attention to their progress, and compared this with that of other players. For both profitability and production errors, performance on day one was similar for SD and NSD, as one might expect (Figs. 2 and 3). It should be noted that as Session 1 is the starting point, these two indices do not become measurable until Session 2. Beyond Session 3 both measures reflected the expected decline due to an increasingly demanding task (see Method). Nevertheless, for both indices and by the final session there were clear differences in performance decrement between the two conditions, whereby profitability during SD changed to a large loss and insolvency, and production errors became catastrophically large. However, even in Session 4 there were developing trends in a greater performance decline during SD (Figs. 2 and 3). It should be noted that profitability for the SD group was marginally higher (nonsignificantly) for the previous (Day 1) session and, because of this, the subsequent fall to Session 4 was greater than for the NSD group. This can be seen in Fig. 2, as can the substantial fall in profitability from Sessions 3 to 4, in the SD group. Profitability revealed no significant effect of sleep condition, a significant main effect of session (F 9.39 [27,3], p .001), and a significant interaction between sleep condition and session (F 2.92 [27,3], p .05). It is apparent from Fig. 2 that this interaction was due to a sudden decline in performance at Session 5 only during the SD condition. Post hoc paired t tests between NSD

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

137

(n

FIG. 2. Mean (SD) Profitability (% of income retained as profit) for SD and NSD conditions 10). For session 5, profitability was significantly less during SD compared with NSD (see text).

and SD for this session revealed significant differences in profitability (t 1.73 [9], p .05). Production errors showed a significant condition effect (F 8.1 [9,1], p .002), a significant session effect (F 11.04 [27,3], p .001), and a significant interaction (F 6.13 [27,3], p .003). Again, the interpretation of this interaction can be determined from Fig. 3, where there are more obvious differences in the rate of decline between NSD and SD conditions at Sessions 4 and 5. Post-hoc paired t tests comparing corresponding sessions between conditions

FIG. 3. Mean (SD) Production misjudgments throughout SD and NSD (n 10). During Sessions 4 and 5, participants made significantly greater production misjudgements in the SD condition (see text).

138

HARRISON AND HORNE

showed significant differences for Sessions 4 (t 3.31 [9], p .005).

2.08 [9], p

.05) and 5 (t

Response style. Masterplanner requires players to monitor the influence of both computer simulation and competitors decisions in interpreting the impact of their own decisions on their companys performance. Accordingly, the player is often required to develop novel and untried approaches. A gradual increase in spending and repetition of previous actions is rarely appropriate and unlikely to be successful. To determine whether participants had been flexible in their decision making, Pearsons correlations were performed using each individuals mean deviation in spending (increase or reduction) between adjacent sessions. The higher the correlation, the more rigid the behavior and the less inclined was the participant to change from previous strategies. Insignificant correlations (Table 1) were found under NSD from Sessions 3 to 4 and 4 to 5, indicating that spending decisions deviated from previous decisions. In contrast, the SD participants showed similar patterns of change for Session 3 onward, leading to high and significant correlations (Table 1a). A similar pattern was evident in total spending (Table 1b) for adjacent sessions during the later stages of SD, again contrasting with NSD and further indicating a reluctance to develop untried spending plans. During NSD, participants were more inclined to modify previous strategies (i.e., increased variability in spending). Updating responses in the light of new information. The ability to shift or update response strategy on the basis of performance feedback and strategic guidelines for play was affected by SD. Participants were free to follow a range of options in pursuing their goals, although implicit in the guidelines prepared for each session was the suggestion that simply repeating the decisions of the previous session was unlikely to work to ones advantageous. Framing the guidelines for each of the decision areas in relative terms between sessions (see Appendix 2) reinforces the inappropriateness of this strategy further. An analysis of guideline observations or otherwise (errors) was performed by categorizing each of the 11 decisions as (a) perseverative type errors (PE)fails to shift in accord with guidelines while repeating the decision of previous session; (b) nonperseverative-type errors (NPE) again breaks the guideline by failing to shift in the direction suggestion but without repetition of previous

TABLE 1
Masterplanner Correlations between Adjacent Sessions for (a) Spending Deviations and (b) Total Spending (see Text for Details) NSD (a) Spending deviations Session 3 to 4 (Days 12) Session 4 to 5 (Day 2) (b) Total spending Session 3 to 4 (Days 12) Session 4 to 5 (Day 2) SD

.12 ( p .25 ( p .57 ( p .44 ( p

.36) .24) .04) .1)

.57 ( p .76 ( p .95 ( p .85 ( p

.04) .006) .000) .001)

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

139

response, i.e., produces novel response; and (c) correct response shifts (CRS) adjusts decision in accordance with guidelines. The outcome is shown in Fig. 4 as the proportion of errors/correct response shifts for all eleven strategic guidelines from Sessions 4 to 5 (see Appendix 2). Paired t tests assessed the differences between the related scores. Compared with NSD, SD resulted in significantly fewer correct response shifts (t 4.9; df 9; p .005 two-tailed) and significantly more perseverative errors (t 4.1; df 9; p .005 two-tailed). There were no differences in nonperseverative errors among conditions. We also observed that in a number of cases SD participants exaggerated the likely effect of observing particular guidelines. For example, spending in accordance with guideline 7 (a recommendation to increase shareholders dividends in the final session of the game) increased by an average of 53% during NSD compared with 174% for SD. However, as the number of errors and inappropriate responses generally increased with SD, the effect of such an exaggerated stance was limited. Subjective reports. Content analysis of the subjective data revealed a number of interesting points, further suggesting increased perseveration toward the end of the SD condition. Table 2 shows the proportion of all observational statements making a direct reference to a strategic guideline. The number of these statements did not change across sessions, as participants made approximately five or six observations for each report. However, by the fifth session of SD the focus for these observations shifted, with the proportion of observations making direct reference to the strategic rules falling to approximately half that of the NSD condition.

FIG. 4. Mean (SD) proportion of perseverative errors (PE), nonperseverative errors (NPE) and correct response shifts (CRS) during Session 5 for both the SD and NSD conditions (n 10). There were significantly more PE and significantly less CRS during the SD condition (see text).

140

HARRISON AND HORNE

TABLE 2
Mean Proportion of Observations Making Reference to Guidelines Included in Subjective Reports (see Text for Details Day 1 Session 2 NSD SD 0.67 0.62 Session 3 0.62 0.60 Session 4 0.57 0.59 Day 2 Session 5 0.61 0.30

Further analysis of the fifth session. Bearing in mind the abrupt and substantial decline in performance during the fifth session for the SD condition, we examined the decisions made, here, under both conditions. In terms of profitability, we were interested in the consequences, if participants in the SD condition had made the same modifications to their spending decisions from Sessions 4 to 5 as they did during the NSD condition. We calculated the relative change (i.e., percentage increase or reduction) between NSD Sessions 4 and 5 for each of the 11 spending decisions and then produced an adjusted set of spending decisions for Session 5 under SD by modifying decisions for Session 4 of SD to an identical extent to those of the corresponding decision in the NSD condition. The simulation of Masterplanner was then rerun for Session 5 of SD, replacing the participants original spending decisions with the adjusted levels. Profitability in Session 5 of SD, which had been in deficit ( 16%), was improved to around 4%, more comparable with that of NSD. This suggested that: (i) the decisions during NSD were indeed aimed at enhancing performance, and (ii) that without sleep loss the SD participants could have produced appropriate response strategies sensitive to the changing demands. That is, for the SD participants the real collapse of the game occurred in Session 5.

DISCUSSION

The emphasis of this investigation was on a real-word task that was complex, dynamic, and intrinsically stimulating. It was able to tap flexible, innovative thinking and the ability to update plans in the light of changing information. Real difficulties in these latter respects were only encountered toward the end of the sleep loss period when SD participants became unable to comprehend and act on a rapidly changing situation, and their play collapsed. A substituted simulation showed that had these players used the same strategies at this crucial point, as was evident during NSD, then they would have succeededthat is, up to Session 4 (about 30 h of SD) they were coping fairly well despite some signs of deterioration. Interestingly, Banderet et al. (1981), in their battle game, also found a marked deterioration of play at the 36-h point. By this fifth session SD participants showed reduced sensitivity to obvious impending changes in the simulation (e.g., with consumer demand), responded less appropriately, relied more heavily on previously developed strategies which

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

141

were no longer effective or appropriate, and failed to produce innovative solutions to an increasingly critical situation. To maintain their position or remain solvent they neglected the guidelines and employed more aggressive and exaggerated tactics, but to little obvious effect. In most cases these factors contributed towards a total loss of competence for the task. Those SD participants who had competed well up until this fifth session became unable to maintain their profitability and incurred severe financial losses or bankruptcy. In contrast, during NSD, participants remained in profit throughout all sessions. Subjective reports of the final session revealed the SD participants were less interested in the key issues and increasingly distracted by peripheral concerns. During the final session of SD it wes not clear whether participants simply failed to recognize the need to change strategies, or whether they were unable to produce alternatives to a tried (previously successful) approach. The former might be viewed as a problem with attention, loss of interest, etc., whereas the latter could indicate difficulties with innovation or flexible thinking. The latter would seem more likely, as even though the sessions lasted for 1 h, pilot studies and debriefing suggested that participants found Masterplanner to be stimulating and challenging, with interest maintained throughout all sessions. Friendly and motivating rivalry between participants pervaded throughout all sessions, even in the final session. Owing to the applied nature of Masterplanner, it is difficult to interpret the data from the analytic and perhaps traditional perspectives of more basic psychological performance metrics. Tasks of logical or deductive thinking, commonly used in sleep loss studies (Dinges & Kribbs, 1992), are unlikely to match the difficulty of Masterplanner, as they are relatively simple to understand and unchallenging for the intellectual level of our participants. GMAT Critical Reasoning may have been more appropriate in this respect; however, we do not claim it was of comparable difficulty to Masterplanner. Also, the two tasks differed in length and frequency of trials. Critical Reasoning remained unaffected by one night of SD; which is a finding in agreement with the scientific literature (c.f. Horne, 1988b; Wilkinson, 1992). This suggests that during SD, our participants failure at Masterplanner was not so much due to an inability to assimilate the material generated by the game, but that they were unable to act on it successfully. Parallels can be drawn between our Critical Reasoning task and Brown et al.s (1994) study in which qualified medics were able to comprehend lengthy, detailed, and complex articles taken from surgical journals despite suffering from acute sleep loss. Inasmuch that successful play at Masterplanner requires innovation and divergent thinking, there is an interesting theoretical underpinning to the collapse of play resulting from SD. This centers on the prefrontal region of the cerebral cortex (PFC) which we, as well as others, have previously shown to be particularly vulnerable to one night of SD (Herscovitch et al., 1980; Horne, 1988b; Thomas, Sing, & Belenky, 1993; Wimmer et al., 1992; Harrison & Horne, 1997, 1998; Petiau et al., 1998). The PFC is particularly associated with dealing with divergent type thinking, innovation, novelty and the unexpected. It directs, sustains, and focuses attention to the task in hand by disregarding

142

HARRISON AND HORNE

competing distraction and is the executive coordinator of many cortical events. Increased effort by the sleep deprived participant to compensate for these impairments are of only limited success (Harrison & Horne, 1998). In contrast, the PFC is not particularly associated with logical or critical reasoning, especially if the tasks dependent on these skills are familiar or well rehearsed. These skills are not affected by one night of SD. In conclusion, our findings indicate that people working for extended periods of time, who are required to make decisions necessitating innovation, flexibility, and the ability to update plans in the light of new information, and presented under rapidly changing situations should avoid sleep loss beyond 3236 h.
APPENDIX 1 Decision Areas and Effects

At each session participants had to make spending decisions for each of the 11 areas (see text).
Decision area Production Price Advertising Deals Quality, design, & cost Loan Invest Share issue Pay dividends Effect Sets level of production Sets retail price Increases sales through media exposure Promotional activitieseffectively lowers price All involved in product development Raises cash To gain return on excess cash Raises cash Improves shareprice

APPENDIX 2 Product Life Cycle (PLC) and Guidelines for Play

Across the sessions Masterplanner assumes the dynamics of a realistic marketing scenario by simulating the concepts of the PLC. It is for the players to determine which category of customer is dominant at any particular stage of play (by monitoring the consequences of their own and competitor behaviours) and to modify their own marketing behaviours accordingly. Players were introduced to these ideas during training using the following instructions: The Product Life Cycle The market has three types of customers: (1) Innovators: Drawn to new products and technology (2) Imitators: Influenced by fashion/trends (3) Repeaters: Second and third time buyers When a product is launched, innovators are the likeliest customers. As the

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

143

game progresses innovators become less important and their numbers decline as a proportion of the market, whereupon imitators, and then repeaters become more important. Each of these customer types responds differently to marketing decisions. As this customer mix changes one should also change tactics to appeal to the dominant group at any particular time. The following is a rough guide to the sensitivity of each customer category to the various aspects of marketing strategy (i.e., the importance of these factors in deciding to buy):
Innovators Price Advertising Deals Quality Design & Cost Low High Low High Medium Imitators Medium Medium Medium Medium Low Repeaters High Low Medium Medium Medium

As one can see, at the launch of a product the market is quite insensitive to price yet responds very positively to advertising and quality improvements. This contrasts with the later stages when dominance of the repeaters makes pricing an important component. By this stage, quality is assumed to be high anyway, as the technology is old and familiar, hence purchase decisions tend to be based on price differences between companies. Guidelines In addition to the above, players are presented with a series of guidelines which describe methods for optimising overall performance at each particular session. Note, these guidelines are presented as suggestions for relative and not absolute values, the onus is still very much on individual players to decide how much is to be spent on a particular strategy. For example, during sessions 4 and 5 the following guidelines to play are suggested: (number in parenthesis session). Decision Area
Maximize sales Directly Set Price (5) lower than Price (4) Advertising spending (5) lower than Advertising spending (4) Deals offered (5) higher than Deals offered (4) Quality (5) higher than Quality (4) Design (5) higher than Design (4) Cost (5) higher than Cost (4) Dividends paid (5) higher than Dividends paid (4)

Research and development

Improve shareprice

REFERENCES
Angus, R. G., & Heslegrave, R. J. (1985). Effect of sleep loss on sustained cognitive performance during a command and control simulation. Behavoural Research Methodology and Instrumentation, 17, 5567.

144

HARRISON AND HORNE

Bandaret, L. E., Stokes, J. W., Francesconi, R., Kowal, D. M., & Naitoh, P. (1981). Artillery teams in simulated sustained combat: Performance and other measures. In L. C. Johnson, D. J. Tepas, W. P. Colquhoun, & M. J. Colligan (Eds.), Biological rhythms, sleep and shiftwork (pp. 459477). New York: Spectrum. Blagrove, M., Alexander, C., & Horne, J. A. (1995). The effects of chronic sleep reduction on the performance of cognitive tasks sensitive to sleep deprivation. Applied Cognitive Psychology, 9, 2140. Brown, B. J., Van Susternen, T., Onsager, D. R., Simpson, D., Salaymeh, B., & Condon, R. E. (1994). Influence of sleep deprivation on learning among surgical house staff and medical students. Surgery, 115(5), 604610. Deary, I. J., & Tait, R. (1987). Effects of sleep disruption on cognitive performance and mood in medical house officers. British Medical Journal, 295, 15131516. Denisco, R. A., Drummond, J. N., & Gravenstein, J. S. (1987). The effect of fatigue on the performance of a simulated anaesthetic monitoring task. Journal of Clinical Monitoring, 3(1), 2224. Dinges, D. F., & Kribbs, N. B. (1992). Performing while sleepy: effects of experimentally-induced sleepiness. In T. M. Monk (Ed.), Sleep, sleepiness and performance (pp. 97128). London: Wiley. Friedman, R. C., Bigger, J. T., & Kornfeld, D. S. (1971). The intern and sleep loss. New England Journal of Medicine, 285, 201203. Friedman, R. C., Kornfeld, D. S., & Bigger, J. T. (1973). Psychological problems associated with sleep deprivation in interns. Journal of Medical Education, 48, 436441. Goldman, L. T., McDonough, M. T., & Rosemond, G. P. (1971). Stresses affecting surgical performance and learning. Journal of Surgical Research, 12, 8386. Jaffe, E. D. (1989). How to prepare for the graduate management admission test GMAT. New York: Barrons. Harrison, Y., & Horne, J. A. (1996). Long-term extension to sleep-are we really chronically sleep deprived? Psychophysiology, 33, 2230. Harrison, Y., & Horne, J. A. (1997). Sleep deprivation affects speech. Sleep, 20, 871877. Harrison, Y., & Horne, J. A. (1998). Sleep loss impairs short and novel language tasks having a prefrontal focus. Journal of Sleep Research, 7, 95100. Haslam, D. R. (1981). The military performance of soldiers in continuous operations. In L.C. Johson, D. J. Tepas, W. P. Colquhoun, & M. J. Colligan (Eds.), Biological rhythms, sleep and shiftwork. New York: Spectrum. Herscovitch, J., Stuss, D., & Broughton, R. (1980). Changes in cognitive processing following short-term cumulative partial sleep deprivation and recovery oversleeping. Journal of Clinical Neuropsychology, 2, 301319. Horne, J. A. (1988a). Why we sleep. Oxford: Oxford Univ. Press. Horne, J. A. (1988b). Sleep loss and divergent thinking ability. Sleep, 11, 528536. Horne, J. A., & Pettitt, A. N. (1985). High incentive effects on vigilance performance during 72 hours of total sleep deprivation. Acta Psychologia, 58, 133139. Leung, L., & Becker, C. E. (1992). Sleep deprivation and House Staff performance: update 1984 1991. Journal of Occupational Medicine, 34(12), 11531160. May, J., & Kline, P. (1987). Measuring the effects upon cognitive abilities of sleep loss during continuous operations. British Journal of Psychology, 78, 443455. Nelson, C. S., DellAngela, K., Jellish, W. S., Brown, I. E., & Skaredoff, M. (1995). Residents performance before and after night call as evaluated by an indicator of creative thought. Journal of the American Osteopathic Association, 95(10), 600603. Percival, J. E., Horne, J. A., & Tilley, A. J. (1982). Effects of sleep deprivation on tests of higher cerebral functioning. In W. P. Koella (Ed.), Sleep (pp. 390391). Karger: Basel. Petiau, C., Harrison, Y., Delfiore, G., Degueldre, C., Luxen, A., Franck, G., Horne, J. A., & Maquet,

SLEEP LOSS IMPAIRS INNOVATIVE AND FLEXIBLE THINKING

145

P. (1998). Modification of fronto-temporal connectivity during a verb generation task after a 30 hour total sleep deprivation. A PET study. Journal of Sleep Research, 7(Supp2), 208. Pilcher, J. J., & Huffcutt, A. I. (1996). Effects of sleep deprivation on performance: a meta analysis. Sleep, 19, 318326. Pinnock, C., Derbyshire, D., Johnstone, P., & Smith, G. (1985). Effects of sleep deprivation on choice reaction time. British Journal of Anaesthesia, 57, 346. Saunders, J. (1989). Masterplanner. Burbage, Leicestershire, UK: Saunders-Wong Associates. Thomas, M., Sing, H. C., & Belenky, G. (1993). Cerebral glucose utilisation during task performance and prolonged sleep loss. Journal of Cerebral Blood Flow and Metabolism, 13, Suppl 1, S351. Wilkinson, R.T. (1965). Sleep deprivation. In O. G. Edholm & A. Bacharach L. (Eds.), Physiology of human survival (pp. 399430). London: Academic Press. Wilkinson, R. T. (1992). The measurement of sleepiness. In R. J. Broughton & R. D. Ogilvie (Eds.), Sleep, arousal and performance (pp. 254265). Boston: Birkhauser. Wimmer, F., Hoffmann, R. F., Bonato, R. A., & Moffitt, A. R. (1992). The effect of sleep deprivation on divergent thinking and attention processes. Journal of Sleep Research, 1, 223230. Received April 24, 1998

You might also like