- Email: [email protected]
Effects of fat content, weight, and acceptability of the meal on postlunch changes in mood, performance, and cardiovascular function
Physiology& Behavior,Vol. 55, No. 3, lap.417--422,1994 Copyright© 1994ElsevierScienceLtd Printedin the USA.All rightsreserved 0031-9384/94$6.00 + .00
Pergamon 0031-9384(94)E0003-M
Effects of Fat Content, Weight, and Acceptability of the Meal on Postlunch Changes in Mood, Performance, and Cardiovascular Function A N D R E W S M I T H , *l A N N A K E N D R I C K , * A N D R E A M A B E N * A N D J E N N Y SALMON~f
*Health Psychology Research Unit, School of Psychology, University of Wales College of Cardiff, P.O. Box 901, Cardiff CF1 3YG, Wales ?~School of Home Economics and Institutional Management, University of Wales College of Cardiff, Wales Received 28 April 1993 SMITH, A. P., A. KENDRICK, A. L. MABEN AND J. SALMON. Effectsoffat content, weight, and acceptabilityof the meal on postlunch changesin mood, performance, and cardiovascularfunction. PHYSIOL BEHAV 55(3) 417-422, 1994.--This study examined the effects of fat content and meal size on postlunch changes in mood, performance, and cardiovascular function. Forty-six subjects (20 males, 26 females) were tested before and after lunch. Subjectswere assignedto one of the followinglunch conditions: a) low fat (23 g), large meal (860 g); b) low fat (18 g), small meal (600 g); c) high fat (84 g), large meal (840 g); d) high fat (79 g), small meal (530 g). The results showed only small effects of fat composition and meal size, with no cardiovascular effects being observed and no evidence of fat content or the weight of the meal influencing performance of logical reasoning or cognitive vigilancetasks. A few effects of meal type were significant in the mood data, but given the large number of analyses conducted, these could represent chance effects. Results from two selectiveattention tasks showed that subjects given the highfat meals responded more slowlybut more accurately,which differs from the effects of carbohydrate, protein, and calorie content reported in earlier papers. Weight of the meal influenced the degree of distraction from near and far distractors and also the accuracy of responses to central and peripheral targets. However, both the effects of fat and meal size were modified by task parameters, and further research is required before firm conclusionscan be drawn about the functional importance of the influences of nutrient content and meal size on performance. The high-fat and large meals were rated as more acceptable than the low-fat and small meals. These differencesin acceptability could not, however, account for the changes in performance observed after consumption of the different meal types. Lunch
Fat
Meal weight
Acceptability
Performance
THERE has been considerable recent interest in the effects of fat consumption on health. In addition to studying the effects of fat intake on physical health it is clearly important to examine whether changes in the consumption of fats influence mental functioning. Information must be provided on both the longterm and acute effects of fat consumption, and the primary aim of the present study was to provide information on the effects of fat content on changes in behaviour observed after lunch. The influence of meal composition on postlunch changes in performance efficiency and mood has been investigated before. The effects of high protein and high carbohydrate meals on mood and performance have been examined (9), and the results showed that the effects of meal composition depended upon the age and sex of the person consuming the meal. However, in this study there was no premeal baseline and, as age and gender may directly influence mood and performance, it is difficult to interpret the results. Another study (7) compared the effects of high carbohydrate and high protein meals on performance of two atten-
Mood
Cardiovascularfunction
tion tasks and on mood and cardiovascular functioning. There was no effect of meal composition on mood, but selective effects were found in the attention tasks, with high starch and high sugar meals slowing responses to peripheral targets, whereas the high protein lunch increased susceptibility to distracting stimuli close to the target. These results suggest that different nutrients produce subtle qualitative changes in attention. There are problems in interpreting results from studies that change nutrient content. If one wishes to compare high carbohydrate and high protein meals, and keep fat content constant and the different meals isocaloric, then the high carbohydrate meal has to be a low protein meal, which means that it is ditficult to determine whether any changes reflect the increase in carbohydrate or decrease in protein. Similarly, although the different meals may be isocaloric, they may differ in weight or in energy density. In other words, there may be correlated attributes of the meals which, as well as the factors of interest, may produce behavioural changes. These correlated attributes will not only
Requests for reprints should be addressed to Andrew Smith, Department of Psychology, University of Bristol, 8 Woodland Road, Bristol B58 ITH, England. 417
418 reflect physical characteristics of the meal, but will clearly also include factors such as meal acceptability. Ideally, one needs to manipulate a range of meal parameters in the same experiment so that one can look not only at the main effects of these factors but also at the interactions between them. Such studies will be time consuming and expensive and an intermediate strategy involves a series of converging operations to provide preliminary information as to whether particular features of a meal are related to subsequent changes in behaviour. As an initial step, we have examined changes in nutrient composition and meal size (as changes in one often lead to variation in the other). The evidence for effects of nutrient content in isocaloric lunches has already been described, and one must now consider whether meal size has an effect on postlunch changes in mood, performance, and physiology. One of the problems is that meal size can be defined in several ways. Nutritionists are usually concerned with the caloric content of meals, and the first question to ask is whether this is important in determining the effects of lunch. A recent study (3) compared the effects of a 1,000 kcal meal with those produced by a single sandwich. Meal size was only an important factor for subjects who had a different meal to normal, with subjects who normally had a small lunch being impaired by a large meal, and those who normally had a large meal showing a postlunch improvement after a sandwich. Further research (8) examined meal size in relation to the metabolic rate of the subjects and their usual eating pattern. The results showed no effect of meal size on mood but demonstrated that subjects who had a larger lunch than normal made more errors on focused attention and search tasks (especially the search task) than those who had a normalsized lunch or one that was smaller than normal. The results from both of these studies show that the effects of caloric content (an increase in momentary lapses of attention) are different from those produced by changes in nutrients (qualitative changes in selective attention). The problems associated with interpretation of changes in nutrient content have already been discussed. A review of the area (3) makes the point that studies of the effects of meal size are difficult to interpret as "regrettably, the relevant studies on this do not permit a distinction between meal volume and caloric value, the two components of quantity." There are clearly more than two ways of defining meal size but, for a start, it is important to examine whether changes in caloric value produce different effects to other changes, such as the weight of the meal. The aim of the present study was, therefore, to continue the investigation of the effects of the nutrient content and size of meals by examining whether changes in fat content and weight of the meal altered postlunch mood, performance, and cardiovascular function. Meals were produced, therefore, that were either high or low in fat content and which differed in weight by 250-300 g. The low-fat meals had a lower caloric value than the high-fat meals, which, on the basis of the present data alone, makes interpretation of effects difficult. However, previous results suggest that increases in calories are associated with momentary lapses of attention, and if the high-fat meals also produce this effect then it is reasonable to assume that this may reflect their calorific value. If, on the other hand, different effects of the highfat meals emerge, then it is difficult to interpret such results solely in terms of higher numbers of calories. The effects of weight of the meal could be more easily assessed, as this was varied with calorie content kept constant. As this was done for both the high- and low-fat meals, it is also possible to examine whether energy density is an important factor (the small/high-fat meal had the greatest energy/g, the large/low-fat meal the least energy/ g, and the other two meals fell in between these extremes). As mentioned earlier, changes to the nutrient content and meal size
SMITH ET AL.
may influence acceptability. This was measured in the present study so that it was possible to determine whether acceptability, rather than fat content or size, was responsible for any effects. EXPERIMENT
Method A between-subject design was used with subjects being randomly assigned to one of the four groups formed by combining high/low-fat and large/small meal conditions.
Subjects The subjects were 46 university students (20 males, 26 females). They were paid for participating in the study.
Procedure In the week prior to the experiment subjects were practiced at the tasks and filled in questionnaires measuring personality (Eysenck Personality Inventory: introversion, impulsivity, sociability, and neuroticism; Spielberger Trait Anxiety Inventory; Horne and Ostberg morningness questionnaire; Cognitive Failures Questionnaire), recent minor psychoneurotic symptoms (a modified version of the Middlesex Hospital Questionnaire), and questionnaires measuring usual eating and drinking habits. The main reason for collecting this data was to check that the different groups of subjects were comparable prior to the meal manipulation. On the day of the experiment, each subject was tested before lunch and then again 90 min after the start of the meal. Two groups of six subjects were tested each day, with the early group being tested about 45 min before the second group. Times for the early group are shown throughout the paper. The premeal session started at 1130, lunch was at 1230, and the postlunch session at 1400. In each session the following measures were recorded: I. pulse and blood pressure; 2. mood was assessed using 18 visual analogue scales with bipolar adjectives (e.g., drowsy-alert, tense-calm) at the ends (4); 3. performance tests--subjects carried out four tests a) a logical reasoning task, b) a repeated digits vigilance task, c) a focused attention task, and d) a categoric search task. Details of the tasks are shown below. Half of the subjects carried out the performance tasks in the order shown above and the rest did them in the reverse order. The tests were chosen because they covered a range of functions (working memory, sustained attention and selective attention) known to be influenced by changes in state.
Logical Reasoning Task This test was developed by Baddeley (1), and the subjects were shown statements about the order of the letters A and B followed by the letters AB or BA (e.g., A follows B: BA), The subjects had to read the statement and decide whether the sentence was a true description of the order of the letters. If it was, the subject pressed the T key on the keyboard, if it wasn't, they pressed the F key. The sentences ranged in syntactic complexity from simple active to passive negative (e.g., A is not followed by B). Subjects carried out the task for 3 min.
Repeated Digits Vigilance Task This task was developed by Smith (6), and the subjects were shown three digit numbers on the screen at the rate of 100 per min. Each number was normally different to the preceeding one,
EFFECTS OF FAT CONTENT, MEAL SIZE, AND ACCEPTABILITY
but occasionally (eight times a min) the same number was presented on successive trials. Subjects had to detect these repeats and respond as quickly as possible. The number of hits, reaction times for hits, and false alarms were recorded. The task lasted for 8 min. The above tasks measure working memory and sustained attention. Selective attention tasks have also been shown to be sensitive to changes in state, and two relatively new tasks are now available that measure a range of different aspects of selective attention. These tasks have the advantage that they are linked to established models of selective attention, are sensitive measuring instruments, and have been shown to be influenced by changes of state produced in a variety of ways (e.g., time of day, noise, pharmacological changes). The two tasks measure two fundamental aspects of attention. The first, often called "filtering", requires the subject to select a single sensory feature (e.g., a specific location) and to ignore events that do not possess it. The other paradigm, called "selective set" or pigeon holing, requires the person to select any member of a category of events. The difference between these two paradigms is very important, and this view is supported by results that show that variables within a task have different effects, depending on whether selection is by filtering or selective set. In addition, changes in the state of the person (e.g., induced by noise, or by testing at different times of day) also influence the two paradigms in different ways. These results suggest that selective attention does not consist of constant and invariant processes that are insensitive to the nature of the task, but that it involves flexible and adaptive systems influenced by context. This means that it is necessary to compare filtering and selective set paradigms and to study the influence of task parameters within these paradigms. The filtering paradigm used here involved a focused attention task where a target letter was always presented in a fixed location.
By presenting stimuli on either side of the target letter it was possible to measure a) the effects of an irrelevant stimulus, b) distraction by agreeing and disagreeing stimuli, c) the extent to which attention was focused or set at a wide angle. In addition to these measures of selective attention, the task examined factors known to be important in response organisation in choice reaction time (e.g., response repetition--alternation). The accuracy of performing this task is usually high, but occasional errors also provide an index of momentary lapses of attention.
Focused Attention Task This task was developed by Broadbent (2). The target letters were upper case As and Bs. On each trial three warning crosses were presented on the screen, the outside crosses being separated from the middle one by either 1.02 or 2.60 ° . The subjects were told to respond to the letter presented in the centre and ignore any distractors presented at the side. The crosses were on the screen for 500 ms and were then replaced by the target letter. The central letter was either accompanied by a) nothing, b) asterisks, c) letters that were the same as the target, d) or letters that differed--the two distractors were identical and the targets and accompanying letters were always A or B. The correct response to A was to press a key with the forefinger of the left hand, and to B a different key with the forefinger of the right hand. Subjects were given 10 practice trials followed by five blocks of 64 trials. In each block there were equal numbers of near/far conditions, A or B responses, and equal numbers of the four distractor conditions. The nature of the previous trial was controlled, as reaction times on the current trial are influenced by the time taken to respond on the previous trial. Selective set or pigeon-holing was measured using a categoric search task. In the focused attention task the subject knew where
TABLE 1 THE FOUR MEALSGIVEN TO THE SUBJECTS (A) LARGE, LOW FAT wholemeal bread roll Tagliatelle bolognaise Caramel oranges with cream
Nutrient Composition: Protein 34 g Fat 23 g (23% energy) Carbohydrate 144 g Energy 880 kcal Weight of meal 860 g (C) SMALL, LOW FAT Fish cakes and tomato ketchup Jacket baked potato Baked beans Meringue with ice cream nuts and fruit sauce Nutrient Composition: Protein 29 g Fat 18 g (19% energy) Carbohydrate 148 g Energy 840 kcal Weight of meal 600 g
419
(B) LARGE, HIGH FAT Steak and kidney pie Creamed potato Peas, Carrots Sliced green beans Apple strudel and cream Nutrient Composition: Protein 28 g Fat 84 g (58% energy) Carbohydrate 118 g Energy 1300 kcal Weight of meal 840 g (D) SMALL, HIGH FAT Pate and wholemeal toast 2 egg omelet Chips Apple pie and cream Nutrient Composition: Protein 47 g Fat 79 g (55% energy) Carbohydrate 103 g Energy 1290 kcal Weight of meal 530 g
420
S M I T H ET AL.
the target letter was going to be presented and the discrimination only involved which letter is displayed. In the categoric search task, the location of the letter was not known in advance and so it involved two stages: 'Where is the target?' and 'What is the target?'. In addition, this task measured the importance of the presence of distracting stimuli, the spatial separation of the two possible locations, and the tendency to orientate to the location in which the previous target was displayed (the place repetition effect). Response organisation was examined by comparing trials that were repeats of the previous response and those that led to alternations, and by comparing responses that were spatially compatible or incompatible with the location in which the stimulus was presented. Again, error rates provided an indication of the frequency of momentary lapses of attention.
Categoric Search Task This task was also developed by Broadbent (2). Each trial started with the appearance of two crosses in the positions occupied by the nontargets in the focused attention task, i.e., 2.04 or 5.20 ° apart. The subject did not know in this task which of the crosses would be followed by the target. The letter A or B was presented alone on half the trials and was accompanied by a digit (1-7) on the other half. Again, the number of near/far stimuli, A vs. B responses and digit/blank conditions were controlled. Half of the trials led to compatible responses (i.e., the letter A on the left of the screen, or letter B on the right), whereas the others were incompatible. The nature of the preceeding trial was also controlled. In other respects (practice, number of trials, etc.) the task was identical to the focused attention task.
Details of the Meals The four meals are described in Table 1. After the meal, subjects rated the acceptability of the meal on a visual analogue scale. They were given a decaffeinated coffee to drink after the meal. RESULTS
Comparability of the Subjects in Different Meal Conditions Analyses of variance showed that the subjects in the different meal conditions did not differ significantly in personality, recent minor symptoms, or regular eating and drinking habits.
Cardiovascular Measures, Mood, and Performance Analyses of covariance were carried out on the postlunch data with prelunch scores as covariates. The between-subject factors were fat content, meal size, time of testing (early vs. late), and order of performance tests.
Cardiovascular Measures No significant main effects of fat content or meal size were found.
Mood (Scores shown are in arbitrary units from 1-50, from the lefthand end of the scale). A significant effect of fat content was found in the tense-calm ratings, F(I, 29) = 5.5, p < 0.05, with
TABLE 2 EFFECTS OF FAT CONTENT AND MEAL SIZE ON THE FOCUSED ATTENTION TASK (1) Effects of fat: (a) Fat X nature of current distractor: accuracy High Fat
Low Fat Blank 89.2
Current distractor: Blank * Percent correct 94.4 93.2 (b) Fat X nature ofdistractor on previous trial: accuracy (A = Agreeing; D = Disagreeing; B = Blank; *) Letter Letter
A 93.0
D 92.6
B 94.8
91
Low Fat
High Fat Percent correct
*
* 94.6
(2) Effects of meal size: Funnel vision effect--accuracy (score = (near-far) disagreeing distractors). Large meal -3.0%
A 89.2
D 92.2
B 88.2
* 90.2
Small meal 1.0%
(3) Fat x meal size: (a) Accuracy Low Fat
High Fat Large meal Small meal Percent correct 93.2 94.2 (b) Fat x meal size x alternation (A)/repeat (R): Speed.
Large meal 91.2 Low Fat
High Fat Large meal A 384
R 363
Small meal 88.8
Small meal A 378
R 366
Large meal A 356
R 338
Small meal A 374
R 332ms
EFFECTS OF FAT C O N T E N T , MEAL SIZE, A N D ACCEPTABILITY
421
TABLE 3 EFFECTS OF FAT CONTENT AND MEAL SIZE ON THE CATEGORIC SEARCH TASK (a) Fat:
Fat X response compatibility X alternation (A)/repeat (R): accuracy Incompatible
Compatible A 94.4% 93.6
High Fat Low Fat (b) Meal size:
R 95.6 93.6
A 92.6 89.0
R
92.6 91.6
Meal size X location (near/far): Accuracy Large meal
Small meal Near 94.4
Near Far Percent correct: 96.0 89.0 (c) Fat X meal size: Fat X meal size x alternation (A)/repeat (R): accuracy High Fat Large meal Percent correct
A 92.8
Low Fat
Small meal
R 94.2
A 94.4
the high-fat group becoming calmer after the meal (high-fat adjusted mean = 36.6; low-fat adjusted mean = 31.3). Meal size influenced how sociable subjects felt [withdrawn-sociable: F(I, 29) = 8.3, p < 0.01, with subjects receiving the larger meal feeling more sociable (large meal adjusted mean = 36.4; small meal adjusted mean = 29.9).
Performance Tasks Logical reasoning task. Neither fat content nor meal size had a significant effect on the speed and accuracy of performing this task. Repeated digits vigilance task. Fat content and meal size had no significant effect on the n u m b e r of targets detected, false alarm rate, or speed of responses. Focused attention task--effects offat. The general trend was for subjects given the high-fat meals to respond more slowly but more accurately. However, the effects of fat were modified by task parameters, with the extent of the changes in accuracy varying according to whether a distracting asterisk was present or not [fat X blank/*: F(1, 40) = 5.85, p < 0.05]. Subjects who had consumed the high-fat meals made about 5% fewer errors than the low-fat meals groups on the trials where only the target letters
TABLE 4 EFFECTS OF FAT AND MEAL SIZE ON ACCEPTABILITYRATINGS Low Fat/ Small Meal
Low Fat/ Large Meal
High Fat/ Small Meal
High Fat/ large Meal
46.4 (16.5)
66.7 (20.6)
60.9 (20.5)
74.6 (14.7)
Scores are mean ratings on a visual analogue scale of 0-100, with higher scores representing greater acceptability. Standard deviations are in parentheses.
Far 92.0
R 94.0
Large meal A 92.8
R 91.4
Small meal A 90.8
R 93.8
were presented. This effect was halved when distracting asterisks accompanied the target letters. An interaction between fat content and the nature of the distractor on the previous trial was also significant [fat X nature of previous trial: F(3, 122) = 4.01, p < 0.01]. Again, subjects given the high-fat meals were 4-5% more accurate in all conditions except those where a disagreeing letter had been a distractor on the previous trial. These effects are shown in Table 2. Focused attention task--effects of meal size. The effects of meal size were restricted to accuracy and also varied according to whether distracting letters were presented or not. Consumption of the large meal was associated with greater accuracy when distractors were far from the target but reduced accuracy when they were closer to the target [near/far X meal size, distracting letter conditions only: F( 1, 40) = 5.49, p < 0.05]. This is shown in Table 2. Focused attention task--fat × meal size. The difference in accuracy between the high- and low-fat groups was greater in the smaller meal conditions [fat X meal size: F(1, 40) = 5.93, p < 0.05]. This is shown in Table 2. An interaction between fat content, meal size, and response alternation/repetition was significant in the analysis of reaction times, F(I, 40) = 5.51, p < 0.05. Subjects given the high-fat meals showed comparable differences between alternations and repeats in the large- and smallmeal conditions, whereas the low-fat, small-meal group were much slower on alternations than the low-fat, large-meal group. This is also shown in Table 2. Categoric search task--effects offat. Again, the trend was for the high-fat group to perform more slowly and accurately, although the magnitude of this effect varied when parameters influencing response organisation were manipulated [fat X response compatibility X alternation/repetition: F(I, 39) = 4.20, p < 0.05]. The largest effect of fat content was found for responses that were incompatible alternations and the smallest effect for compatible alternations (with the other two conditions falling between these). This is shown in Table 3.
422
SMITH ET AL.
Categories search task--effects of meal size. Meal size again influenced the accuracy of responses to central and peripheral stimuli [meal size × near/far: F(I, 39) = 4.94, p < 0.05], with those consuming the large meals being more accurate in their responses to near targets but less accurate when the targets were in the periphery. Categories search task--fat × meal size. The effects of fat and meal size again varied with task parameters. For example, subjects given the large meal showed a greater effect of fat content when responses were repetitions than when they were alternations, whereas the smaller meal group showed larger fat effects on the alternations than the repetitions [fat × meal/size × alternation/repeat: F(I, 39) = 5.11, p < 0.05]. Acceptability of the Meals There were significant differences between the acceptability ratings given for the different meals. This is shown in Table 4. An analysis of variance revealed significant main effects of fat, F(2, 42) = 4.31, p < 0.05, and meal size, F(1, 42) = 9.88, p < 0.005, which reflected the higher acceptability ratings for the high-fat and heavier meals. Acceptability was included as a factor in further analyses of performance, mood, and cardiovascular function (subjects were subdivided into those who gave the meal a high acceptability rating and those who gave it a low one on a median split). None of the effects of fat or meal size reported earlier could be attributed to differences in acceptability. DISCUSSION
The results from the present study showed that the effects of fat and meal size were very small and largely restricted to the two selective attention tasks. It should be noted that both the present studies and earlier ones (7,8) only examined effects occuring soon after (60 min after) consumption of the meal. It is quite possible that other effects may appear later in the afternoon, and this must be considered in further studies. In addition, it should be noted that the changes in fat content also changed the amount of carbohydrate and protein in the meal. The results suggest, however, that the effects obtained here cannot be attributed to the changes in carbohydrate or protein. This is because the pattern of results obtained here is very different from those observed in the early study which manipulated protein and carbohydrate (7). Carbohydrate influenced response speed to peripheral targets, whereas protein was associated with slower response times when distracting stimuli were close to the target. Neither of these effects was produced by the fat manipulation in the present experiment. Indeed, the fat effect was largely a change in the speed/error tradeoff, with high fat leading to slower but more accurate performance. Furthermore, fat interacted with factors known to influence response organisation (alternation
vs. repetition; S-R compatibility) rather than the different attentional effects described by Broadbent (2). For the same reasons it is difficult to explain the effects of fat in terms of differences in energy value of the meals, since the profile of performance changes observed differed from the meal size effect reported in an earlier study (8). The effects of weight of the meal were largely restricted to the accuracy of responding. In the focused attention task the larger meal was associated with greater accuracy when distractors were close to the target, but reduced accuracy when they were further away. A similar effect was obtained in the categoric search task. Meal weight influenced the magnitude of the fat effects, with these being greater in the small meal conditions. This demonstrates the importance of examining different combinations of nutritional manipulations rather than studying single factors in isolation. Fat content and meal size were also associated with differences in acceptability, with the high fat and large meals being rated more acceptable than the low fat and small meals. These differences in acceptability could not account for the behavioural differences obtained after consumption of the various meal types. Although the present study has demonstrated effects of fat content and meal size, it tells us little about the mechanisms involved. Indeed, it will probably be necessary to develop new techniques for assessing the impact of nutrient composition before we can learn more about the mechanisms involved. Both this study and previous research (7,8) have shown that nutrient content and meal size can have an effect. However, the effects have been very small, difficult to interpret, and only obtained by using very sensitive measures of performance. Indeed, further research is needed to determine whether the effects are reliable or merely chance effects observed because of the large number of analyses carried out. In contrast to this, psychoactive substances found in beverages, such as caffeine, produce much larger effects which in freeliving situations may easily mask the more subtle effects of meal composition. It is possible, however, that the effects observed here may be much smaller than the long-term effects of different nutrients, and future studies must now examine the long-term effects of different diets on behaviour. One reason for examining the long-term effects of fat consumption is that recent data suggests that low cholesterol diets are associated with an increased suicide rate (5). Whether the long-term effects of fat consumption are apparent in a wide range ofbehaviour in the majority of the population, or whether the effects are more restricted and selective is something that only further research can answer. ACKNOWLEDGEMENT
This research was supported by a grant from the Agriculture and Food Research Council.
REFERENCES
1. Baddeley,A. D. A three-minutereasoningtest based on grammatical transformation. Psychonom. Sci. 10:341-342; 1968. 2. Broadbent, D. E.; Broadbent, M. H. P.; Jones, J. L. Performance correlates of self-reported cognitivefailure and obsessionality.Br. J. Clin. Psychol. 25:285-299; 1986. 3. Craig,A. Acuteeffectsof mealson perceptualand cognitiveefficiency. Nutr. Rev. 44:163-171; 1986. 4. Herbert, M.; Johns, M. W.; Dore, C. Factor analysis of analogue scales measuring subjectivefeelingsbefore sleep and after sleep. Br. J. Med. Psychol. 49:373-379; 1976. 5. Muldoon, M. F.; Manuck, S. B.; Matthews, K. A. Loweringcholesterol concentrationsand mortality:A quantitative review of primary prevention trials. Br. Med. J. 301:309-314; 1990.
6. Smith, A. P. Acute effects of noise exposure: An experimental investigation of the effects of noise and task parameters on cognitive vigilancetasks. Int. Arch. Occup. Environ. Health 60:307-310; 1988. 7. Smith, A.; Leekam, S.; Ralph, A.; McNeill,G. The influenceof meal composition of postlunch changes in performance efficiency and mood. Appetite I0: 195-203; 1988. 8. Smith, A.; Ralph, A.; McNeill, G. Influencesof meal size on postlunch changes in performance efficiency,mood and cardiovascular function. Appetite 16:85-91; 1991. 9. Spring, B. J.; Mailer, O.; Wurtman, J.; Digman, L.; Cozolino, L. Effects of protein and carbohydrate meals on mood and performance: Interactions with sex and age. J. Psychiatr. Res. 17:155167; 1983.
Pergamon 0031-9384(94)E0003-M
Effects of Fat Content, Weight, and Acceptability of the Meal on Postlunch Changes in Mood, Performance, and Cardiovascular Function A N D R E W S M I T H , *l A N N A K E N D R I C K , * A N D R E A M A B E N * A N D J E N N Y SALMON~f
*Health Psychology Research Unit, School of Psychology, University of Wales College of Cardiff, P.O. Box 901, Cardiff CF1 3YG, Wales ?~School of Home Economics and Institutional Management, University of Wales College of Cardiff, Wales Received 28 April 1993 SMITH, A. P., A. KENDRICK, A. L. MABEN AND J. SALMON. Effectsoffat content, weight, and acceptabilityof the meal on postlunch changesin mood, performance, and cardiovascularfunction. PHYSIOL BEHAV 55(3) 417-422, 1994.--This study examined the effects of fat content and meal size on postlunch changes in mood, performance, and cardiovascular function. Forty-six subjects (20 males, 26 females) were tested before and after lunch. Subjectswere assignedto one of the followinglunch conditions: a) low fat (23 g), large meal (860 g); b) low fat (18 g), small meal (600 g); c) high fat (84 g), large meal (840 g); d) high fat (79 g), small meal (530 g). The results showed only small effects of fat composition and meal size, with no cardiovascular effects being observed and no evidence of fat content or the weight of the meal influencing performance of logical reasoning or cognitive vigilancetasks. A few effects of meal type were significant in the mood data, but given the large number of analyses conducted, these could represent chance effects. Results from two selectiveattention tasks showed that subjects given the highfat meals responded more slowlybut more accurately,which differs from the effects of carbohydrate, protein, and calorie content reported in earlier papers. Weight of the meal influenced the degree of distraction from near and far distractors and also the accuracy of responses to central and peripheral targets. However, both the effects of fat and meal size were modified by task parameters, and further research is required before firm conclusionscan be drawn about the functional importance of the influences of nutrient content and meal size on performance. The high-fat and large meals were rated as more acceptable than the low-fat and small meals. These differencesin acceptability could not, however, account for the changes in performance observed after consumption of the different meal types. Lunch
Fat
Meal weight
Acceptability
Performance
THERE has been considerable recent interest in the effects of fat consumption on health. In addition to studying the effects of fat intake on physical health it is clearly important to examine whether changes in the consumption of fats influence mental functioning. Information must be provided on both the longterm and acute effects of fat consumption, and the primary aim of the present study was to provide information on the effects of fat content on changes in behaviour observed after lunch. The influence of meal composition on postlunch changes in performance efficiency and mood has been investigated before. The effects of high protein and high carbohydrate meals on mood and performance have been examined (9), and the results showed that the effects of meal composition depended upon the age and sex of the person consuming the meal. However, in this study there was no premeal baseline and, as age and gender may directly influence mood and performance, it is difficult to interpret the results. Another study (7) compared the effects of high carbohydrate and high protein meals on performance of two atten-
Mood
Cardiovascularfunction
tion tasks and on mood and cardiovascular functioning. There was no effect of meal composition on mood, but selective effects were found in the attention tasks, with high starch and high sugar meals slowing responses to peripheral targets, whereas the high protein lunch increased susceptibility to distracting stimuli close to the target. These results suggest that different nutrients produce subtle qualitative changes in attention. There are problems in interpreting results from studies that change nutrient content. If one wishes to compare high carbohydrate and high protein meals, and keep fat content constant and the different meals isocaloric, then the high carbohydrate meal has to be a low protein meal, which means that it is ditficult to determine whether any changes reflect the increase in carbohydrate or decrease in protein. Similarly, although the different meals may be isocaloric, they may differ in weight or in energy density. In other words, there may be correlated attributes of the meals which, as well as the factors of interest, may produce behavioural changes. These correlated attributes will not only
Requests for reprints should be addressed to Andrew Smith, Department of Psychology, University of Bristol, 8 Woodland Road, Bristol B58 ITH, England. 417
418 reflect physical characteristics of the meal, but will clearly also include factors such as meal acceptability. Ideally, one needs to manipulate a range of meal parameters in the same experiment so that one can look not only at the main effects of these factors but also at the interactions between them. Such studies will be time consuming and expensive and an intermediate strategy involves a series of converging operations to provide preliminary information as to whether particular features of a meal are related to subsequent changes in behaviour. As an initial step, we have examined changes in nutrient composition and meal size (as changes in one often lead to variation in the other). The evidence for effects of nutrient content in isocaloric lunches has already been described, and one must now consider whether meal size has an effect on postlunch changes in mood, performance, and physiology. One of the problems is that meal size can be defined in several ways. Nutritionists are usually concerned with the caloric content of meals, and the first question to ask is whether this is important in determining the effects of lunch. A recent study (3) compared the effects of a 1,000 kcal meal with those produced by a single sandwich. Meal size was only an important factor for subjects who had a different meal to normal, with subjects who normally had a small lunch being impaired by a large meal, and those who normally had a large meal showing a postlunch improvement after a sandwich. Further research (8) examined meal size in relation to the metabolic rate of the subjects and their usual eating pattern. The results showed no effect of meal size on mood but demonstrated that subjects who had a larger lunch than normal made more errors on focused attention and search tasks (especially the search task) than those who had a normalsized lunch or one that was smaller than normal. The results from both of these studies show that the effects of caloric content (an increase in momentary lapses of attention) are different from those produced by changes in nutrients (qualitative changes in selective attention). The problems associated with interpretation of changes in nutrient content have already been discussed. A review of the area (3) makes the point that studies of the effects of meal size are difficult to interpret as "regrettably, the relevant studies on this do not permit a distinction between meal volume and caloric value, the two components of quantity." There are clearly more than two ways of defining meal size but, for a start, it is important to examine whether changes in caloric value produce different effects to other changes, such as the weight of the meal. The aim of the present study was, therefore, to continue the investigation of the effects of the nutrient content and size of meals by examining whether changes in fat content and weight of the meal altered postlunch mood, performance, and cardiovascular function. Meals were produced, therefore, that were either high or low in fat content and which differed in weight by 250-300 g. The low-fat meals had a lower caloric value than the high-fat meals, which, on the basis of the present data alone, makes interpretation of effects difficult. However, previous results suggest that increases in calories are associated with momentary lapses of attention, and if the high-fat meals also produce this effect then it is reasonable to assume that this may reflect their calorific value. If, on the other hand, different effects of the highfat meals emerge, then it is difficult to interpret such results solely in terms of higher numbers of calories. The effects of weight of the meal could be more easily assessed, as this was varied with calorie content kept constant. As this was done for both the high- and low-fat meals, it is also possible to examine whether energy density is an important factor (the small/high-fat meal had the greatest energy/g, the large/low-fat meal the least energy/ g, and the other two meals fell in between these extremes). As mentioned earlier, changes to the nutrient content and meal size
SMITH ET AL.
may influence acceptability. This was measured in the present study so that it was possible to determine whether acceptability, rather than fat content or size, was responsible for any effects. EXPERIMENT
Method A between-subject design was used with subjects being randomly assigned to one of the four groups formed by combining high/low-fat and large/small meal conditions.
Subjects The subjects were 46 university students (20 males, 26 females). They were paid for participating in the study.
Procedure In the week prior to the experiment subjects were practiced at the tasks and filled in questionnaires measuring personality (Eysenck Personality Inventory: introversion, impulsivity, sociability, and neuroticism; Spielberger Trait Anxiety Inventory; Horne and Ostberg morningness questionnaire; Cognitive Failures Questionnaire), recent minor psychoneurotic symptoms (a modified version of the Middlesex Hospital Questionnaire), and questionnaires measuring usual eating and drinking habits. The main reason for collecting this data was to check that the different groups of subjects were comparable prior to the meal manipulation. On the day of the experiment, each subject was tested before lunch and then again 90 min after the start of the meal. Two groups of six subjects were tested each day, with the early group being tested about 45 min before the second group. Times for the early group are shown throughout the paper. The premeal session started at 1130, lunch was at 1230, and the postlunch session at 1400. In each session the following measures were recorded: I. pulse and blood pressure; 2. mood was assessed using 18 visual analogue scales with bipolar adjectives (e.g., drowsy-alert, tense-calm) at the ends (4); 3. performance tests--subjects carried out four tests a) a logical reasoning task, b) a repeated digits vigilance task, c) a focused attention task, and d) a categoric search task. Details of the tasks are shown below. Half of the subjects carried out the performance tasks in the order shown above and the rest did them in the reverse order. The tests were chosen because they covered a range of functions (working memory, sustained attention and selective attention) known to be influenced by changes in state.
Logical Reasoning Task This test was developed by Baddeley (1), and the subjects were shown statements about the order of the letters A and B followed by the letters AB or BA (e.g., A follows B: BA), The subjects had to read the statement and decide whether the sentence was a true description of the order of the letters. If it was, the subject pressed the T key on the keyboard, if it wasn't, they pressed the F key. The sentences ranged in syntactic complexity from simple active to passive negative (e.g., A is not followed by B). Subjects carried out the task for 3 min.
Repeated Digits Vigilance Task This task was developed by Smith (6), and the subjects were shown three digit numbers on the screen at the rate of 100 per min. Each number was normally different to the preceeding one,
EFFECTS OF FAT CONTENT, MEAL SIZE, AND ACCEPTABILITY
but occasionally (eight times a min) the same number was presented on successive trials. Subjects had to detect these repeats and respond as quickly as possible. The number of hits, reaction times for hits, and false alarms were recorded. The task lasted for 8 min. The above tasks measure working memory and sustained attention. Selective attention tasks have also been shown to be sensitive to changes in state, and two relatively new tasks are now available that measure a range of different aspects of selective attention. These tasks have the advantage that they are linked to established models of selective attention, are sensitive measuring instruments, and have been shown to be influenced by changes of state produced in a variety of ways (e.g., time of day, noise, pharmacological changes). The two tasks measure two fundamental aspects of attention. The first, often called "filtering", requires the subject to select a single sensory feature (e.g., a specific location) and to ignore events that do not possess it. The other paradigm, called "selective set" or pigeon holing, requires the person to select any member of a category of events. The difference between these two paradigms is very important, and this view is supported by results that show that variables within a task have different effects, depending on whether selection is by filtering or selective set. In addition, changes in the state of the person (e.g., induced by noise, or by testing at different times of day) also influence the two paradigms in different ways. These results suggest that selective attention does not consist of constant and invariant processes that are insensitive to the nature of the task, but that it involves flexible and adaptive systems influenced by context. This means that it is necessary to compare filtering and selective set paradigms and to study the influence of task parameters within these paradigms. The filtering paradigm used here involved a focused attention task where a target letter was always presented in a fixed location.
By presenting stimuli on either side of the target letter it was possible to measure a) the effects of an irrelevant stimulus, b) distraction by agreeing and disagreeing stimuli, c) the extent to which attention was focused or set at a wide angle. In addition to these measures of selective attention, the task examined factors known to be important in response organisation in choice reaction time (e.g., response repetition--alternation). The accuracy of performing this task is usually high, but occasional errors also provide an index of momentary lapses of attention.
Focused Attention Task This task was developed by Broadbent (2). The target letters were upper case As and Bs. On each trial three warning crosses were presented on the screen, the outside crosses being separated from the middle one by either 1.02 or 2.60 ° . The subjects were told to respond to the letter presented in the centre and ignore any distractors presented at the side. The crosses were on the screen for 500 ms and were then replaced by the target letter. The central letter was either accompanied by a) nothing, b) asterisks, c) letters that were the same as the target, d) or letters that differed--the two distractors were identical and the targets and accompanying letters were always A or B. The correct response to A was to press a key with the forefinger of the left hand, and to B a different key with the forefinger of the right hand. Subjects were given 10 practice trials followed by five blocks of 64 trials. In each block there were equal numbers of near/far conditions, A or B responses, and equal numbers of the four distractor conditions. The nature of the previous trial was controlled, as reaction times on the current trial are influenced by the time taken to respond on the previous trial. Selective set or pigeon-holing was measured using a categoric search task. In the focused attention task the subject knew where
TABLE 1 THE FOUR MEALSGIVEN TO THE SUBJECTS (A) LARGE, LOW FAT wholemeal bread roll Tagliatelle bolognaise Caramel oranges with cream
Nutrient Composition: Protein 34 g Fat 23 g (23% energy) Carbohydrate 144 g Energy 880 kcal Weight of meal 860 g (C) SMALL, LOW FAT Fish cakes and tomato ketchup Jacket baked potato Baked beans Meringue with ice cream nuts and fruit sauce Nutrient Composition: Protein 29 g Fat 18 g (19% energy) Carbohydrate 148 g Energy 840 kcal Weight of meal 600 g
419
(B) LARGE, HIGH FAT Steak and kidney pie Creamed potato Peas, Carrots Sliced green beans Apple strudel and cream Nutrient Composition: Protein 28 g Fat 84 g (58% energy) Carbohydrate 118 g Energy 1300 kcal Weight of meal 840 g (D) SMALL, HIGH FAT Pate and wholemeal toast 2 egg omelet Chips Apple pie and cream Nutrient Composition: Protein 47 g Fat 79 g (55% energy) Carbohydrate 103 g Energy 1290 kcal Weight of meal 530 g
420
S M I T H ET AL.
the target letter was going to be presented and the discrimination only involved which letter is displayed. In the categoric search task, the location of the letter was not known in advance and so it involved two stages: 'Where is the target?' and 'What is the target?'. In addition, this task measured the importance of the presence of distracting stimuli, the spatial separation of the two possible locations, and the tendency to orientate to the location in which the previous target was displayed (the place repetition effect). Response organisation was examined by comparing trials that were repeats of the previous response and those that led to alternations, and by comparing responses that were spatially compatible or incompatible with the location in which the stimulus was presented. Again, error rates provided an indication of the frequency of momentary lapses of attention.
Categoric Search Task This task was also developed by Broadbent (2). Each trial started with the appearance of two crosses in the positions occupied by the nontargets in the focused attention task, i.e., 2.04 or 5.20 ° apart. The subject did not know in this task which of the crosses would be followed by the target. The letter A or B was presented alone on half the trials and was accompanied by a digit (1-7) on the other half. Again, the number of near/far stimuli, A vs. B responses and digit/blank conditions were controlled. Half of the trials led to compatible responses (i.e., the letter A on the left of the screen, or letter B on the right), whereas the others were incompatible. The nature of the preceeding trial was also controlled. In other respects (practice, number of trials, etc.) the task was identical to the focused attention task.
Details of the Meals The four meals are described in Table 1. After the meal, subjects rated the acceptability of the meal on a visual analogue scale. They were given a decaffeinated coffee to drink after the meal. RESULTS
Comparability of the Subjects in Different Meal Conditions Analyses of variance showed that the subjects in the different meal conditions did not differ significantly in personality, recent minor symptoms, or regular eating and drinking habits.
Cardiovascular Measures, Mood, and Performance Analyses of covariance were carried out on the postlunch data with prelunch scores as covariates. The between-subject factors were fat content, meal size, time of testing (early vs. late), and order of performance tests.
Cardiovascular Measures No significant main effects of fat content or meal size were found.
Mood (Scores shown are in arbitrary units from 1-50, from the lefthand end of the scale). A significant effect of fat content was found in the tense-calm ratings, F(I, 29) = 5.5, p < 0.05, with
TABLE 2 EFFECTS OF FAT CONTENT AND MEAL SIZE ON THE FOCUSED ATTENTION TASK (1) Effects of fat: (a) Fat X nature of current distractor: accuracy High Fat
Low Fat Blank 89.2
Current distractor: Blank * Percent correct 94.4 93.2 (b) Fat X nature ofdistractor on previous trial: accuracy (A = Agreeing; D = Disagreeing; B = Blank; *) Letter Letter
A 93.0
D 92.6
B 94.8
91
Low Fat
High Fat Percent correct
*
* 94.6
(2) Effects of meal size: Funnel vision effect--accuracy (score = (near-far) disagreeing distractors). Large meal -3.0%
A 89.2
D 92.2
B 88.2
* 90.2
Small meal 1.0%
(3) Fat x meal size: (a) Accuracy Low Fat
High Fat Large meal Small meal Percent correct 93.2 94.2 (b) Fat x meal size x alternation (A)/repeat (R): Speed.
Large meal 91.2 Low Fat
High Fat Large meal A 384
R 363
Small meal 88.8
Small meal A 378
R 366
Large meal A 356
R 338
Small meal A 374
R 332ms
EFFECTS OF FAT C O N T E N T , MEAL SIZE, A N D ACCEPTABILITY
421
TABLE 3 EFFECTS OF FAT CONTENT AND MEAL SIZE ON THE CATEGORIC SEARCH TASK (a) Fat:
Fat X response compatibility X alternation (A)/repeat (R): accuracy Incompatible
Compatible A 94.4% 93.6
High Fat Low Fat (b) Meal size:
R 95.6 93.6
A 92.6 89.0
R
92.6 91.6
Meal size X location (near/far): Accuracy Large meal
Small meal Near 94.4
Near Far Percent correct: 96.0 89.0 (c) Fat X meal size: Fat X meal size x alternation (A)/repeat (R): accuracy High Fat Large meal Percent correct
A 92.8
Low Fat
Small meal
R 94.2
A 94.4
the high-fat group becoming calmer after the meal (high-fat adjusted mean = 36.6; low-fat adjusted mean = 31.3). Meal size influenced how sociable subjects felt [withdrawn-sociable: F(I, 29) = 8.3, p < 0.01, with subjects receiving the larger meal feeling more sociable (large meal adjusted mean = 36.4; small meal adjusted mean = 29.9).
Performance Tasks Logical reasoning task. Neither fat content nor meal size had a significant effect on the speed and accuracy of performing this task. Repeated digits vigilance task. Fat content and meal size had no significant effect on the n u m b e r of targets detected, false alarm rate, or speed of responses. Focused attention task--effects offat. The general trend was for subjects given the high-fat meals to respond more slowly but more accurately. However, the effects of fat were modified by task parameters, with the extent of the changes in accuracy varying according to whether a distracting asterisk was present or not [fat X blank/*: F(1, 40) = 5.85, p < 0.05]. Subjects who had consumed the high-fat meals made about 5% fewer errors than the low-fat meals groups on the trials where only the target letters
TABLE 4 EFFECTS OF FAT AND MEAL SIZE ON ACCEPTABILITYRATINGS Low Fat/ Small Meal
Low Fat/ Large Meal
High Fat/ Small Meal
High Fat/ large Meal
46.4 (16.5)
66.7 (20.6)
60.9 (20.5)
74.6 (14.7)
Scores are mean ratings on a visual analogue scale of 0-100, with higher scores representing greater acceptability. Standard deviations are in parentheses.
Far 92.0
R 94.0
Large meal A 92.8
R 91.4
Small meal A 90.8
R 93.8
were presented. This effect was halved when distracting asterisks accompanied the target letters. An interaction between fat content and the nature of the distractor on the previous trial was also significant [fat X nature of previous trial: F(3, 122) = 4.01, p < 0.01]. Again, subjects given the high-fat meals were 4-5% more accurate in all conditions except those where a disagreeing letter had been a distractor on the previous trial. These effects are shown in Table 2. Focused attention task--effects of meal size. The effects of meal size were restricted to accuracy and also varied according to whether distracting letters were presented or not. Consumption of the large meal was associated with greater accuracy when distractors were far from the target but reduced accuracy when they were closer to the target [near/far X meal size, distracting letter conditions only: F( 1, 40) = 5.49, p < 0.05]. This is shown in Table 2. Focused attention task--fat × meal size. The difference in accuracy between the high- and low-fat groups was greater in the smaller meal conditions [fat X meal size: F(1, 40) = 5.93, p < 0.05]. This is shown in Table 2. An interaction between fat content, meal size, and response alternation/repetition was significant in the analysis of reaction times, F(I, 40) = 5.51, p < 0.05. Subjects given the high-fat meals showed comparable differences between alternations and repeats in the large- and smallmeal conditions, whereas the low-fat, small-meal group were much slower on alternations than the low-fat, large-meal group. This is also shown in Table 2. Categoric search task--effects offat. Again, the trend was for the high-fat group to perform more slowly and accurately, although the magnitude of this effect varied when parameters influencing response organisation were manipulated [fat X response compatibility X alternation/repetition: F(I, 39) = 4.20, p < 0.05]. The largest effect of fat content was found for responses that were incompatible alternations and the smallest effect for compatible alternations (with the other two conditions falling between these). This is shown in Table 3.
422
SMITH ET AL.
Categories search task--effects of meal size. Meal size again influenced the accuracy of responses to central and peripheral stimuli [meal size × near/far: F(I, 39) = 4.94, p < 0.05], with those consuming the large meals being more accurate in their responses to near targets but less accurate when the targets were in the periphery. Categories search task--fat × meal size. The effects of fat and meal size again varied with task parameters. For example, subjects given the large meal showed a greater effect of fat content when responses were repetitions than when they were alternations, whereas the smaller meal group showed larger fat effects on the alternations than the repetitions [fat × meal/size × alternation/repeat: F(I, 39) = 5.11, p < 0.05]. Acceptability of the Meals There were significant differences between the acceptability ratings given for the different meals. This is shown in Table 4. An analysis of variance revealed significant main effects of fat, F(2, 42) = 4.31, p < 0.05, and meal size, F(1, 42) = 9.88, p < 0.005, which reflected the higher acceptability ratings for the high-fat and heavier meals. Acceptability was included as a factor in further analyses of performance, mood, and cardiovascular function (subjects were subdivided into those who gave the meal a high acceptability rating and those who gave it a low one on a median split). None of the effects of fat or meal size reported earlier could be attributed to differences in acceptability. DISCUSSION
The results from the present study showed that the effects of fat and meal size were very small and largely restricted to the two selective attention tasks. It should be noted that both the present studies and earlier ones (7,8) only examined effects occuring soon after (60 min after) consumption of the meal. It is quite possible that other effects may appear later in the afternoon, and this must be considered in further studies. In addition, it should be noted that the changes in fat content also changed the amount of carbohydrate and protein in the meal. The results suggest, however, that the effects obtained here cannot be attributed to the changes in carbohydrate or protein. This is because the pattern of results obtained here is very different from those observed in the early study which manipulated protein and carbohydrate (7). Carbohydrate influenced response speed to peripheral targets, whereas protein was associated with slower response times when distracting stimuli were close to the target. Neither of these effects was produced by the fat manipulation in the present experiment. Indeed, the fat effect was largely a change in the speed/error tradeoff, with high fat leading to slower but more accurate performance. Furthermore, fat interacted with factors known to influence response organisation (alternation
vs. repetition; S-R compatibility) rather than the different attentional effects described by Broadbent (2). For the same reasons it is difficult to explain the effects of fat in terms of differences in energy value of the meals, since the profile of performance changes observed differed from the meal size effect reported in an earlier study (8). The effects of weight of the meal were largely restricted to the accuracy of responding. In the focused attention task the larger meal was associated with greater accuracy when distractors were close to the target, but reduced accuracy when they were further away. A similar effect was obtained in the categoric search task. Meal weight influenced the magnitude of the fat effects, with these being greater in the small meal conditions. This demonstrates the importance of examining different combinations of nutritional manipulations rather than studying single factors in isolation. Fat content and meal size were also associated with differences in acceptability, with the high fat and large meals being rated more acceptable than the low fat and small meals. These differences in acceptability could not account for the behavioural differences obtained after consumption of the various meal types. Although the present study has demonstrated effects of fat content and meal size, it tells us little about the mechanisms involved. Indeed, it will probably be necessary to develop new techniques for assessing the impact of nutrient composition before we can learn more about the mechanisms involved. Both this study and previous research (7,8) have shown that nutrient content and meal size can have an effect. However, the effects have been very small, difficult to interpret, and only obtained by using very sensitive measures of performance. Indeed, further research is needed to determine whether the effects are reliable or merely chance effects observed because of the large number of analyses carried out. In contrast to this, psychoactive substances found in beverages, such as caffeine, produce much larger effects which in freeliving situations may easily mask the more subtle effects of meal composition. It is possible, however, that the effects observed here may be much smaller than the long-term effects of different nutrients, and future studies must now examine the long-term effects of different diets on behaviour. One reason for examining the long-term effects of fat consumption is that recent data suggests that low cholesterol diets are associated with an increased suicide rate (5). Whether the long-term effects of fat consumption are apparent in a wide range ofbehaviour in the majority of the population, or whether the effects are more restricted and selective is something that only further research can answer. ACKNOWLEDGEMENT
This research was supported by a grant from the Agriculture and Food Research Council.
REFERENCES
1. Baddeley,A. D. A three-minutereasoningtest based on grammatical transformation. Psychonom. Sci. 10:341-342; 1968. 2. Broadbent, D. E.; Broadbent, M. H. P.; Jones, J. L. Performance correlates of self-reported cognitivefailure and obsessionality.Br. J. Clin. Psychol. 25:285-299; 1986. 3. Craig,A. Acuteeffectsof mealson perceptualand cognitiveefficiency. Nutr. Rev. 44:163-171; 1986. 4. Herbert, M.; Johns, M. W.; Dore, C. Factor analysis of analogue scales measuring subjectivefeelingsbefore sleep and after sleep. Br. J. Med. Psychol. 49:373-379; 1976. 5. Muldoon, M. F.; Manuck, S. B.; Matthews, K. A. Loweringcholesterol concentrationsand mortality:A quantitative review of primary prevention trials. Br. Med. J. 301:309-314; 1990.
6. Smith, A. P. Acute effects of noise exposure: An experimental investigation of the effects of noise and task parameters on cognitive vigilancetasks. Int. Arch. Occup. Environ. Health 60:307-310; 1988. 7. Smith, A.; Leekam, S.; Ralph, A.; McNeill,G. The influenceof meal composition of postlunch changes in performance efficiency and mood. Appetite I0: 195-203; 1988. 8. Smith, A.; Ralph, A.; McNeill, G. Influencesof meal size on postlunch changes in performance efficiency,mood and cardiovascular function. Appetite 16:85-91; 1991. 9. Spring, B. J.; Mailer, O.; Wurtman, J.; Digman, L.; Cozolino, L. Effects of protein and carbohydrate meals on mood and performance: Interactions with sex and age. J. Psychiatr. Res. 17:155167; 1983.