Previous research has found that cognitive ability is related to health and mortality, but not much is known about why this relationship exists . Various authors suggested that cognitive ability is related to behaviours that affect health [14, 24, 59]. This study was set up to examine this possibility. Three main issues were studied. First, it was investigated whether there was a relationship between cognitive ability, reaction time and memory span and health related behaviour and body weight; second, it examined whether self-control was related to health related behaviour and body weight and, third, whether self-control mediated the relationship between cognitive ability, reaction time and memory span on the one hand and health related behaviour and body weight on the other hand.
The study was based on a sample of 201 adolescents. During adolescence, individuals become more independent and more often decide themselves about their dietary habits. The consolidation of health behaviours, such as food intake and the amount of exercise, starts in childhood and early adolescence and is relatively stable thereafter [60–62]. Despite the importance of healthy nutrition and exercise in adolescence, research often shows that relatively large groups of adolescents are involved in risky health related behaviours .
The sample seems to be a good representation of Dutch adolescents. The occupation of the primary wage earner is similar to a sample of the general population of the same age: the mean in the present sample is 4.7 (SD = 1.1), in the sample of the general population, this was 4.6 (SD = 1.04) . The scores of the RSPM in the present sample were comparable to what might be expected based on other samples (present sample: mean: 46.37 (SD: 7.01). In Ravens norms the scores for fifteen year olds were 47 (SD not mentioned) (Raven, 2000). The interrelationships between the three measures of cognitive ability were also comparable to those found in previous studies . These findings indicate that the present sample is a good representation of a general sample of Dutch adolescents, and very similar to samples studied abroad.
As hypothesized, it was found that cognitive ability was positively related to health related behaviours. Adolescents with a high score on general cognitive ability had healthier dietary habits, they more often had breakfast during the weekend, ate more vegetables and less unhealthy foods, and they engaged more often in physical activity. They also spend less money on unhealthy foods and drinks (candy, snacks and soft drinks). To the best of our knowledge, no previous studies investigated these relationships.
These findings are in line with expectations formulated by several researchers in the field who suggested that the association between cognitive ability and health is mediated by the impact of cognitive ability on health related behaviours which, in turn, affects health and mortality [1–3, 7–15, 65].
In contrast, no relationships were found between reaction time and health related behaviour or BMI. This stands in contrast to previous, but scarce, research that reported a relationship between reaction time and higher mortality risk  and lower levels of physical exercise . Similarly, memory span was not related to health related behaviour. Again, this contrasts with what was expected, as two previous studies found a relationship between working memory, cognitive ability and mortality [6, 66]. These findings do not support Deary & Der 's hypothesis that agues that reaction time instead of general cognitive ability might be the most important predictor of health and mortality. A possible explanation is that reaction time and memory span are related to health related behaviours that are different from the specific health related behaviours investigated in this study.
Self-control was also positively related to healthy eating patterns (higher frequency of having breakfast during week and weekend and lower intake of crisps and snacks) and physical activity. Adolescents with high self-control also spend less money on unhealthy foods and drinks and have lower BMI's. These findings replicate what has been found in previous research [39–41, 44, 57]. These findings also are in line with studies reporting a relationship between conscientiousness and longevity as the concept of conscientiousness, meaning '..self-discipline, carefulness and thoroughness' , is similar to that of self-control [68, 69].
The analysis showed that the relationship between self-control and health behaviour and the relationship between cognitive ability and health behaviour are approximately of equal strength with β 's varying for cognitive ability between β = .14 and β = .19 and for self-control between β = .17 and β = .24.
Although in this study cognitive ability and self-control are both positively related to health behaviours, no relationship was found between cognitive ability and self-control. Therefore, self-control does not explain the relation between cognitive ability and health related behaviour in the present study. These findings suggest that self-control and cognitive ability are both independent predictors of health related behaviour.
Controlling for the variables age, gender and family income made no difference for the significance of the results. Interestingly, when controlled for family income, the amount of money spend on unhealthy foods is still related to cognitive ability and self-control. That is, less intelligent adolescents and adolescents with less self-control spend more money on unhealthy foods and drinks independent of family income.
A relation between cognitive ability and BMI was expected, since former studies found that more intelligent people are less likely to be overweight and have problems with obesity [30, 70]. No such relation was found in the present study. Explanation for the lack of a relation may relate to the age of the participants and the validity of BMI development in adolescents.
There was no relationship between self-control and cognitive ability. Previous findings describe a relationship between self-control and school performance  and therefore a relationship between self-control and cognitive ability was hypothesized. A recent meta-analysis found a mild relationship between self-control and cognitive ability: r = -23 . However, in this meta-analysis the age range was very large, from mean age 4 to 45.18. Furthermore, although the researchers controlled for age, this may have influenced the findings, as age is strongly related to self-control as well as cognitive ability scores . Further, in the meta-analysis, self-control was measured through tests of delay aversion and not by a questionnaire as was the case in the present study, and this may also explain the difference in the findings. Although self-reports are amenable to biases, it is also possible that self-control as measured by tests measures to some extent 'general performance on tests' and, also, delay aversion. Subjects with relatively high cognitive ability may perform better on delay aversion tasks (show more delay), while not showing more behavioural control in real life.
The present study did not control for school type of the subjects as a possible confounder of cognitive ability. This was based on the fact that educational achievement and cognitive ability are strongly related to each other so that they are, to some extent, measures of the same construct. Several studies showed that 'ability and schooling are so strongly dependent that it is not possible, over a wide range of variation in schooling and ability, to independently vary these two variables and estimate their separate impacts' [, p. 1] and that 'ability and schooling effects appear to be inseparable' [, p. 11].
This is especially true in the Netherlands. In the Netherlands, at the end of the primary school, all children are tested, usually with the 'CITO-test'. The CITO-test is a compulsory national test for cognitive achievement comparable to the SAT, but age adjusted  [For an introduction of the Dutch educational system for Americans: see http://www.fulbright.nl/cache/30/30dbd0481349e7a6188e372c5b049e51/31dutchsecondaryeducation.pdf] The results of the CITO-test are used by primary schools in order to advise parents as to the type of secondary education most suited to their child . This means that the school type chosen by Dutch children after primary education is heavily influenced by their cognitive abilities as measured through the CITO-test.
This system has several consequences. In a Dutch study, a relationship was found between educational achievement as measured by the CITO-test and cognitive ability of .63 at age 12. In the present study school type and cognitive ability are strongly correlated: r = .54 (p < .001). This means that controlling for school type when analyzing the relationship between cognitive ability and health related behaviour is almost the same as controlling for cognitive ability.
In an additional analysis, the regression analysis (presented in table 2, table 3 and table 4) was repeated and, in a fourth step, three dummy variables were added that coded for school type [We added 3 dummies coding for VMBO, MBO, HAVO, VWO was the reference category.]. It appeared that in seven of the eight outcome measures, the fourth step was not statistically significant. Only for 'breakfast during weekends', the fourth step was statistically significant (with p set on .05). In this case, the measure of cognitive ability (RSPM) became non significant and results show that subjects in the MBO school type have breakfast less often during weekends. It is concluded that even after controlling for school type, the findings of this study do not change.
The present study was subjected to limitations. First, the data on self-control and health behaviours was based on self-reports. The use of self-reports may have affected the results because of misinterpretation of the questions and social desirability. Findings should be replicated using other measures of self-control. Second, more intelligent participants might be better educated about healthy foods. Therefore, the results of the questionnaire might not only be representing actual health-related behaviours, but could be influenced by knowledge about healthy eating. Third, health behaviour as well as cognitive ability follows a socioeconomic gradient . Although this study controlled for socioeconomic status, other factors such as parents' dietary patterns have an influence on health behaviour. Fourth, malnutrition during childhood can affect intellectual development . Fifth, we used a relatively small sample. Last, because of the cross-sectional nature of the design, no causal implications can be drawn.