Testing the effects of nutrient deficiencies on behavioral performance

The association between specific nutrient deficiencies and poor performance on behavioral tests has been documented for several nutrients. The determination of causality, however, remains elusive. This paper presents the essential criteria for a valid test of causality. Findings from experimental studies in which a nutritional treatment was randomly allocated can be summarized in a statistical statement about the probability that the nutrient treatment caused the behavioral response. Criteria for assessing the internal validity of these studies are examined in terms of whether alleviation of a nutrient deficiency did or did not produce a detectable behavioral response. The plausibility of such a causal inference is dependent on its congruency with known or theorized biological and behavioral mechanisms. External validity describes the extent to which inferences from internally valid studies may be applicable to other populations or circumstances. In addition to these scientific considerations, some of the ethical issues of nutrient-treatment trials are also discussed. All of these considerations provide a better basis for judging whether public health action would be worthwhile than do observed associations that could actually be due to other causes.