Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Infant and preschool psychological development

Patrice L. Engle, Kathleen Gorman, Reynaldo Martorell, and Ernesto Pollitt



The INCAP longitudinal study (1969-1977) was designed to test the hypothesis that nutrition supplementation improves the cognitive test performance of infant and preschool children. The hypothesis was based on a series of arguments about effects of supplementation on brain development that have not been supported by subsequent research. Both early reports of these studies and current re-examination of the data suggest effects during the preschool years, but of modest magnitude.


For the past two decades researchers have examined the effects of early nutrition supplementation on the course of children's cognitive development using data collected by the INCAP longitudinal study on early supplementary feeding (1969-1977). The investigation was designed to test the hypothesis that protein supplementation would improve the cognitive functioning of low-income, nutritionally at-risk children. This paper summarizes theoretical research that influenced the design of the psychological component of the study, the methodology of the infant and preschool psychology batteries, design considerations related specifically to psychological assessment, and a summary of results from analyses published to date. Other studies on the same topic are reviewed elsewhere [1].

Historical overview of hypotheses

The main hypothesis of the INCAP longitudinal study—that improved protein ingestion during early childhood would be causally related to cognitive development—must be understood within the context of the psychological and nutritional knowledge of the period. The planning and design of the study took place from 1965 through 1968; the supplementation centres opened in 1969. During this period in the United States, the War on Poverty and concern for hunger in that country had lifted malnutrition from an obscure problem of developing nations to centre stage. At the time the INCAP study was initiated, researchers believed that protein was essential for adequate cognitive development [2]. It was reasonable that the major hypothesis to be tested was that the provision of supplementary protein would result in higher test performance in 7-year-olds.

Two theories about mechanisms linking malnutrition (either protein or protein-energy) to retarded cognitive development proposed during this period were (1) that malnutrition influences mental functioning directly through biological insult (the neural hypothesis), and (2) that malnutrition affects a child's energy and receptivity to learning, which in turn affects mental development (the developmental hypothesis).


The neural hypothesis

A series of international conferences from 1967 through 1974 brought together researchers from medical, nutrition, psychological, and anthropological backgrounds to improve their understanding of the mechanisms through which malnutrition could affect mental development, and a number of investigations of the effects of malnutrition on cognitive development were launched during the decade. In a 1967 conference the neural hypothesis was advanced on the argument that lack of protein would impair brain growth, particularly during the first postnatal months of life [3-5].

The initial evidence for this hypothesis came from studies with severely malnourished children, often with little experimental control. These children were described as apathetic, listless, and unresponsive. Low test scores were reported in 14 severely malnourished Chilean children [5], and 20 severely malnourished South African children differed from matched controls on non-verbal tests [6], which the authors claimed was similar to what one would expect from brain-damaged children. Significantly smaller head sizes and reduced brain weight were observed among children who died of malnutrition before the age of 5 years because of both protein and protein-energy malnutrition [7].

In a 1974 conference in New Delhi, the hypothesis that early malnutrition had a direct biological effect on the developing (human or other animal) brain, and that protein was the most important nutrient, was accepted by many [8, 9]. It was postulated that these effects occurred during the critical period of the brain growth spurt in humans, presumed to begin during the third trimester of pregnancy and to continue through 18 months postnatally [10, 11]. During this period there is rapid multiplication of glial cells and myelination of the neurons.

The neural hypothesis found parallels in psychobiology, where a consensus existed among researchers that particular stressful events in early life resulted in neuro-developmental deficits in later life. For example, low birth weight was found to have negative sequelae in later life, particularly neuro-behavioural effects [12].

It was argued on the basis of the neural hypothesis that undernutrition would result in a lack of neurosensory or intersensory integration [13]. Intersensory integration, a concept based in behavioural psychology, was considered one of the most important indicators of adaptive capacity in children 6-12 years of age and to be directly related to reading and writing [14]. To test the hypothesis that malnourished children would be inferior in intersensory integration, the tallest and shortest quartiles of an impoverished Guatemalan Indian sample and of a middle-class Guatemalan sample were assessed using a tool of intersensory integration [13]. The child was allowed to feel an object without seeing it and then was shown several objects and asked to select the one that matched the one that was felt. The rural group differed on intersensory integration as a function of quartile of height, but the wealthy urban group did not. The authors interpreted these findings as confirming their hypothesis, since only in the rural, and presumably malnourished, group did height relate to test performance. Unfortunately, since no other test was given, it is not possible to conclude from this study that a specific deficit in "intersensory integration" is associated with shortness.


The developmental hypothesis

Doubts about a strictly neural hypothesis were raised by several investigators during this period. The research designs were questioned; given the crucial importance of environmental stimulation for mental development, it was essential to extricate the effects of poor environmental stimulation and poor nutrition, ideally with an intervention design instead of the correlational results being presented [15, 16]. Much of the research was based on non-humans, and extensions from animal models to humans was questionable, given the complexity of human development [17].

A number of studies had showed that environmental stimulation could reduce the negative effects of malnutrition on the exploratory behaviour of rats [18-20]. It was suggested that improved nutrition might have the same kinds of effects on cognitive development as improved environmental stimulation [16]. Evidence from studies of concentration-camp victims had indicated that the weight of an adult's brain and heart was reduced only 3% under starvation conditions, whereas the weight of other organs might be reduced up to 60%, a process called brain sparing [7]. Similarly, an adult human brain would be protected against food limitation; whether brain sparing would occur for children was unknown. Most of these investigations were done with severely malnourished animals and humans; how much the more common mild to moderate malnutrition might affect intellectual development was unknown.

Finally, criticisms of the focus on protein were raised [21]. Protein deficiency was hard to measure reliably. Scientists observed that not all nutritional diseases could be treated by nutritional means [22]; rather than focus simply on the supply of protein-adequate foods, it was necessary to examine also the conditions under which the food was prepared and distributed, the role of infection and disease, and the complex social reactions to an increased supply of protein foods [21].

The hypothesis that lowered energy and less responsiveness caused by malnutrition would result in reduced learning opportunities (an indirect effect) was advanced during the late 1960s and early 1970s. Piaget's influential research underlined the significance of experience and the child's active role in the development of cognitive processes. If these children were apathetic and listless, they would be less able to create for themselves the interactions with the environment necessary for optimum cognitive development.

The behaviours displayed by malnourished children are thought to represent a device for energy conservation [23]. To maintain energy balance, the allocation of energy to motor activity and exploration is reduced [24]. The cost could be a delay in the development of cognition, since energy-conserving behaviours would prevent children's learning from environmental contingencies.

Long-term effects are likely to be seen in areas of attentiveness, curiosity, activity, motivation, and social responsiveness rather than in the cognitive domain [17]. Theoretically, these changes in learning style would in turn influence children's ability to incorporate experiences. Finally, it was postulated that the effects of undernutrition would reduce the rats' ability to learn non-essential information, resulting in a kind of "functional isolation" in which incidental learning was limited but necessary information was acquired [20].

Design issues in the INCAP study

The INCAP longitudinal study was intended to test the neural hypothesis by controlling for all possible confounding variables through good experimental design. It compared children from two villages randomly assigned to receive atole (containing protein and calories) with children from two villages that had been randomly assigned to receive fresco (with calories only), while providing all the children with medical care and the social stimulation of attending a feeding centre.

This design controlled for two of the confounding variables that were most likely to influence cognitive development: social stimulation received during supplementation, and the possibility of self-selection. Because the supplement was provided at a central feeding station, where there were round tables so that people could converse while eating, the possibility that children would receive more stimulation during the sessions was quite likely. Self-selection was a potential problem for both the psychological and the growth data if wealthier families were more likely to have children attend the centre than poorer families. For the fresco villages to function as an effective control, attendance patterns had to be similar in the two village types. A test of this assumption is discussed elsewhere [1].

Another qualification of an experimental study is initial equality of the treatment groups on the dependent variable, particularly since villages rather than individuals were randomly assigned to treatment. Initial equivalence of the villages in anthropometry and home diet has been reported [25], but initial equivalence in mental test performance and socio-economic status, both assessed in 1968, has not. These comparisons are reported elsewhere in this issue [26]. Analyses of initial equivalence on a few cognitive tests were performed in 1968, but those data are no longer available to the research team.



All children under the age of 7 living in the villages at the time the study began and all those who were born in the villages or moved into them during the period of the study were potential subjects. The coverage rates for children living in the villages were above 90%, although children not living in the villages at the time of their assessment were not measured. Children were assessed at birth, at 6 and 15 months, and annually from 24 months on. In all, 1,614 children were tested, with the sample size ranging from 400 for infants to over 800 for 7-year-olds.

The maximum exposure cohort was defined as all children who could have received supplementation from the beginning of conception through at least 2 years of age, the hypothetical sensitive period for cognitive outcomes. This cohort includes all children who were conceived after the supplementation began in January 1969 and who had reached 2 years of age by the time it ceased in February 1977.



The infant battery

The test developed to measure infant functioning was based on items from the Bayley, Gesell, Psyche Cattell, and Merrill-Palmer scales. These items were adapted to the Guatemalan population. The categorization of items into either mental or motor was based on the Bayley scales of infant development [27], described more fully elsewhere [28, 29].

Tests were administered by trained Guatemalan women in an adobe room similar to the children's homes. The mother or other care-giver was present for the infant tests. All children were tested within two weeks of their birthdays. Standardization between testers took place every six months. Test-retest reliability and inter-observer reliability were adequate [29].

The preschool battery

The preschool battery consisted of 22 separate tests administered to each child within a month of his or her birthday at ages 3, 4, 5, 6, and 7 years. Initially there were 12 tests in the battery; 10 more were added in 1971 (table 1). All the tests were given to the 5-7-year-olds, with a smaller set given to the 3- and 4-year-olds. The tests are described in the Appendix (p. 209).

TABLE 1. The preschool test battery

Test Agesa Dateb Measure Test-retest reliability
Embedded figures  
version 1 3 1969 correct responses .66 good
version 2 4-7 1969 correct responses .41-.65 adeq.
Verbal inferences 3-7 1969 correct responses .95 good
Vocabulary naming and recognition  
naming 3-7 1969 items named .89-.91 good
recognition 3-7 1969 items recognized .55-.77 good
Memory for digits 3-7 1969 digits recalled .65 good
Memory for sentences 3-7 1969 words recalled .60 good
Draw a line slowly 3-7 1969 velocity .65 good
Persistance on impossible puzzle 3-7 1969 time persisting .37-.97 adeq.
Reversal discrimination learning  
version 1 3-4 1969 trials to criterion .00-.48 poor
version 2 5-7 1969 trials to criterion .42-.73 adeq.
Memory for objects 3-7 1971 items correct .20-.70 adeq.
Knox cubes  
slow 3-7 1971 series correct .54 adeq.
fast 3-7 1971 series correct .42 poor
Memory for designs 5-7 1969 point score .77 good
Incidental and intentional learning  
incidental 5-7 1969 correct responses .22 - .52 adeq .
intentional 5-7 1969 correct responses .59- .71 good
Haptic-visual matching 5-7 1969 correct responses .12-.50 poor
Matching familiar figures 5-7 1969 correct responses .00-.25 poor
Block design 5-7 1971 point score .54-.83 adeq.
Animal houses 5-7 1971 WISC score .49-.70 adeq.
Elimination of odd figure 5-7 1971 correct responses .45-.83 adeq.
Face-hands touching 5-7 1971 correct responses .41-.81 adeq.
Incomplete figures 5-7 1971 correct responses .28-.75 adeq.
Conservation of matter 5-7 1971 0, 1, 2 code .51-.90 adeq.
Conservation of area 5-7 1971 0, 1, 2 code .09-.81 poor
Conservation of continuous quantity 5-7 1971 0, 1, 2 code .02-.71 poor

a. Age range of children to whom test was administered.
b. Date of initiation of test: March 1969, or June 1971.
c. Adeq. = adequate.

Testing began in March 1969 in the large fresco village, in April in the large atole village, in May in the small fresco village, and in June in the small atole village. Testing with the enlarged battery began in June 1971.

In order to identify specific skills that might be affected by malnutrition, the battery consisted of separate tests rather than a composite measure. Tests designed to address specific hypotheses were the haptic-visual matching test (intersensory integration), the matching familiar figures test and the draw a line slowly test (impulsivity) [30], the incidental learning test (functional isolation), and persistence on the impossible puzzle (motivation and persistence).

Other tests were included because they represented specific skills of interest in the literature: the embedded figures test, measuring field independence [31]; reversal discrimination learning [32]; the face -hands touching test, measuring the child's perception of the midline [33]; the memory for objects test, measuring children's ability to categorize objects in order to enhance recall; and three Piagetian tests of conservation. Others were derived from intelligence tests: short-term memory (memory for digits, sentences, and designs), performance items from standard IQ tests (animal houses and elimination of the odd figure from the Wechsler Preschool and Primary Scale of Intelligence, block design from the Wechsler Intelligence Scale for Children, Knox cubes, and incomplete figures from Stanford-Binet), and verbal items from standard IQ tests (vocabulary and verbal analogies). There were far fewer verbal tests than performance tests and no expressive language tests.

As with the infant battery, the tests were administered by trained Guatemalan women, who were standardized every six months. Testing took place in the same adobe rooms as the infant battery. Because of the number of tests, the child was brought in on several different days to complete them. The testers rotated between atole and fresco villages.


Reliability and validity

Although inter-observer reliability for the tests was high, test-retest reliability varied considerably (table 1). Some tests that had quite low test-retest reliability were too difficult for many of the children, and they should probably not be included in evaluating the hypothesis. For example, the mean score for the matching familiar figures test at age 5 indicated that children were on average performing at chance level.

Several studies were carried out to evaluate the emic validity of the battery. Village adults were asked to rate children according to their concept of listura, or "brightness," and these ratings were correlated with test scores [34]. Children's observed behaviours, such as carrying out complex sequences of commands (chores and errands), were also associated with test performance [35]. Finally, performance on the battery was correlated with the child's age at enrolment in school and school achievement [36].

Contents - Previous - Next