Previous - Next
Notwithstanding these possible alterations in cognitive flexibility and susceptibility to proactive interference, the animal data seem to suggest that specific cognitive functions are, for the most part, quite normal in previously malnourished subjects, independent of effects related to altered motivation and/or emotion. For example, reasoning and higher order functions appear intact on the basis of the many instances of normal acquisition of complex tasks exhibited by rats malnourished early in life. Examples include a negative patterning task (Tonkiss et al. 1993), the Hebb-Williams maze (for example, Celedon et al. 1979, Slob et al. 1973, Smart et al. 1973), the Morris maze (Bedi 1992, Campbell and Bedi 1989, Castro and Rudy 1989), 8- and 12-arm radial mazes (Hall 1983) and various pattern discriminations (reviewed in Smart and Tonkiss 1985)8. The exceptions appear attributable to altered motivation and/or emotionality, as discussed above.
8Although some investigators reported impaired acquisition in visual discrimination tasks the impairment does not appear to reflect a deficit in associational ability but rather differences in emotionality and/or motivation (see Smart and Tonkiss 1985), attention (Castro and Rudy 1989) or possibly visual acuity (reviewed in Smart and Tonkiss 1985 Tonkiss et al. 1991).
Similarly, the weight of the evidence suggests that short-term memory is intact in previously malnourished animals when tested as adults (discussed below). Although the data on long-term memory are more equivocal (reviewed below), it appears that if there are deficits in this function, they are subtle.
Can we therefore conclude that, with the possible exception of alterations in proactive interference and cognitive flexibility, purely cognitive deficits are not produced by early malnutrition? The answer to this question depends on whether the appropriate cognitive processes were targeted in these studies and whether the tests and test parameters were constructed in such a way that a sensitive index of the target process was provided. The first of these issues will be discussed below; the second will be addressed only within the context of specific examples that are integral to the conclusions reached. It is beyond the scope of this paper to deal with this issue in detail, despite its importance.
Framework for assessing the long-term effects of an early insult in animal models. How does one know, a priori, which particular cognitive processes to assess when using animal models to study human cognitive pathology? Two types of information can be useful in this regard. First, clinical reports or experimental studies with humans can identify cognitive processes or testing conditions that are likely to reveal lasting deficits levee if the human data, considered alone, are inconclusive due to concerns about confounding factors). Second, scrutiny of the ways in which specific neuroanatomical and/or neurochemical systems were affected can lead to informed hypotheses about the cognitive processes that are likely to be affected, drawing upon the increasing database linking particular brain regions and neurochemical systems to specific cognitive functions.
Before considering the ways in which each of these two sources of information can guide test selection in the case of malnutrition, it may be necessary to address the question of why this type of approach is needed. Why can't the experimenter just pick a representative learning or memory test and assume that deficits would be seen if the animal is cognitively impaired? The simple reason for eschewing this latter approach is that it is likely to fail because it does not take into account the biological distinctiveness of different cognitive processes. A particular insult may profoundly impair one cognitive process but leave another entirely intact. The case of global amnesia after damage to the limbic system is illustrative. Despite the profound amnesia for events of everyday life (declarative or explicit memory), implicit memory (for example, the acquisition of cognitive and motor skills; priming) remains unscathed (for further discussion, see Squire 1992, Squire and Zola-Morgan 1991). Another example is the double dissociation seen with respect to allocentric and egocentric spatial memory after parietal and medial frontal lesions (Kesner et al. 1989). This concept has important implications for the study of malnutrition, illustrating that significant cognitive impairment can easily be missed if the appropriate processes are not tapped. This conceptual framework also suggests that it is erroneous to think that one can create a unidimensional sensitivity scale for cognitive tests. The nature of the underlying brain damage will determine which tests are most sensitive in any particular case.
Clues from the clinical literature. Scrutiny of the clinical literature reveals several areas of functioning that are likely to be vulnerable to the enduring effects of early malnutrition, some of which have not been explicitly targeted in studies of previously malnourished subject (either animals or humans). These are each briefly discussed.
Transfer of learning. Small deficits in IQ have consistently been found in previously malnourished humans (for example, Grantham-McGregor et al. 1987; reviewed in Grantham-McGregor 1989, Levitsky and Strupp 1995a, Pollitt et al. 1993, Ricciuti 1993). Notwithstanding the controversy over the cause of this deficit, this finding suggests that the propensity to transfer learning from one situation to another may be an important measure to include in future studies of malnutrition in both animals and humans. Not only is transfer propensity one of the most profoundly affected processes in mental retardation (Brooks et al. 1987, Campione and Brown 1984), but in addition, it correlates with IQ scores in the normal to supranormal range (Campione et al. 1985). Moreover, unlike many correlates of IQ, transfer of learning is readily assessed in animal models (discussed below).
The potential importance of including measures of transfer propensity in the animal studies of early malnutrition is illustrated by the literature on animal models of two mental retardation (MS) syndromes: early hypothyroidism and hyperphenylalaninemia (HP; a model of PKU). Although both of these insults produce profound mental retardation in humans, only subtle impairments were generally detected in studies with animal models. This apparent disparity between the human and animal conditions may stem at least in part from the failure to assess transfer of learning in the animal studies (see Strupp and Levitsky 1990). In these studies, as is common in studies of animal cognition, the animals were generally administered a single learning task under conditions in which they had had no relevant learning experiences. This type of testing situation would entirely miss a deficit in the ability to benefit from related learning experiences, a hallmark deficit in mentally retarded individuals.
To test the idea that this classic testing situation may have underestimated the degree of impairment in animal models of MR, we studied the ability of rats exposed to lactational HP la model of PKU) to master a learning set task (Strupp et al. 1990~. Although the HP and control rats did not differ in the rate at which they mastered the first task, the HP group did not improve across the three successive tasks, unlike the controls. As a result, the HP animals were significantly inferior to their controls on both the second and third discriminations. More recent research on prenatal HP la model of maternal PKU), using a more extensive 10-problem series, supports this idea that the assessment of cumulative learning provides a more sensitive model of IQ deficits than does learning rate of a single task (see Strupp et al. 1994).
Recent data by Pollitt and colleagues (1993) on a supplementation program are consistent with the idea that the propensity to accumulate and transfer learning may be particularly vulnerable to poor nutrition. This study revealed robust effects of an early supplementation pro gram in tests of academic achievement (which are cumulative in nature), in contrast to relatively small effects on tests of specific cognitive functions. The animal data showing a decreased rate of improvement across a successive reversals (Bresler et al. 1975) also provide suggestive evidence of impaired transfer of learning after early malnutrition. It is also possible that the observed impairment of prenatally malnourished animals in two visual discrimination tasks reflects impaired transfer of learning because deficits were observed on the second and third discriminations but not the first (Tonkiss et al. 1991). However, because experimentally naive animals were not tested on these same discrimination tasks, this conclusion must be viewed as tentative.
Executive functions (planning, control of attention, inhibition). Several lines of converging evidence suggest that careful assessment of executive functioning is critical before a conclusion regarding the effects of early malnutrition is reached. First, at least one study of previously malnourished children reports attentional dysfunction and impulsivity (Galler et al. 1983), a behavioral pattern indicative of deficits in executive functioning. A second line of reasoning derives from the theory that executive functions, along with phonological processes, are the most vulnerable to disruption by developmental insults, based on the fact that deficits in these two areas are disproportionately represented in developmental disorders (Pennington 1991a). It is argued that because the prefrontal cortex - the brain region linked to executive functions - evolved relatively recently, it is more likely to be subject to genetic and environmental variation than older, more highly conserved brain regions, such as subcortical structures (see Pennington 1991 a for further discussion). This theoretical framework may explain why two other early insults that produce diffuse and subtle brain changes - developmental exposure to lead or alcohol - also manifest as syndromes in which executive function deficits are prominent (Bellinger et al. 1994, Streissguth et al. 1986). A third reason for an emphasis on executive functions in future studies of previously malnourished individuals is the evidence that impairment in this area can be produced in the absence of IQ deficits (for example, Diamond et al. l992, Fuster 1989), the most commonly used index in past studies of early malnutrition. For example, phenylalanine levels in early treated phenylketonurics were not found to correlate with IQ but are inversely related to measures of executive function (Diamond et al. 1992).
Selection of tests of executive/prefrontal function. The selection of tasks of prefrontal function for future studies of previously malnourished individuals may be guided by several related areas of research. The first pertains to tests that are sensitive to attention deficit hyperactivity disorder (ADHD), a disorder that has been linked to both prefrontal dysfunction (Zametkin et al. 1990, Zametkin et al. 1993) and early malnutrition (Galler et al. 1993). A second guide is provided by tests of specific attentional processes that are commonly used in the cognitive neuroscience literature. The use of these tasks would enable the results concerning malnourished children to be interpreted within a growing wealth of neurological and clinical data. The third guide for task selection in this area pertains to the role of testing conditions in detecting ADHD-like effects. In a final component of this section, this framework is applied to future animal studies, with the presentation of guidelines for selecting tasks of prefrontal dysfunction.
Tests of executive functions that have proven sensitive in revealing deficits in children diagnosed with ADHD may provide a useful guide in selecting tests of executive functions for future studies of malnourished children. Such tests include the following: 1) vigilance tasks such as the Continuous Performance test (CPT); 2) perceptual search tasks, such as the Matching Familiar Figures Test; 3) logical search tasks, such as Raven's Progressive Matrices; 4) motor control and visuomotor tasks, such as the Porteus mazes and the Bender Gestalt Test; 5) tests of cognitive flexibility, such as the Wisconsin Card Sorting Task; 6) planning tasks such as the Tower of Hanoi task; and 7) go-no go tasks (for review, see Pennington 1991b). The literature on maternal social drinking suggests that subtle attentional deficits are more robustly detected by the CPT when the AX subtest rather than the X test is used (Streissguth et al. 1986).
Several lines of evidence point to the importance of assessing attentional functioning in studies of previously malnourished children, particularly those aspects of attention linked to prefrontal cortex. The frameworks proposed by Posner and Mirsky may be useful in this regard. Posner has identified three neuroanatomically distinct subsystems within the realm of visual attention (relating, respectively, to the engage process, the disengage process and the movement of attentional focus; see Posner 1992, Posner and Petersen 1990 for further discussion). The identification of these subsystems is based on data concerning the performance of individuals with specific brain lesions and normal individuals on a simple visual attention task, often in conjunction with PET scans. Mirsky has identified several attentional factors based on factor analytic studies of a neuropsychological test battery (Mirsky 1987, Mirsky 1989, Mirsky et al. 1991). This framework was recently used in a study of low level lead exposure (Bellinger et al. 1995). The fact that both of these frameworks identify biologically distinct components of attention should aid in detecting an attentional impairment in previously malnourished children if it exists. In addition, because these schemes are being used in diverse areas of cognitive neuroscience, their use would provide a rich database to aid in interpreting results obtained with previously malnourished individuals.
An examination of factors that affect the detection of deficits in children with ADHD may help design the optimal testing conditions for detecting subtle deficits in executive functioning. It has been found that many factors affect whether a given task will elicit deficits in ADHD children, including reinforcement schedule Parry and Douglas 1983), number of items and distracters and processing speed [reviewed in Pennington 1991b). In addition, clinical reports suggest that observations in natural settings may be more likely to reveal subtle attentional deficits than the often artificial conditions of the lab. The strategy of observing children in the home was used successfully in detecting subtle attentional deficits in the offspring of social drinkers (Landesman-Dwyer et al. 1981).
This line of reasoning suggests that future studies of early malnutrition in animals should include tasks known to be sensitive to prefrontal damage. Several tasks that fall in this category were used in prior malnutrition research, such as spatial mazes and delayed spatial alternation tasks. As discussed above, one of the most severe impairments reported to date in rats exposed to lactational malnutrition was observed in a delayed spatial alternation task (Castro et al. l989, Castro and Rudy 1993). Other tasks sensitive to prefrontal lesions include delayed response tasks and species-typical behaviors such as food hoarding and nest building (reviewed in Kolb 1990a). Tasks in this latter category may be particularly sensitive to early postnatal lesions in light of the evidence that they reveal minimal recovery of function, unlike spatial mazes, when the lesions are made at the time of maximal sparing postnatal d 10; reviewed in Kolb 1990b). Finally, the rationale implicating executive function deficits would also auger for the sensitivity of vigilance tasks. Although not used frequently in the animal literature on prefrontal lesions, such tasks were developed for both rodents (Carl) et al. 1983, Pang et al. 1992, Strupp and Bunsey 1992, Strupp et al. 1993) and nonhuman primates (Astor-Jones et al. 1994, Kubiak et al. 1992, Rajkowski et al. 1992~. We are aware of no studies in which attentional processes were specifically assessed in previously malnourished animals, with the exception of an ongoing study in our lab (Barnabe, J., Strupp, B. J., and Levitsky, D. A., unpublished observations). However, the pattern of findings obtained in a study of visual discrimination learning in previously malnourished rats is consistent with the hypothesis of impaired attention (Castro and Rudy 1989).
Clues from the reported neuroanatomical and neurochemical changes, With increasing knowledge in behavioral neuroscience, it is possible for behavioral data to suggest the locus of underlying neural damage and, reciprocally, for neural changes to suggest likely functional deficits. Even though the brain changes produced by early malnutrition are diffuse, rendering unlikely specificity in the resulting dysfunction, the nature of the brain changes observed can suggest cognitive processes that should be tapped. An added benefit of this approach is the opportunity to test for correlations between the neural and cognitive changes, which may suggest underlying mechanisms of action. The potential benefits of such knowledge include optimal therapeutic intervention as well as increasing knowledge of brain and behavior relationships.
The application of this strategy to the study of early malnutrition is exemplified here, with a focus on the following three types of enduring brain changes that have been observed: 1) changes in hippocampal structure and electrophysiology; 2) changes in number and/or sensitivitity of beta adrenergic receptors; and 3) cerebellar changes. This list is not intended to be exhaustive but, rather, illustrative of the approach..*
Hippocampal alterations. Enduring changes in hippocampal structure and neurophysiology have been reported after either prenatal or early postnatal malnutrition in rodents (for example, Austin et al. l 986, Bronzino et al. 1990, Jordan et al. 1982, Jordan and Clark 1983, reviewed in Levitsky and Strupp 1995b, Morgane et al. 1993). It has, therefore, seemed reasonable to predict that previously malnourished animals would exhibit impairments in tasks sensitive to hippocampal dysfunction. Several such tasks have been used, primarily those designed to tap short-term, or working, memory.
The effects of prenatal malnutrition on working memory are more consistent in this regard than are those of lactational malnutrition. In the former case, the studies uniformly reported a lack of impairment in spatial working memory, as evidenced by intact performance in the Morris maze (Tonkiss et al. 1994) and a T-maze alternation task (see Tonkiss and Galler 1990). As discussed above, the available evidence concerning lactational malnutrition, albeit less consistent, also points to a lack of impairment in working memory in adulthood, although developmental delays in this function have been reported Castro et al. 1989)
These studies, therefore, permit the conclusion that if there are deficits in working memory after early malnutrition, they are subtle. This caveat, that is, that it is not possible to entirely rule out a subtle deficit, is necessitated by several considerations. First, none of the radial maze or Morris maze studies imposed lengthy retention intervals during which the animal must hold the critical information in memory. It has been shown that some manipulations that alter hippocampal function (such as pharmacological blockade of NMDA receptors) only impair memory in an 8-arm radial maze when delays of at least 15 min are imposed after half of the arms have been chosen (Butelman 1989). It is unlikely that the continuous choice procedure used in the malnutrition studies would have detected an effect of these drugs in an 8-arm maze. Second, the evidence of intact performance of prenatally malnourished rats in spatial alternation tasks may not provide strong evidence of intact working memory, in light of the evidence that the performance drop seen with increasing retention interval may largely reflect proactive interference, rather than a memory deficit per se (Fusser 1989, Strupp and Alber 1994); therefore, delayed alternation tasks may not provide a sensitive index of a subtle memory impairment. Finally, the extent to which any of these tasks, as implemented,9 is sensitive to mild hippocampal dysfunction is unknown, an issue that is critical in light of the fact that the hippocampal changes produced by malnutrition are subtle. Often the difference between animals with hippocampal lesions and controls in studies using these tasks barely achieves statistical significance, thereby precluding the detection of less severe impairment. In fact, lesioning techniques that damage only the hippocampus proper were not always found to impair performance on classic tests of hippocampal function (see Jarrard 19931. For this reason, future studies of early malnutrition would benefit from the inclusion of positive control conditions known to produce subtle hippocampal dysfunction to allow the maximal conclusions to be reached in the event that differences between the control and previously malnourished groups are not observed.
9The literature on hippocampal lesions illustrates that small changes in task parameters or testing conditions can alter whether a given task is sensitive to hippocampal lesions (see Eichenbaum et al. 1992, Jarrard 1993).
The effects on long-term memory function are more equivocal: Two studies observed long-term memory impairment in previously malnourished animals (Celedon et al. 1982, Wetzel et al. 1979), but several other studies did not detect this type of deficit (Bedi 1992, Halas et al. 1979, Nagy and Porada 1991, Rogers et al. 1986). The evidence for enduring long-term memory impairment in previously malnourished animals, notwithstanding the reports to the contrary, suggests that a more thorough analysis of this function is warranted. Converging support for this suggestion is provided by the data, noted above, that pharmacological disruption of hippocampal function was found to impair long-, but not short-term, declarative memory (also see Kesner and& Dakis 1993). Future research should explicitly target long-term memory for events (that is, declarative memory!, as opposed to long-term memory for skills or procedures (that is, nondeclarative memory), on the basis of the evidence that only the former type of long-term memory is dependent on the hippocampus (for example, Squire 1992). Several past studies of long-term memory in previously malnourished animals appear to have tapped nondeclarative memory.
Changes in noradrenergic activity. There is evidence of enduring changes in central noradrenergic activity after early malnutrition. The effects tend to be in the direction of increased norepinephrine (NE) levels coupled with down regulation of beta adrenergic receptors (reviewed in Levitsky and Strupp 1995b). This alteration in central noradrenergic activity provides converging evidence that previously malnourished individuals are likely to react differently to stress. Central noradrenergic activity is intimately related to the organism's ability to respond adaptively during times of stress (Astor-Jones 1985). These alterations in the NE system therefore suggest that differences between previously malnourished individuals and controls may be particularly evident when stressors are imposed.
These observed NE alterations also provide converging support for the postulate that attentional processes are likely to be altered in previously malnourished individuals. Recent data suggest that the relationship between activity of the locus coeruleus (the source of cortical NE) and performance in attention tasks follows an inverted U-shaped dose-response curve, with either too much or too little activity associated with suboptimal performance. This conclusion is based on electrophysiological studies of LC activity in monkeys during vigilance testing (Kubiak et al. 1992, Rajkowski et al. 1992) and pharmacological stimulation of LC activity in rats performing a distraction task (Strupp and Bunsey in press). Both sets of data indicate that altered NE activity (either increased or decreased) is associated with increased distractibility, particularly in tasks involving response inhibition.
One approach that may be useful in revealing behavioral consequences of these receptor changes is to pharmacologically challenge the NE system. One such drug is idazoxan, an alpha2-adrenergic antagonist that at certain doses increases LC firing. This drug has been shown to modulate distractibility in both rats (Strupp and Bunsey in press). and humans (Sahakian et al. 1994, Smith et al. 1992). The observed changes in beta adrenergic receptors suggest that previously malnourished subjects may exhibit a shifting of the dose-response curve in these tasks. This type of result would not only demonstrate a functional effect of altered NE activity that may not be evident in the nondrug state, but, in addition, it would provide converging evidence that previously malnourished subjects respond differently during times of stress.
Cerebellar changes. Early postnatal malnutrition produces enduring changes in the cerebellum of the rat. Compared with well-fed controls, the cerebellum is smaller, contains less DNA and exhibits an altered ratio of granule:Perkinje cells (Bedi et al. 1980a, Bedi et al. 1980b, McConnell and Berry 1978, 1981, Warren and Bedi 1988). These findings suggest that behavioral tests sensitive to cerebellar function may reveal changes in animals malnourished during the early postnatal period. Examples include tests of motor coordination and tests of procedural learning, such as eyeblink conditioning (see Thompson 1989). Consistent with this prediction, enduring changes in gait were reported in previously malnourished rats (Clarke et al. 1992, Gramsbergen and Westerga 1992). although altered reactivity to the test situation cannot be excluded as a possible basis of the observed differences. Tests of: procedural learning were not explicitly studied in previously malnourished animals. However, the fact that many complex learning tasks did not reveal impaired performance of previously malnourished rats, despite the requirement for procedural memory (for example, rule learning and conditioning) indicates that if there are deficits in this aspect of functioning, they are subtle. Nonetheless, it is notable that two skills that were found to be improved with supplementation (for example, Gorman 1995, Pollitt et al. 1993) - reading and writing - are largely procedural tasks, the type that would be expected to be vulnerable to cerebellar damage (discussed in Thompson 1989). As discussed in the previous section on hippocampal changes, it is possible that the tasks used in the animal studies were not sufficiently sensitive to detect subtle alterations in function. Alternatively, it is possible that behavioral function is essentially normal, despite the subtle enduring changes in cytoarchitecture.
Summary and conclusions. Questions remain concerning the functional consequences of early malnutrition. One point of clear consensus, however, is that animals exposed to early malnutrition exhibit lasting changes in the realm of emotionality, motivation and/or anxiety. Researchers have historically discounted these alterations as nuisance factors that complicate the assessment of cognition. However, because such changes affect every sphere of behavioral functioning - from interpersonal relations to cognition - it is important for future research to focus on the nature of these changes, rather than control for them.
The functional integrity of specific cognitive processes is less clear. There is no question that the aforementioned alterations in emotionality and motivation can profoundly affect information processing. But the extent to which cognition is altered, independent of these affective changes, has yet to be established. The fact that previously malnourished animals have been found to perform as well as controls on a variety of complex learning and memory tasks suggests that the ability to reason, remember and solve problems may be largely intact, at least when the testing conditions are such that increased emotionality and/or motivation are not evoked or do not adversely affect performance. This conclusion is consistent with the evidence indicating that previously malnourished children exhibit small, if any, IQ deficits when raised in advantaged circumstances. The only cognitive processes for which enduring changes have been shown in nutritionally rehabilitated animals - outside of effects mediated by these affective changes - are cognitive flexibility and, possibly, susceptibility to proactive interference. However, the inference that these are the only processes affected is not warranted on the basis of the evidence that several cognitive processes likely to be affected have not been fully assessed. On the basis of an appraisal of the available clinical and experimental behavioral data as well as reported neural changes, we identified several cognitive/behavioral effects likely to be most vulnerable to early malnutrition, notably including executive functions linked to the prefrontal cortex (for example, attentional processes, response inhibition and planning). The fact that executive functions appear to be among the processes most vulnerable to early malnutrition contributes to the view that the functional consequences of this insult may have been underestimated. The cognitive tests most frequently used in studies of malnutrition in both humans and laboratory animals are not sensitive to subtle deficits in these functions. For example, children with attention deficit hyperactivity disorder often perform normally on IQ tests, despite the problems in sustained attention, impulsivity and interpersonal relations that dramatically affect their lives.
When considering the implications of the animal studies, it is also important to emphasize the types of effects that such studies would likely miss but that are surely operative in the analogous human condition. First, in the human case, the nutritional insult does not occur in isolation but rather within the context of an impoverished environment, in terms of other biological risk factors (for example, lead exposure, parasitic infection and iron deficiency), adverse psychosocial circumstances (child neglect, poor quality schools) and socioeconomic risk factors (for example, lack of medical care). Consequently, although the study of malnutrition as an isolated variable is instructive, it is also likely to underestimate the consequences of malnutrition in humans where these other risk factors likely synergize in curtailing optimal cognitive development. Second, regardless of whether or not enduring changes in information processing persist after nutritional rehabilitation, previously malnourished children would be expected to exhibit developmental delays as a result of reduced acquisition of information during the period of malnutrition (that is, due to the concurrent effects of malnutrition). In contrast, in most animal studies focusing on the enduring effects of early malnutrition, potentially important information is not available during the period of malnutrition. Although this type of animal study is important in determining if the brain mechanisms underlying the acquisition of information are irreversibly altered, such studies would underestimate the enduring consequences of a protracted period of malnutrition during critical periods of development in childhood. On the other hand, the remarkable resiliency of the brain suggested by these animal studies bodes well for the success of remediative intervention strategies If or example, psychosocial stimulation).
It is hoped that future research will attempt to further elucidate the nature of these unequivocal emotional/motivational changes as well as specifically assess these cognitive processes that appear likely to be most vulnerable to early malnutrition. As discussed above, many of these were not fully tapped in past research. Such efforts should allow a more definitive conclusion concerning the functional consequences of early malnutrition, information that is important for providing the optimal therapeutic intervention.