Electrophysiological Assessment of Pre-Explicit Memory

In At-Risk 6- And 8- Month-Old infants


Research focusing on the developmental outcome of children who experienced pre- or perinatal brain insult has demonstrated that children who are asymptomatic during the newborn period nevertheless often show cognitive deficits later in life. Behavioral measures of memory and cognitive ability during the infancy period have failed to consistently identify this subpopulation of at-risk infants. We designed two experiments that each reflect operations of the medial temporal robe and reflect pre-explicit memory. In the current protocol, electrophysiological techniques were used to examine the neural processing underlying recognition memory during the first year of life. Infants of diabetic mothers and infants suffering from growth retardation were tested on visual recognition memory paradigms at 6 and 8 months of age. Results suggest that fetal neurologic insults do result in abnormal patterns of electrophysiologic response to familiar and novel stimuli during the first year. (Also see Poster #G09; Neurophysiologic Assessment of Auditory Recognition Memory in Newborn Infants of Diabetic Mothers).


The current research protocol was designed to study cognitive brain function using a population of neurologically asymptomatic infants who are nevertheless at-risk for cognitive developmental delay, based on fetal neurologic insult. Such insult may include chronic fetal hypoxia, protein-energy malnutrition, iron deficiency. and repetitive metabolic insults The 6- and 8-month-old infants in these experiments are from a longitudinal study examining infant of diabetic mothers (IDMs) and intrauterine growth retarded infants (IUGRs). Although the neurologic sequelae of these pre-natal clinical syndromes are likely multifactorial, previous research has demonstrate that iron deficiency during the first six post-natal months can result in permanent cognitive and transient-motor sequelae. In this protocol, cord blood ferritin level at birth is used as a marker of fetal iron stores and also as a general marker of neurologic risk, i.e., is correlated with worse hypoxia, more acidosis. Long-term studies of IDM and IUGR children (Rizzo et al., 1994; Winer and Tejani, 1994) have shown that a subgroup of these infants, who are asymptomatic for neurologic sequelae at birth, will demonstrate cognitive deficits, including learning disabilities later in life. These latter findings are buttressed by our own animal models (Rao, 1997) in which prenatal iron deficiency increases the vulnerability of the hippocampus to birth asphyxia. In addition, prenatal iron deficiency selectively decreases the iron content of the hippocampus (deUngria, 1997). Based on these findings we hypothesized that:

1. Overall, infants of diabetic mothers (IDM) and intrauterine growth retarded infants (IUGR) will display abnormal patterns of recognition memory as assessed by measures of event-related potentials when compared to gestational age-matched controls.

2. Given the severity of risk factors associated with IUGR (e.g., protein energy malnutrition, ischemia), the ERP responses of these infants will show more dramatic differences from the control group than will the IDM’s.

3. Clinical and behavioral differences among groups will become more apparent with each step in the longitudinal study, consistent with the subtle nature of the later cognitive deficits.


A. Participants

Participants were recruited pre-natally through physician referral and a Twin Cities based HMO. Control participants were healthy infants born between 38-42 weeks gestation. IUGR infants were selected based on ultrasound criteria indicative of slow growth in utero. The IDM group included both insulin dependent and gestational diabetic mothers.

Etiological Risk factors

1. Hypoxia
2. Iron Deficiency
3. Acidosis
4. Reactive Hypoglycemia

1. Hypoxia
2. Ischemia
3. Protein Energy Malnutrition
4. Iron Deficiency

B. Method

Infants are tested at 6-months and 8-months of age using behavioral measures and event-related potentials (ERP’s). The ERP’s are recorded from 16 electrodes placed at midline and lateral standard electrode locations. Visual ERPs testing recognition memory were recorded to familiar and novel stimuli. Infants were also tested behaviorally at both sessions in a visual paired comparison paradigm.

6-Month Visual Recognition Memory

Infants are shown a familiar face (mother) and a novel face (stranger). Mother’s face is paired with a dissimilar looking stranger. Stimuli are presented for 500 ms, with equal presentations of both faces.

8-Month Cross-Modal Recognition Memory

Infants ave allowed to feel a geometric block for 60 seconds without being able to see it. Infants are then tested visually with the “familiar” block and a novel block. Stimuli are presented for 500 ms, with equal presentations of both blocks.

6- and 8-Month Old Visual Paired Comparison

Infants are familiarized to a pair of identical stimuli (faces) for a fixed period of time (60 sec), and then presented with two counterbalanced, 10 second test trials: the familiar picture paired with a novel picture (reversing left/right positions across trials). We presented infants with two problem sets (i.e., two sets of novel and familiar stimuli).


6 Month ERP Data

The control infants show a larger Nc to the mother’s face than to the stranger’s face. This effect is observed across the midlines and in particular, is greater at T4 than T3, consistent with the hypothesis that the right temporal robe plays a prominent role in face recognition.

The IDM infants reveal an Nc effect at the midline leads that is similar to the controls; however, there are no differences in this component for mother vs. stranger at the lateral temporal leads.

Thus, IDM infants are, in Fact, recognizing the mother’s face and discriminating it from the stranger’s face, but may well be using different neural generators than the control infants, as inferred from the lack of a lateralized effect (right > left).

The latency to the Nc observed to the mother’s face (Ct lead) is longer in the IDM infants (580.83 msec) than in the control infants (539.67 msec), suggesting that the IDM infants are taking longer to recognize the mother’s face. A reasonable interpretation of this finding is the possibility of delayed myelination due to low brain iron (Larkin, E.C., and Rao, G.A., 1990).

The Nc response we are observing in the IUGR infants is opposite to that of the control and IDM infants; specifically, they show a larger Nc to the stranger’s face than to the mother’s face. Moreover, there are no differences between the lateral leads.

Our data suggest that the IUGR infants are showing differential brain activity to the two faces but are unable to recognize their mother’s face. Consistent with this interpretation is the observation that the IUGR infants are showing a positive slow wave (PSW) to the mother’s face, which we interpret as reflecting memory updating for a partially encoded stimulus. (Nelson, C.A., 1994).

8-Month ERP Data

The control infants show recognition of the familiar stimulus based on the return to baseline following the negative component at midline leads; these infants also show evidence of partial encoding and memory updating of the novel stimulus observed by a positive slow wave.

However, the IDM infants are failing to show evidence of discriminating between the novel and familiar stimuli at midline leads, and thus there is no evidence of recognition memory or cross modal transfer.

The IUGR infants are showing a PSW to the familiar stimulus, indicating that they are updating their memory for this stimulus during this test. Thus, the IUGR infants most likely failed to encode this stimulus during familiarization.


Visual Paired Comparison

Set 1:Mean62.03*58.1958.47
Set 2:Mean56.7563.17*53.28

Results indicate that the control infants are showing a novelty preference in Set 1 (p<.01), however, by Set 2, it appears that these infants have mastered the task and thus show a smaller percentage to looking time to the novel stimulus. The IDM infants do not demonstrate a novelty preference in Set 1, however, by Set 2, they have “learned” the protocol and show a significant
preference for novelty (p<.01). The IUGR infants do not indicate any attainment of a novelty preference. From this pattern of findings, we conclude that the control infants show evidence of an intact: memory system. The IDM infants, in contrast, require additional time to master the task requirements.

The IUGR infants show a pattern similar to that of their ERP, in that they fail to show evidence of recognition memory. Again, as with the ERP data, a possible explanation is hippocampal damage due to multifactorial risk.


Visual Paired Comparison

Set 1:Mean58.98*54.0552.62
Set 2:Mean55.8348.2443.18

Results indicate that the control infants are showing a novelty preference in Set 1 (p<.01), however, by Set 2, it appears that these infants have mastered the task and thus show a smaller percentage to looking time to the novel stimulus. The IDM and IUGR infants do not demonstrate any novelty preference in Set 1 or Set 2. In fact, results of Set 2 suggest a familiarity preference indicating that these infants have not fully encoded the familiar stimulus during the familiarization period. For both the IDM and IUGR groups, the failure to show a novelty preference indicates an encoding deficit, consistent with what is evidenced in the ERP data.

Degree of Pre-Natal Neurologic Risk

Using the cord ferritin levels as an indicator of composite neurologic risk, these findings in the low ferritin group are indicative of atypical recognition memory processing. A low serum ferritin at birth is indicative of 1) the degree of fetal iron deficiency (Georgieff, 1990, 199 7) and 2) the degree of chronic intrauterine hypoxia (Georgieff; 1987, 1990). As such, the ferritin reflects the overall degree of abnormal fetal metabolism during diabetes or IUGR. It is of interest that the infants with a newborn serum ferritin of less than 60 ng/mL show waveform morphology similar to that of the IUGR group. The infants with a newborn serum ferritin of greater than 60 ng/mL demonstrate waveforms which appear to differentiate between the familiar and novel stimuli, more consistent with the control group.


Overall, the results of our studies appear to provide support for the hypothesis that prenatal deficiency of essential nutrients such as iron (in the case of IDM infants) and oxygen (in the case of the IUGR infants and possibly the IDM infants) selectively damages structures in the medial temporal lobe (e.g., hippocampus) that are involved in recognition memory. As our longitudinal study progresses, we intend to examine the relation between our early measures of memory and later outcome, as revealed, for example, by functional (e.g., neurophysiological, fMRI) and structural (e.g., MRI) imaging.

IDM Risk Factors

Maternal hyperglycemia—->maternal ketoacidosis—->fetal acidosis—->
fetal hyperglycemia—->fetal hyperinsulanemia—->chronic hypoxia—->
increased red blood cells—->increased iron for red blood cells—->
decreased tissue (e.g., brain) iron—->decreased serum ferritin (less than 60)

IUGR Risk Factors

Placental insufficiency—->decreased oxygen (hypoxia)
 decreased protein energy transport
 decreased iron transport
 increased red blood cells
 increased iron for red blood cells
 decreased tissue (e.g., brain) iron