Introduction: Enhanced models for assessing cognitive function in the neonatal period are imperative in higher animals. Postnatal motor deficits, characteristic of cerebral palsy, emerge in newborn kits within our prenatal rabbit model of hypoxia-ischemia (HI). In humans, prenatal HI leads to intellectual disability and cerebral palsy. In a study examining cognitive function in newborn rabbits, we explored several questions. Is there a distinction between conditioned and unconditioned kits? Can the kits discern the human face or the laboratory coat? Do motorically normal kits, born after prenatal HI, exhibit cognitive deficits? Methods: The conditioning protocol was randomly assigned to kits from each litter. For conditioning, the same human, wearing a laboratory coat, fed the rabbit kits for 9 days before the cognitive test. The 6-arm radial maze was chosen for its simplicity and ease of use. Normally appearing kits, born after uterine ischemia at 79% or 92% term in New Zealand White rabbits, were compared to naïve kits. On postpartum day 22/23 or 29/30, the 6-arm maze helped determine if the kits recognized the original feeder from bystander (test 1) or the laboratory coat on bystander (test 2). The use of masks of feeder/bystander (test 3) assessed confounding cues. A weighted score was devised to address variability in entry to maze arms, time, and repeated-trial learning. Results: In conditioned kits, both naïve and HI kits exhibited a significant preference for the face of the feeder but not the laboratory coat. Cognitive deficits were minimal in normal-appearing HI kits. Conclusion: The weighted score was amenable to statistical manipulation.

It is crucial to identify more effective models in higher animals for assessing cognitive function during the newborn period [1]. This is particularly needed in the investigation of prenatal hypoxia-ischemia (HI), a condition linked to intellectual disability and cerebral palsy [2, 3]. Our laboratory has been dedicated to studying the origins of motor deficits [4‒7]. Additionally, there is an indication that prenatal and perinatal risk factors might contribute to autism [8]. While various animal models have enhanced our understanding of the fetal origins of mental disorders, including autism [9], concerns exist about the reliability of rodent neurobehavioral tests for assessing autism spectrum disorders. Tests involving rodents, such as Y-maze spontaneous alteration, open field novel object recognition for long-term memory, and social approach and interactions with an unfamiliar rat [10], may not accurately represent the manifestation of the human disease. Urgently required are alternative approaches using higher mammal phenotypes to advance our understanding of the mechanisms underlying neurodevelopmental disabilities, especially those tests that exhibit face validity [1].

Our rabbit model of antenatal uterine ischemia leads to fetal global hypoxia mirroring acute placental insufficiency states in humans, such as placental abruption [11, 12]. Global hypoxia in turn leads to HI in fetal brains. Post-birth, some newborn rabbits exhibit motor deficits characteristic of cerebral palsy. As cognitive tests necessitate relatively intact motor function, evaluating cognitive abilities in newborn rabbits is feasible only in those displaying either mild deficits or normal motor function. Similarly, in humans, newborns who exhibit no immediate motor deficits after intrapartum asphyxia face a risk of later cognitive deficits [3].

Recognition of visual stimuli in infants forms a foundational aspect of future cognitive development [13]. In our model of cerebral palsy in rabbits [14], preliminary videos indicated that newborn kits with intact motor function might be recognizing their human caretaker. It has long been observed that 4-week-old rabbit kits readily approach a human hand [15]. However, it remained unclear whether newborn rabbits would exclusively approach their caretakers and not others. To address this question, we employed a straightforward operant conditioning paradigm, using a human face as a discriminative stimulus. While classical conditioning paradigms like eye blink [16, 17] or nictitating membrane conditioning [18] are commonly employed in cognitive testing for rabbits, we opted against them partly because (1) operant conditioning engages selected prefrontal and striatum synapses as opposed to the various hippocampal synapses in classical conditioning [17], and (2) we aimed to develop cognitive testing methods with relevance to intellectual disability and autism in humans.

Herein, we describe the experiments in our rabbit model of prenatal HI and postpartum testing of the newborn rabbits that were exposed to a human caretaker. We tested the following hypotheses: (1) newborn rabbits did not recognize the human face of the caretaker and/or the laboratory coat, and (2) kits with normal motor outcome born to dams subjected prenatal HI would be no different from naïve kits.

Animal Surgery

The surgical procedure has been described in detail previously [7, 19, 20]. Briefly, pregnant New Zealand White dams underwent uterine ischemia at either 25 days of gestation (79% term; E25) or 29 days of gestation (92% term; E29), term being 31.5 days. Dams were anesthetized with an initial regimen of intravenous fentanyl (75 μg/kg/h) and droperidol (3.75 mg/kg/h) followed by epidural anesthesia using 0.75% bupivacaine with continuous infusion of about one-third lower initial intravenous anesthetic dose. A 4F Fogarty arterial embolectomy balloon catheter was inserted into the left femoral artery and the balloon placed above the uterine arteries and below the renal arteries. The balloon was then inflated with saline, and the resultant occlusion of aorta stopped blood flow to the uterus. Body core temperature was monitored with a rectal temperature probe and maintained at 37.5 ± 0.3°C with a water blanket laid under the back, which was connected to a temperature-controlled heating pump. The length of time for inflation of balloon was 32 min in E29 dams or 40 min in E25 dams, taking into account the difference in mortality at different ages with the same amount of uterine ischemia [20]. For purposes of this study, only normally appearing newborn kits following antenatal HI were included in the HI group because the cognitive test performance depended upon intact motor performance.

Randomization and Conditioning

Conditioning of the kits began 9 days before testing on postpartum day 22/23 (P22/23) or postpartum day 29/30 (P29/30). For sake of brevity, these groups are labeled as P22 and P29, respectively. On the first day of conditioning, kits were randomized according to body weight in groups of two. One kit per group was selected randomly to undergo conditioning. During feeding, the feeder administered the kit’s food: milk via a pipette and hay and pellets via hand. To keep the appearance of the original feeder consistent throughout the conditioning process, the human caretaker wore the same laboratory coat and scarf, without any perfume. However, if the kit refused to eat, the feeder would spend the remaining time holding the rabbit. When all the kits completed their conditioning for the day, they were returned to their respective cages. The comparison was made with the kits that were not conditioned.

Feeding by Human Caretaker

Each kit was held for a total of 20 min, and the human caretaker attempted to feed the kit during this time, once a day, for 9 consecutive days, excluding weekends. During conditioning, kits were offered three food items: milk, pellets, and Timothy hay. The milk was created with a 1:10 ratio of rabbit breast milk to Australian Wombaroo (i.e., rabbit milk replacer powder mix). We added rabbit breast milk for the beneficial effect of pheromones on newborn rabbit feeding [21]. It is important to note that the same feeder fed the kits for the 9 days and also conducted the cognition test (author N.S.).

Cognitive Recognition Tests

We chose the simple 6-arm radial maze for the testing apparatus for its ease of use and simplicity in mathematical analysis. Each kit underwent a series of three tests, with five trials conducted per test. The arm containing the positive stimulus of food was a priori labeled as the south side (arm 5 in Fig. 1), and the opposite arm without any food was labeled the north side (arm 2 in Fig. 1). To begin each test, the kit was placed in the center with a hand over its eyes. Once the kit stopped moving, the hand was removed from its eyes and the timer was started. When the kit reached the end of the north or south end, the timer was stopped and its location in reference to the respective stimuli was recorded. In test 1, on the south side (represented by #5), the original feeder was wearing a laboratory coat and the original head scarf with food in their hand. On the north side (represented by #2), the bystander was wearing different head scarf and street clothes, without any food.

Fig. 1.

6-arm maze. Newborn kit placed in the center with head toward west. South is 5 and north is 2. The head of the kit is positioned with head toward west. Weighted scores for every action of the kit in either direction with scores increasing toward arm 5 and decreasing toward arm 2. Maximum score is 50, and minimum score is 0.

Fig. 1.

6-arm maze. Newborn kit placed in the center with head toward west. South is 5 and north is 2. The head of the kit is positioned with head toward west. Weighted scores for every action of the kit in either direction with scores increasing toward arm 5 and decreasing toward arm 2. Maximum score is 50, and minimum score is 0.

Close modal

Scoring of Tests

Entries into Arm 5

To test whether responses of conditioned kits were different from those of unconditioned kits, we just counted the number of entries into arm 5 over 5 trials. However, a score of 0 does not differentiate the near misses from the clearly negative reactions.

Weighted Scoring

We surmised that each entry into the maze arms held significance, with particular attention to the time taken to reach arm 5. In order to address the variability in entry across different maze arms, time variations, and the effects of repeated-trial learning, we devised a weighted score for each test. This scoring system draws inspiration from the cognitive index philosophy employed in mice for the Barnes Maze, where various ordinal scores were assigned based on how the mice reached the goal. However, a limitation was observed due to the almost infinite combinations between ordinal scores [22]. Our weighted score aims to comprehensively account for every outcome in the 6-arm maze. It rewards directional responses toward arm 5 and penalizes responses away from arm 5.

Consequently, a maximum score of 50 is awarded if the kit enters arm 5 on the first trial in less than the median time of all attempts. Conversely, a minimum score of 0 is assigned if it enters arm 2 on the fifth trial in less than the median time (Fig. 1). A priori, we added or subtracted (weighted) the score based on how the kit reached arm 5. Starting from the kit’s head facing west, arm 6 entry incurred the least penalty as the kit maintained the same direction as arm 5. Traveling to arm 4 required passing arm 5, and the failure to enter arm 5 resulted in a larger penalty. Arm 1 incurred a greater penalty as it was farther away from arm 5. To reach arm 3, the kit had to bypass arm 5 and move in the opposite direction, thus incurring the maximum penalty. Since there were 5 trials, one point was deducted for each subsequent trial after the first to account for learning from the test trials themselves. A penalty of 10 was subtracted from the weighted score for arm 5 if the time exceeded the median time. Correspondingly, the score was increased by 10 in arm 2 if the time exceeded the median time. We opted for a threshold scoring system with a single cutoff due to uncertainties regarding the implications of a more granular interval-based system on the developing rabbit kit, especially in terms of its influence on exploration and reward-seeking behavior [23]. The complete scoring guide and the weighting system are summarized in Figure 2, and the rationale for devising the system is included in the online supplementary material (for all online suppl. material, see https://doi.org/10.1159/000538607).

Fig. 2.

Weighted scoring system. Directionality of every action of the kit is scored, and time and number of attempts are factored in.

Fig. 2.

Weighted scoring system. Directionality of every action of the kit is scored, and time and number of attempts are factored in.

Close modal

In test 2, on the south side, the original feeder was wearing street clothes and original head scarf. On the north side, the bystander was wearing a laboratory coat and a different head scarf. This set of experiments tested if the kits recognized the laboratory coat and not the face of the feeder.

In test 3, on the south side, the original feeder was wearing a laboratory coat, a mask with the printed bystander’s face, and food. On the north side, the bystander was wearing a mask with the printed original feeder’s face and wearing a laboratory coat. The use of masks was to factor out confounding effects such as possible recognition by the kit from the odor of the feeder. We studied 27 naïve kits (11 conditioned and 16 unconditioned) and 45 normal-appearing HI kits (23 conditioned and 22 unconditioned), of which 35 were tested at P22 and 37 at P29.

Statistical Analysis

The experiments were designed to answer the following questions: (1) is there any difference between conditioned and unconditioned kits? (2) Do the kits recognize the human face of the caretaker? (3) Or the laboratory coat? (4) Do kits with normal motor outcome following prenatal HI manifest cognitive deficits? We initially assessed the normality of the distribution of our data points (results are not included here for the sake of brevity). The normality was indicated through box and whisker plots, which also illustrated the actual distribution of the data points. The pie chart, bar chart, PROC UNIVARIATE from SAS, and the mean and 95% confidence intervals were also employed. We determined statistical significance using Fisher’s exact tests, paired and two-sample t tests, and two-way ANOVA. To address type I error in multiple t tests and ANOVA comparisons, we applied the Bonferroni correction for multiple tests. If the type 1 error was <0.05 and more than the Bonferroni correction, we labeled the significance as a trend. All statistical tests were done by SAS v9.4 M7 (SAS, Cary, NC, USA).

Is There Any Difference between Conditioned and Unconditioned Kits?

We initially thought the simplest way to analyze cognitive function was to take the aggregate of entries into arm 5 over the 5 trials, with each trial being either 0 (no entry) or 1 (entry in arm 5). Arm 5 was where the original feeder would be seated with food in the 6-arm maze. Figure 3 demonstrates the response to positive and negative stimuli among kits that were unconditioned and conditioned in both the HI and naïve groups during cognitive testing. The unconditioned kits ought to have no preference to either stimuli in arms 2 or 5, as they have not been trained to recognize one over the other. Interestingly, the range of means entering arm 5 in the naïve unconditioned group was 1–3.8 out of maximum 5. This variability is in agreement with the literature of neurobehavioral studies [24‒27] and meant that we had to have enough controls to tease out effects. Conditioning showed a trend for an effect on cognitive function at P22 in all tests (p < 0.05, Fig. 3a–c) and a significant effect at P29 in test 2 (p < 0.004, Bonferroni correction, Fig. 3e). This would suggest that the rabbit kits could recognize the original feeder. As expected, P29 seems to be better than P22 to detect mild deficits of HI using conditioning, as there was trend to have an interaction between age, conditioning, and HI in tests 1 and 2 (p < 0.05, Fig. 3a vs. d and Fig. 3b vs. e) but not for test 3.

Fig. 3.

Arm 5 entry over 5 trials depicted on the ordinate. The three tests depicted in the left column. Rows show the results of each test. The two right columns show the data from P22 (a–c) and P29 (d–f) as box and whisker plots of all the means of the 5 trials shown as black circles. The abscissa shows 4 groups of animals, naïve and post-hypoxia-ischemia (HI), unconditioned (-U) and conditioned (-C). By chance, naïve kits would enter 1–3.8 times in arm 5 (range of means). Conditioning significantly affects the response to entry of arm 5 (*p < 0.05, **Bonferroni correction for multiple comparisons, p < 0.004). Simple main effects analysis showed that both conditioning and HI had a statistically significant effect at P22 in test 1, conditioning alone at P22 and P29 in test 2, and HI alone at P22 and P29 in test 3. There is a statistically significant interaction between the conditioning and HI at P29 in test 1, and at P22 for test 3 (#p < 0.05 by two-way ANOVA, SAS PROC GLM). The directionality of this interaction shown by dashed light green lines for naïve and solid lines for HI. c, d Both suggest that cognition is worsened following HI. Statistically significant interaction between age, conditioning, and HI was found in tests 1 and 2 (§p < 0.05, three-way ANOVA, SAS PROC GLM), suggesting P29 to be better age to detect effects of HI on cognition. Power of statistical tests in ANOVA in f 7–38%.

Fig. 3.

Arm 5 entry over 5 trials depicted on the ordinate. The three tests depicted in the left column. Rows show the results of each test. The two right columns show the data from P22 (a–c) and P29 (d–f) as box and whisker plots of all the means of the 5 trials shown as black circles. The abscissa shows 4 groups of animals, naïve and post-hypoxia-ischemia (HI), unconditioned (-U) and conditioned (-C). By chance, naïve kits would enter 1–3.8 times in arm 5 (range of means). Conditioning significantly affects the response to entry of arm 5 (*p < 0.05, **Bonferroni correction for multiple comparisons, p < 0.004). Simple main effects analysis showed that both conditioning and HI had a statistically significant effect at P22 in test 1, conditioning alone at P22 and P29 in test 2, and HI alone at P22 and P29 in test 3. There is a statistically significant interaction between the conditioning and HI at P29 in test 1, and at P22 for test 3 (#p < 0.05 by two-way ANOVA, SAS PROC GLM). The directionality of this interaction shown by dashed light green lines for naïve and solid lines for HI. c, d Both suggest that cognition is worsened following HI. Statistically significant interaction between age, conditioning, and HI was found in tests 1 and 2 (§p < 0.05, three-way ANOVA, SAS PROC GLM), suggesting P29 to be better age to detect effects of HI on cognition. Power of statistical tests in ANOVA in f 7–38%.

Close modal

We found that kits varied in their responses in the quality of the response. A proportion of kits would enter either of the stimulus arms right away: 19% to arm 2 and 21% to arm 5 (Fig. 4). There were kits who entered the other four arms first (Fig. 4). After a few experiments, we undertook a detailed investigation of multiple entries. The data in Figure 5 constitute a subset of Figure 4. The number of times kits entered the non-stimulus arms varied, with a sizable number entering the four arms other than arms 2 or 5, some of them as many as 4 arms (Fig. 5).

Fig. 4.

Preliminary arm entry depicted as a pie chart. Total number of entries = 943.

Fig. 4.

Preliminary arm entry depicted as a pie chart. Total number of entries = 943.

Close modal
Fig. 5.

Multiple entries to arms other than arms 2 and 5.40% entered arms 2 or 5 and stayed there, but others entered other arms. Some kits would enter the other arms as many as 4 times. Total number of entries = 918.

Fig. 5.

Multiple entries to arms other than arms 2 and 5.40% entered arms 2 or 5 and stayed there, but others entered other arms. Some kits would enter the other arms as many as 4 times. Total number of entries = 918.

Close modal

The time taken to enter arm 5 was skewed in distribution and not normally distributed (Fig. 6). Relying solely on arm entry in arm 5, as depicted in Figure 3, thus fails to provide any indication of the quality of the response or the inherent variability associated with it.

Fig. 6.

Descriptive statistics of the time before entering the stimulus arms. Graphs generated by PROC UNIVARIATE, SAS. Ordinate. The time was censored at 1,966 s, and testing was then stopped at that time. Also the skewed distribution of time in the box and whisker plot. Two-factor ANOVA shows no significant interaction between conditioning and HI in any age or test.

Fig. 6.

Descriptive statistics of the time before entering the stimulus arms. Graphs generated by PROC UNIVARIATE, SAS. Ordinate. The time was censored at 1,966 s, and testing was then stopped at that time. Also the skewed distribution of time in the box and whisker plot. Two-factor ANOVA shows no significant interaction between conditioning and HI in any age or test.

Close modal

Using the weighted score had the advantage that each trial could be scored, and we did not have to use the aggregate of 5 trials, with obvious advantages of the n reflecting the true number of attempts. Now, conditioning shows a significant effect on cognitive behavior, with Bonferroni correction for multiple comparisons of the α error in three of the six figures in Figure 7 as opposed to only one in Figure 3 (main effects comparison in ANOVA). This implies that with conditioning, kits preferred the feeder. We then decided to determine what the origins of this preference are.

Fig. 7.

Weighted score shown as box and whisker plots of all the means of the 5 trials shown as black circles. The setup of three tests depicted in the left column. Columns show age, and rows show the results of each test. The abscissa shows 4 groups of animals, naïve and post-hypoxia-ischemia (HI), unconditioned (-U) and conditioned (-C). Conditioning significantly affects the response to entry of arm 5 (*p < 0.05, **Bonferoni correction for multiple comparisons, p < 0.004). Simple main effects analysis showed both conditioning and HI had a statistically significant effect at P22 in test 1 (a), conditioning alone at P22 and P29 in test 2 (b, e) (ANOVA). There is a statistically significant interaction between the conditioning and HI at both ages in test 2, and at P22 for test 3 (#p < 0.05 by two-way ANOVA, SAS PROC GLM). Dashed light green lines for naïve and solid lines for HI show directionally of interaction. In b, it seems that cognition is improved, while in c and e, cognition is worsened following HI. There is also statistically significant interaction between age, conditioning, and HI in all the tests (§p < 0.05, §§p < 0.004 Bonferroni correction, three-way ANOVA, SAS PROC GLM). Interestingly, there was significant interaction between age and conditioning only in test 3. Power of statistical tests in ANOVA in 5–73% (d) and in 12–27% (f).

Fig. 7.

Weighted score shown as box and whisker plots of all the means of the 5 trials shown as black circles. The setup of three tests depicted in the left column. Columns show age, and rows show the results of each test. The abscissa shows 4 groups of animals, naïve and post-hypoxia-ischemia (HI), unconditioned (-U) and conditioned (-C). Conditioning significantly affects the response to entry of arm 5 (*p < 0.05, **Bonferoni correction for multiple comparisons, p < 0.004). Simple main effects analysis showed both conditioning and HI had a statistically significant effect at P22 in test 1 (a), conditioning alone at P22 and P29 in test 2 (b, e) (ANOVA). There is a statistically significant interaction between the conditioning and HI at both ages in test 2, and at P22 for test 3 (#p < 0.05 by two-way ANOVA, SAS PROC GLM). Dashed light green lines for naïve and solid lines for HI show directionally of interaction. In b, it seems that cognition is improved, while in c and e, cognition is worsened following HI. There is also statistically significant interaction between age, conditioning, and HI in all the tests (§p < 0.05, §§p < 0.004 Bonferroni correction, three-way ANOVA, SAS PROC GLM). Interestingly, there was significant interaction between age and conditioning only in test 3. Power of statistical tests in ANOVA in 5–73% (d) and in 12–27% (f).

Close modal

Do the Kits Recognize the Human Face of the Caretaker?

There was a preference for the face of the feeder in the HI-conditioned group at P22 in tests 1 and 2 (p < 0.05, Fig. 7a, b) but significant again at P29 in test 2 (p < 0.004, Bonferroni correction, Fig. 7e). With the mask of the feeder in arm 2, the difference between conditioned and unconditioned became nonsignificant in test 3 at P22 when compared to test 2 (Fig. 7c). This would suggest the rabbit kit was influenced by the mask to a milder degree than the actual face and with possible mild influence of smell in the wrong direction. There was a significant interaction between age, conditioning, and HI in test 2 (p < 0.004, Bonferroni correction, Fig. 7b vs. e), and trend for an interaction in tests 1 and 3 (p < 0.05, Fig 7a vs. d and Fig. 7c vs. f). This confirms the conclusion from Figure 3 that P29 is better than P22 to detect mild deficits of HI using conditioning.

Next, we examined the trend observed in each successive trial as depicted in Figure 8. When focusing on the conditioned groups, it becomes apparent that rabbits appear to recognize the face of the feeder in tests 1 and 2. In test 3, the face mask is now situated in arm 2. A lower score indicates that the kits are drawn toward the mask. The cumulative findings across the three tests strongly suggest that the kits’ preference for the feeder originates from recognizing the feeder’s face. The weighted score did partially consider the impact of repeated trials. No significant changes were noted between trial 1 and trial 5 in any of the tests at any age, considering the Bonferroni correction for multiple tests for α error. A noteworthy trend of differences between trial 1 and trial 5 was identified in HI for test 2 at P22, suggesting that the kit learned less (p < 0.05, Fig. 8e). Conversely, an opposing trend, indicating that kits learned with trials and exhibited a greater preference for arm 2, was observed in test 3 for HI_C at both P22 and P29 (p < 0.05, Fig. 8f, l), even with the weight against repeated trials in the weighted score. While this effect might be partially attributed to the food being held by the bystander in arm 2 in test 3, the lower scores of trial 5 versus 1 in the Naïve_U group, with no feeder influence, argue against this explanation.

Fig. 8.

Effect of Trials. Weighted score depicted on the ordinate. Rows show the results of each test. The columns show the data from P22 and P29 with means (circles) and 95% confidence intervals (a–l, shaded portion). The abscissa shows trials. A neutral score is 25. Higher than 25 favors the feeder in tests 1 and 2, and a score <25 favors the feeder in test 3. The arrows depict the implications of deviation from 25. Yellow lines joining yellow circles depict lines joining the means in unconditioned groups, while blue lines joining blue circles depict conditioned group means’ line. The conditioned groups are offset a little for better clarity. Naïve groups are in light colors and HI in darker shades. There was no significant change between trial 1 versus trial 5 using Bonferroni correction for α error. However, * for p < 0.05 is shown for trends, shown in (e, d, l). The trend in (e) if anything indicates the rabbit is learning less with more trials, while the trends in (f) and (l) indicate the rabbit could be learning with more trials. Also, most of the area in the confidence intervals in blue is over 25 in tests 1 and 2, and lower than 25 in test 3 in the conditioned groups (compared to unconditioned groups), indicating that the kits seem to recognize the face of the feeder, confirming the findings in Figure 7.

Fig. 8.

Effect of Trials. Weighted score depicted on the ordinate. Rows show the results of each test. The columns show the data from P22 and P29 with means (circles) and 95% confidence intervals (a–l, shaded portion). The abscissa shows trials. A neutral score is 25. Higher than 25 favors the feeder in tests 1 and 2, and a score <25 favors the feeder in test 3. The arrows depict the implications of deviation from 25. Yellow lines joining yellow circles depict lines joining the means in unconditioned groups, while blue lines joining blue circles depict conditioned group means’ line. The conditioned groups are offset a little for better clarity. Naïve groups are in light colors and HI in darker shades. There was no significant change between trial 1 versus trial 5 using Bonferroni correction for α error. However, * for p < 0.05 is shown for trends, shown in (e, d, l). The trend in (e) if anything indicates the rabbit is learning less with more trials, while the trends in (f) and (l) indicate the rabbit could be learning with more trials. Also, most of the area in the confidence intervals in blue is over 25 in tests 1 and 2, and lower than 25 in test 3 in the conditioned groups (compared to unconditioned groups), indicating that the kits seem to recognize the face of the feeder, confirming the findings in Figure 7.

Close modal

Test 2: Was the Laboratory Coat the Discriminative Stimulus Influencing Preferences?

During the conditioning process, the original feeder consistently wore a laboratory coat. Test 2 sought to eliminate the laboratory coat as a positive stimulus that would prompt the kits to choose the original feeder over the bystander. A comparison of Figure 7a and b, as well as Figure 7d and e, suggests that there was no reduction in preference for the original feeder (arm 5) when the laboratory coat was worn by the bystander (arm 2). We employed paired t tests to specifically analyze the influence of the laboratory coat alone. The results revealed no significant preference for the laboratory coat in the Δ Score (or difference) between two tests, which would show up with more data points in the white area rather than the green area, indicating a decrease of Δ Score below zero in test 2-1 and test 2–3, or a subsequent further increase of the data points in the green area, i.e., above zero in test 1–3 compared to test 2–3 (Fig. 9). In fact, the opposite is true, i.e., more data are in the green area, above zero, in tests 2-1 and 2–3, and no further increase in test 1–3 from test 2–3 (Fig. 9). The preference for the human face over the laboratory coat provides confirmation that kits recognized the human face.

Fig. 9.

Paired comparisons of tests. Box and whisker plots with distribution of data in black circles. The difference between the weighted scores of 2 tests is depicted on the ordinate. The columns show the data from P22 and P29. Naïve groups on the top row and HI on the bottom row. The zero line represents a neutral result. The green shaded area shows the preference for the face of the feeder. The implications of the changes are shown in C with arrows. A preference for the face is observed mostly in conditioned kits by the significant deviations from zero (*p < 0.05, **p < 0.004 Bonferroni correction, paired t test). a, c The significant effects of conditioning are confirmed in P22 (#p < 0.05, two-sample t test). There does not seem to be a preference for the laboratory coat (b, d). A significant interaction of age and conditioning was only observed for tests 1–3 (§p < 0.05, three-way ANOVA, SAS PROC GLM) but not for other pairings.

Fig. 9.

Paired comparisons of tests. Box and whisker plots with distribution of data in black circles. The difference between the weighted scores of 2 tests is depicted on the ordinate. The columns show the data from P22 and P29. Naïve groups on the top row and HI on the bottom row. The zero line represents a neutral result. The green shaded area shows the preference for the face of the feeder. The implications of the changes are shown in C with arrows. A preference for the face is observed mostly in conditioned kits by the significant deviations from zero (*p < 0.05, **p < 0.004 Bonferroni correction, paired t test). a, c The significant effects of conditioning are confirmed in P22 (#p < 0.05, two-sample t test). There does not seem to be a preference for the laboratory coat (b, d). A significant interaction of age and conditioning was only observed for tests 1–3 (§p < 0.05, three-way ANOVA, SAS PROC GLM) but not for other pairings.

Close modal

Do Kits with Normal Motor Outcome following Prenatal HI Manifest Cognitive Deficits?

The effect of prenatal HI on cognitive deficits utilizing the 6-arm maze test showed mixed results. At P22, in test 2, there is an effect of HI to have a trend to greater preference for the face over the laboratory coat with conditioning (p < 0.05, interaction between HI and conditioning for weighted scores, Fig. 7b). In contrast, arm 5 entry shows the opposite trend at P29 for test 1 (p < 0.05 for interaction between HI and conditioning, Fig. 3d), even though there was not a trend in the weighted scores (Fig. 7d).

Interestingly, time to completion was significantly decreased in HI at P22 in test 2 (data not shown for brevity, p < 0.004, Bonferroni correction, two-way ANOVA). However, the comparison of findings of test 2-1 and test 2–3 in Figure 9a and c do not confirm HI to have a greater preference for the face, and the main effects showed no trend for HI for face preference. At P29, in test 2, HI resulted in a trend for less preference for the face over the laboratory coat (p < 0.05 for interaction between HI and conditioning, Fig. 7e). There is a significant interaction between age, conditioning, and HI in the three-way ANOVA, by Bonferroni correction, indicating P29 to have less preference for the face in HI-conditioned kits compared to P22. This finding could be explained by (1) a possible brain injury that manifests at P29 but not at P22, or (2) at P29, the kits had a greater awareness of the laboratory coat as another visual cue. However, the second explanation is not borne out by the paired testing where test 1–3 scores should be greater than 2–3 scores in Figure 9 if the laboratory coat played any role. At the very least, it suggests that for testing any possible cognitive damage from HI, P29 would be the better age to conduct the cognitive testing. At P22, in test 3, there was a trend for interaction between HI and conditioning, suggesting that HI resulted in less preference of the mask of the feeder (Fig. 7c). Overall, normally appearing HI kits show either mild deficits or no difference compared to naïve kits.

Do Sex Differences Explain the HI Difference with Naïve?

The percentage of females at P22 in naïve and HI were 53 and 50%, respectively; for P29 in naïve and HI, it was 56 and 64%, respectively. We did not find any statistically significant difference in sex scores, comparing HI with naïve kits (see Fig. 10). Nor did we find any interaction for sex, age, conditioning, and HI for weighted scores in any of the tests.

Fig. 10.

a-d No difference with sex. Weighted score in ordinate. Pink circles are female and blue circles represent male. Yellow lines represent naïve, and purple lines represent HI. For convenience, the dashed yellow line shows the neutral score of 25. Means and standard errors of mean depicted. There was no significant interaction with sex, age, HI, and conditioning.

Fig. 10.

a-d No difference with sex. Weighted score in ordinate. Pink circles are female and blue circles represent male. Yellow lines represent naïve, and purple lines represent HI. For convenience, the dashed yellow line shows the neutral score of 25. Means and standard errors of mean depicted. There was no significant interaction with sex, age, HI, and conditioning.

Close modal

Which Age Is Better for Testing Cognitive Deficits?

To summarize, statistically significant interaction between age, conditioning, and HI was found in weighted score in test 2 (§§p < 0.004, Fig. 7b vs. e). There was a trend for the interaction between age, conditioning, and HI for weighted score in tests 1 and 3, and for arm 5 entry in tests 1 and 2 (§p < 0.05), suggesting P29 to be better age to detect effects of HI on cognition. Using paired comparisons, significant interaction of age and conditioning was only observed for test 1–3 (§p < 0.05) but not for other pairings.

This study marks the first demonstration of newborn rabbits’ capacity to recognize the human face and exhibit a preference for it over the laboratory coat. This assertion is substantiated by several factors: (1) notable changes in arm 5 entry and weighted score in P29 naïve kits during test 2 (p < 0.004, Fig. 3e, Fig. 7e) following conditioning, (2) visual confirmation from the confidence intervals predominantly exceeding 25 in tests 1 and 2, and falling below 25 in test 3 for the conditioned groups as opposed to the unconditioned groups in Figures 3 and 8 significant deviations from zero at P22 in test 2–3 for Naïve_C and test 1–3 for HI_C. While the ability of newborn rabbits to recognize the feeder has been empirically suspected based on their behavior during feeding times, it has never been systematically demonstrated until now. This study also provides insight into how rabbit kits approach cognitive tasks, representing a preliminary step in assessing the translational implications of rabbit cognitive abilities [28].

In this study, a weighted scoring system was employed to account for various confounding factors, providing an unbiased summary of the overall behavior of newborn rabbits. This approach facilitates the use of parametric statistical tests for the analysis of cognitive behavior. The weighted score enables readers to easily discern the general inclination toward either favoring the face or the laboratory coat, as indicated by the threshold of 25 in Figures 7 and 8. Additionally, it allows the utilization of a Δ score, offering insights into the overall preference as depicted in Figure 9. The key advantage of using weighted scoring lies in its ability to mitigate investigator bias in selecting an endpoint that demonstrates significance. This approach also circumvents the challenges associated with transforming various endpoints into a z score [29], which is limited to normalized, interval-distribution data and may exaggerate small differences.

The predetermined amount of weighting in the weighted score was established a priori in our study. Further refinement and increased precision in the weighting could be achieved with the accumulation of additional data. In hindsight, the weighting against the learning from repeated trials appears to have been appropriately balanced, as indicated by two instances of greater learning counterbalanced by one instance of reduced learning in Figure 8. However, one limitation of our weighted score likely lies in the time to completion weight, as we only utilized a median cutoff for the actual times. This concept of weighting could be extended to an Open Field test, but a substantially larger database would be necessary to precisely determine the appropriate amount of weighting. The notably shorter time taken for completion in test 2 by P22 HI kits is intriguing. The question of whether cognitively damaged kits make quicker, but erroneous decisions remains unanswered.

The primary aim of this study was not originally centered around investigating the developmental timeline of cognitive abilities; instead, our focus was on determining whether newborn rabbit kits could be effectively tested through a conditional paradigm. Initially, we anticipated that conditioning would have a more pronounced impact on P29 kits compared to P22 kits. This expectation was validated by the significant interaction observed between age, conditioning, and HI in the weighted score during test 2 (p < 0.004). Consequently, P29 may be a more optimal age for assessing cognitive deficits than P22. Nevertheless, the distinctions between the two age groups are not highly pronounced, and there is the additional consideration of the increased cost associated with ensuring the kits survive an extra week.

Previous researchers have employed the T-maze to evaluate cognitive function in 4-week-old rabbits, implementing a 15-min interval between a sample run and a test run [26]. The T-maze is purportedly designed to investigate early spatial learning and short-term memory. A commonality between this study and the current one is the observed variability in unweighted endpoints across repeated trials. However, the six-arm maze provides more choices than the two in a T-maze, reducing the likelihood of entering arm 5 in our study being a random occurrence. In our study, we did not see any progression over the five trials or any significant difference between test 1 and test 5. There is increased likelihood of chance misrepresentation of the cognitive state when there are only two trials [26]. The wide confidence intervals in two consecutive trials in Figure 8 support this likelihood. It is conceivable that the discrepancy in findings could be attributed to the use of a different rabbit species, as the previous investigators utilized a hybrid of New Zealand White/Black and Flemish Giant rabbit [26].

Determining endpoint variables in cognitive tests poses a challenge due to the inherent variability of responses [24‒27]. Consequently, investigators are often compelled to selectively choose endpoint variables. For instance, in the Open Field test, only specific elements such as time spent with a moved object, the percentage of investigated time in a new position, and the discriminative index (the ratio of the difference between time spent in novel and familiar locations to total time spent on both locations) demonstrated differences between term and preterm born kits [30], among various theoretical elements that could be tested. In the Barnes Maze, ordinal scores are assigned based on endpoints like direct targeting, corrected targeting with a mild detour, focused searching with a moderate detour, long corrections with an initial approach to the opposite side, serial scoring based on a circular route, or random/failure where no pattern is discerned by the human eye [22]. Unfortunately, the ordinal scores may not capture all infinite combinations accurately. The time spent at each position in the Open Field or Barnes Maze may not depend upon spatial recognition but could be influenced by other factors, for example, laziness. Additionally, it may not be appropriate to apply parametric tests to populations that combine negative and positive discriminative indices. In cases where marked preferences for a novel object are offset by marked preferences for a familiar object, caution is needed to avoid equating such marked preferences with the absence of preference entirely. Similar considerations apply to the modified Barnes Maze test when using endpoints like time spent and latency. Notably, the data from the three repetitive training trials were not utilized in the assessment of the Barnes Maze test in a prior study [30].

Investigators studying cognition have grappled with various questions, such as (1) the duration allocated for observation in each trial. For instance, a 20-min time limit was imposed in studies involving mice, and mice failing to explore for at least 20 s were also excluded [31]. In a study of similar aged rabbits, a 10-min time limit was applied for the Open Field test, with a 2-min limit for training and 90 s for evaluation in the Barnes Maze [30]. (2) Another question concerns the variability of the readout. Due to the inherent variability in responses, it is common for responses to be categorized into ordinate classifications of high and low performance [31], which, in turn, poses challenges in utilizing the statistical power of linear statistics. The skewed responses observed in the time to enter arm 5 (Fig. 6) also precluded the use of time-based cognitive indices [30], such as the Discrimination Index used in 70-day rabbits [24] or P27–30 rabbits [30], which assume a linear response.

Some cognitive tests used for rodent evaluation may not be suitable for the rabbits. Rabbits are perinatal brain developers [32] compared to postnatal brain development in rodents. Rabbits usually freeze when confronted with an unfamiliar caregiver [33]. It would have been attractive to use a cognitive test that did not depend on intact motor function, such as the touchscreen test in rodents [34]. However, the challenges of using the touchscreen technology in rabbits are freezing behavior, smell deficits in HI [35, 36], larger size of rabbits, and prohibitive cost [34]. Also, the 50 days needed to train for eligibility for testing [37], and the high threshold of 80% total accuracy, 20% omission rate, and 100–150 trials completed in 60 min [38] may render the test unsuitable for newborn rabbit neurodevelopmental testing. Of note also, the very touchscreen testing procedure acts as a cognitive enhancer [37], which may thus mask cognitive deficits from the experimental group.

In our previous work, we demonstrated that newborn rabbits possess a keen sense of smell, and after antenatal HI events, they exhibit deficits in olfactory sensory function [35, 36]. In our study, there was always the possibility that cognitive recognition of the feeder may occur through olfactory rather than visual cues. If the rabbit relies predominantly on olfactory cues, then the mask of the bystander in test 3 should have no impact. However, in Figure 9, test 1–3 displays a positive trend across all conditioned groups, suggesting that newborn rabbits primarily use visual cues for cognitive recognition, While some instances of a negative trend in test 1–3 suggest potential reliance on olfactory cues, an evaluation of test 2–3 confirms the predominant use of visual cues, specifically toward the face of the feeder. It is plausible that newborn rabbits utilize both visual and olfactory cues, but when faced with contradictory cues, the preference is for visual cues. A potential avenue for future research could involve conditioning rabbits entirely in the dark to assess olfactory cognitive recognition.

The HI kits in our study were normally appearing on motor examination at P1–P11. This self-selection of normal kits thus excluded all the kits with mild-to-severe motor deficits and perinatal deaths following antenatal uterine ischemia in the dams. So, it is not surprising that this subgroup of HI kits showed mild effects if any when compared to naïve kits. In humans, the risk ratio of obtaining a Stanford-Binet IQ of <70 at 4 years of age in placental abruption is only 1.66 (confidence intervals 1.17, 2.36) [39]. However, 80% of the placental abruption have an IQ above 84, not different from the 78% in non-placental abruption newborns [39]. The initial newborn presentation in humans has only a weak correlation with later motor deficits and cognitive deficits. In our study, P29 kits in the HI-conditioned group showed a diminished preference for the face compared to P22 kits. Additionally, our findings indicate that conditioning had a beneficial impact on both naive and HI kits, enhancing cognitive recognition. This suggests that conditioning could potentially mitigate cognitive damage in the context of HI. The proper execution of the 6-arm maze investigation relies on intact motor function. As we specifically selected HI kits with normal motor function immediately after birth for this study, it is likely that these kits experienced either very mild or no brain injury, or they had recovered from the initial HI insult, exhibiting normal cognitive function during later testing. It is unknown how the more severely brain-injured kits would perform on cognitive tests. In cases of motor impairment, adjustments to the weighting score might be necessary, considering that the time required to reach arm 5 or arm 2 could be longer. A future study could focus on a specific population with mild-to-moderate motor deficits, which has enough motor function to ambulate. Most unilateral and bilateral spastic CP patients have normal IQ [40], so even moderately severe motor deficits in rabbits could have normal cognitive function.

Interestingly, if our cognitive tests are dependent on the recognition of the face then these tests could possibly be used to investigate prosopagnosia. Prosopagnosia or face blindness and zooagnosia (blindness to animals) can be congenital in origin or acquired later. Developmental prosopagnosia is often observed in autism [41] and Asperger’s syndrome [42]. The ability to test human face recognition may be useful for testing prosopagnosic tendencies in newborn animals, which may be useful for investigators of pediatric stroke, pediatric trauma in addition to investigating developmental origins of autism or Asperger syndrome. The present cognitive tests may also be used for other models leading to newborn neurobehavioral deficits [43‒46].

In conclusion, our unbiased study demonstrates that conditioned newborn rabbits not only recognize the human face but also exhibit a preference for it over the laboratory coat, and potentially, conflicting olfactory cues. Newborn rabbits born to mothers who underwent antenatal HI generally display minimal to no deficits at 3–4 weeks of age. Additionally, we introduce a weighted scoring system for the 6-arm maze, which accounts for the contribution of all experimental trials and enables linear statistical analysis. The weighted scoring holds promise for future investigations into cognitive function.

This study protocol was reviewed and approved by the IACUC of Wayne State University, approval number (IACUC -19-06-1161).

The authors have no conflicts of interest to declare.

S.T. received NIH funding support (NINDS 1R01NS081936, NS114972, and NS117146), and N.S. received ReBUILDetroit WSU Program support. These funding sources are not involved in the preparation of data or the manuscript.

Zhongjie Shi: designed the study, acquisition of data for the work, reviewed the manuscript critically for important intellectual content, final approval of the version to be published, and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Nadiya Sharif: acquisition and analysis of data for the work, reviewed the manuscript critically for important intellectual content, final approval of the version to be published, and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Kehuan Luo: acquisition of data for the work, reviewed the manuscript critically for important intellectual content, final approval of the version to be published, and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Sidhartha Tan: designed the study, drafting the work, final approval of the version to be published, and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The data that support the findings of this study are not publicly available due to IACUC regulations but are available from the corresponding author upon reasonable request.

1.
Silverman
JL
,
Thurm
A
,
Ethridge
SB
,
Soller
MM
,
Petkova
SP
,
Abel
T
, et al
.
Reconsidering animal models used to study autism spectrum disorder: current state and optimizing future
.
Genes Brain Behav
.
2022
;
21
(
5
):
e12803
. .
2.
Vexler
ZS
,
Ferriero
DM
.
Molecular and biochemical mechanisms of perinatal brain injury
.
Semin Neonatol
.
2001
;
6
(
2
):
99
108
. .
3.
Rennie
JM
,
Hagmann
CF
,
Robertson
NJ
.
Outcome after intrapartum hypoxic ischaemia at term
.
Semin Fetal Neonatal Med
.
2007
;
12
(
5
):
398
407
. .
4.
Vasquez-Vivar
J
,
Shi
Z
,
Luo
K
,
Thirugnanam
K
,
Tan
S
.
Tetrahydrobiopterin in antenatal brain hypoxia-ischemia-induced motor impairments and cerebral palsy
.
Redox Biol
.
2017
;
13
:
594
9
. .
5.
Vasquez-Vivar
J
,
Shi
Z
,
Jeong
JW
,
Luo
K
,
Sharma
A
,
Thirugnanam
K
, et al
.
Neuronal vulnerability to fetal hypoxia-reoxygenation injury and motor deficit development relies on regional brain tetrahydrobiopterin levels
.
Redox Biol
.
2020
;
29
:
101407
. .
6.
Shi
Z
,
Luo
K
,
Deol
S
,
Tan
S
.
A systematic review of noninflammatory cerebrospinal fluid biomarkers for clinical outcome in neonates with perinatal hypoxic brain injury that could be biologically significant
.
J Neurosci Res
.
2022
;
100
(
12
):
2154
73
. .
7.
Shi
Z
,
Luo
K
,
Jani
S
,
February
M
,
Fernandes
N
,
Venkatesh
N
, et al
.
Mimicking partial to total placental insufficiency in a rabbit model of cerebral palsy
.
J Neurosci Res
.
2022
;
100
(
12
):
2138
53
. .
8.
Kolevzon
A
,
Gross
R
,
Reichenberg
A
.
Prenatal and perinatal risk factors for autism: a review and integration of findings
.
Arch Pediatr Adolesc Med
.
2007
;
161
(
4
):
326
33
. .
9.
Wang
B
,
Zeng
H
,
Liu
J
,
Sun
M
.
Effects of prenatal hypoxia on nervous system development and related diseases
.
Front Neurosci
.
2021
;
15
:
755554
. .
10.
Vanderplow
AM
,
Kermath
BA
,
Bernhardt
CR
,
Gums
KT
,
Seablom
EN
,
Radcliff
AB
, et al
.
A feature of maternal sleep apnea during gestation causes autism-relevant neuronal and behavioral phenotypes in offspring
.
PLoS Biol
.
2022
;
20
(
2
):
e3001502
. .
11.
Tan
S
,
Drobyshevsky
A
,
Jilling
T
,
Ji
X
,
Ullman
LM
,
Englof
I
, et al
.
Model of cerebral palsy in the perinatal rabbit
.
J Child Neurol
.
2005
;
20
(
12
):
972
9
. .
12.
Derrick
M
,
Drobyshevsky
A
,
Ji
X
,
Tan
S
.
A model of cerebral palsy from fetal hypoxia-ischemia
.
Stroke
.
2007
;
38
(
2 Suppl l
):
731
5
. .
13.
Rose
SA
,
Feldman
JF
,
Jankowski
JJ
.
The building blocks of cognition
.
J Pediatr
.
2003
;
143
(
4 Suppl l
):
S54
61
. .
14.
Derrick
M
,
Luo
NL
,
Bregman
JC
,
Jilling
T
,
Ji
X
,
Fisher
K
, et al
.
Preterm fetal hypoxia-ischemia causes hypertonia and motor deficits in the neonatal rabbit: a model for human cerebral palsy
.
J Neurosci
.
2004
;
24
(
1
):
24
34
. .
15.
Pongrácz
P
,
Altbäcker
V
.
The effect of early handling is dependent upon the state of the rabbit (Oryctolagus cuniculus) pups around nursing
.
Dev Psychobiol
.
1999
;
35
(
3
):
241
51
. .
16.
Thompson
LT
,
Moyer
JR
Jr
,
Disterhoft
JF
.
Transient changes in excitability of rabbit CA3 neurons with a time course appropriate to support memory consolidation
.
J Neurophysiol
.
1996
;
76
(
3
):
1836
49
. .
17.
Gruart
A
,
Leal-Campanario
R
,
López-Ramos
JC
,
Delgado-García
JM
.
Functional basis of associative learning and its relationships with long-term potentiation evoked in the involved neural circuits: lessons from studies in behaving mammals
.
Neurobiol Learn Mem
.
2015
;
124
:
3
18
. .
18.
Schreurs
BG
,
Smith-Bell
CA
,
Burhans
LB
.
Incubation of conditioning-specific reflex modification: implications for post traumatic stress disorder
.
J Psychiatr Res
.
2011
;
45
(
11
):
1535
41
. .
19.
Derrick
M
,
Drobyshevsky
A
,
Ji
X
,
Chen
L
,
Yang
Y
,
Ji
H
, et al
.
Hypoxia-ischemia causes persistent movement deficits in a perinatal rabbit model of cerebral palsy: assessed by a new swim test
.
Int J Dev Neurosci
.
2009
;
27
(
6
):
549
57
. .
20.
Derrick
M
,
Englof
I
,
Drobyshevsky
A
,
Luo
K
,
Yu
L
,
Tan
S
.
Intrauterine fetal demise can be remote from the inciting insult in an animal model of hypoxia-ischemia
.
Pediatr Res
.
2012
;
72
(
2
):
154
60
. .
21.
Keil
W
,
von Stralendorff
F
,
Hudson
R
.
A behavioral bioassay for analysis of rabbit nipple-search pheromone
.
Physiol Behav
.
1990
;
47
(
3
):
525
9
. .
22.
Souza
KA
,
Powell
A
,
Allen
GC
,
Earnest
DJ
.
Development of an age-dependent cognitive index: relationship between impaired learning and disturbances in circadian timekeeping
.
Front Aging Neurosci
.
2022
;
14
:
991833
. .
23.
O’Doherty
JP
,
Cockburn
J
,
Pauli
WM
.
Learning, reward, and decision making
.
Annu Rev Psychol
.
2017
;
68
:
73
100
. .
24.
Illa
M
,
Eixarch
E
,
Batalle
D
,
Arbat-Plana
A
,
Muñoz-Moreno
E
,
Figueras
F
, et al
.
Long-term functional outcomes and correlation with regional brain connectivity by MRI diffusion tractography metrics in a near-term rabbit model of intrauterine growth restriction
.
PLoS One
.
2013
;
8
(
10
):
e76453
. .
25.
Gümüş
HG
,
Agyemang
AA
,
Romantsik
O
,
Sandgren
R
,
Karlsson
H
,
Gram
M
, et al
.
Behavioral testing and litter effects in the rabbit
.
Behav Brain Res
.
2018
;
353
:
236
41
. .
26.
van der Merwe
J
,
van der Veeken
L
,
Inversetti
A
,
Galgano
A
,
Toelen
J
,
Deprest
J
.
Earlier preterm birth is associated with a worse neurocognitive outcome in a rabbit model
.
PLoS One
.
2021
;
16
(
1
):
e0246008
. .
27.
van der Merwe
J
,
van der Veeken
L
,
Inversetti
A
,
Galgano
A
,
Valenzuela
I
,
Salaets
T
, et al
.
Neurocognitive sequelae of antenatal corticosteroids in a late preterm rabbit model
.
Am J Obstet Gynecol
.
2022
;
226
(
6
):
850.e1
850.e21
. .
28.
Gazes
RP
,
Lazareva
OF
.
Does cognition differ across species, and how do we know? Lessons from research in transitive inference
.
J Exp Psychol Anim Learn Cogn
.
2021
;
47
(
3
):
223
33
. .
29.
Meng
B
,
Li
X
,
Lu
B
,
Liu
R
,
Yuan
H
,
Zhai
X
, et al
.
The investigation of hippocampus-dependent cognitive decline induced by anesthesia/surgery in mice through integrated behavioral Z-scoring
.
Front Behav Neurosci
.
2019
;
13
:
282
. .
30.
Sharma
DR
,
Cheng
B
,
Jaiswal
MK
,
Zhang
X
,
Kumar
A
,
Parikh
N
, et al
.
Elevated insulin growth factor-1 in dentate gyrus induces cognitive deficits in pre-term newborns
.
Cereb Cortex
.
2023
;
33
(
10
):
6449
64
. .
31.
Bonds
JA
,
Tunc-Ozcan
E
,
Dunlop
SR
,
Rawat
R
,
Peng
CY
,
Kessler
JA
.
Why some mice are smarter than others: the impact of bone morphogenetic protein signaling on cognition
.
eNeuro
.
2023
;
10
(
1
):
ENEURO.0213-22.2022
. .
32.
Harel
S
,
Shapira
Y
,
Hartzler
J
,
Teng
EL
,
Quiligan
E
,
Van Der Meulen
JP
.
Neuromotor development in relation to birth weight in rabbits
.
Biol Neonate
.
1978
;
33
(
1–2
):
1
7
. .
33.
Miller
AL
,
Leach
MC
.
Pain recognition in rabbits
.
Vet Clin North Am Exot Anim Pract
.
2023
;
26
(
1
):
187
99
. .
34.
Dumont
JR
,
Salewski
R
,
Beraldo
F
.
Critical mass: the rise of a touchscreen technology community for rodent cognitive testing
.
Genes Brain Behav
.
2021
;
20
(
1
):
e12650
. .
35.
Drobyshevsky
A
,
Robinson
AM
,
Derrick
M
,
Wyrwicz
AM
,
Ji
X
,
Englof
I
, et al
.
Sensory deficits and olfactory system injury detected by novel application of MEMRI in newborn rabbit after antenatal hypoxia-ischemia
.
Neuroimage
.
2006
;
32
(
3
):
1106
12
. .
36.
Drobyshevsky
A
,
Yu
L
,
Yang
Y
,
Khalid
S
,
Luo
K
,
Jiang
R
, et al
.
Antenatal insults modify newborn olfactory function by nitric oxide produced from neuronal nitric oxide synthase
.
Exp Neurol
.
2012
;
237
(
2
):
427
34
. .
37.
Shepherd
A
,
Zhang
T
,
Hoffmann
LB
,
Zeleznikow-Johnston
AM
,
Churilov
L
,
Hannan
AJ
, et al
.
A preclinical model of computerized cognitive training: touchscreen cognitive testing enhances cognition and hippocampal cellular plasticity in wildtype and alzheimer’s disease mice
.
Front Behav Neurosci
.
2021
;
15
:
766745
. .
38.
Haddad
FL
,
Ghahremani
M
,
De Oliveira
C
,
Doornaert
EE
,
Johnston
KD
,
Everling
S
, et al
.
A novel three-choice touchscreen task to examine spatial attention and orienting responses in rodents
.
eNeuro
.
2021 Jul-Aug
;
8
(
4
):
ENEURO.0032-20.2021
. .
39.
Ananth
CV
,
Friedman
AM
,
Lavery
JA
,
VanderWeele
TJ
,
Keim
S
,
Williams
MA
.
Neurodevelopmental outcomes in children in relation to placental abruption
.
BJOG An Int J Obstet Gynaecol
.
2017
;
124
(
3
):
463
72
. .
40.
Stadskleiv
K
.
Cognitive functioning in children with cerebral palsy
.
Dev Med Child Neurol
.
2020
;
62
(
3
):
283
9
. .
41.
Fry
R
,
Li
X
,
Evans
TC
,
Esterman
M
,
Tanaka
J
,
DeGutis
J
.
Investigating the influence of autism spectrum traits on face processing mechanisms in developmental prosopagnosia
.
J Autism Dev Disord
.
2023
;
53
(
12
):
4787
808
. .
42.
Barton
JJ
,
Cherkasova
MV
,
Hefter
R
,
Cox
TA
,
O’Connor
M
,
Manoach
DS
.
Are patients with social developmental disorders prosopagnosic? Perceptual heterogeneity in the Asperger and socio-emotional processing disorders
.
Brain
.
2004
;
127
(
Pt 8
):
1706
16
. .
43.
Shi
Z
,
Vasquez-Vivar
J
,
Luo
K
,
Yan
Y
,
Northington
F
,
Mehrmohammadi
M
, et al
.
Ascending lipopolysaccharide-induced intrauterine inflammation in near-term rabbits leading to newborn neurobehavioral deficits
.
Dev Neurosci
.
2018
;
40
(
5–6
):
534
46
. .
44.
Vasquez-Vivar
J
,
Shi
Z
,
Tan
S
.
Tetrahydrobiopterin in cell function and death mechanisms
.
Antioxid Redox Signal
.
2022
;
37
(
1–3
):
171
83
. .
45.
Dean
JM
,
Shi
Z
,
Fleiss
B
,
Gunn
KC
,
Groenendaal
F
,
van Bel
F
, et al
.
A critical review of models of perinatal infection
.
Dev Neurosci
.
2015
;
37
(
4–5
):
289
304
. .
46.
Tan
S
,
Shi
Z
.
Commentary to the in-focus issue “Perinatal brain injury leading to later neurodevelopmental disorders: early detection and treatment options”
.
J Neurosci Res
.
2022
;
100
(
12
):
2109
11
. .