An Animal Model Of Adolescent Harsh Physical Abuse


The philosophy of science seeks to understand how scientists and their dynamics in a community devise experiments to generate data and models by analyzing their underlying biases, how they interpret data, and career and community influences, among other things. This text is a critical assessment of the scientific barriers to studying brain impairments evoked by harsh physical child abuse and rape in an experimental animal model. It also explains why we need the financial support of the survivor community.

Neurobiology involves the interdisciplinary exchange of information and ideas that seek to explain aspects of brain function and disease. It has always been puzzling to me that there is no neurobiology of child physical abuse. While there are extensive epidemiological studies, case reports, and brain imaging, there are no studies of changes in the molecular, single neuron, circuitry or other information that can only be achieved with experimental animals. The answer to these questions lies in how scientists conceptualize abuse and their biases as to what can and cannot be studied with such a model.

We begin with the general issues of why and how a model is devised with particular reference to abuse realizing from the start that this is not an exact replication of human experiences as no animal model can perform to this level; animals are unlikely to have similar perceptual responses to humans. We will consider statements from the investigators themselves through their commentaries on grant applications and manuscripts between 2011 and 2016 in section "Scientific commentary on the model". Following this exercise, we will be able to understand why this area of neurobiology has yet to develop and will help us challenge the underlying assumptions about what survivors of physical abuse, rape and war experience and how it undermines their brain function.

Why employ an experimental animal model?

Experimental neuroscientists extensively use animal models of human diseases because we are unable to intervene directly in the human brain and make manipulations that test specific hypotheses. Human imaging has evolved quickly over the past decades with well-designed behavioral tests, higher resolution with increased field strength magnets of 3-7 Tesla, and better head movement correction methods; yet there are many impediments to analyzing the molecular and cellular mechanisms of a disease with such tools. Let us consider four examples of this problem.

First, neurochemistry with proton magnetic resonance spectroscopy provides a measure of some neurochemicals such as glutamate and metabolites. However, most neurochemical changes that involve complex pathways that regulate neuronal excitability cannot be accessed without tissue removal. Brain banks provide postmortem tissue with valuable information; however, such an approach is of little value when studying an adolescent disorder as it will be as many as 7 decades before the tissue is available; well after the initial abusive insult and impairing cause and effect conclusions.

Second, magnetic resonance imaging (MRI) can be employed to assess structural increases or decreases in tissue volume. However, there are many potential sources of such changes such as neuron somata or dendrite atrophy or hypertrophy or patent neuron loss, glial responses and loss or gain of axonal projections. These questions can only be addressed following tissue removal at a relevant time point in longitudinal studies.

Third, connectivity studies with correlational functional magnetic resonance imaging (fMRI) and diffusion weighted imaging (DTI) is a means of assessing connectivity and changes in the brain of a survivor. However, these approaches come with many caveats and testing connectivity directly in experiments with lesions, optogenetics and other techniques are only possible in experimental animals.

Fourth, there are only rare instances in which single neuron recording can be made with intracerebral electrodes in patients with epilepsy, obsessive-compulsive disorder (Williams et al., 2004) and other disorders that can discount them as reflective of normal human subjects. With animal studies, the onset, duration and amplitude of an intervention can be standardized and neurophysiological recordings made in relevant sites in healthy animals.

Thus, in terms of child abuse, it is possible to show sites of under- or overactive tissue with fMRI (Ringel et al., 2008) and child abuse-related PTSD (van Harmaelen et al. 2010; Thomaes et al., 2010); yet the onset, amplitude, and duration of abuse cannot be controlled, there is no information on what transmitter systems or circuits are specifically damaged or how neuronal discharges are altered. Without such information, rationale therapeutic interventions are not possible. These are the many reasons why a model of harsh physical child abuse is necessary.

How is an animal model devised including one of Child Abuse Disorder?

An animal model is developed according to a number of criteria. Face validity addresses the extent to which it appears to encompass key features of the human condition. Construct validity defines how well a test or experiment measures up to its claims. It refers to whether the operational definition of a variable actually reflects the true theoretical meaning of a concept. Finally, predictive validity is demonstrated with human-relevant therapeutic interventions; a subject that is not yet relevant to the current stage of this research.

At the outset it should be stated that no animal model is an exact replication of any human disorder but rather is used to study aspects of a disease. Mice overexpressing amyloid-β peptide and showing memory deficits is a model of Alzheimer’s disease that allows one to assess the synthesis and degradation of this peptide and potential relationships with neurodegeneration and altered behavior; yet no one would claim that this is equivalent to the human disease. This does not negate the model and can be viewed as a strength as some confounding variables in humans may be avoided as is the case with our abuse model. For example, although adolescent harsh abuse is influenced by socioeconomic factors (Imbierowicz and Eagle, 2003), abuse is initiated with intense physical pain and stress, usually by a male predator on a young female, and it is not known what occurs in the brain independent of socioeconomic factors. The physical parameters of the primary abuse event can be simulated experimentally with controlled frequency, duration and amplitude of noxious/stressful stimuli, assessment of single neuron responses and postmortem brains and would show the effects of harsh physical abuse without confounding human psychosocioeconomic factors. This has been done by CNSI in its first publication on the subject (Vogt et al., 2017; #66 in the Library).

All mammals can be abused and it is not limited to humans. My son’s rescue dog was abused to some extent as demonstrated by her early responses to the human hand and some normally non-threatening situations. Thus, there is no reason to exclude the possibility that our experimental animal of choice, the rabbit, can be abused. The only question is; what constitutes abuse and can its effects be demonstrated behaviorally?

(We use rabbits because of the structure of their cingulate cortex including the size of their neurons that are about 4 times larger than those in commonly used experimental rats allowing for long periods (hours) of holding neurons at the end of an electrode for recording responses to peripheral stimulation. Its cortex is also about 8 times larger volumetrically than in rats providing more flexibility for placing independently moveable electrode carriers, substantially more tissue per animal for histological analysis and we have large data sets and histological tissues from previous studies.)

It is important to emphasize that this modelling effort does not seek to replicate the entire human experience and follow it over many decades (noted above); exclusion of socioeconomic factors is a virtue of the animal model to the extent that we seek to understand how the primary abuse event itself alters brain function. The abuse protocol was derived from the physical parameters of harsh physical abuse and rape as experienced by adolescents during predatory or disciplinary assault. Severe physical abuse involves being hit with an object, burning, or forced penetrative sex at least once a month for at least a year (Bremner et al., 1999). A review by The New York City Alliance Against Sexual Assault (web site) states, “adolescents are more likely to experience sexually violent crimes than any other age. Nationally, 7-10% of girls aged 12-17 have experienced … sexual assault, rape, or sexual abuse … Rape of males is any sexual assault involving forced penetration of the anus or mouth by a penis or other object.” Thus, sexual abuse is forced sex involving vaginal or anal intercourse and only life-threatening force predicts adult health status (Leserman et al., 1996, 1997). Rape is a form of physical abuse, is always stressful and abuser gender does not predict symptomatology (Briere and Elliot, 2003). As children are severely affected by harsh physical abuse and rape (Keiley et al., 2001), rabbits in late childhood and early adolescence were used. Finally, although females are more often raped than males, males experience more profound impairments following rape (Kang et al., 2005; Vogt DS et al., 2005; Street et al., 2008). Thus, both genders were employed with a common conditioning protocol.

Adolescence in New Zealand White rabbits begins at about 11-12 weeks when the testes descend. Childhood is the time between weening at about 4 weeks and lasts to week 12. We targeted childhood during the 3rd postnatal month which may be approximately related to human years 9-13. We use noxious colorectal distension (nCRD) because it can be employed in both sexes without involving vaginal stimulation as in females only. It is important to note that we seek to limit the total conscious exposure to pain and stress where possible by anesthetizing the animal when the colorectal distender is inserted, not moving the distender in and out as during sexual encounters and provide 1 min rest periods between each 2 min nCRD; i.e., this is not the maximal abuse as experienced by human children. The frequency of nCRD was determined with comparison to the human early lifespan in years compared to the rabbit early lifespan in months. For the human this is ages 8-11 for 4 years, while for the rabbit it is the 3nd month. Human harsh physical abuse is defined as at least one experience/month for at least one year (above) and involves at least 48 exposures over 4 years. As we are converting ~4 years of human life to 6 months of rabbit life, 8 or 9 nCRD exposures may equate to the human harsh physical abuse experience.

To summarize, the model of the physical parameters of child abuse includes 1) an age approximating late childhood/early adolescence, 2) painful stimulation (nCRD, 60 mmHg) with the distender simulating the male penis during anal intercourse, 3) short duration exposure of ~21 min to simulate time to male ejaculation, 4) forced stimulation by holding balloon in place, and 5) repeated events (3X/week for 3 weeks). The rabbit’s behavior during the procedure is as would be expected from a child under similar circumstances (below). Thus, the protocol engages conscious pain systems for only a total of two hours over 3 weeks in contrast to that experienced by animals in chronic pain studies with chronic nerve constriction injury, for example, over three weeks that are exposed to ~500 hours of constant noxious stimulation. We propose that chronic pain studies are substantially more abusive than this protocol by exposing them to ~250X more conscious pain and stress.

Animal behavior is critical to interpreting their internal states

Animals exposed to 3 weeks of the nCRD protocol had normal weight gain (initial: 1.66±0.06 kg for controls vs 1.66±0.13 kg for nCRD conditioned; after 3 weeks weights were 2.3±0.01 kg in controls vs 2.26±0.21 kg in nCRD conditioned). Stools were coded (hard-1, soft-2, fluid-3) and all animals had a normal pellet discharge (scores=1) indicating no changes in bowel habits over 3 weeks. There was no evidence for abdominal pain when not in the sling as would be indicated by writhing, stretching, abdominal probing or hair pulling. Quantitative measures of visceral pain are clearly needed in future studies.

If the animals were experiencing ongoing pain, they might have shortened withdrawals of their rear paws from water baths as occurs in chronic visceral pain but this was not the case. Thermal nociceptive testing with water baths showed no differences between controls and nCRD-conditioned animals in the threshold for withdrawal at initial testing (50.8±0.9ºC for controls vs 51.5±0.9ºC for nCRD conditioned) or after 3 weeks of nCRD conditioning (49.6±1ºC for controls vs 50.6±0.5ºC for nCRD conditioned). The withdrawal temperature before and after nCRD-conditioning on the last day of conditioning also did not change (50.3±0.5ºC). These temperatures were higher than for other species likely because rabbits have an immobility response during handling.

If the animals were not experiencing abuse, they would lay quietly in the sling. However, during conditioning, rabbits always tried to excrete the distender by arching their backs, tail lifting, rear leg extension or walking motions, and tugging on their perineal and abdominal muscles. The head/ears and all legs were tense and fully extended and the animals often turned their heads to either side to look back. None of these nocifensive (pain avoidance) activities were observed in control animals while in the sling. Following the first few days of conditioning, the animals showed anticipation of pain and stress by trying to avoid being removed from the carrying box (nose poking into bottom corners versus controls rearing to look out) and in some instances they struggled to avoid being placed in the sling. These findings suggest the nCRD-conditioned animals soon understood and feared the context of their abuse much as patients with PTSD avoid traumatic triggers.

To document rabbit avoidance and fear suggested by these qualitative observations, contextual fear was tested with the sling in full view of a start box. While the 5 control animals acclimated to the task over the 5 testing sessions (response times declined from 95±29 on day 1 to 17±3 sec on day 7; p=0.04), the 7 nCRD-conditioned animals had longer response times that did not normalize over successive testing days. Although ANOVA analyses showed no overall significant differences by group and session (p=0.056), day 7 showed a significant difference from controls (p<0.03) and trends on days 5 (p=0.08) and 9 (p=0.09). Further consideration of the 7 conditioned animals showed that they did not express the same level of contextual fear; 4 of them were fearful of their environment (vulnerable), while 3 were not (resilient) with very low scores. A three group analysis showed there were significant differences among the group means (ANOVA p=0.005; Means±SEM: Control 53±70, Vulnerable 437±78, Resilient 55±90). Post-hoc mean analysis showed significant differences between both the control versus vulnerable and vulnerable versus resilient contrasts (p=0.008 and 0.016, respectively), while that between control and resilient groups was not significant (p=0.5).


Severe adolescent physical abuse, as defined in the human research literature, was applied to rabbits of a similar relative age and with a similar amplitude and duration to evoke abuse. This is not a model of chronic visceral pain or irritable bowel syndrome as argued by Vogt et al. (2017). The animal’s behavior in the sling, qualitative observations and contextual fear testing clearly demonstrate that they were abused. Interestingly, there is not yet evidence that they experience any ongoing pain as in chronic pain models. The direct relevance of the model to human experience is exemplified by a case report by Terr (1991) who documented a girl that was sexually abused by her father between ages 5-15. At 38 she feared sex with her husband unless she initiated it and she avoided positions taken by her father as they stimulated fear, pain and revulsion. This latter response is one of contextual fear as likely experienced by our rabbits.

  • ​Bremner JD, Randall P, Scott TM, Capelli S, Delaney R, McCarthy G, Charney DS (1995) Deficits in short-term memory in adult survivors of childhood abuse. Psychiatry Res 59:97-107.
  • Briere J, Elliot DM (2003) Prevalence and psychological sequelae of self-reported childhood physical and sexual abuse in a general population sample of men and women. Child Abuse Neg 27:1205–1222.
  • Kang H, Dalager N, Mahan C, Ishii E (2005) The role of sexual assault on the risk of PTSD among Gulf War veterans. Ann Epidemiol 15:191-195.
  • Keiley MK, Howe TR, Dodge KA, Bates JE, Pettit GS (2001) The timing of child physical maltreatment: A cross-domain growth analysis of impact on adolescent externalizing and internalizing problems. Develop Psychopathol 13:891-912.
  • Leserman J, Drossman DA, Li Z, Toomey TC, Nachman G, Glogau L (1996) Sexual and physical abuse history in gastroenterology practice: How types of abuse impact health status. Psychosom Med 58:4-15.
  • Leserman J, Li Z, Drossman DA et al. (1997) Impact of sexual and physical abuse dimensions on health status: development of an abuse severity measure. Psychosom Med 59:152-160.
  • Ringel Y, Drossman DA, Leserman JL, Suyenobu BY, Wilber K, Lin W, Whitehead WA, Naliboff BD, Berman S, Mayer EA (2008) Effect of abuse history on pain reports and brain responses to aversive visceral stimulation: An fMRI study. Gastroenterology 134:396–404.
  • Street AE, Stafford J, Mahan CM, Hendricks A (2008) Sexual harassment and assault experienced by reservists during military service: Prevalence and health correlates. J Rehab Res Dev45:409–420
  • Terr LC (1991) Childhood trauma: An outline and overview. Am J Psychiatry 148:10-20.
  • Thomaes K, Dorrepaal E, Draijer N, et al. (2010) Reduced anterior cingulate and orbitofrontal volumes in child abuse-related complex PTSD. J Clin Psychiatry 71:1636-1644.
  • van Harmelen A-L, van Tol M-J, van der Wee NJA, et al. (2010) Reduced medial prefrontal cortex volume in adults reporting childhood emotional maltreatment. Biol Psych 68:832-838.
  • Vogt BA, Vogt LJ, Sikes RW (2017) A nociceptive-stress model of adolescent physical abuse induces contextual fear and cingulate nociceptive neuroplasticities. Brain Struc Func,
    (doi: 10.1007/s00429-017-1502-3).
  • Vogt DS, Pless AP, King LA, King DW (2005) Deployment stressors, gender, and mental health outcomes among Gulf War I veterans. J Traum Stress 18:115-127.
  • Williams ZM, Bush G, Rauch SL, Cosgrove GR, Eskandar EN (2004) Human anterior cingulate neurons and the integration of monetary reward with motor responses. Nat Neurosci7:1370-1375.