Good memories of bad events in infancy: Ontogeny of conditioned fear and the amygdala

Abstract

If a helpless newborn infant does not form an attachment to its care-giver, even an abusive one, its chances of survival diminish, so evolution should strongly favour attachment by the infant, regardless of the quality of care-giving¹. As a part of the brain called the amygdala is critical for learned fear in adult animals^{2, 3, 4}, we investigated whether the development of learned avoidance behaviour could be delayed by late maturation of amygdala function. We found that very young rat pups exposed to various odours associated with shock treatment learn an approach response to that odour, whereas older pups learn odour avoidance. We show that the origin and development of learned odour-avoidance behaviour is associated with enhanced neural responses in the amygdala during odour-shock conditioning.

Attachment in infant rats is learned predominantly postnatally, relying mainly on the olfactory system — learned odour cues are necessary for successful nipple attachment and for orientation to the mother and litter⁵. As a model for early attachment to an abusive care-giver, we used odour–shock pairings in neonatal rats (fear-conditioning paradigm^{6, 7}) that result in odour approach by very young rats but an odour aversion in older rats. Pups were trained in a classical-conditioning paradigm in which odour is paired with shock, or they were exposed only to the odour. Four hours after training, pups were tested for relative odour preference in a two-odour choice test.

Figure 1a shows that the conditioning of approach to odours paired with aversive shock occurs during a period that ends before postnatal day (PN) 10. Pups trained after this sensitive period learn to avoid odours paired with shock (analysis of variance (ANOVA), conditioning group × postnatal age, F_2,54 = 9.29, P< 0.001; Fisher post-hoc tests indicate that the PN9 paired pups are significantly different from the PN10 and PN11 pups, P < 0.01). This striking difference in the learned behavioural responses of PN9 and older pups occurs despite their similar responses to shock (vocalization, vigorous physical response, wall climbing^{6, 7}; Fig. 1c, d). Furthermore, both PN9 and PN10 pups try to escape from the shock (results not shown).

Figure 1: Ontogeny of learned odour aversions corresponds to the ontogeny of amygdala activation.

a,Postnatal-day (PN)-9 pups trained with paired odour–shock demonstrated a subsequent odour preference compared with PN9 controls. Older pups showed learned odour aversions (n, 8–12 per group). b, Pups trained on PN9 continued to show an odour preference for an odour paired with shock, even at an age when they could learn an odour aversion (n = 8). c,d, PN9 pups find shock as aversive as PN10 pups (as measured by shock-induced vocalizations, in c, and behavioural activation in d; n, 5–6). e, Association of odour and shock produces enhanced neural activity within the amygdala of PN12 pups compared to controls, whereas the same training does not differentially affect amygdala activity in PN8 pups (n, 6–8). CS, conditioning stimulus; 2-DG, 2-deoxyglucose.

Once acquired, a learned odour-approach response to odours paired with shock before PN10 continues to be expressed as an approach even in older pups that are capable of learning an aversion (good memory of a bad event). We trained pups with odour–shock pairings on PN9 and tested them after either 4 or 24 hours (on PN10). As shown in Fig. 1b, pups trained on PN9 and tested either on PN9 or PN10 showed a relative odour preference for the odour paired with shock (ANOVA, main effect of conditioning group, F_1,33 = 17.94, P < 0.001; Fisher post-hoc tests indicate that each paired group is significantly different from each control group, P < 0.01).

To determine the role of the amygdala in the differential response to odour–shock training, we injected PN8 and PN12 pups with ¹⁴C-2-deoxyglucose and subjected them to a 45-min odour–shock or control classical-conditioning procedure. Amygdala uptake of labelled 2-deoxyglucose was expressed relative to uptake in the corpus callosum, which did not vary across conditioning groups. The relative uptake by the amygdala was higher in PN8 pups than in PN12 pups, but did not vary with conditioning group in PN8 pups. However, it was significantly enhanced in PN12 pups trained in odour–shock pairings compared with control pups (ANOVA, training group × age interaction, F_2,35 = 3.37, P < 0.05; post-hoc Fisher tests revealed PN12 paired pups were significantly different from both PN12 control groups, P < 0.05).

Our results show that the appearance of a learned odour aversion using this conditioning paradigm originates with the initial appearance of an aversive conditioning-induced enhancement of neural activity within the amygdala. We cannot infer a causal connection between amygdala development and ontogeny of avoidance conditioning, however. In fact, illness-induced odour aversions and conditioned taste aversions can be induced in pups during their first postnatal week, although illness-induced aversions may involve a different mechanism from fear conditioning^{7, 8, 9}. Nonetheless, our results indicate that a unique behavioural¹⁰ and neural response to aversive conditioning may exist in helpless newborns during a time when aversion to a care-giver could be fatal.