Context learning before birth: evidence from the chick embryo

Abstract

Learning contextual information to form associative memories with stimuli of interest is an important brain function in both human and non-human animals. Intuitively, one would expect that such a sophisticated cognitive skill develops postnatally, as the organism starts exploring the surrounding environment to search for significant contingencies among stimuli. Here we show, instead, that even before hatching, domestic chicks are capable of forming associative memories between discrete alerting sounds and the surrounding context, as attested by the fact that habituation of the freezing response to the sounds is affected by the context of stimulation. This finding indicates that, while in the egg, chicks recognize and learn the context in which they are stimulated. Hence, context learning in chicks is an innate brain function already active before birth, which can provide an immediate survival advantage to the newborns of this precocial avian species.

1. Introduction

Motivationally significant stimuli, like for instance, food or mates, are always experienced in a given context, namely integrated into a background of other objects or events, which are automatically encoded as a unified configural representation. Hence, the cognitive ability to encode context identity and to learn its contingency with meaningful stimuli is of paramount importance to animals. Context, indeed, can affect cognition, emotion and behaviour of animals in different ways [1,2]. Context-fear conditioning, for example, shows that animals learn to anticipate a defensive response to a dangerous stimulus when the aversive context is recognized [3]. Through context learning, a fruit fly can select the appropriate response to conflicting sensory information [4]; in honeybees, memory consolidation and retrieval are enhanced by the appropriate odour context [5]. In monkeys, context confers rewarding properties to neutral stimuli [6], and context can affect the oculomotor behaviour [7] and the decision-making strategy of the animal [8]. Finally, in humans, context features guide attention toward the relevant parts of the scene [9] or are used to filter visual distractors [10], and more recent studies recording neural spiking activity have demonstrated that the response in the early visual cortex is affected by contextual factors [11].

So far, the influence of context on behavior and cognition has been mainly investigated in adult animals, perhaps because it would seem natural to expect that this cognitive skill develops after birth, when the organism starts to explore the surrounding environment and learns the existence of reliable contingencies between significant stimuli and their context. However, given the highly adaptive value of context learning for survival, a fascinating alternative is that natural selection may have endowed animals with the innate ability to encode and learn contextual information even before birth. Alongside this possibility, previous studies have shown that amphibians are capable of temporal learning during the embryonic developmental stage and to use this information after hatching [12]. However, on the one hand, temporal learning might not necessarily involve contextual information, as shown by temporal conditioning, whereas on the other hand, it might be a very specific case of contextual learning. Hence, we decided to further explore the issue of prenatal context learning, and specifically, we addressed whether, in the chick embryo, habituation of the freezing response to a threatening sound is context-specific, which implies that the auditory stimuli are integrated with broad contextual information, and crucially, that context learning occurs prenatally. Previous studies have shown that habituation, which consists of a response decrement as a result of repeated stimulation [13], can be context-specific [14–16]. Notably, habituation of the freezing response to a sudden loud noise has been found to be context-specific also in newborn chicks [17]. Specifically, chicks repeatedly stimulated with an alerting sound in two consecutive days in the same context showed habituation of the freezing response across days. By contrast, chicks receiving the second day of stimulation in a different context exhibited an increased rate of freezing, a result consistent with a recovery of the habituated response caused by the context change [17].

By using an analogous paradigm, we explored whether in the domestic chick (Gallus gallus) context-specific habituation can take place even before birth. The paradigm consisted of an exposure phase followed by a test phase. During the exposure phase, animals were exposed, either before or after hatching, to repeated alerting sounds (bursts of white noise) in the same context of the next test phase, or in a different one. Specifically, chicks in the same-context group were stimulated in the running-wheel both in the exposure phase (within 24 h from hatching) and then in the test phase (within the next 24 h); the different-context group was stimulated in the incubator after hatching during the exposure phase (0–24 time window), and then in the running-wheel in the test phase (24–48 time window); the in ovo group was stimulated prenatally in the incubator within 24 h before hatching (exposure phase), and then in the running-wheel in the test phase (24–48 time window). Finally, a control group was directly tested in the running-wheel (24–48 time window) (figure 1).

Figure 1.

Schematic representation of the experimental design and conditions. The exposure (−24/+24) and the test (+48) phases could take place in the same or different contexts. The freezing response to the bursts of white noise (90 dB) was measured as a stop of the wheel-running behaviour. Each condition consisted of 30 chicks. (Online version in colour.)

Habituation was evaluated by measuring the amount of the chick's freezing response to the alerting sounds while animals were running in a running-wheel in an attempt to reach an artificial companion (see material and methods for details). The freezing response was defined as a stop of the wheel-running behaviour [17].

The performance of the in ovo group is crucial to establish the presence of prenatal context learning, as for this group of chicks prenatal (exposure) and postnatal (testing) environments should represent different contexts. Consequently, for the in ovo group to show context-specific habituation in the test phase, chicks should exhibit a higher rate of freezing (reduced habituation) compared to the same-context group. Conversely, if before hatching no context learning takes place, the in ovo group should compare to the same-context group, because both groups receive the same amount of stimulation; if the pre-hatching stimulation was instead ineffective (i.e. no habituation), the rate of freezing in the in ovo group should be similar to that of the control group. In addition, the lowest rate of freezing should emerge in the same-context group, and the highest in the control group, which cannot benefit from habituation in the exposure phase. Finally, the different-context and the in ovo groups should exhibit similar rates of freezing, because for both groups the exposure and test environments were different.

2. Material and methods

(a). Subjects

Domestic chicks (N = 120; males = 58) of the Ross 308 (Aviagen) broiler strain hatched singly in individual opaque compartments of 10 cm³ from fertilized eggs incubated in our laboratory under controlled temperature (37.7°C) and humidity (about 50–60%) conditions. The hatching moment was recorded by a camera set inside the incubator. The auditory stimuli were administered by two loudspeakers positioned on the ceiling of the incubator. In the exposure phase, only for the different-context group the temperature of the incubator was lowered to 31.5°C. Illumination was kept constant across conditions at 0 lx.

(b). Apparatus

The test setting consisted of a running-wheel (30 cm in diameter) located on the rear end of a black arena (45 × 50 × 160 cm, width, height, depth). A red cylinder (6 × 7.5 cm, diameter, height) was hung from above in front of the running-wheel to elicit the chicks' running behaviour [18]. In the testing room, the temperature was 28°C; illumination within the apparatus varied from 3 lx in the running-wheel to 14 lx in the proximity of the red cylinder. The auditory stimuli were delivered by two loudspeakers positioned on the top of the running-wheel, at about 30 cm from the chick's head. Time, distance and direction of the runs within the wheel were computed by an Arduino circuit and displayed on a monitor. Both the running-wheel and the monitor were recorded by a video camera from above the arena.

(c). Stimuli and procedure

Each sequence of stimulation consisted of five bursts of 250 ms white noise (90 dB SPL) delivered at a pseudo-random inter-stimulus-interval, ranging from 30 to 60 s. Four experimental groups of 30 chicks each were used: same context, different context, in ovo and control. In the exposure phase, all groups except the control one received two sessions of stimulation 1 h apart, with each session consisting of a sequence of stimulation. However, the groups of chicks differed as a function of the context in which they received the acoustic stimulation: in the same-context group, each chick was individually stimulated in the running-wheel (as in the test phase); chicks in the different-context group were collectively stimulated in the incubator within 24 h after hatching; chicks in the in ovo group were collectively stimulated in the incubator in the last 24 h before hatching; chicks in the control group were not submitted to the exposure phase.

During the test phase, all chicks were tested individually in the running-wheel from 24 to 48 h after hatching. The acoustic stimulation was identical to the one delivered in the exposure phase. Hence, in the test phase, the context of stimulation was identical for all chicks. The stimulation in the running-wheel was started manually by the experimenter once the chick, in the attempt to reach the red object in front of it, showed a consistent wheel-running behaviour. The criterion to start the acoustic stimulation was that chicks had to run for a minimum distance of 1000 cm, which took approximately 8 min (also see [17]). During this period, the animals had also the possibility to familiarize with the running-wheel context.

Habituation to the acoustic stimulation was evaluated by scoring the number of the freezing responses to the bursts of white noise during the test phase. For chicks in the same-context group, the same type of data was collected also during the exposure phase.

3. Results

The main results are illustrated in figure 2a. Since the proportions of freezing responses were not normally distributed, data were analysed with non-parametric statistical tests (Kruskal–Wallis and Wilcoxon signed-rank tests).

Figure 2.

(a) Amount of freezing in chicks exposed to the bursts of noise. In the test phase, data are from the first session of stimulation (material and methods). The rate of freezing increased in chicks stimulated in a different context (in ovo or after hatching in the incubator) during the previous exposure phase. (b) During the test phase, all groups showed a reliable habituation of the freezing response across the two sessions of stimulation.

The same-context group showed a reliable habituation of the freezing response from the exposure to the test phase (p < 0.001, r = 0.73). Furthermore, the rate of freezing in the first session of the test phase was different among groups (χ² (3) = 41.897, p < 0.001, η² = 0.33, Kruskal–Wallis H test). Post hoc comparisons showed that the different-context group had a higher freezing rate compared to the same-context group (p < 0.001, r = 0.40), which confirmed previous results of context-specific habituation in chicks [16]. Crucially, the same result emerged also for the in ovo group (p < 0.001, r = 0.33), indicating that, for the chick embryo, habituation was context-specific, namely that even before hatching, chicks were capable of encoding the surrounding context forming associations with the acoustic stimuli. In addition, both groups showed a similar freezing rate (p = 0. 854, r = 0.02), but a lower rate compared to the control group (different-context group: p = 0.010, r = 0.22; in ovo group: p = 0.012, r = 0.21), meaning that during the exposure phase, chicks partially habituated to the sounds regardless of context. The lower freezing rate of the in ovo group, compared to the control group, is a key finding to rule out the possibility that the higher freezing rate of the in ovo group, compared to the same-context group, was not just due to a failure to habituate.

We interpreted the overall pattern of results as evidence that context learning took place in the chick embryo, but an alternative possibility is that, in fact, context learning did not occur in ovo. According to this view, the response at the test of the in ovo group was only coincidentally similar to that of the different-context group, and higher than that of the same-context group because of the longer interval between the exposure and test phase in the in ovo group, which caused habituation to decay more rapidly. We are sceptical about this interpretation because it requires more assumptions than the context-learning one to explain the overall pattern of results, but at present we cannot rule it out completely. An alternative related manipulation to eliminate the role of temporal differences between exposure and test phases among groups, if any, would be to have a further in ovo group stimulated in the running-wheel, whose performance could be directly compared with that of the in ovo group stimulated in the incubator. Further studies will be needed to clarify these issues.

Finally, as shown in figure 2b, habituation was effective for all chick groups also between the two sessions of the test phase (same-context group: p < 0.001, r = 0.79; different-context group: p < 0.001, r = 0.78; in ovo group: p < 0.001, r = 0.76; control group: p < 0.001, r = 0.79), thus confirming that our paradigm was adequate to elicit a reliable habituation of the freezing response in this avian species.

4. Discussion

Our results indicate that, while still in the egg, chicks learned to familiarize with the acoustic stimulation, and that, crucially, they formed associations between the representation of the stimuli and that of the surrounding context. An interesting question is what kind of information defines the context for the chick embryo. Since chicks in ovo can have only undetailed sensory experiences of the external world, it is likely that their context representation was influenced by the overall level of ambient light or sound perceived through the eggshell, or the perceived temperature. In addition, other factors like the hormonal, physiological and proprioceptive states may have played a major role in context definition. Notably, all these parameters changed after hatching, and particularly when chicks were moved from the egg in the incubator to the running-wheel.

Prenatal learning for specific discrete stimuli has been already documented in different taxa [19,20]. Some previous evidence of prenatal learning might be interpreted as cases of embryonic context learning [12,21]. For example, woodfrog embryos, exposed to predator cues in two fixed time windows during the day, use this temporal information after hatching to adjust their antipredator response in the same day periods [12]. It is not clear, though, whether this remarkable ability of temporal learning in embryonic amphibians is an instance of temporal conditioning, or an instance of temporal context learning. However, here we have shown that prenatal context learning is not restricted to temporal information, which was the same for all our groups of chicks, and therefore could not account for the results.

Our study showed that, in chicks, prenatal cognition extends to the ability to automatically encode the complex ensemble of background stimuli over which significant stimuli are perceived and to form associative memories between the two kinds of information. The sophisticated in ovo cognition that we have documented is in agreement with previous evidence showing ‘waking-like’ brain function in the late stage of the chick's embryonic life [22]. Furthermore, our findings suggest that, among the numerous innate building blocks of cognition already documented in this avian species [23], we can now count the capacity to associate incoming discrete sensory information with context.

Ethics

This experiment complies with the current European Community and Italian laws for the ethical treatment of animals and has been approved by the Organismo Preposto al Benessere Animale of the University of Trieste and licensed by the Italian Health Ministry (permit 88/2019-PR).

Data accessibility

The dataset supporting this article has been uploaded as part of the electronic supplementary material.

Authors' contributions

M.T. and C.C. conceived the study; C.C. designed and supervised the experiment; A.D. collected the data; A.D. and C.C. performed the analysis; M.T. drafted the manuscript, C.C. and A.D. revised the manuscript. All authors approved the final version of the manuscript and agree to be held accountable for the content therein.

Competing interests

We have no competing interests.

Funding

The study was partially supported by grants from University of Trieste (uniTs-FRA2015) and FFABR (Finanziamento delle Attività Base di Ricerca) to C.C.