Are You Better Than Me? Social Comparisons in Carrion Crows (Corvus corone)

doi:10.21203/rs.3.rs-2500378/v1

Comparing oneself to others is a key process in humans that allows individuals to gauge their performances and abilities and thus develop and calibrate their self-image. Very little is known about its evolutionary foundations. A key feature of social comparison is the sensitivity to other individuals’ performance. Recent studies on primates produced equivocal results, leading us to distinguish a ‘strong’ variant of the social comparison hypothesis formulated for humans from a ‘weak’ variant found in non-human primates. Here, we focus on animals that are distantly related to primates but renowned for their socio-cognitive skills, birds from the family Corvidae. We were interested in whether crows’ task performances were influenced i) by the presence of a conspecific co-actor performing the same discrimination task and ii) by the simulated acoustic cues of a putative co-actor performing better or worse than themselves. Crows reached a learning criterion quicker when tested simultaneously as compared to when tested alone, indicating a facilitating effect of social context. The performance of a putative co-actor influenced their performance: crows were better at discriminating familiar images when their co-actor was better than they were. Standard extremity, i.e., how pronounced the difference was between the performance of the subject and that of the co-actor, and category membership (i.e., affiliation status and sex), of the putative co-actors had no effect on their performance. Our findings are in line with the ‘weak’ variant of social comparison and indicate that elements of human social comparison can be found outside of primates.

Comparing oneself to others and evaluating the self are amongst the most prevalent processes of the human psyche (Wheeler and Miyake 1992) and an important part of our everyday life (Festinger 1954) Such social comparisons are not only ubiquitous and virtually inevitable (Gilbert et al. 1995), they also have powerful affective (Gilbert et al. 1995), motivational (Lockwood and Kunda 1997) and self-evaluative (Mussweiler and Strack 1999) consequences for the self. Recent social psychological research has demonstrated that these social comparison consequences can take one of two forms. In particular, self-evaluations may either be assimilated towards a given standard, i.e., another person, or contrasted away from it, whereby its direction is determined by factors such as standard extremity or category membership (‘selective accessibility model’, Mussweiler 2003). For instance, humans were found to improve their performance in a discrimination task when they were faced with a slightly (as opposed to extremely) better participant (Seta 1982). Typically, humans tend to assimilate towards moderate standards and towards in-group members, whereas they contrast away from extreme standards and from out-group members (Brewer and Weber 1994; Dijksterhuis et al. 1998; Mussweiler and Bodenhausen 2002; Mussweiler et al. 2004). How humans see themselves, how they feel about themselves and how they perform thus strongly depends on how they compare to others.

Sensitivity to other individuals’ performance has also been described for several species of non-human animals in a variety of contexts, ranging from mate choice (e.g. fish (Bateson and Healy 2005), frogs (Gasparini et al. 2013), fiddler crabs (Reaney 2009)) to rival assessment during conflicts (e.g. carnivores (McComb et al. 1994), primates (Fischer et al. 2004; Wilson et al. 2001)) and reward distribution in foraging tasks (e.g. inequity aversion in primates (Brosnan and de Wal 2003; Talbot et al. 2018), dogs (Range et al. 2009) corvids (Wascher and Bugnyar 2013)). Worth mentioning is that these effects have been studied mainly in the framework of reproduction, communication or cooperation, leaving open to what extend they can be linked to the comparison processes studied in humans (Mussweiler 2003). Possibly, social comparison is a phylogenetically old trait that is essential for judging one’s own skill relative to those of conspecific competitors (e.g., for male/male competition, female choice or access to other resources), and hence is widespread in the animal kingdom. Alternatively, the phenomena described in animals are based on different cognitive processes, and specifically, human comparison skills are based on sophisticated mechanisms that are linked to the concept of self and self-other distinction (Mussweiler 2003). In the latter case, studies on closely related species, i.e., non-human primates, may be informative about the evolutionary roots of human social comparison.

Recent experiments on long-tailed macaques, Macaca fascicularis (Schmitt et al. 2016) and baboons, Papio papio (Dumas et al. 2017) tested predictions from social comparison theory, using a co-action set-up in a discrimination task similar to the one Seta (1982) used with human participants. While the focal subjects performed a discrimination task on a touch screen (TC) computer, they could hear a series of computer beeps indicating the positive and negative choices of a putative co-actor, the comparison standard, in an adjacent compartment. The ratio of positive to negative beeps was experimentally controlled and resembled the performance of a conspecific that was moderately or extremely better or worse than the performance of the focal subject (see also Dumas et al. 2017 for a slightly different method in categorizing pairings as ‘self better’ and ‘other better’). To test for possible effects of category membership, the individuals’ affiliation status to the putative co-actor was varied in macaques (‘friend’ or ‘non-friend’) and the sex composition was taken into account in baboons (pair members of same or different sex). According to the selective accessibility model of human social comparison (Mussweiler 2003), monkeys should perform better with co-actors that are slightly better than themselves (moderate standard) and/or that belong to the same social category (friends; same sex). In contrast, they should perform worse with co-actors that are much better than themselves (extreme standard) and/or belong to a different social category (non-friend; different sex).

The macaques performed indeed better in the presence of a ‘friend’ compared to a ‘non-friend’, but only in the baseline condition when no feedback of the co-actor’s performance was audible. Even more surprisingly, the simulated differences in the performance of co-actors in the test did not affect the monkeys’ discrimination results and only marginally their latency to make a choice. These findings are in accordance with the view that the social comparison skills in question exceed the cognitive realm of long-tailed macaques, and possibly non-human animals in general (Schmitt et al. 2016). This interpretation has been challenged by Dumas and colleagues (2017), who found a significant effect in the predicted direction with moderate standards in baboons, possibly because of refinements in methods and a much larger data set. However, within this data set, only 24 out of 397 dyads revealed scores that were large enough to meet the authors’ criteria to qualify as perceivably different. Thus, although the model estimates are based on a substantial data set, a relatively small fraction supports the critical range of data, making interpretations difficult. Taken together, results on non-human primates in this set-up are equivocal, and further studies are needed to test monkeys’ sensitivity to another’s task performance, if possible, in a set-up that is more ecologically relevant to them than working on computers.

In a recent study, Keupp and colleagues (2019) presented long-tailed macaques with a co-action food extraction task: two monkeys could access a feeding apparatus from opposing sides and pull drawers within their reach. One monkey was defined as ‘partner’ and its performance was manipulated either by attaching weights to the drawers or by changing the food distribution; while the former manipulation affected the partner’s effort to pull and thus its speed of food acquisition, the latter manipulation affected the level of food competition, i.e., whether or not subjects were exploiting the same resource. The authors found no indication that focal subjects compared and adapted their performance to the performance levels of their partner. However, the monkeys paid close attention to their partner’s actions in the competitive condition, when the other’s behaviour was consequential for food availability. These findings suggest that, if long-tailed macaques were capable of engaging in social comparison processes, these should manifest in situations involving direct competition. Social comparisons in humans, in contrast, manifest also in the absence of direct competition, for example in evaluative contexts (Corcoran and Mussweiler 2009; Hoorens and van Damme 2012) or out of social motivations such as conforming to group norms (Baldwin and Mussweiler 2009).

These recent findings in primates provide first insights into possible similarities and differences in social comparison processes between humans and their close relatives and lead us to propose two variants of the original hypothesis formulated for humans (Festinger 1954; Mussweiler 2003): while the ‘strong’ variant assumes that non-human animals engage in social comparison in the same way as humans and hence show the same pattern of assimilation and contrast, a ‘weak’ variant assumes that non-human animals show some elements of social comparison but do not show the same patterns of assimilation and contrast as humans, as these are driven by higher order cognition (i.e. similarity and dissimilarity testing).

Another possibility to shed light on the evolution of social comparison is to test species that are phylogenetically distant to the primate line but face similar social challenges in daily life, so that similar selection pressures could have favoured the convergent evolution of these socio-cognitive skills. Corvids are particularly suited for such an approach, as these birds excel in a variety of cognitive tasks (review in Güntürkün and Bugnyar 2016), often performing comparable to primates (Emery and Clayton 2004; Marler 1996). For instance, ravens Corvus corax selectively choose cooperation partners (Asakawa-Haas et al. 2016), reciprocate favours (Fraser and Bugnyar 2012) and retaliate cheating (Massen et al. 2015); they judge competitors for food based on what those can or cannot see (Bugnyar et al. 2016) and extract information about others’ relationships from a third-party perspective (Massen et al. 2014). Furthermore, ravens remember their relationship valence to other ravens over years (Boeckle and Bugnyar 2012) and discriminate ‘fair’ from ‘unfair’ human experimenters after weeks (Müller et al. 2017). Most importantly for the current study, ravens and carrion crows Corvus corone respond to inequity in reward quality and working effort in a token exchange task, indicating sensitivity to their own and others’ behaviour and payoff in an experimental setting (Wascher and Bugnyar 2013).

We here focused on carrion crows, as these birds can easily be trained on TC computers (Braun 2013), allowing us to use a set-up similar to that of Seta (1982) and almost identical to that of Schmitt et al. (2016). In fact, the second experiment of the current study was designed in line with the study of Schmitt et al. (2016) as part of a joint program. Before this collaborative study, we conducted an exploratory experiment, Experiment 1, in which we investigated whether a social setting would affect the crows’ performance in a discrimination task at all. With this Experiment 1, we also aimed to distinguish between two hypotheses according to which the social context would either facilitate (compare Caldwell and Whiten 2003) or inhibit (compare Giraldeau and Lefebvre 1987) the focal subject’s performance. We further investigated whether an effect of social context was merely due to the other’s presence or its actions, by giving the focal subject a head start (conspecific is merely present) or a delayed start (conspecific is already performing the task). In Experiment 2, we utilized the procedure of the social comparison study on macaques (Schmitt et al. 2016) and explored in what way and to what extent the performance of a conspecific co-actor would influence the crows’ subsequent performance (standard direction and extremity) and whether the identity of the co-actor would matter (category membership) by testing the birds with their mated pair partner and non-affiliated individuals (non-partners). If crows are capable of human-like social comparison (Mussweiler 2003), the direction of social comparison effects should depend on the extremity of the standard performance and the category membership. Specifically, we would predict that subjects assimilate towards moderate standards, i.e., adjust their own performance (not necessarily consciously) to become similar to the one they can hear, and contrast away from extreme ones. The other’s influence should be more pronounced in affiliated dyads than non-affiliated ones. As most findings in primates do not support these predictions of a ‘strong’ variant of social comparison, we were prepared to test also the newly proposed idea of a ‘weak’ variant. Accordingly, crows should respond to others’ actions but would not show the same patterns of assimilation and contrast as humans. The ‘strong’ variant of our social comparison hypothesis thus predicts i) that subjects should assimilate towards moderate standards and contrast away from extreme ones and ii) that the other’s influence should be more pronounced in affiliated dyads than non-affiliated dyads. The ‘weak’ variant predicts that crows show some elements of social comparison, but do not show the same patterns of assimilation and contrast as humans (i.e., similarity and dissimilarity testing).

Subjects and set-up

In total, nine carrion crows (5 male, 4 female) participated in this study. They were all donated from private owners or collected from sanctuaries over the course of five years to form a captive colony at the Konrad Lorenz Forschungsstelle (KLF) in Grünau, Austria. Specifically, three crows arrived as nestlings and were hand-raised to fledging at KLF; four crows were hand-raised by other persons and arrived shortly after fledging at KLF, and two other crows arrived as adults at KLF (Table S1 in Supplementary Information).

All birds were individually marked with coloured leg bands. From 2007 to 2012, they were housed socially in one group at KLF and each bird’s dominance status was determined during that period (i.e., prior to splitting them up into pairs), once per year via approach/retreat interactions, displacements at food, and show-off displays. Within a pair, male crows are generally dominant over females. In 2012, the crows moved to a newly built aviary complex at the Cumberland Wildpark Grünau, Austria (Fig. 1a). The aviary complex consisted of five outdoor holding compartments (24 m² each) and a central indoor area for experiments (circa 21 m²). At the time of Experiment 1 (2012), three birds had formed a pair bond with a partner of the opposite-sex and thus were housed in pairs in three visually separated holding compartments; the others were housed in one social group in the remaining holding compartments. At the time of Experiment 2 (2014), all birds were housed in pairs. All holding compartments were connected via doors to the indoor experimental area with thick wooden walls, which blocked any view and attenuated most sounds. The experimental area could be further divided into two equally-sized test compartments (Fig. 1a, b). Crows were fed twice a day (morning, afternoon) with a mixture of meat, fruit, vegetables and milk products. The amount of food given to the birds per day did not change during the experiments, but the preferred items were shifted to the afternoon feeding in order to keep up the crows’ motivation to work for treats throughout the day.

All birds were well habituated to brief individual separation for experimental testing and entered the experimental area on a voluntary basis after being called by name. Prior to the experiments, they were all trained on a TC computer, i.e., they had to discriminate between depicted images in a two-choice task by pecking the rewarded stimulus. The images were displayed next to each other, in the centre of the screen on white background. When choosing a positive image, a ‘beep’ (frequency = 300 Hz, duration = 100 ms) was played back as acoustical reinforcement, and the subjects received a food reward; choosing a negative image produced an error sound (frequency = 150 Hz, duration = 300 ms), the presentation of a red screen for one second and no food reward. If the subject had chosen a negative image in the previous trial, a ‘correction trial’ followed, i.e., the same stimulus pairing was displayed again, allowing the birds to learn from their mistakes.

Five of the crows had already participated in another discrimination study on a TC (involving depicted objects or conspecifics; Braun 2013). To ensure that all crows were at a similar level of performance at the onset of the current studies (75% correct), they received additional TC training before Experiment 1 and 2, respectively (see Supplementary Information for details). We used dried dog food as a reward in the experiments. Dried dog food is highly preferred by the crows and not part of their standard daily diet.

Procedure

Experiment 1

We equipped both indoor compartments of the experimental area with identical TC computers (Elo Touch Screen; distance between computer screens: approx. 70 cm). The partition between the two compartments consisted of wire mesh so that we could test two subjects in physical separation but in full visual and auditory contact. Subjects could thus work on their own computer and observe their neighbour working on the other computer.

We exposed eight crows (4 m, 4 f) to a series of eight discrimination tasks using images of non-human animals of different taxonomic groups (Fig. S1 in Supplementary Information). Per task, the crows had to discriminate a small number of images of two taxonomic groups (e.g., 4 butterflies and 4 bees, half of the subjects were shown butterflies as S+, the other half bees). The resulting eight different images were presented in different combinations three times per experimental session (n = 24 trials/session). Crows were tested on a given task until they reached criterion (75% correct at first choice, over three consecutive sessions). They could solve the task by different means, e.g., rote learning or categorization. We were specifically interested in how long it took them to reach criterion under different social settings (four experimental conditions, see below), irrespective of the cognitive mechanism used.

Crows were tested in fixed dyads consisting of one male and one female each, who were not mated partners but familiar with each other (Table S1 in Supplementary Information). Each dyad was presented with the eight discrimination tasks in four settings: condition (1), ‘simultaneous start’ (both subjects were present in adjacent compartments and could simultaneously work on the task), condition (2), ‘focal head start’ (the focal subject performed the entire task while the dyad partner was present in the adjacent compartment; then the dyad partner performed the task while focal was present), condition (3), ‘focal waits’ (the dyad partner performed the entire task while the focal subject was present in the adjacent compartment; then the focal subject performed the task while the dyad partner was present), condition (4), ‘alone’ (each bird performed the task without the other being present in the experimental area). All four conditions were conducted in a randomised order per dyad. To control for the possibility that images of certain taxonomic groups were easier or more difficult to discriminate than others, we ran the four conditions twice with different taxa to discriminate (series 1: fish, reptiles, canines, insects; series B: molluscs, rodents, spiders, amphibians; Fig. S1 in Supplementary Information).

To test whether crows differed in the number of trials to reach criterion across the four experimental conditions, we calculated a generalized linear mixed model (GLMM) (family = negative binomial) with condition, status (dominant or subordinate), and series as fixed factors and individual and pair as random factors. Subsequently, we calculated Tukey Contrasts (multiple comparisons of means) as post-hoc tests on the fixed factor condition (Table 1).

Experiment 2

We tested six crows (3 male, 3 female), five of which had participated in Experiment 1 (Table S1 in Supplementary Information). The other three birds who had participated in Experiment 1 were not available anymore (two died due to parasite infections, one was transferred to a different facility). The setting was the same as in Experiment 1, except that the partition between the two indoor compartments was now opaque, i.e., in the experimental area, birds were visually, but not acoustically, separated from each other (Fig. 1b). This allowed us to apply auditory feedback from putative conspecific co-actors similarly to the set-up used by Seta (1982) for humans and by Schmitt et al. (2016) for long-tailed macaques. The auditory feedback was given twice in Experiment 2, so that it combined the head start condition (feedback first) and the simultaneous condition (feedback during own session) of Experiment 1. Further details on the methods of Experiment 2 can be found in the Supplementary Information.

Because of a time gap of more than a year between the two experiments, all birds received additional training on the two-choice task procedure as described above. This pre-step ensured that all birds were accustomed with the basic procedure (choosing the positive out of two images) and with receiving acoustical feedback upon their choices (positive ‘beep’ with 300 Hz, 100 ms; error sound with 150 Hz, 300 ms plus red screen and correction trial; this procedure was identical to that of Experiment 1). During this initial step, we also audio-recorded the birds’ pecking sounds when choosing correct or incorrect images using a direction microphone (Sennheiser) and audio recorder (Marantz Solid State Recorder PMD661).

In the second training step, all subjects were presented with novel categories of images, i.e., images of male and female humans (Fig. S2 in Supplementary Information). We opted for this type of discrimination task because i) discriminating 2D human images is within the cognitive capacity of birds (Loidolt et al. 2003; Troje et al. 1999) and ii) recognizing specific humans (researchers, caretakers) out of otherwise anonymous people (visitors to game park) is an ecologically relevant problem for our crows. As we planned the study together with the first macaque study (Schmitt et al. 2016), we could also use the same images for discrimination. We presented 20 images per session in a randomised order (8 passport cut-outs, 6 half-body, 6 whole-body images); half of the birds were rewarded for choosing male images, the other half for choosing female images. We set a relatively moderate criterion of 70% of correct first choices in three out of six consecutive sessions to allow enough variance and thus also room for improvement in performance for the following test. This learning criterion was reached in an average of 32.5 sessions (range: 16-40).

After the criterion was reached and right before the test, the crows conducted three daily sessions in the exact same way as in the subsequent test (see below), but without a second bird being present in the experimental area and without any sound being played back. These pre-test sessions served to establish the performance level of each crow (proportion of correct choices), which the playbacks in the test were then based on (for details on how playbacks were assembled see Table S2 in Supplementary Information).

In test sessions, the crows faced a putative co-actor, which was either their mate they shared an aviary with (‘partner’) or a conspecific from an adjacent compartment (‘non-partner’; Table S1 in Supplementary Information).

Subjects could determine the identity of the co-actor at the beginning of a session, when both birds were let into the indoor area of the aviary together. They were then placed separately into the two visually isolated compartments with the TC computers from where they could hear but not see each other (Fig. 1b). Out of sight of the subject, the co-actor was then returned to the housing compartment and a playback mimicked its performance in the experiment. To ensure the subject would believe the co-actor was still present, we released it quietly and subsequently provided food to the co-actor so as to keep the bird from vocalising. The co-actor never acted in a way that would suggest this had not worked (e.g. calling or searching for the co-actor). Furthermore, the playback sounds did not just include positive or negative feedback to the choices made at the TC, but also individual, pre-recorded pecking sounds the birds had made at the TC. The experimenter (IGF) exposed the focal subject twice to the same playback sequence of 20 trials in which the putative co-actor pecked the TC and received positive and/or negative feedback sounds. In the first round of playback, the focal subject’s TC was switched off so as to ensure that the subject’s attention was not diverted from hearing the playbacks. In the second round, the TC was switched on and the focal subject could perform the task while listening to the playback of the putative co-actor again. The ratio of positive to negative trials in the playback mimicked a performance of the putative co-actor that was moderately or extremely better or moderately or extremely worse than the subject’s discrimination in the pre-test sessions (moderate standards: average number of positive sounds plus or minus one standard deviation; extreme standards: average number of positive sounds plus or minus four standard deviations; see Table S2 in Supplementary Information). The subject thus heard pecking sounds by the putative co-actor and a performance, which had been calculated and assembled based on its own performance. In the control condition, no playback was played.

Test conditions were presented in a pseudo-randomised, counter-balanced order in six blocks (Table S3 in Supplementary Information). Three blocks were conducted with the partner as the putative co-actor, three with a non-partner. Within each block, each condition was tested once, resulting in 30 sessions per bird and thus three sessions per condition with a partner and three with a non-partner. Conditions were sessions with a moderately better or worse standard, an extremely better or worse standard or alone. Typically, one session was conducted per day. Per session, 10 of the displayed stimulus pairs were familiar images from the training sessions; the other 10 pairings consisted of novel images. Interspersing novel images should increase the difficulty of the discrimination and thus the variance in subjects’ performance. The order of the familiar and novel images displayed in each session was randomised.

To test whether standard extremity and/or category membership (relationship to co-actor) had an effect on the birds’ performance (proportion of correct choices) when discriminating stimulus pairs, we fitted Generalised Linear Mixed Models (GLMMs) applying random intercept/random slope models with a Gaussian error structure. We used a beta distribution with a logit link for analysing the birds’ performance when choosing the correct stimulus out of two familiar or two novel stimuli. We fitted a GLMM for all stimulus pairs to examine whether the familiarity of images shown had an influence on the number of correct choices made by the animals. We modelled fractions of correct choices made by the birds, i.e., the percentage of the positively rewarded stimulus (S+) pecked out of the two images in a stimulus pair, the direction (playback played, i.e., better or worse), extremity of the standard (moderate or extreme), the social relationship to co-actor (partner or non-partner) and the stimulus familiarity (familiar or novel image) as fixed effects. According to our hypothesis, we added the interaction of direction with extremity and direction with relationship to the co-actor; bird ID and test block were used as random effects. The data and analysis code can be found in the Supplemental Information.

Experiment 1

Being together with a conspecific significantly improved the crows’ performance: birds needed fewer sessions to reach criterion when they were tested simultaneously with a conspecific in the adjacent compartment (M = 1.44, SD = 1.46), as compared to when tested alone (M = 4.06, SD = 4.39; Table 1, Fig. 2).

Table 1. Effects of different predictor variables on the number of sessions needed to reach criterion in Experiment 1 (above) and results of post-hoc Tukey Contrasts (below). Estimates for the predictor variables, standard errors, z- and p-values were calculated with a GLMM. ‘Series’ stands for the two different sets of stimulus pairs used per condition (see Fig. S1 in the Supplementary Information). Significant effects are indicated by asterisks (*** p<0.001, ** p<0.01, * p<0.05).

coefficients	estimate	std. error	z-value	p-value
(intercept)	1.498	0.251	5.970	<0.001 ***
condition ‘simultaneous start’	-1.003	0.288	-3.480	0.001 ***
condition ‘focal head start’	-0.384	0.253	-1.520	0.130
condition ‘focal waits’	-0.205	0.245	-0.830	0.405
status (subordinate)	-0.142	0.218	-0.650	0.515
series 2	-0.237	0.190	-1.250	0.212

Tukey Contrast comparisons	estimate	std. error	z-value	p-value
simultaneous - alone	-1.003	0.289	-3.476	0.002 **
head start - alone	-0.384	0.253	-1.516	0.400
waiting - alone	-0.205	0.245	-0.833	0.822
head start - simultaneous	0.619	0.391	1.583	0.361
waiting - simultaneous	0.798	0.384	2.081	0.145
waiting – head start	0.179	0.359	0.500	0.954

Being tested sequentially (focal subject either first or second) did not have a significant effect; the crows’ performance, i.e., the number of sessions to reach criterion, was in between those of the alone and simultaneous conditions (focal head start: M = 2.63, SD = 1.88; focal wait: M = 3.13, SD = 2.22; Table 1, Fig. 2).

Experiment 2

The ‘strong’ variant of our social comparison hypothesis predicts i) that subjects should assimilate towards moderate standards and contrast away from extreme ones and ii) that the other’s influence should be more pronounced in affiliated dyads than non-affiliated dyads. Contrary to these predictions, we found no significant effect of either interaction (direction with extremity, direction with relationship to co-actor) on the crows’ performance (Table 2a). The model did reveal a strong effect of stimulus familiarity (whether the images shown were familiar or novel) and some effect of standard extremity (Table 2a), however.

As we could not find (the predicted) interaction effects, we calculated the same model, but without interaction terms (Table 2b), performed based on its AIC value equally good (-240.2 vs -241.0). In this latter model, the fractions of correct choices made were best explained as a function of direction and stimulus familiarity (Fig. 3). Thus, how well crows performed at choosing the correct stimulus out of two images depended on the performance of the putative co-actor and on whether the images shown were familiar or novel. Specifically, the crows performed better, when their co-actor performed better than when the co-actor performed worse than themselves and they performed better with familiar images as compared to novel images (on average, 72.2% correct with direction better vs 66.4% correct with direction worse with familiar images; 65% correct with direct better vs 64% correct with direction worse with novel images).

Table 2. Effects of different predictor variables on choosing the correct stimulus in Experiment 2 (a) with interactions and (b) without interactions. Both interaction terms (marked in bold) had no significant effect and did not increase the AIC compared to the model without interactions. Estimates for the predictor variables, standard errors, z- and p-values were calculated with a GLMM. Significant effects are indicated by asterisks (*** p<0.001, ** p<0.01, * p<0.05).

(a)
Coefficients	estimate	std. error	z-value	p-value
(Intercept)	1.114	0.241	4.63	< 0.001***
Direction (better)	0.125	0.174	0.72	0.47
Extremity (moderate)	-0.307	0.141	-2.17	0.03*
Relationship (partner)	-0.133	0.147	-0.91	0.36
Stimulus familiarity (novel)	-0.458	0.102	-4.49	< 0.001***
*Direction extremity**	0.223	0.201	1.11	0.27
*Direction relationship**	0.29	0.201	1.44	0.15

(b)
Coefficients	estimate	std. error	z-value	p-value
(Intercept)	0.959	0.246	3.90	< 0.001 ***
Direction (better)	0.325	0.114	2.85	0.004 **
Extremity (moderate)	-0.207	0.171	-1.21	0.225
Relationship (partner)	-0.003	0.133	-0.02	0.981
Stimulus familiarity (novel)	-0.342	0.094	-3.63	< 0.001 ***

Carrion crows were influenced by the presence and actions of a conspecific when solving discrimination tasks on a TC computer. Experiment 1 revealed a facilitating rather than inhibiting effect of social context: crows needed a lower number of sessions to reach criterion when tested simultaneously next to each other compared to when tested alone. Experiment 2 showed that the performance of a putative co-actor had an effect on the crows’ performance, but we did not find the predicted interaction between affiliation status, direction and extremity of comparison standard. The same model without interactions (more parsimonious) was equally good (based on the AIC values), and showed that crows performed better when their co-actor was better than when their co-actor was worse than themselves and when they discriminated between familiar images rather than novel images. Standard extremity (moderate or extreme) showed a significant effect only in the statistically less powerful model including interactions, and category membership of the putative co-actors (mated partner vs. non-affiliated individual) had no effect on their performance. Hence, we find some support for a ‘weak’ variant of the social comparison hypothesis but not for the ‘strong’ variant discussed for humans.

Our results are clearly in line with the assumption that social settings affect information processing in crows. One plausible mechanism could be a shift in motivation, referred to as social facilitation (Zajonc 1965; Zentall 2004). Such a motivational shift typically results in increased activity levels or in reduced fear in the presence of a conspecific, both of which may facilitate individual learning (Morrison and Hill 1967). Social facilitation has been shown to influence task performance in humans (Seta 1982) and non-human animals ranging from capuchin monkeys (Dindo et al. 2009) to rats, Rattus norvegicus (Levine and Zentall 1974) and cockroaches, Blatta orientalis (Zajonc et al. 1969). The performance of crows in Experiment 1 fits well with the idea of increased activity levels through social facilitation. Note that the motivational shift could be found in the simultaneous condition only, hinting at scramble competition for food, which is a typical feature in wild crows’ foraging behaviour (Miller et al. 2014). A fear-reducing role of social facilitation is unlikely to be at play in the current set-up, because all birds were well accustomed to the procedure due to individual training.

In Experiment 2, the crows only had visual access to their putative co-actor just before the experiment, when they were let into the test compartments. Receiving auditory information about the other’s behaviour in the experiment seems to be enough to trigger a motivational shift characteristic for social facilitation; it also rules out alternative forms of social learning that rely on visual information such as observational conditioning or action imitation (Hoppitt and Laland 2008). Interestingly, the motivational shift is specifically evident in the birds’ categorization of familiar stimuli known from training, suggesting that it affects selective attention and/or a better retrieval of learned information rather than an increased activity per se. Moreover, this effect could be explained by the difficulty of categorizing novel images; in the study by Zajonc (1969), social facilitation also resulted in an improved performance in simple tasks, but a decreased performance in difficult tasks. The apparent discrepancy between the results of Experiment 1 (subjects learned a discrimination faster) and Experiment 2 (subjects responded to already learned images only) might be explained by the task affordances, as Experiment 1 required crows to distinguish 8 images out of two distinguished taxonomic groups, whereas Experiment 2 required them to categorize 10 familiar and 10 novel images of female and male humans per session (gender discrimination).

In Experiment 2, the crows were influenced by the standard direction, i.e., whether their putative co-actor was performing better or worse than they were. The playback thus seems to serve not only a motivational function - that an adjacent conspecific is receiving food - but also an informational function - how this conspecific is performing in the task. Hence, the findings of Experiment 2 cannot be explained by social facilitation alone, but point towards some form of social comparison process, i.e., the ‘weak’ variant of our hypothesis. Contrary to the typical findings with human participants (Mussweiler 2003), the direction of social comparison effects in the tested crows did not depend on the extremity of the standard performance. On the one hand, this context-dependent switch might be cognitively demanding, possibly exceeding the capacities of crows. On the other hand, defining meaningful standards on the basis of behaviour seems to be difficult, specifically in these first studies on non-human animals (Dumas et al. 2017; Keupp et al. 2019; Schmitt et al. 2016). Note that even in the study by Seta (1982), which inspired the design of the recent studies on non-human animals, subjects improved their performance only when they experienced a slightly better-performing participant, but not when they experienced a participant that was performing extremely better, worse, or on an equal level. Possibly, this type of set-up (which represents a rather untypical set-up for human social comparison tests) could have biased subjects to pay attention to moderate standards. As a final point, we acknowledge that the effects in our study are not strong (compare Figure 3), most likely due to substantial individual variation and a small sample size. Such constraints could be circumvented in the future by using automatized training and test procedures as in the study of Dumas et al. (2017), allowing researchers to work with several individuals of a colony at a time. If subjects are able to freely choose to work on the TC, this may result in a lack of control in who is working in whose presence, however. Future studies should also include a motivational control condition where auditory feedback is provided, but the presence of a conspecific is not simulated. This should rule out the effect of a positive auditory feedback on the crows’ motivational state and thus performance. Although we do not believe this played a role in the current studies and this was also not included in the original human studies (Seta 1982), it would add an additional piece to the puzzle.

Taking a comparative perspective, our study supports the assumption that aspects of social comparison might evolve in social systems with a certain degree of complexity, independently of the species’ phylogenetic relatedness to humans. At first glance, the apparent contrast to our almost identical study on long-tailed macaques (Schmitt et al. 2016) seems puzzling. Both crows and macaques live in structured social groups (macaques: Thierry et al. 2004; crows: Baglione et al. 2002; Miller et al. 2014) that are expected to favour the ability to compare one’s own performance to that of another. In line with this assumption, both species have previously been found to be averse to inequity in reward distribution and/or working effort in exchange tasks (Massen et al. 2012; Wascher and Bugnyar 2013). The macaques were also sensitive to the presence of a co-actor in the currently used set-up, as they performed significantly better when an affiliated than when a non-affiliated co-actor was present (Schmitt et al. 2016). Interestingly, they showed this effect of membership category only in the social control condition, when no auditory feedback was given. In test conditions with playback, however, the macaques did not respond to a co-actor’s performance. It is possible that they failed to recognise the auditory feedback from the adjacent compartment as being indicative of the behaviour of a co-actor and/or they did not pay attention to the available information. Thanks to the two-step approach of the current study, we can rule out similar methodological ambiguities for crows. While Experiment 1 confirms that crows were influenced by the other’s visual presence and actions, Experiment 2 indicates that auditory feedback was enough to trigger a response, at least with familiar stimuli. Hence, unlike the macaques, our crows seemed to pay attention to the relevant cues in the current set-up. Note that in the most recent study on macaques, Keupp et al. (2019) avoided the auditory modality, and subjects were always able to see their partners’ reward; still, this had hardly any effect on their social comparison process, as long as there was no direct, or anticipated, competition for the reward. Unexpectedly, category membership of their co-actors had no significant effect on the birds’ performance. The fact that crows were treating partners similar to non-partners could indicate that they were not aware of the identity of their putative co-actors in our set-up. Alternatively, crows could have been equally attentive towards members of both social categories, which has also been found in a study testing for attention patterns in jackdaws (Scheid and Bugnyar 2008).

In conclusion, we show that carrion crows are not only influenced by the behaviour of others but likely engage in a social comparison process that resembles some aspects of the self-evaluation of humans (responsiveness to standard direction) but do not show the same patterns of assimilation and contrast as humans, probably as these are driven by higher-order cognition (similarity and dissimilarity testing). Our findings are in line with the formulation of a ‘weak’ social comparison hypothesis for non-human animals derived from recent studies on primates (Dumas et al. 2017; Keupp et al. 2019; Schmitt et al. 2016). To our knowledge, this is the first study indicative of a comparison process in birds. Further species need to be tested in order to shed more light on the biological relevance and evolution of such a process, as well as its underlying mechanism.

Ethical Approval

Research presented here was conducted in line with the ASAB/ABS Guidelines for the Use of Animals in Research (Animal Behaviour, 2006, 71, 245-253), and in compliance with the Austrian Animal Experiments Act (§ 2, Federal Law Gazette No. 114/2012; use of playbacks: Animal Experimentation Permit Number 66006/0019-WF/II/3b/2014).

Competing Interests

The authors have no competing interests.

Authors’ Contributions

IGF participated in the design of the study, carried out experimental work for Exp. 2, participated in data analysis and drafted the manuscript; VS participated in the design of the study; RS and ML conducted data analysis; CR and AB carried out experimental work for Exp. 1; CM participated in the design of the study; JF participated in the design of the study and participated in data analysis; TM conceived the study, participated in the design of the study and coordinated the study; TB conceived the study, participated in the design of the study and drafted the manuscript. All authors gave final approval for publication.

Acknowledgements and Funding

This study was financially supported by DFG (Forschergruppe FOR 2150 “Relativity in Social Cognition“ (Fi707/18-1)), Leibniz Price to T. Mussweiler (Mu 1500/5-1) and the FWF (Austrian Science Fund, START prize Y366-B17) to T. Bugnyar; the ‘Verein der Förder KLF’ provided permanent support and the Cumberland Wildpark Grünau provided logistical support. We are grateful to the teams at the Social Cognition Center Cologne, University of Cologne, the Cognitive Ethology Lab at the DPZ Göttingen and the Cognitive Biology Department, University of Vienna, for fruitful discussions. Furthermore, we thank the team at the Konrad Lorenz Forschungsstelle, Grünau, for their help with the birds and Nadja Kavcik for assembling Figure 1b.

Availability of data and materials

The data and analysis code can be found in the Supplemental Information.

Asakawa-Haas K, Schiestl M, Bugnyar T, Massen JJM (2016) Partner choice in raven (Corvus corax) cooperation. PLoS ONE 11:e0156962 doi:10.1371/journal.pone.0156962
Baglione V, Marcos JM, Canestrari D, Ekman J (2002) Direct fitness benefits of group living in a complex cooperative society of carrion crows, Corvus corone corone. Animal Behaviour 64:887-893 doi:10.1006/anbe.2002.2007
Baldwin M, Mussweiler T (2018) The culture of social comparison. Proceedings of the National Academy of Sciences of the United States of America 115:E9067–E9074 doi:10.1073/pnas.1721555115
Bateson M, Healy SD (2005) Comparative evaluation and its implications for mate choice. Trends in Ecology and Evolution 20:659-664 doi:10.1016/j.tree.2005.08.013
Boeckle M, Bugnyar T (2012) Long-term memory for affiliates in ravens. Current Biology 22:801-806 doi:10.1016/j.cub.2012.03.023
Braun A (2013) Social complexity in corvid non-breeder aggregations., University of Vienna, Austria
Brewer MB, Weber JG (1994) Self-evaluation effects of interpersonal versus intergroup social comparison. Journal of Personality and Social Psychology 66:268-275 doi:10.1037//0022-3514.66.2.268
Brosnan SF, de Waal FBM (2003) Monkeys reject unequal pay. Nature 425:297-299 doi:10.1038/nature01963
Bugnyar T, Reber SA, Cameron Buckner C (2016) Ravens attribute visual access to unseen competitors. Nature Communications 7:10506 doi:10.1038/ncomms10506
Caldwell CA, Whiten A (2003) Scrounging facilitates social learning in common marmosets, Callithrix jacchus. Animal Behaviour 65:1085-1092 doi:10.1006/anbe.2003.2145
Corcoran K, Mussweiler T (2009) The efficiency of social comparisons with routine standards. Social Cognition 27:939-948 doi:10.1521/soco.2009.27.6.939
Dijksterhuis A et al. (1998) Seeing one thing and doing another: Contrast effects in automatic behavior. Journal of Personality and Social Psychology 75:862-871 doi:10.1037/0022-3514.75.4.862
Dindo M, Whiten A, de Waal FBM (2009) Social facilitation of exploratory foraging behavior in capuchin monkeys (Cebus apella). American Journal of Primatology 71:419–426 doi:10.1002/ajp.20669
Dumas F, Fagot J, Davranche K, Claidière N (2017) Other better versus self better in baboons: an evolutionary approach of social comparison. Proceedings of the Royal Society B 284:1-7 doi:10.1098/rspb.2017.0248
Emery NJ, Clayton NS (2004) The mentality of crows: Convergent evolution of intelligence in corvids and apes. Science 306:1903–1907 doi:10.1126/science.1098410
Federspiel IG et al. (2023) Social Comparisons in Carrion Crows.
Festinger L (1954) A theory of social comparison processes. Human Relations 7:117-140 doi:10.1177/001872675400700202
Fischer J, Kitchen D, Seyfarth R, Cheney D (2004) Baboon loud calls advertise male quality: acoustic features and their relation to rank, age, and exhaustion. Behavioral Ecology Sociobiology 56:140-148 doi:10.1007/s00265-003-0739-4
Fraser O, Bugnyar T (2012) Reciprocity of agonistic support in ravens. Animal Behaviour 83:171-177 doi:10.1016/j.anbehav.2011.10.023
Gasparini C, Serena G, Pilastro A (2013) Do unattractive friends make you look better? Context-dependent male mating preferences in the guppy. Proceedings of the Royal Society B 280:133–139 doi:10.1098/rspb.2012.3072
Gilbert DT, Giesler RB, Morris KA (1995) When comparisons arise Journal of Personality and Social Psychology 69:227–236 doi:10.1037/0022-3514.69.2.227
Giraldeau L-A, Lefebvre L (1987) Scrounging prevents cultural transmission of food-finding behaviour in pigeons. Animal Behaviour 35:387–394 doi:10.1016/S0003-3472(87)80262-2
Güntürkün O, Bugnyar T (2016) Cognition without cortex. Trends in Cognitive Sciences 20:291-303 doi:10.1016/j.tics.2016.02.001
Hoorens V, Van Damme C (2012) What do people infer from social comparisons? Bridges between social comparison and person perception. Social and Personality Psychology Compass 6:607–618 doi:10.1111/j.1751-9004.2012.00451.x
Hoppitt W, Laland K (2008) Social processes influencing learning in animals: A review of the evidence. In: Brockmann HJ, Roper R, Naguib M, Wynne-Edwards K, Barnard C, Mitani J (eds) Advances in the Study of Behavior, vol 38. 1st edn. Elsevier, pp 105-165. doi:10.1016/S0065-3454(08)00003-X
Keupp S, Titchener R, Bugnyar T, Mussweiler T, Fischer J (2019) Competition is crucial for social comparison processes in long-tailed macaques. Biology Letters 15:20180784 doi:10.1098/rsbl.2018.0784
Levine JM, Zentall TR (1974) Effect of a conspecific’s presence on deprived rats’ performance: social facilitation vs distraction/imitation. Animal Learning & Behavior 2:119-122 doi:10.3758/BF03199135
Lockwood P, Kunda Z (1997) Superstars and me: Predicting the impact of role models on the self. Journal of Personality and Social Psychology 73:91–103 doi:10.1037/0022-3514.73.1.91
Loidolt M, Aust U, Meran I, Huber L (2003) Pigeons use item-specific and category-level information in the identification and categorization of human faces. Journal of Experimental Psychology: Animal Behavior Processes 29:261–276 doi:10.1037/0097-7403.29.4.261
Marler P (1996) Social cognition: are primates smarter than birds? In: Nolan V, Ketterson ED (eds) Current ornithology, vol 13. Plenum Press, New York, USA, pp 1-32
Massen JJM, Pašukonis A, Judith Schmidt J, Bugnyar T (2014) Ravens notice dominance reversals among conspecifics within and outside their social group. Nature Communications 5:3679 doi:10.1038/ncomms4679
Massen JJM, Ritter C, Bugnyar T (2015) Tolerance and reward equity predict cooperation in ravens (Corvus corax). Scientific Reports 5:15021 doi:10.1038/srep15021
Massen JJM, Van Den Berg LM, Spruijt BM, Sterck EHM (2012) Inequity aversion in relation to effort and relationship quality in long-tailed macaques (Macaca fascicularis). American Journal of Primatology 74 doi:10.1002/ajp.21014
McComb K, Packer C, Pusey A (1994) Roaring and numerical assessment in contests between groups of female lions, Panthera leo. Animal Behaviour 47:379-387 doi:10.1006/anbe.1994.1052
Miller R, Schiestl M, Andrew Whiten A, Schwab C, Bugnyar T (2014) Tolerance and social facilitation in the foraging behaviour of free-ranging crows (Corvus corone corone; C. c. cornix). Ethology 120:1248-1255 doi:10.1111/eth.12298
Morrison BJ, Hill WF (1967) Socially facilitated reduction of the fear response in rats raised in groups or in isolation. Journal of Comparative and Physiological Psychology 63:71-76 doi:10.1037/h0024175
Müller JJA, Massen JJM, Bugnyar T, Osvath M (2017) Ravens remember the nature of a single reciprocal interaction sequence over 2 days and even after a month. Animal Behaviour 128:69–78 doi:10.1016/j.anbehav.2017.04.004
Mussweiler T (2003) Comparison processes in social judgement: Mechanisms and consequences. Psychological Review 110:472-489 doi:10.1037/0033-295X.110.3.472
Mussweiler T, Bodenhausen G (2002) I know you are but what am I? Self-evaluative consequences of judging in-group and out-group members. Journal of Personality and Social Psychology 82:19-32 doi:10.1037//0022-3514.82.1.19
Mussweiler T, Rüter K, Epstude K (2004) The man who wasn't there: Subliminal social comparison standards influence self-evaluation. Journal of Experimental Social Psychology 40:689-696 doi:10.1016/j.jesp.2004.01.004
Mussweiler T, Strack F (1999) Hypothesis-consistent testing and semantic priming in the anchoring paradigm: A selective accessibility model. Journal of Experimental Social Psychology 35:136-164 doi:10.1006/jesp.1998.1364
Range F, Horn L, Viranyi Z, Huber L (2009) The absence of reward induces inequity aversion in dogs. Proceedings of the National Academy of Sciences of the United States of America 106:340-345 doi:10.1073/pnas.0810957105
Reaney LT (2009) Female preference for male phenotypic traits in fiddler crab: do females use absolute or comparative evaluation? Animal Behaviour 77:139-143 doi:10.1016/j.anbehav.2008.09.019
Scheid C, Bugnyar T (2007) When, what, and whom to watch? Quantifying attention in ravens (Corvus corax) and jackdaws (Corvus monedula). Journal of Comparative Psychology 121:380-386 doi:10.1037/0735-7036.121.4.380
Schmitt V et al. (2016) Do monkeys compare themselves to others? Animal Cognition 19:417–428 doi:10.1007/s10071-015-0943-4
Seta JJ (1982) The impact of comparison processes on coactors' task performance. Journal of Personality and Social Psychology 42:281-291 doi:10.1037/0022-3514.42.2.281
Talbot CF, Parrish AE, Watzek J, Essler JL, Leverett KL, Paukner A, Brosnan SF (2018) The influence of reward quality and quantity and spatial proximity on the responses to inequity and contrast in capuchin monkeys (Cebus [Sapajus] apella). Journal of Comparative Psychology 132:75–87 doi:10.1037/com0000088
Thierry B, Singh M, Kaumanns W (2004) Macaque societies: A model of the study of social organization. Cambridge University Press, Cambridge, UK
Troje NF, Huber L, Loidolt M, Aust U, Fieder M (1999) Categorical learning in pigeons: The role of texture and shape in complex static stimuli. Vision Research 39:353–366 doi:10.1016/S0042-6989(98)00153-9
Wascher CAF, Bugnyar T (2013) Behavioral responses to inequity in reward distribution and working effort in crows and ravens. PLOS One 8:e56885 doi:10.1371/journal.pone.0056885
Wheeler L, Miyake K (1992) Social comparison in everyday life. Journal of Personality and Social Psychology 62:760–773 doi:10.1037/0022-3514.62.5.760
Wilson ML, Hauser MD, Wrangham RW (2001) Does participation in intergroup conflict depend on numerical assessment, range location, or rank for wild chimpanzees? Animal Behaviour 61:1203-1216 doi:10.1006/anbe.2000.1706
Zajonc RB (1965) Social facilitation. Science 149:269–274 doi:10.1126/science.149.3681.269
Zajonc RB, Heingartner A, Herman EM (1969) Social enhancement and impairment of performance in the cockroach. Journal of Personality and Social Psychology 13:83-92 doi:10.1037/h0028063
Zentall TR (2004) Action imitation in birds. Learning & Behavior 32:15-23 doi:10.3758/BF03196003

No competing interests reported.

Are You Better Than Me? Social Comparisons in Carrion Crows (Corvus corone)

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Materials and Methods

Results

Discussion and Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1