Introduction
Good, bad, right, and wrong are some of the central concepts in moral philosophy and psychology. Socrates, Aristotle and the Stoics discussed the distinction between apparent and real good and they were puzzled by the fact that generally all people wish to do well, the question is “how we are to do so” (Boeri, Reference Boeri2004). Many philosophical works in the analytical tradition focused on the meanings of ‘good’ and related value-words, such as ‘right’ and ‘ought’, as they are used in moral judgements (e.g., Hare, Reference Hare1952). Good and bad are not necessarily linked to the moral domain, even when they refer to positive or harmful events. A tornado is very bad, but it is not morally wrong (Nichols & Mallon, Reference Nichols and Mallon2006).
The word ‘good’ is used in a variety of contexts with different meanings. It might occur as an adjective, a noun, an interjection, and an adverb. Thus, there will be different lexical readings for each of these types of occurrence (Katz, Reference Katz1964). The word ‘good’ is a very frequent word in spoken and written language, but its semantic complexity may have discouraged developmental psychologists from investigating the acquisition of its meanings. The developmental literature is rich in studies concerning the acquisition of concrete noun and adjectives (Bloom, Reference Bloom2002), but very few studies looked specifically at the early acquisition of the moral meaning of ‘good’.
By 24 months, typically developing children use a number of adjectives in their everyday speech to refer to the properties of things (e.g., Mintz & Gleitman, Reference Mintz and Gleitman2002). Infants appear to start to comprehend the word ‘good’ by 13 months and start to use it by 22 months (Fenson, Dale, Reznick, Bates, Thal, & Pethick, Reference Fenson, Dale, Reznick, Bates, Thal and Pethick1994). Multiple single case studies found that some children may start to use ‘good’, ‘bad’, ‘naughty’, and ‘nice’ as early as 19 months (e.g., Bloom, Lightbown, & Hood, Reference Bloom, Lightbown and Hood1975). Snow (Reference Snow, Kagan and Lamb1987) conducted an analysis on one child's moral use of three target words, i.e., ‘good’, ‘bad’, and ‘should’, and found that at around 30 months the words ‘good’ and ‘bad’ were used in moral contexts to refer to actions (e.g., “good thing to do”), “spiritual states” (e.g., “good boy”), or mental products (e.g., “good idea”). Wright and Bartsch (Reference Wright and Bartsch2008) found similar results in the conversations of two children with adults, suggesting the moral use of some target words, including ‘good’, from the second year.
The first two years of life may be crucial for understanding the origins of moral cognition. A rich series of studies showed that infants display early socio-moral evaluations of helping and hindering agents (e.g., Bloom, Reference Bloom2013; Dunfield & Kuhlmeier, Reference Dunfield and Kuhlmeier2010; Hamlin, Reference Hamlin2013, Reference Hamlin2015; Hamlin & Wynn, Reference Hamlin and Wynn2011; for a meta-analysis see Margoni & Surian, Reference Margoni and Surian2018). For example, after seeing a wooden square enable a circle to reach the top of a hill, and a triangle hinder the circle's climb, infants as young as 10 months reach for the helping agent in preference to the one who hinders (Hamlin, Wynn, & Bloom, Reference Hamlin, Wynn and Bloom2007). Their preferences reveal that they attribute positive and negative values to helping and hindering actions, respectively (Hamlin et al., Reference Hamlin, Wynn and Bloom2007; Kuhlmeier, Wynn, & Bloom, Reference Kuhlmeier, Wynn and Bloom2003). In another study, Hamlin, Wynn, Bloom, and Mahajan (Reference Hamlin, Wynn, Bloom and Mahajan2011) showed 8-month-old infants short events with helping or hindering puppets. Then, infants saw a new puppet who acted prosocially (i.e., giving a ball) or antisocially (i.e., taking away the ball), towards the helping and hindering puppet. Infants selectively choose the puppet who acted positively toward prosocial individuals and the puppet who acted negatively towards antisocial individuals.
Infants also display an ability to evaluate agents by taking into account their distributive actions (e.g., Burns & Sommerville, Reference Burns and Sommerville2014; Geraci & Surian, Reference Geraci and Surian2011; Lucca, Pospisil, & Sommerville, Reference Lucca, Pospisil and Sommerville2018; Meristo & Surian, Reference Meristo and Surian2013, Reference Meristo and Surian2014; Ziv & Sommerville, Reference Ziv and Sommerville2017; see Blake, McAuliffe, & Warneken, Reference Blake, McAuliffe and Warneken2014, and Sommerville, Reference Sommerville2018, for reviews). For example, Geraci and Surian (Reference Geraci and Surian2011) presented infants aged 12 to 18 months with two distributors, two recipients, and an observer. One distributor gave each recipient a multicolor disk, while the other distributor gave two disks to one receiver and nothing to the other one. Infants showed a significant tendency to pick the fair rather than the unfair distributor. Infants’ preferences for fair agents were also found by Sommerville, Schmidt, Yun, and Burns (Reference Sommerville, Schmidt, Yun and Burns2013), in a reaching task, and by Surian and Franchin (Reference Surian and Franchin2017a), using a selective helping task. Schmidt and Sommerville (Reference Schmidt and Sommerville2011) presented actual people instead of animal puppets and found that 15-month-old infants looked longer at the unfair distribution, suggesting that infants found this division unexpected. Furthermore, Meristo and Surian (Reference Meristo and Surian2013) found that even at 10 months this evaluation affects infants’ expectations about the likelihood that another agent will reward the distributor. Infants looked longer when a reward was given to an unfair rather than a fair agent: they expect third parties to act positively toward fair donors who have distributed attractive resources equally between two recipients, rather than toward unfair donors who made unequal distributions.
Non-verbal measures such as looking times and preferential reaching suggest that infants by the end of the first year expect resources to be allocated equally to similar recipients (Meristo, Strid, & Surian, Reference Meristo, Strid and Surian2016; Schmidt & Sommerville, Reference Schmidt and Sommerville2011), and in their second year they expect unequal distributions performed towards recipients that differ in merit (Sloane, Baillargeon, & Premack, Reference Sloane, Baillargeon and Premack2012; Surian & Franchin, Reference Surian and Franchin2017b) or dominance (Enright, Gweon, & Sommerville, Reference Enright, Gweon and Sommerville2017). They prefer agents that perform fair distributions and expect other agents to affiliate with fair over unfair distributors (Geraci & Surian, Reference Geraci and Surian2011; Sommerville et al., Reference Sommerville, Schmidt, Yun and Burns2013; Surian & Franchin, Reference Surian and Franchin2017b). Overall, these results on infants’ reactions towards distributing agents and events concerning resources distributions point to an early emerging sense of fairness in human infants.
By 13 months of age, infants also appear to see fair and unfair distributive behaviors as differentially linked to praises and admonishments. DesChamps, Eason, and Sommerville (Reference DesChamps, Eason and Sommerville2016) showed movies of two agents that distributed equally or unequally some crackers, and then showed again, side by side, the two distributors while either some praises (“She is a good girl”, “She did a good job”, spoken with a positive vocal affect) or admonishments (“She is a bad girl”, “She did a bad job”, spoken with a negative vocal affect) were presented from a central speaker. Infants at 13 months looked significantly longer at the unfair distributor while hearing admonishments, but they did not look reliably longer to one of the two agents while hearing praise. At 15 months, infants looked significantly longer at the unfair distributor while hearing praise, but did not look reliably longer at one of the two agents while hearing admonishments. This outcome provides initial evidence of infants’ ability to link the words ‘good’ and ‘bad’ to different distributive actions. However, it is possible that the effect was mainly driven by the positive and negative intonation used in the praise and admonishments, respectively, rather than the words ‘good’ and ‘bad’ per se.
In the present study, the main goal was to investigate whether toddlers map the word ‘good’ to the domains of help (Experiment 1) and fairness (Experiment 2). Infants were exposed to movies or puppet shows starring two agents who helped or hindered others or who distributed resources fairly or unfairly. Infants were asked to pick the good one. Two follow-up studies were carried out to test toddlers’ preference for fair individuals (Experiment 3) and their knowledge of good in a non-moral domain (food; Experiment 4).
Experiment 1
Here, we tested whether 20- and 30-month-olds take the word ‘good’ as referring to a helping rather than a hindering agent.
Method
Participants
Sixty-five healthy full-term infants participated. Infants were divided into two age groups: 26 20-month-olds (14 females; M = 20 months, 81 days; age range: 17 months, 6 days to 25 months, 10 days) and 39 30-month-olds (23 females; M = 31 months, 1 day; age range: 26 months, 25 days to 33 months, 25 days). Twelve additional 20-month-olds were excluded from the study due to their refusal to participate (n = 8), interference of the educator (n = 3), or procedural error (n = 1). Eight additional 30-month-olds were excluded from the study due to their refusal to participate (n = 4), interference of the educator (n = 2), or procedural error (n = 2). With an a priori power analysis performed with G*Power 3.1 (Faul, Erdfelder, Buchner, & Lang, Reference Faul, Erdfelder, Buchner and Lang2009), we calculated the minimum sample size, for each age group, that was necessary to reach a statistical power of 0.80 for a generic binomial test, setting an alpha index at 0.05, and the proportions values at 0.5 (p1) and 0.8 (p2). The sample size obtained was 20.
In this and the following experiments, infants were recruited at local public and private preschool nurseries in Rovereto (Northern Italy), and they were tested only after their parents had given their informed consent. The ethical committee of the University of Trento approved the research project.
Stimuli and procedure
Infants were tested in a quiet room in the nurseries. They sat on an educator's lap or on a children's chair, centered in front of the monitor used to display the stimuli (at a distance of about 60–80 cm). Two cameras (Canon, Legria HF R 26; Sony Handycam DCR-HC27E) were used to record the testing sessions; one was placed behind the monitor and the other behind the participant. Infants were presented with four familiarization events, followed by a test trial as in Hamlin et al. (Reference Hamlin, Wynn and Bloom2007). In the familiarization events two helping and two hindering events were shown. The familiarization events started by showing a climber (a red circle with eyes and mouth) that jiggled up and down at the bottom of a hill and then attempted twice to climb it, but each time it fell back to the bottom. On the third attempt, in the helping events, the climber was bumped up the hill by a different geometrical shape (a square or a triangle), the ‘helper’, whereas in the hindering events the climber was bumped down the hill by a different geometric shape, the ‘hinderer’. The helper entered the scene always from the lower right, and the hinderer from the upper left. The helper and the hinderer characters were played by a blue square and a yellow triangle, both with eyes and mouth. The identity of the characters and the order presentation of the helping and hindering events were counterbalanced. Total events duration was 160 sec (40 sec for each event).
At the end of four familiarization events, in the test trial infants were presented with a white board where there were the helper shape and the hinderer shape (25 cm apart). Using a neutral intonation, the experimenter said to infants: “Pick the good one” (Italian: Prendi quello buono). The side of presentation of the two shapes was counterbalanced.
Infants’ responses coding
The experimenter noted on a sheet the infants’ responses and later she examined the video-recordings to remove any doubt. Furthermore, the experimenter coded the infants’ response time, i.e., the time it took from the experimenter's request to pick the good character for the infants to reach with the hand one character. An independent coder examined all infants’ responses and agreed with the original experimenter on 100% of trials.
Results
Infants’ responses are represented in Figure 1. In the 20-month-olds, the responses were equally distributed in the sample: 13 infants picked the helper, and 13 the hinderer. Infants’ response time to choose the helper was 9.38 sec (SD = 21.54) and for infants who chose the hinderer it was 13.85 sec (SD = 18.88). There was no significant difference between the two response times (z = –1.28, p = .20).
Number of infants who chose the helping or the hindering agents (Experiment 1) and the fair or the unfair distributor (Experiments 2 and 3).
In the 30-month-olds, 27 infants (out of 39) picked the helper rather than the hinderer (p = .024, two-choice binomial test, two-tailed). Infants’ response time to choose the helper was 6.37 sec (SD = 6.66) and for infants who choose the hinderer it was 6.33 sec (SD = 6.01). No significant difference resulted from the two response times (z = –0.05, p = .964).
Note that the sample size of the 30-month-old infants was about twice the size of the 20-month-olds’ sample. We decided to double the number in this age group because it was less difficult to access children of this age group, and in this way we could provide a more representative sample. However, if we analyse the data by having equal number of infants in each age group, thus selecting the first 26 30-month-old infants, the results are similar to those obtained with the complete sample. Twenty infants (out of 26) picked up the helper rather than the hinderer (p = .009, two-choice binomial test, two-tailed). Infants’ response time to choose the helper was 5.15 sec (SD = 4.86), and for infants who choose the hinderer it was 5.17 sec (SD = 4.26). The two response times were not reliably different (z = –0.05, p = .929).
Experiment 2
In this experiment we investigated whether toddlers at 20 and 30 months of age map the word ‘good’ to the domain of fairness.
Method
Participants
Sixty-six healthy full-term infants participated. Infants were divided into two age groups: 24 20-month-old infants (12 females; M = 20 months, 15 days; age range: 17 months, 19 days to 23 months, 21 days; 6 additional infants were excluded from the study because they refused to participate) and 42 healthy full-term 30-month-old infants (24 females; M = 30 months, 24 days; age range: 26 months, 25 days to 34 months, 3 days; 10 additional infants were excluded from the study, 8 because they refused to participate and 2 due to the interference of the educator). The sample sizes of the two age groups was decided a priori based on Experiment 1 sample size.
Stimuli and procedure
Infants sat on an educator's lap in a quiet room of the nursery. An experimenter kneeled in front of the infant and introduced him/her to two teddy-bear families, each one composed by a mother-bear and two cubs (see Figure 2). The two families were differentiated by the color (green or blue) of a ribbon bow tied around the neck of the mother and her cubs. Mother-bears were approximately 40 cm high, while the cubs were 25 cm high. All bears were hidden inside a black rigid bag. Two cameras (Canon, Legria HF R 26; Sony Handycam DCR-HC27E) were used to record the testing sessions: one was placed behind the experimenter and the other behind the participant.
The two teddy-bear families with fair (on the right) and unfair (on the left) distributions of biscuits.
The experimenter started by presenting the first family, she raised a mother-bear and said: “Let's play, I introduce you to mother-bear Lisa [placing it on the floor], and her two cubs [the experimenter took two small bears from the bag]: Carlo [the experimenter showed one of the two cubs and put it in front of the mother-bear on her right] and Marco [the experimenter showed the other cub and put it in front of the mother-bear on her left].” Two biscuits were hidden behind each mother-bear. The experimenter took the biscuits and, showing them one by one to mother-bear Lisa, she mimicked an equal distribution saying: “Mother-bear Lisa gives one cookie to Carlo and another to Marco.” Then the experimenter took the second mother-bear from the bag and, acting out a similar presentation to the first family, she said: “There is also another bear family: Mother-bear Maria with her cubs, Antonio and Alessio.” The experimenter toke the two biscuits hidden behind Maria and mimicked an unequal distribution of the biscuits saying: “Mother-bear Maria gives two cookies to Antonio and nothing to Alessio.” The experimenter then put the two mother-bears in front of the baby (at 50–60 cm) and said, with a neutral intonation: “Pick the good mother-bear” (It. Prendi la mamma orsa buona). Once the baby took or pointed to one mother-bear, the experimenter thanked him/her and let the child play a bit with the bears. Although the educator was instructed before the study not to talk to the infants or point to the bears, if the infants were reluctant to choose, a second experimenter prompted the educator with a slight touch of the shoulder to help the infant to get up to go towards the bears.
The following factors were counterbalanced across infants: order of the distribution (fair first vs. fair second), side of fair distribution (right vs. left), and color of the ribbon bows of the bear families (blue vs. green). The entire session lasted approximately 5 minutes.
Infants’ responses coding
The coding was the same as followed in Experiment 1. In this case, for the infants’ response time the experimenter coded the time it took from the experimenter's request to take the good bear for the infants to reach or point to one mother-bear. An independent coder recoded all infants’ responses and agreed with the original experimenter on 100% of trials.
Results
Infants’ responses to the request of picking “the good mother-bear” are represented in Figure 1. Seven 20-month-old infants (out of 24, 29.2%) selected the fair mother and the rest choose the unfair mother (70.8%) (p = .064, two-choice binomial test, two-tailed). Infants’ response time to choose the fair mother-bear was on average 53.6 sec (SD = 48.09) and did not differ significantly from the response time infants took to choose the unfair mother-bear (M = 24.9 sec, SD = 18.5, z = –1.3, p = .21). Infants’ responses showed no evidence that they have mapped to the word ‘good’ the property of being fair in a distributive action.
In the 30-month-olds, 18 infants (out of 42) chose the fair mother (p = .441, two-choice binomial test, two-tailed). Response time to choose the fair mother-bear was 25.56 sec (SD = 16.51), while the response time to choose the unfair mother-bear was 34.87 sec (SD = 43.33, z = –0.70, p = .484).
As in Experiment 1, in this experiment the sample size of the 30-month-old infants was about twice the size of the 20-month-olds’ sample. We decided to double the number also in this case, in order to have a balanced comparison between the samples of Experiments 1 and 2. However, if we analyse the data by having equal numbers of infants in each age group, thus selecting the first 24 30-month-old infants, the results are similar to those obtained with the complete sample. Eight infants (out of 24) chose the fair mother (p = .152, two-choice binomial test, two-tailed). Response time to choose the fair mother-bear was 22.63 sec (SD = 15.49), while the response time to choose the unfair mother-bear was 41.56 sec (SD = 49.75, z = 0.01, p = 1.0).
Experiment 3
In Experiment 3, 20-month-old infants were presented with the same puppet show scenario as used in Experiment 2, but in this case the experimenter asked them only to “pick a mother bear”, leaving out the word ‘good’. Previous results suggest that they should prefer the fair rather than the unfair agent (Geraci & Surian, Reference Geraci and Surian2011; Surian & Franchin, Reference Surian and Franchin2017b). However, there were several differences in the scenarios and procedures used in these previous studies and the present one. The aim of this experiment was to rule out the possibility that the failures found in Experiment 2 were due to the infants’ inability to generate social preferences and evaluations of agents based on their distributive behaviors.
Method
Participants
Twenty-four healthy full-term 20-month-old infants participated (12 females; M = 20 months, 2 days; age range: 17 months, 25 days to 24 months, 16 days). Fourteen additional infants were excluded from the study due to their refusal to participate (n = 8), interference of the educator (n = 4), or procedural error (n = 2). The recruitment method was the same as Experiment 1. The sample size was decided a priori based on the Experiments 1 and 2 sample size.
Stimuli and procedure
For stimuli and procedure see Experiment 2. Here, at the end of the presentation of the bear families, the experimenter asked the infants: “Pick one mother-bear” (It. Prendi una mamma orsa).
Infants’ responses coding
The coding was the same as in Experiment 2. The independent coder agreed with the original experimenter on 100% of trials.
Results and discussion
Infants’ responses to the request of pick “one mother-bear” are represented in Figure 1. Eighteen infants (out of 24) preferred the fair mother who equally divided resources between her cubs (p = .023, two-choice binomial test, two-tailed). Infants’ response time to choose the fair mother-bear, either by pointing or grasping, was 23.61 sec (SD = 27.04), while the response time for infants who choose the unfair mother-bear was 31.67 sec (SD = 26.42). No significant difference resulted from the two response times (z = –0.70, p = .49). Infants’ preference for the fair agent confirmed previous results, providing further support for an emerging sense of fairness in the second year of life.
Experiment 4
In Experiment 4, infants were tested for their knowledge of the word ‘good’ in the non-social domain of food preferences.
Method
Participants
Twenty-three healthy full-term 20-month-old infants participated (13 females; M = 20 months, 28 days; age range: 17 months, 5 days to 25 months, 10 days). Nine additional infants were excluded from the study due to their refusal to participate. The sample size was decided a priori based on the three previous experiments with same-age infants.
Stimuli and procedure
Infants were presented with a choice test with five pairs of 15 cm width × 21 cm colored pictures depicting different kinds of food. In each pair, there was one item that children typically like (Plasmon biscuit, ice cream, candies, Oro Saiwa biscuits, pizza) and one that they typically dislike (broccoli, fennel, brussels sprouts, lettuce, and radicchio). The side of presentation of the liked and disliked item was counterbalanced across participants. Using a neutral intonation, the experimenter asked the infant to pick or touch “the good one”. The response coding procedure was as in Experiment 1. The two coders agreed on 100% of the trials.
Results and discussion
In three trials, infants preferred a Plasmon biscuit, ice cream, and pizza rather than broccoli, fennel, and radicchio (for all pairs p < .01, two-choice binomial test, two-tailed). In another trial (Oro Saiwa biscuits vs. lettuce), infants’ preference for the biscuits approached significance (p = .052, two-choice binomial test, two-tailed). Finally, for the remaining trial (candies vs. brussels sprouts), infants did not show any significant preference. A possible explanation (suggested by some educators) is that many parents do not allow their children to eat candies and tell them that “candies are bad”. Most infants selected the items that are typically liked by young children (M = 16.2, SD = 3.5) rather than the food typically disliked (M = 3.8 DS = 3.7) (t(8) = 4.87, p = .001), suggesting that, by 20 months, the term ‘good’ is correctly mapped to positive qualities in the food domain.
Toddlers systematically chose their preferred food when asked to pick the ‘good one’. While this is suggestive that they understood the word ‘good’ in this context, it may be argued that the same pattern of choices would also have been observed in the absence of any request. Therefore, it would be useful to run a control study to confirm that the choices observed in Experiment 4 are the result of correct interpretations of the verbal request.
General discussion
Infants were presented with a helping and a hindering agent and asked to pick the ‘good’ one (Experiment 1). Twenty-month-olds’ choices were at chance whereas 30-month-olds selectively picked the helper, suggesting that, by 30 months, infants link the word ‘good’ to helping actions and agents. When presented with a fair and an unfair distributor and asked to pick the ‘good’ one, both 20- and 30-month-olds were at chance (Experiment 2). However, when asked to “pick one mother-bear”, leaving out the word ‘good’, infants preferred the fair agent (Experiment 3). The preference found in Experiment 3 is in line with previous findings showing a preference for fair rather than unfair distributors (Burns & Sommerville, Reference Burns and Sommerville2014; Geraci & Surian, Reference Geraci and Surian2011; Surian & Franchin, Reference Surian and Franchin2017b). Finally, at 20 months, infants’ choices suggest that they interpreted correctly the word ‘good’ as referring to their preferred food as opposed to food typically disliked by children (Experiment 4).
If one takes failure to perform above chance in the present study as suggesting lack of lexical knowledge, taken together the present findings suggest that by 20 months infants know that ‘good’ may express evaluations of food, and by 30 months it refers also to evaluations of some prosocial actions. What infants have not yet discovered is that ‘good’ can be used also to refer to positive qualities in the domain of distributive fairness. However, it is possible that they have already mapped the word ‘good’ to fairness, but failed to show it due to contextual factors. In Experiment 2, the distributive agents were mother-bears, and toddlers may have assumed, by default, that all mothers/mother-bears are good, especially to their children/cubs, regardless of their distributive actions. Under this assumption, the agents’ distributive behavior would have no weight on infants’ responses to the test question. To test this possibility, it would be necessary to run other studies in which the distributors are not mothers of the recipients.
The results found by DesChamps et al.’s (Reference DesChamps, Eason and Sommerville2016) study are consistent with the possibility outlined above: distributors were not presented as mothers of the recipients and infants at 15 months looked preferentially at one of the distributors when hearing the praise “She is a good girl”. It is possible that infants in that study were able to show a preference because no explicit request was asked of them, making the task easier by lowering executive function requirements (Baillargeon et al., Reference Baillargeon, Scott, He, Sloane, Setoh, Jin, Mikulincer, Shaver, Borgida and Bargh2015). Notice, however, that infants in DesChamps et al.’s (Reference DesChamps, Eason and Sommerville2016) study looked longer at the unfair distributor, not at the fair one, and this casts some doubts on their grasp of the word meaning. The reported looking-time differences could be due to infants’ noticing a mismatch between the praise and the unfair agent, but it is not clear whether noticing this mismatch reflects knowledge of the word ‘good’. In fact, the mismatch could also be detected simply by attending to the positive intonation used in the praise, rather than to the word meanings.
The present study has a number of limitations. First, cartoon animations were used in Experiment 1 (on Helping/Hindering events) whereas live events staged using puppets were used in Experiments 2 and 3 (on distributive fairness). In future studies it would be useful to test the replicability of the present findings by using more homogeneous sets of stimuli. However, a recent meta-analysis showed that this factor does not affect infants’ performance on tasks tapping their socio-moral evaluations (Margoni & Surian, Reference Margoni and Surian2018).
The procedure used in Experiments 2 and 3 has probably contributed to the slowing down of the responses compared to the responses found in Experiment 1. It is likely that children's choices in Experiments 2 and 3 were much slower because infants were presented with the two bears on the carpet and they had to get up from the ground to take one of them, whereas in Experiment 1 a tray was put in front of the child and she had just to lift her arm to touch or pick one of the shapes. This slowed down the execution of the response and one may worry whether such slowing down was also responsible for the differences in response patterns. We believe this is unlikely since the information about the distributive fairness or unfairness of the two mother-bears always remained visible to the child, minimizing the memory requirements of the task.
Another issue that is left open by the current results concerns the processes underlying the successes found in Experiment 1. Infants could have chosen the helping agent because they know that ‘good’ refers to also his or her prosocial actions, or because they know that it cannot refer to antisocial actions such as hindering. Further studies employing neutral agents, paired with helping agents, would allow a clarification of this issue.
In the present study we focused on just one of the most frequent evaluative words, ‘good’. We should consider the possibility that, in the domain of fairness, this may not be the first evaluative word learned by children. Infants might start by mapping a different lexical entry such as ‘just’, ‘right’, or ‘fair’. In Italian they may hear bene, bravo, or giusto. Future studies may find more positive results by employing alternative lexical items that are commonly associated with distributive justice. It would also be very interesting to study the input received by children, to assess contextual effects in the use of different words.
This study is a first step towards an understanding of when and how children acquire the different meanings of the term ‘good’. Clearly, this goal is strictly dependent on moral development in general and, more specifically, on the acquisition of the concept MORALLY GOOD. Our findings suggest that a link between the word ‘good’ and prosocial actions, such as helping, is acquired before the link with distributive fairness. If this is indeed the case, lexical development would show a parallel with the acquisition of socio-moral preferences, in which preferences for helpers over hinderers (e.g., Hamlin et al., Reference Hamlin, Wynn and Bloom2007, Reference Hamlin, Wynn, Bloom and Mahajan2011) appear to emerge earlier than preferences for fair agents. Why there should be such a developmental asynchrony? One possibility has to do with the processing load involved in identifying helping agents and fair distributors. It is plausible that the former are easier to identify and track than the latter because of the greater complexity of the sequence of actions involved in distributive contexts. Another possible explanation points to the communicative relevance of exchanging information about helpers and fair distributors. It may be more relevant for young children to comment and communicate about the former than about the latter, since young children are routinely facing helping agents, but only occasionally do they face distributive contexts. This difference should be attenuated in the case of children with siblings of similar age, allowing for future empirical tests of this account.
Acknowledgements
We thank the infants, parents, and nurseries of Rovereto and Trento who participated. We are grateful to Immacolata Giuliani, Nataliia Greiman, and Flavia Pascale for their help with data collection. The authors declare no conflict of interest. When the study was implemented, Federica Savazzi was affiliated to the Department of Psychology and Cognitive Sciences, University of Trento, Italy.
