Infants’ Moral Evaluations of Third-Party
Interactions: A Potential Role for
Attachment
Nettalie Cohen S1076272 Bachelor-Thesis Child and Family Studies
Leiden University Supervisor: Dr. S. Biro
2 Table of contents
Abstract 3
1. Introduction 4
1.1 Understanding actions in infancy 4
1.2 Moral evaluations in infancy 5
1.3 Early experiences and attachment 7
1.4 The current study 9
2. Method 11
2.1 Participants 11
2.2 General procedure 11
2.2.1 Eye-tracking stimuli 12
2.2.2 Procedure of eye-tracking and object choice experiment 13
2.2.3 Strange Situation Procedure 14
2.3 Measures 14
2.3.1 Eye-tracking measure 15
2.3.2 Choice measures 15
2.3.3 Attachment classification 16
2.4 Data analysis plan 16
3. Results 16
3.1 Preliminary analyses 16
3.1.1 Comparing included and excluded infants 16
3.1.2 Testing the main effects of background and condition variables 17
3.2 Hypothesis 1: Agent choice 17
3.3 Hypothesis 2: Agent choice & attachment 19
4. Discussion 20 4.1 Summary of results 20 4.2 General discussion 21 4.2.1 Interaction effects 21 4.2.2 Animated interactions 23 4.2.3 Assumptions 24
4.3 Limitations & recommendations 27
4.4 Conclusion 27
3 Abstract
The capacity to evaluate others’ behavior and to attribute positive or negative values to actions is called moral evaluation (Hamlin, 2013; Bloom, 2010). The origins of this capacity, innate, through socialization or both, is investigated in the current study. Infants’ general capacity to make moral evaluations about responsive and unresponsive behavior in third-party interactions is examined. Also, individual differences in moral evaluation based on infants’ attachment classification is investigated. We presented 147 12-month-old infants with eight animated third-party interactions with two outcomes. In half of the interactions, a big agent acts responsive to a small agent, while in the other half of the interactions the big agents acts unresponsive to the small agent. The two responses of the big agents are marked by different colors. Infants are then presented with two objects in the same shape and color as the two big agents and encouraged to choose the object of their preference. Infants’ preference is taken as an indicator of their moral evaluation. It was found that infants in general chose the
‘responsive’ object as often as the ‘unresponsive’ object. Securely attached infant chose the ‘responsive’ object as often as insecurely attached infants. Interaction effects indicate that gender and condition variables might be related to infants’ preference. The current study found no evidence for infants’ capacity to make moral evaluations about responsive and unresponsive behavior. The results are discussed in light of earlier and future research. Keywords: Infant social cognition, third-party interactions, moral evaluation, attachment
4 1. Introduction
The way infants are viewed has changed dramatically in the last 200 years (Bloom, 2009). Whereas Rousseau viewed the infant as an ignorant creature, nowadays developmental psychologists agree that even very young infants possess considerable knowledge about the world they live in (Bloom, 2009; Johnson, 2010). They have fundamental understanding about various aspects of the physical and mental world. The current study investigates infants’ capacity to make moral evaluations about the behavior of interacting agents depicted in
animations. We explore whether the origins of this capacity is innate, learnt through socialization, or both. In order to understand how infants interpret and evaluate third-party interactions, it is necessary to first describe their impressive early skills in interpreting others’ behavior. Following this, previous research into moral evaluation in infancy will be presented and the main view about the origins of this capacity is discussed. Next, the role of early social emotional experiences in infants’ social and cognitive development is described and a
potential role for attachment in moral evaluation in infancy is suggested. Finally, our current study on moral evaluation in infancy is presented.
1.1 Understanding actions in infancy
It has been suggested that reasoning about human actions require a specific system that represents agents and their actions (Leslie, 1995). Evidence supports the existence of such a system in infancy, that remains present into adult life (Kinzler & Spelke, 2007). Even very young infants can discriminate between objects and agents. Infants represent the goals of others and interpret an agent’s action as goal-directed (Woodward, 1998; Csibra & Gergely, 2007). Infants expect agents to move efficiently and actions to be performed in a rational way (Gergely, Nádasdy, Csibra & Biró, 1995; Gredebäck & Melinder, 2010). From birth, infants have the ability to follow gaze direction and use this to interpret actions of others (Hood, Willen & Driver, 1998; Kinzler & Spelke, 2007). Interestingly, agents do not need to look like real human beings. Infants are sensitive for several behavioral cues, such as self-propelled movement, efficient goal realization, equifinality of actions, and biological movement, that help them to categorize an entity as an intentional agent (Premack & Premack, 1997; Csibra, Biró, Koós & Gergely, 2003; Muir & Hains, 1999). Thus abstract characters, even with no obvious featural similarity to humans, can be interpreted as intentional agents as long as they act toward a goal in a self-propelled and efficient way (Shimizu & Johnson, 2004).
Infants are also able to reason about the interactions of others (Kuhlmeier, Wynn & Bloom, 2003). From 3-months of age, infants are sensitive for subtle differences in movement
5 of abstract disks that either define a social interaction such as a chasing scenario or (non-social) ‘random’ movement (Rochat, Morgan & Carpenter, 1997). Infants have expectations about interacting agents, namely that their interaction is contingent and reciprocal (Kinzler & Spelke, 2007). Older infants also understand that agents can have different roles and
intentions when interacting, where one for example is ‘the chaser’ while the other is ‘the one being chased’ (Rochat, Striano & Morgan, 2004). The emotional sound that accompanies an interaction can also influence how infants interpret an observed scenario (Biro, Alink, van IJzendoorn & Bakermans-Kranenburg, 2014). Infants reasoning about the interaction of others is an important skill for social development. It provides valuable information about possible social partners. Infants can use this information to predict others’ behavior and evaluate who is reliable and who is possibly harmful (Hamlin, 2013; Hamlin & Wynn, 2011). 1.2 Moral evaluations in infancy
The capacity to evaluate others’ behavior and to attribute positive or negative values to actions is called moral evaluation (Hamlin, 2013; Bloom, 2010). An early study using a ‘’violation of expectation’’ paradigm based on looking times, explored infants’ evaluation of agents’ actions (Premack & Premack, 1997). Infants were habituated to either a ‘’positive’’ (e.g. helping) or ‘’negative’’ (pushing/hindering) interaction involving two abstract dots on a screen. Following habituation, a different negative interaction was shown to the infants as a test event. It was found that infants dishabituated more (looked longer) to a negative test event when they were habituated to a positive interaction compared to those infants who were habituated to a negative interaction. The difference in looking times between the two groups suggests that infants attributed a value (positive or negative) to the interaction and they
noticed when this value changed. Later studies also investigated and expanded our knowledge about moral evaluation in infancy. Besides looking time paradigms, the so called object choice paradigm is a commonly used method to examine infants’ moral evaluations (Hamlin, 2014). In the object choice paradigm, infants are presented with physical objects and given the opportunity to choose between them. Infants’ reaching or touching one of the objects is an indicator of their choice. Infants are thought to express a preference for the object of choice. Hamlin, Wynn & Bloom (2007) assessed infants’ capacity to evaluate others’ behavior using a looking time paradigm in combination with an object choice paradigm. In their first
experiment, infants were familiarized with animations showing an agent trying to reach the top of the hill, who was either helped or hindered by another agent. Infants looked longer in the test events in which the climbing agent approached the hindering character than when the
6 climber approached the helping character, indicating that this was surprising for them. When offered to choose for either the helping object or the hindering object, they more often chose the helping object. Hamlin, Wynn & Bloom (2007) thus concluded that infants preferred the helper character over the hinderer character. In another experiment, a neutral character was introduced in the animations who moved up and downhill without interacting with the other agent. Infants preferred the neutral character over the hindering character and preferred the helping character over the neutral character. These results again pointed in the same direction; infants not only make a positive moral evaluation about the helping character, but also make a negative moral evaluation about a hindering character. Using the same measures, Hamlin & Wynn (2011) showed that infants’ capacity to morally evaluate the behaviors of others could be generalized in a variety of third-party interactions. They found that infants preferred an agent that was helpful to someone in opening a box over an agent that was hindering someone to open a box. In a second experiment infants showed a preference for the agent that returned a lost ball to someone over an agent that took a ball away from somebody. Again, infants thus preferred the helping agents over the hindering agents. Geraci & Surian (2011) examined how infants evaluated agents distributing colored disks to receiving agents. One distributing agent equally distributed the disks to the receiving agents while the other distributing agent made unequal distributions of the disks. Looking times and object choices indicated that infants were capable of evaluating the two distributing agents and had a preference for the equal distributor. A study presenting infants with agents’ failed and successful attempts to help or harm added to the findings that it is not only the outcome that infants take into account in their moral evaluations (Hamlin, 2013). Infants also consider intentions in their evaluation and prefer an agent with good intentions over an agent with bad intentions regardless of the outcome (failed or successful). In sum, infants are able to attribute value to others’ action and when doing so, they are sensitive for agents’ intentions. Infants have a preference for
prosocial behavior and equal distributions, while having a dislike for antisocial behavior and unequal distributions. The studies described above confirm the capacity in infancy to make moral evaluations.
The origins of infants’ capacity to make moral evaluations is still debated, but most researchers agree with an evolutionary perspective that views the capacity as a process of biological adaptation (Govrin, 2014; Hamlin, 2013; Sloane, Baillargeon & Premack, 2012; Hamlin, Wynn & Bloom, 2007; Hamlin, Wynn, Bloom & Mahajan, 2011). From this perspective, moral evaluations are critical for survival and support group cooperation.
7 that infants have learned to evaluate and distinguish good from bad through socialization (Hamlin & Wynn, 2011; Sloane, Baillargeon & Premack, 2012). Thus, infants’ early
experiences are thought not to influence the basic capacity (Govrin, 2014). They conclude that infants’ capacity to make moral evaluations is universal, innate and unlearned.
1.3 Early experiences and attachment
In contrast, in many other dimensions of development the importance of infants’ early social emotional experiences is extensively studied and its influence has been repeatedly
demonstrated (Govrin, 2014; Goldberg, 2000). Especially the early caregiver-child
relationship is found to be of significance in a range of developmental aspects and throughout life (Goldberg, 2000). The attachment theory provides a theoretical framework that explains the importance of early caregiver-child relationships and its influence on development. Attachment is the psychological and specific bond between an infant and a caregiver that serves as a foundation from which an infant can explore and return to in stressful situations. From birth, infants can signal to their parent to elicit the parents’ proximity and care
(Goldberg, 2000; Benoit, 2004). Behaviors as crying and laughing are examples of such signals. Parents’ accurate responding to infants’ signaling impacts the quality of developing attachment relationship. There are three types of organized attachment patterns, one secure and two insecure types. Secure attachment (B) develops when a parent responds to an infant quickly in a consistent and sensitive way. The parent comforts the infant when necessary while also leaving room for exploration. As a result, infants with a secure attachment
approach the parent in a stressful situation, use them as a ‘safe haven’ but also feel secure to explore in a non-threatening situation (Prior & Glaser, 2006; Benoit, 2004). When a parent repeatedly rejects the infants’ distress by ignoring, becoming agitated or not noticing the infants’ needs, the infant develops an insecure-avoidant attachment (A). Infants with this type of attachment have learned that expressing distress is pointless, because the parent is not available for them. They therefore avoid the parent, overly show exploratory behavior and deal with distress themselves (Goldberg, 2000). A parent who acts unpredictable and inconsistent towards the infant is very likely to have an infant with an insecure-resistant attachment (C) (Benoit, 2004). Infants with this attachment-type express excessive negative emotion as a means to attract the attention of the unstable parent. They show little exploration and instead ‘cling’ on to their parent (Goldberg, 2000; Prior & Glaser, 2006). Research into attachment quality has found that secure attachment is a protective factor against social and emotional problems, but that it is also independently related to social competence and optimal
8 emotional and cognitive development (Benoit, 2004; Goldberg, 2000). Insecure attachment, either resistant or avoidant, is a risk factor for psychopathology and associated with
disadvantage in cognitive development (Allen, Hauser, Borman-Spurrell, 1996; Goldberg, 2000). At a physiological level, insecure attachment is related to dysregulation of the stress system. Attachment quality, developed via early experiences with the parent, thus is related to a broad range of developmental outcomes.
Even though researchers suggest that infants’ early experience are unlikely to influence the basic capacity of moral evaluation, its role has not been studied sufficiently (Govrin, 2014). Govrin (2014) proposed a hypothetical model in an attempt to integrate infants’ early experiences with the innate capacity to make moral evaluations. In this attachment based model, infants’ basic capacity to make moral evaluations is likely to be innate, but an environmental trigger is necessary in order to develop this capacity properly. Infants early experiences with the parent are thought to influence the way moral situations are interpreted and evaluated. Through early experiences and interaction with the attachment figure, infants develop internal working models that serve as a blueprint for representing relationships and interpreting social aspects of life (Govrin, 2014; Johnson, 2010; Thompson, 1999; Goldberg, 2000). Govrin (2014) proposed that infants use expectations and
representations stored in their internal working models to make moral evaluations about agents seen in third-party interactions.
Internal working models represent expectations based on infants’ actual experiences with the attachment figure (Bowlby, 1969; George, 1996). Infants who are securely attached thus are likely to have developed internal working models that represent sensitive and consistent caregivers that are responsive to their needs. On the other hand, infants who are insecurely attached, are likely to have developed internal working models that represent more inconsistent and less sensitive caregivers that are more often unresponsive to their needs. Expectations about caregivers’ behavior are formed based on infants’ representations of caregivers. Johnson and colleagues (2007; 2010) tested the presence of internal working models in infancy using the ‘’violation of expectation’’ method based on looking times. The studies examined infants’ expectations about caregiver-infant interaction in animations with two abstract agents. The animations involved a small agent and a big agent, representing an infant and a caregiver respectively. In the first experiment, the big agent climbs a hill while the small agent fails to climb this hill. As a response to the small agents failure, the big agent either acted responsive, by returning to the small agent, or unresponsive, by moving further away from the small agent. In a second experiment, the small agent was crying at the bottom
9 of the hill and as a response the big agent returned downhill (Johnson et al., 2010). The small agent then either approached the big agent or moved away from the big agent. Infants’ attachment classification was determined using the Strange Situation Procedure. The study found that infants’ expectations about the caregiver-infant interaction differed depending on their attachment classification. Infants who were securely attached expected the big agent to act responsive and the small agent to approach the responsive big agent. In contrast, infants who were insecurely attached expected the big agent to act unresponsive to the small agent. While insecure-resistant infants expected the small agent to approach the responsive agent, insecure-avoidant infants expected the small agent to move further away from the responsive agent. In sum, infants expected agents to respond in a way that fit their representation of caregiver-infant relationships. The studies confirm that infants interpret social situations based on their internal working models of attachment. Thus, Govrins’ proposal that infants moral evaluations may also be based on infants’ internal working models is a plausible suggestion that research needs to address.
1.4 The current study
The current study aims to investigate moral evaluation in infancy. We do this by examining infants’ preference for a responsive or an unresponsive agent, assuming that responsive and unresponsive behavior can be considered as prosocial and antisocial behavior respectively. We presented 12-month-old infants with animations similar to those used by Johnson and colleagues (2007; 2010). The animations involved a small agent and a big agent. The small agent separates from the big agent and signals the big agent with crying and laughing sounds. The big agent in half of the animations acts responsive to the small agent by returning to it. In the other half of the animations, the big agent acts unresponsive by moving further away. The two different responses of the big agent (responsive or unresponsive) are marked by
contrastive colors (red or blue) and counterbalanced across infants. We used a within-subject design. Infants all watched a total of eight animations of which four showed the responsive big agent and four showed the unresponsive big agent. Immediately after the animations were played, infants’ preference for the responsive big agent or the unresponsive big agent was assessed with an object choice paradigm. Infants were presented with two colored oval shaped objects on a carton board representing the responsive agent and the unresponsive agent from the animations. Infants were encouraged to make a choice for one of the two. Relying on earlier research we assumed that infants would choose the object they preferred and that their preference is an indicator of their moral evaluation (Geraci & Surian, 2011; Hamlin, 2014;
10 Bloom, 2010). The Strange Situation Procedure was used to assess infants’ type of
attachment.
Our research questions and hypotheses are two folded by investigating infants’ general capacity as well as individual differences in their preference. First, we examine whether infants in general show a preference for a responsive or an unresponsive big agent seen in the interactions. Previous studies into moral evaluation have found that in different situations, infants in general are capable of showing a preference for agents’ behavior in third-party interactions. They prefer helpers over hinderers and equal distributors over unequal distributors. Overall, they prefer agents who display prosocial behavior over agents who display antisocial behavior. The current study aims to test if infants also prefer agents who act responsive over agents who act unresponsive. This study therefore will add to the literature on the kinds of social behavior that infants show a preference for, namely responsive or
unresponsive behavior. Our first hypothesis is formulated as follows: Infants will show a preference for the agent that acts responsive over the agent that acts unresponsive. In the object choice paradigm, we expect that infants will overall more often choose the colored object that represents the responsive big agent than the colored object that represents the unresponsive big agent.
Second, we examine possible individual differences in infants’ preference for the responsive or unresponsive big agent seen in the interactions based on their attachment classification. To our knowledge, no studies have empirically examined the relation between attachment type and infants’ capacity to make moral evaluations. Previous research has however showed that type of attachment, via internal working models, influences a range of developmental outcomes as well as expectations about third-party interactions. Securely attached infants are likely to have developed internal working models that contain stronger expectations for responsive caregiving behavior, while insecurely attached infants are likely to have developed internal working models with somewhat weaker expectations for
responsive caregivers. Govrins’ proposal, although not empirically tested, also gives us plausible reasons to investigate a possible influence of type of attachment on infants’ capacity to make moral evaluations. Based on this, our second hypothesis is described as follows: Although in general infants will prefer the responsive agent over the unresponsive agent, type of attachment may account for the variation in preference. We expect that securely attached infants will prefer the responsive agent more often than insecurely attached infants. In the object choice paradigm, securely attached infants will choose the colored object that represents the responsive big agent more often than insecurely attached infants. We do not
11 have different expectations for the two types of insecure attachments (A or C), since they are likely to have similar representations about the responsiveness of caregivers.
2. Method 2.1 Participants
The participants of the current study are part of an ongoing longitudinal research project about attachment and infants’ social cognition. Our sample consists of 147 infants and their
mothers. The infants were 12 months old (M = 12,33 months, SD = 0,37 months ), 74 (50,3%) males and 73 (49,7%) females. The mothers were on average 33,29 years old (SD = 4,19 years) old. Mothers were asked to fill in an online questionnaire about family background information. It was missing for seven (4,8%) of the participants. More than 81% of the mothers reported to have completed an academic degree, 12% had finished secondary education and 2% had no or a primary educational degree. Most (135, 91,8%) infants lived with both biological parents. One infant (0,7%) lived with only one biological parent, one infant (0,7%) lived with a biological parent and a stepparent and two (1,4%) mothers
indicated that their infant had an altered family structure. Eighty-eight (59,9%) of the infants were an only child, forty-six (31,3%) had one sibling and six (4%) had two or more siblings. Mothers reported that twenty-five (17%) of the infants either had a chronic illness, a physical handicap, used medicine or had undergone surgery.
The mothers and their infants were recruited through the local municipality. Addresses were provided of all the families with a newborn infant, who were contacted via mail and follow-up phone calls. Mothers who were willing to participate with their infant were scheduled for an appointment at the University of Leiden. Mothers all signed
informed-consent forms before participation and were financially compensated for their travel expenses. All infants received a small gift and ‘baby diploma’ after participation. The ethics review board of the Department of Education and Child Studies at the University of Leiden approved this research project.
2.2 General procedure
Mothers and their infants came to the laboratory for either their first, second or third visit as part of the longitudinal Attachment and Social Cognition project. They first participated in an eye-tracking experiment which presented the infants with the animated interactions. This was immediately followed by the object choice experiment. Next, the Strange Situation Procedure was conducted to assess the quality of infants’ attachment. After a short break, temperament
12 measures for joy, anger and fear (Lab-TAB) were conducted. Maternal sensitivity was
observed during break, free play with and free play without toys. The visit ended with an eye-tracking measure of the mother. The current study uses data from the first three measures, which will be described in more detail below.
2.2.1 Eye-tracking stimuli
The infants were shown eight animations (34,5 cm x 25,5 cm) containing abstract social interactions with a small and a big generic agent, representing a child and an adult respectively (see Figure 1). The small generic agent was a 2 cm x 1,5 cm light blue oval shape. The big generic agent was a 3,5 cm x 2,5 cm oval shape and either red or blue, depending on the animation. The animations all started with the small and the big generic agent moving towards a hill. When they arrived at the start of the hill, the big generic agent moved uphill until it reached the horizontal area while the small generic agent failed to move uphill. As a result, the small generic agent and the big generic agent got separated from each other. The small generic agent in half of the animations started crying, whereas it in the other half of the animations started laughing. To indicate that the small generic agent was the source of the sound, it expanded and contracted 2 mm and slightly changed in color three times. In response to the crying or laughing, the big generic agent in half of the animations acted responsive by moving downhill and returning next to the small generic agent (see B in Figure 1). In the other half of the animations, the big generic agent acted unresponsive to the small generic agent by moving further away up the second hill to the horizontal area (see A in Figure 1). The color of the responsive big generic agent in all eight animations always
differed from the color of the unresponsive big generic agent and was counterbalanced across infants. In this way it was possible to associate the type of response (responsive versus unresponsive) with the color of the big generic agent. Infants were either assigned to the red responsive condition or the blue responsive condition. Furthermore, they were assigned to one of four order conditions. The animations started with either two crying or laughing movies which then altered being either responsive or unresponsive. Infants were therefore randomly assigned to one of eight (2 color x 4 order) eye-tracking conditions.
13 Figure 1. Schematic fragments of the animations. A: Start, uphill, separation, leaving
response (unresponsive). B: Start, uphill, separation, returning response (responsive). 2.2.2 Procedure of eye-tracking and object choice experiment
In the testing room, mothers were asked to take place in a small cabin with their infant on their lap. The infants were facing a Tobii T120x eye-tracker TFT monitor (17-inch) which would play the animations. The chair was adjusted in a way that the distance from the infants’ eyes to the monitor would be approximately 60 centimeters. A pair of blinded glasses were given to the mother, so that the eye-tracker would only record the infants’ eye movement and that the mothers would not influence the infant’ reaction. Mothers were asked to try to prevent their infant from moving too much by holding them. Following instruction, the curtain of the cabin was closed and the lights in the testing room were dimmed. After a calibration, the animations were played on the eye-tracker monitor. In between every animation, one of four short attention-getters was played in an effort to remain the infants’ attention. After the infants watched the animations the experimenter adjusted the camera position towards the infants’ body, opened the curtain and presented two wooden colored objects on a carton board (29,4 cm x 29,4 cm). One of the objects, a 10,9 cm x 6,7 cm blue oval, looked like the blue big generic agent and the other object, a 10,9 cm x 6,7 cm red oval, looked like the red big generic agent from the animations. The side of the ‘responsive’ object on the board was counterbalanced across infants, either presented on the left side of the board or on the right side of the board. The infants were encouraged to make a choice between the two objects by
Start Uphill Separation
Response: Responsive Response: Unresponsive
Start Uphill Separation
A
14 the experimenter first simply saying: ‘‘Look at these! Which one do you like?’’. If the infant did not reach for any of the objects for 5 seconds, the mother was instructed to ask the infant the same question. We made sure that no reference was made to the colors of the objects when the child was asked to pick one. The infants’ reaction to the objects presented on the board was videotaped for later coding and was also noted on a form by the experimenter. 2.2.3 Strange Situation Procedure
The standardized Strange Situation Procedure (SSP, Ainsworth et al., 1978) was used to assess infant-mother attachment security. In an unfamiliar room with some toys, the infants were confronted with a strange person and two separations from and reunions with the mother. The procedure is designed to activate attachment behavior, while also eliciting exploratory behavior (Goldberg, 2000). The SSP is divided in eight 3-minute episodes with increasing, but mild, stress for the infant. In the first episode the experimenter instructs the mother about the procedure. The mother is asked to respond to her infant normally when she is approached, but not to initiate activities. The second episode starts when the experimenter leaves the room. The mother reads a magazine, while the infant plays with the toys. In the third episode, a stranger enters the room and sits down with a magazine. After one minute, the stranger starts a conversation with the mother. One minute later, the stranger initiates to play with the infant. In the fourth episode, the mother leaves the room quietly for the first
separation. The stranger and the infant are left alone in the room. The mother returns to the room in the fifth episode for the first reunion. She greets and if necessary comforts her infant. When the infant is calm again, the mother sits down to read the magazine and the stranger leaves the room quietly. In the sixth episode, the mother leaves the room for the second separation, this time saying goodbye to her infant. The infant is now alone in the room. In the seventh episode the stranger enters the room and comforts the infant if needed. The stranger sits down and reads the magazine if comforting is not (or no longer) necessary. The mother returns to the room in the eighth episode and reunites with the infant for the second time. After the reunion, the mother sits down to read the magazine and the stranger leaves the room. If the infant is upset in the separation episodes (4, 6 & 7), the episodes are shortened by either sending the mother back (4, 7) or sending the stranger in (6) to comfort the infant. The entire procedure was videotaped for later coding.
2.3 Measures
The following paragraphs describe the measures used in this study in more detail. First, we explain what measure we used from the eye-tracking session as an inclusion criterion. Second,
15 the object choice measure will be described. Finally, attachment classification from Strange Situation Procedure will be discussed.
2.3.1 Eye-tracking measure
Infants’ total looking at the eight animations in the eye-tracking experiment was used as a rough and general measure of attentiveness. Using a Tobii fixation filter (velocity and distance threshold of 35), total duration of fixation for each animation was calculated and added together. Only infants with a duration of fixation of minimal 40 seconds were included in the analyses (N = 146). This arbitrary criteria was based on the rationale that infants’ choices have to be based on at least having seen two animations, involving both the responsive (20 seconds) as well as the unresponsive big agent (20 seconds). Seventeen
(11,6%) infants had a total duration of fixation lower than 40 seconds and were excluded from the analyses.
2.3.2 Choice measures
Infants could show four possible reactions in the objects choice experiment, namely reaching for the blue object, reaching for the red object, reaching for both object at the same time or not reaching for any of the objects. The videotapes were coded by categorizing the reaching reaction respectively into blue choice, red choice, both or none. Due to an experimenter or technical error, some videos were inconclusive about the infants’ choice. In that case (17,5%) the form was consulted. Object choice data was available for 127 infants. Forty (31,5%) infants chose the blue object, thirty-eight (29,9%) chose the red object, eleven (8,7%) chose both object and thirty-eight (29,9%) did not make a choice for an object. A considerable proportion of infants did not make a choice for one of the two objects. The study could only include infants in the analyses that made a clear choice for either the blue object or the red object. Therefore, we will compare infants who made a clear choice (N = 78) with infants who did not make a clear choice (N = 49) to examine if the two groups differ from each other on other variables.
This study is interested in infants’ preference for one of the two objects representing the responsive big agent or the unresponsive big agent. Infants who made a clear choice were therefore further analyzed for agent choice. It was investigated which agents’ response type (responsive versus unresponsive) corresponded with the color of the object of choice (red versus blue). Infants who chose for the colored object that represented the responsive agent were assigned to the ‘responsive’ choice group (N = 43) and infants who chose for the colored
16 object that represented the unresponsive agent were assigned to the ‘unresponsive’ choice group (N = 35). Agent choice data was available for 78 infants.
2.3.3 Attachment classification
Infants’ behavior in the Strange Situation Procedure was coded by two expert coders for proximity-seeking, contact-maintaining, avoidance of the mother, resistance to comforting, search behavior during separation, and distance interaction with the mother (Goldberg, 2000). Twenty cases were double coded, with an intercoder agreement for these cases of 75% (κ = .50). Attachment data was available for 98 infants.Fifteen (15,3%) infants were classified as insecure-avoidant (A), fifty (51,0%) were classified as secure (B) and thirty-three (33,7%) were classified as insecure-resistant (C). A variable for security category was also created in which all infants categorized as avoidant (A) and all infants categorized as resistant (C) were assigned to the insecurely attached group (N = 48). All infants categorized as secure (B) were assigned to the securely attached group (N = 50). Attachment and agent choice data was available for 60 infants.
2.4 Data analysis plan
All the variables in this study are categorical, therefore non-parametrical chi-square tests will be used to analyze if infants’ agent choice differs significantly from chance and if attachment security category is related to their agent choice. Possible interaction effects are also explored with chi-square comparisons. SPSS statistics 21 is used to analyze the data.
3. Results
3.1 Preliminary analyses
In the first paragraph of the preliminary analyses, included and excluded infants will be compared. In the second paragraph, main effects of background and condition variables on infants’ agent choice will be examined.
3.1.1 Comparing included and excluded infants
We compared, with chi-square tests, infants who made a clear choice for either the blue object or the red objects with those who did not make a clear choice (reached for both objects or for none of the objects). On the background variables gender of the infant, age of the infant, age of mother, educational level of mother, family structure or medical problems of the infant, the two groups did not differ (ps >.30). The two groups did not differ on condition variables either such as the eye-tracking condition, order of the animations and color of the responsive
17 big agent in de animations (ps > .40). Included infants did not differ from excluded infants on attachment classification (ABC) or on security category (secure versus insecure, ps > .40). In sum, infants who made a clear choice for one of the two objects did not differ significantly from infants who did not make a clear choice for one of the two objects. We can thus safely interpret the findings from infants who made a clear choice.
3.1.2 Testing the main effects of background and condition variables
The current studies’ main measure of interest is agent choice, infants’ choice for either the ‘responsive’ object or the ‘unresponsive’ object. Preliminary analyses, chi-square tests, were conducted on agent choice and possible influencing variables. The background variables such as gender of infant, age of infant, age of mother, educational level of mother, family structure or medical problems of the infant, were not significantly related to agent choice (ps > .17). The eye-tracking condition, as well as the order of the animations and the color of the responsive big agent in the animations, were also not significantly related to infants’ agent choice (ps > .35).
To be able to conclude that infants’ agent choices are based on higher cognitive processes rather than a preference for physical aspects of the objects such as color or side preference, the influence of these aspects were also tested. Using chi-square comparisons, it was analyzed if infants had a color preference (red choice versus blue choice) for one of the two objects. The analysis revealed that infants did not have a significant color preference (χ² (1, N = 78) = 0.05, p = .82). The side on which the object of choice was presented on the board (left side choice versus right side choice) was also tested for possible preference, but no significant effect was found (χ² (2, N = 78) = 2.55, p = 0.11). In sum, there were no main effects of any of these variables on infants’ agent choice. Possible interaction effects will however be explored with inclusion of these variables.
3.2 Hypothesis 1: Agent choice
Addressing our first hypothesis, it was examined if infants chose the object representing the responsive agent more often than the object representing the unresponsive agent. A chi-square test was used to analyze agent choice. The analysis revealed that there was no significant main effect of agent choice (χ² (1, N = 78) = 0.82, p = .37). This means that infants chose the ‘responsive’ object as often as the ‘unresponsive’ object.
The data was further analyzed with chi-square tests to detect possible interaction effects. Gender, color of the responsive big agent in the animations, and side of the
18 revealed a three-way interaction between the blue responsive condition, gender and agent choice (χ² (1, N = 39) = 3.81, p = .05). When infants were in the blue responsive condition, gender and agent choice were significantly related. No interaction was found when infants were in the red responsive condition (χ² (1, N = 39) = 0,01, p = .92). This gender effect in the blue responsive condition was further explored by analyzing it separately for male and female infants. A significant effect on agent choice was found only for female infants in the blue responsive condition (χ² (1, N = 16) = 4.000, p =.05). Female infants in the blue responsive condition, chose the ‘responsive’ object more often than the ‘unresponsive’ object (see Figure 2). Male infants in the blue responsive condition did not differ in their agent choice (χ² (1, N = 23) = 0.39, p = .53). These findings need to be interpreted with caution, since the analyses are conducted over a small part of the sample.
Figure 2. Interaction between gender and agent choice in blue responsive condition. Female infants in blue responsive condition significantly chose the ‘responsive’ object more often than the ‘unresponsive’ object (* p =.05).
Furthermore, a three-way interaction between side of the ‘responsive’ object on the board and agent choice was found when infants were in the red responsive condition (χ² (1, N = 19) = 4.26, p = 0.04). Infants in the red responsive condition more often chose the
‘responsive’ object when it was presented on the right side (see Figure 3). Infants’ agent choice did not differ when they were in the red responsive condition and the ‘responsive’ object was presented on the left side (χ² (1, N = 20) = 1.80, p = .18). For infants in the blue
19 responsive condition, side of the ‘responsive’ object on the board was not related to their agent choice (χ² (1, N = 39) = 0.30, p = 0.86). Again, because these analyses are conducted over a small part of the sample, the findings need to be interpreted with caution.
Figure 3. Interaction between agent choice and side of ‘responsive’ object on the board in red responsive condition. Infants in red responsive condition significantly chose the ‘responsive’ object more often than the ‘unresponsive’ object when it was presented to them on the right side (* p = .04).
3.3 Hypothesis 2: Agent choice & attachment
Addressing our second hypothesis, it was examined if infants’ agent choice differed
depending on their attachment classification and in specific if securely attached infants chose the object representing the responsive agent more often than insecurely attached infants did. Chi-square tests were used to analyze if infants’ agent choice was related to attachment classification. First, a chi-square test was carried out with the three attachment categories (secure, avoidant, resistant) and agent choice. The test revealed that there was no difference in infants’ agent choice depending on their attachment classification (χ² (2, N = 60) = 1.67, p = .43). Furthermore, it was examined if the findings changed when agent choice was related to security category (securely attached versus insecurely attached). Security category did not change the findings, the associating between security category and infants’ agent choice was still non-significant (χ² (1, N = 60) = 1.07, p = .30). In sum, infants’ agent choice did not differ depending on their attachment classification. Securely (B) attached infants did not
20 chose the object representing the responsive agent more often than insecurely, either insecure-avoidant (A) or insecure-resistant (C), attached infants.
Interaction effects were only explored with attachment security category (secure versus insecure), since possible effects will only be found, and interpretable, if the sample is not reduced too much. Again, we included gender, color of the responsive big agent in the animations, and side of the ‘responsive’ object on the board in the analyses for security category and agent choice. However, no interaction between security category and agent choice reached significance and will therefore not be interpreted or explored further (ps > .06).
4. Discussion
4.1 Summary of results
The current study aimed to investigate moral evaluation in infancy. Using the object choice paradigm, we examined infants’ preference for one of two objects representing a responsive and unresponsive agent whose behaviors have been previously observed by the infants during animated interactions. We expected infants to prefer the responsive agent over the
unresponsive agent. Our results however did not support this hypothesis. Infants did not chose the object representing the responsive agent more often than the object representing the unresponsive agent. We also hypothesized that attachment classification could account for variation in infants’ preference. In specific, we expected that securely attached infants would chose the object representing the responsive agent more often than insecurely attached infants. This hypothesis was also not supported by our findings.
Gender, the color of the responsive agent in the animation and the side of the ‘responsive’ object on the board however proved to be important variables in infants’
preference for one of the two objects. An interaction effects revealed that the expected pattern for our first hypothesis was found when gender, color of the responsive agent in the
animations and side of the ‘responsive’ object on the board were taken into account. Female infants who had seen a blue responsive agent in the animated interactions more often chose the ‘responsive’ object than the ‘unresponsive’ object in the object choice experiment. Also, infants in the red responsive condition more often chose the ‘responsive’ object than the ‘unresponsive’ object when it was presented to them on the right side of the board. The interaction effects partly confirm our first hypothesis, in which we expected that infants
21 would prefer the object that represented the responsive agent more often than the object that represented the unresponsive agent.
4.2 General discussion
We will now discuss the findings of our study in a broader perspective. We begin with the interaction effects we found and place them in light of earlier and future research. Next, we discuss issues relating to the animations we used and suggest how these issues could explain the lack of findings for preference. In view of earlier commentaries about the use of looking- and reaching paradigms, we then carefully review some of the assumptions in the current study and consider if perhaps some of them are not satisfied.
4.2.1 Interaction effects
Due to the reduced sample size and thus limited power, our interaction effects are exploratory and our interpretation is speculative. We do however address these effects and place them in light of earlier and future research. Under certain conditions the expected pattern, the
preference for the ‘responsive’ over the ‘unresponsive’ object, was found. Right side presentation of the ‘responsive’ object on the board, female gender and color of the responsive agent in the animations were related to the choice for the ‘responsive’ object. Female infants in the blue responsive condition had a preference for the ‘responsive’ object over the ‘unresponsive’ object. Research into gender differences reveals that female infants have some advantages over male infants. First, gender differences are reported in social development (Bayliss, Pellegrino & Tipper, 2005). In contrast to males, female infants are more sensitive to social stimuli and prefer objects with human characteristics over objects with mechanical characteristics (Alexander & Wilcox, 2012). Some studies thus conclude that female infants are better in processing social information than male infants. The stimuli in the current study involves social agents with behavioral characteristics and investigates a
preference for ‘social’ objects. Female infants may therefore have benefited more than male infants. It could thus be argued that our finding that female infants do show a preference for the ‘responsive’ objects while male infants do not, is in line with earlier research. A second explanation for our gender effect could be color blindness. Although we did not include a measure for colorblindness, a convincing amount of literature reports that males (8%) are much more likely to be colorblind than females (0,4%, Birch, 2012). People with color blindness are more likely to confuse colors. Our study relies greatly on the capacity to make a distinction between colors. Infants’ need to remember which color (red versus blue),
22 two objects cannot be interpreted as a preference for the responsive or unresponsive agent. Since males in general are much more likely to be color blind, our male participant are also much more likely to be colorblind. Our finding that female infants more often choose the ‘responsive’ object than male infants, could perhaps be explained by a proportion of colorblind male infants. It is however unclear why this gender effect was only found in the blue responsive condition. Perhaps we did not have enough power to detect an overall clear gender difference. Or, female infants more often have a preference for blue over red, and only when the color preference match their preference for the responsive agent, they more often choose this object. Franklin and colleagues (2010), Bornstein (1975) and Hurlbert & Ling (2007) however all report that no sex differences are found for color preference in infancy. But at a later age, female infants are found to better remember and name colors than boys (Bimler, Kirkland & Jameson, 2004).
Research into color preference in infancy is inconclusive. Little studies have directly compared red versus blue. Some studies show that both blue and red are preferred colors (Franklin, Bevis, Ling & Hurlbert, 2010). There are however also some indications that reddish colors are preferred over greenish colors, with blueish intermediate (Adams, 1987; Franklin, Bevis, Ling & Hurlbert, 2010). Both male and female infants between the ages 3-5 months seem to like red most and green least. Infants aged 6-24 months, when allowed to reach for colored objects, preferred red most and green least with blue and yellow being intermediate. So if there is a difference in color preference for red or blue, red seems to be more preferred. In our study, red responsive condition proved to be a factor in infants’ choice for the ‘responsive’ object.
The interaction effect also revealed that right side presentation of the ‘responsive’ object on the board was related to their choice for the ‘responsive’ object. Although
handedness is variable until 3 years of age, studies have demonstrated that infants are three times more likely to be right-handed than left-handed (Michel, Tyler, Ferre & Sheu, 2006). They thus prefer to reach for an object with their right hand. Our interaction effects can be explained with this knowledge. Infants in our study are more likely to be right-handed than left-handed. So when they reach for an object, it is easier for them to reach for the object on that side. Thus, when their preferred object is presented on the side that is most easy for them to reach for, it is likelier that they choose this object than when their preferred object is presented on the side that is more difficult for them to reach for. In our interaction effect, infants choose the ‘responsive’ object more often when it was presented to them on the right side. This is, by assuming that most of the infants are right-handed, consistent with the
23 literature on handedness. The interaction was however only found in the red responsive condition. This is in line with studies reporting that red is more often preferred than blue (Adams, 1987; Franklin, Bevis, Ling & Hurlbert, 2010).
4.2.2 Animated interactions
One of the possible explanations for our non-significant results could be that the animations did not serve the most ideal basis upon which infants could form a preference for one agent over the other. It perhaps sent out mixed signals that infants found difficult to evaluate. The animations all started with a separation between the small agent and the big agent. Following this, the small agents in half of the movies starts crying while it starts laughing in the other half of the movies. Both crying and laughing are examples of attachment related behaviors that encourage the proximity of the caregiver, so we expected that both sounds would contribute to infants’ preference for one of the two responses of the big agent (Goldberg, 2000). It could however also be argued that while crying is a clear sign of distress, laughing could also be interpreted as playful and thus not encourage the caregivers’ proximity per se (Biro et al., 2014). Attachment behavior is especially triggered in stressful situations
(Goldberg, 2000). Therefore, when a laughing sound was heard, attachment behavior was not activated and as a result, infants did not form a specific expectation or preference for the big agents’ response. Crying at the other hand, is a clear example of distress and infants are expected to prefer the agent who was responsive to this crying by returning to the infant. The inclusion of the laughing sound thus, may have reduced infants’ preference for the responsive agent. The same analogy may explain our non-significant findings for our second hypothesis about a potential role for attachment. If infants’ attachment related behavior was indeed not activated due to the mixed emotional sounds, the fact that in our study securely attached infants did not differ in their preference from insecurely attached infants, is perhaps not surprising.
This suggestion is supported by Biro and colleagues (2014) who investigated the effect of emotional cues on infants’ monitoring of third-party interactions. Their study was part of the same research project about attachment and social cognition as the current study and used exactly the same animated interactions. With an eye-movement measure, they found that emotional cues accompanying a third-party interaction influenced infants’ representation of that interaction. When following separation a crying sound was heard, infants looked longer at the big agent than when a laughing sound was heard. The allocation and duration of their attention on the big agent could indicate that infants have a stronger expectations when a
24 crying sound was heard, because it is a more obvious sign of distress to which the caregiver needs to respond. It was also found that in the animations where the big agent acted
unresponsive to the small agent, infants attention differed depending on the sound that was heard. Infants gradually lost their attention for the animations when a laughing sound was heard. When however a crying sound accompanied the big agents unresponsiveness, infants attention did not change indicating that infants find this response unexpected. This finding is supported by the study of Jin et al (2012, as cited in Baillargeon, Setoh, Sloane, Jin & Bian, 2014), who also found that only when a crying sound was heard, infants expected an adult to respond and comfort an infant. Infants did not form specific expectations about adult behavior when a laughing sound was heard.
The current study differs from the studies by Jin et al (2012) and Biro et al (2014) because the current study investigated infants’ preference instead of infants’ expectations. Earlier studies into moral evaluations in infancy have however repeatedly demonstrated that infants’ preferences are in line with their expectations (Hamlin, Wynn & Bloom, 2007; Hamlin & Wynn, 2011; Geraci & Surian, 2011; Hamlin, 2013; Hamlin, 2014). Infants usually prefer the scenario that fits their representation, and thus expectation, of that scenario. The fact that we did not find a significant preference for the responsive agent over the
unresponsive agent or a difference between securely and insecurely attached infants, can thus possibly be explained by the inclusion of two contradicting emotional sounds. Future research should investigate if infants’ preference for agents’ behavior in third-party interactions indeed differs depending on the emotional sound (laughing or crying) that is heard.
4.2.3 Assumptions
Since infants do not have language to express themselves, methods tapping into social cognitive skills in infancy are always indirect. Usually looking time measures or preferred reaching tasks are used. The use of these methods are however not without controversy. A group of researchers has questioned the use of indirect methods such as looking measures or preferential reaching tasks for tapping into infants’ social cognitive skills (Tafreshi,
Thompson & Racine, 2014; Cohen, 2004; Haith, 1998). They argue that the inference from infants’ looking and reaching behavior to higher order cognitive skills, such as socio-moral reasoning, is to far stretched. In a commentary by Tafreshi, Thompson & Racine (2014), the conclusion that infants’ looking and reaching indicates socio-moral reasoning or evaluation is criticized. They argue that in the reasoning from looking or reaching behavior to higher order cognition, too many and perhaps invalid assumptions are used. Others have disagreed with
25 Tafreshi, Thompson & Racine (2014) and with valid arguments defended the use of these methods in infancy (for a commentary see Hamlin, 2014). However, the suggestion that some of the assumptions in studies using these methods are possibly not satisfied is worth
considering. In the current study we used multiple assumptions. Although many of them have been well established, others are arguable and could perhaps explain our non-significant findings. These assumptions will be discussed below.
Previous studies into moral evaluation in infancy have showed that infants prefer prosocial over antisocial behavior. We assumed that prosocial behavior could be compared to responsive behavior and antisocial behavior to unresponsive behavior and we thus
hypothesized that infants would prefer responsive over unresponsive behavior. An overall significant preference was however not found. It could be that the context we used with responsive and unresponsive behavior cannot be compared to prosocial and antisocial behavior. One could argue that the behaviors in our animated interactions were not really ‘moral’ and perhaps therefore infants did not form a specific preference. Most people would agree that prosocial and antisocial behavior are clear examples of moral behaviors, where one behavior is judged as good while the other is judged as bad. Responsive and unresponsive behavior on the other hand are less clear examples of ‘good’ and ‘bad’ behavior. Most people prefer responsive behavior, but unresponsive behavior is not necessarily ‘bad’. Perhaps this is also why our study did not find a clear preference. Maybe, infants interpreted the animations of showing two possible behaviors of caregivers. They did not judge one as ‘good’ or ‘better’ and thus as ‘preferred’. The same may apply for our findings about attachment security. While securely attached infants may have experienced more responsive behavior than insecurely attached infants, surely all infants have also experienced unresponsive behavior since caregivers cannot always respond to their infant when he or she desires (Richter, 2004). Perhaps therefore infants, either securely attached or insecurely attached, did not form a specific preference for one behavior over the other. The suggestion that our animated
interactions are not clear examples of showing ‘moral’ behavior is thus a possibility that could explain why we did not find a preference while previous studies did.
Another assumption we made is that infants could make a connection between the two dimensional agents they see interacting during the animations and the objects
presented to them on the board after the animations were played. This assumption was based on previous research that also used object choice measures that allowed infants to reach for objects after they have monitored agents interact (Hamlin, Wynn & Bloom, 2007; Hamlin & Wynn, 2011; Geraci & Surian, 2011). The lack of findings for preference however could
26 indicate otherwise. There are some differences between previous research and our study, perhaps accounting for our findings. In the studies by Hamlin and colleagues (2007, 2011) and Geraci & Surian (2011) infants were presented with puppets or puppet-like (shapes with eyes) agents in interactions and immediately hereafter presented with exactly the same agents as they had just seen. The current study used more abstract agents without eyes, since infants have been found to be capable to interpret abstract characters with no eyes (Shimuzu & Johnson, 2004). We presented infants with objects in the same color and form as in the animations, yet they were not the same size or material. The objects resembled the agents in the animations, but they were not exactly the same. Perhaps the transition from the 2D abstract agents in the animations to the 3D objects was too big in our study. Another possibility, also relating to this inference issue, is that in our study the delay between the animations and the presentation of the board was too long for the infants. Although the exact duration before presentation is unknown, it is likely that the delay usually was about 10 seconds, since the experimenter had to adjust the camera position and open the curtain before presenting the board. There are studies indicating that a delay of about 10 seconds can lead to error in a task were 12-month-old infants need to reach for a correct (versus incorrect) object (Diamond, 1998). In the classic A-not-B task infants need to search a hidden object. A longer delay between monitoring the object being hidden and allowed to search for it, increased the number of errors infants made. Infants’ limited memory is in part thought to account for the mistakes in this task (Bjorklund, 2000). Although we used a different task, infants in our study also needed to use their memory. They had to remember which agent they preferred most in the animations and after a short delay choose the object that resembled that agent on the board. It could thus be that infants, by the time the board was presented to them, had forgotten which agent they preferred. This could explain why a preference for the responsive over the unresponsive agent was not found. It should however be noted that it is unknown whether infants form a preference while monitoring the interactions or if a preference is only formed when presented with the objects. Nevertheless, future studies should strive for a minimal delay and include a measure to control for the delay between the animations and the object choice measure.
In sum, it could be that one or perhaps a combination of assumptions that were (partly) not satisfied explain our non-significant findings. The commentary by Tafreshi, Thompson & Racine (2010) criticizing the use of methods investigating infants’ socio-moral reasoning may be too strong. It is however always important in infant research to explore possible invalid premises.
27 4.3 Limitations & recommendations
Some of the limitations of our study have already been mentioned above. There are however more limitations that need to be addressed. Additionally, we will propose recommendations for future research. First, the current study used infants’ overall looking at the animations as an indicator of their attentiveness. Since each animation was 20 seconds, all infants who had looked for less than 40 seconds were excluded. However, it is still possible that some included infants have not seen the minimal of one complete animation with the responsive agent and one complete animation with the unresponsive agent. The measure of attentiveness is a sum variable of infants’ total duration of fixation of all the animations. It is thus possible that infants’ total looking time is composed of repeated short looking, without watching two animations completely. Although repeated short looking by a big group of infants would be unlikely, it could be that a possible effect was reduced by infants without sufficient looking and ‘randomly’ preferring one object over the other. Future research should use more
sensitive measures that calculates infants’ duration of fixation per animation. Second, a great proportion of infants were excluded because they did not make a clear choice for one of the two objects. Although we found no differences in other (background or condition) variables between those who made a clear choice versus those who did not, the loss of almost half of our participants indicates that it was often very difficult for infants to make a choice. Future research should make the circumstances under which infants make a choice more optimal, so that not so many infants will be excluded. As a consequence, our third limitation is the small sample size. It did not allow us to look closer at differences within groups of insecurely attached infants (insecure-ambivalent versus insecure-resistant). More participants would also have made our interaction effects less speculative. Fourth and finally, our sample is not representative for the entire population of infants. All participants of our study were living in or nearby Leiden and the mean educational level of the mothers was highly above average. Generalization of our findings is thus arguable and future studies should use a more
representative sample. 4.4 Conclusion
We finally return to the debate we introduced earlier, whether the capacity to make moral evaluations in infancy is innate, learnt through socialization or both. The current study found no evidence for infants’ general capacity to make moral evaluations about others who behave responsive or unresponsive, nor for a role for attachment. Since we did not find evidence that infants can make moral evaluations at all, we can obviously not provide any clearance about
28 the origins of this capacity. We however do not want to claim in any way that infants do not have the capacity to make moral evaluations, but in our study we did not find evidence that infants can make moral evaluations about responsive and unresponsive behavior. However, as mentioned above, there are multiple (methodological as well as theoretical) explanations that could account for our findings. Future research should aim to clarify the origins of infants’ capacity to make moral evaluations. As suggested by most researchers, there are good reasons to believe that the capacity is innate. However, the role of the environment on this capacity is under investigated and should be explored in future studies.
29 Literature
Adams, R. J. (1987). An evaluation of color preference in early infancy. Infant Behavior and Development, 10(2), 143-150.
Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (2014). Patterns of attachment: A psychological study of the strange situation. Psychology Press.
Alexander, G. M., & Wilcox, T. (2012). Sex differences in early infancy. Child Development Perspectives, 6(4), 400-406.
Allen, J. P., Hauser, S. T., & Borman-Spurrell, E. (1996). Attachment theory as a framework for understanding sequelae of severe adolescent psychopathology: an 11-year follow-up study. Journal of consulting and clinical psychology, 64(2), 254.
Baillargeon, R., Setoh, P., Sloane, S., Jin, K., & Bian, L. (2014). Infant social cognition: Psychological and sociomoral reasoning. In M. S. Gazzaniga & G. R. Mangun (Eds.-in-chief), The cognitive neurosciences (5 th ed., pp. 7-14). Cambridge, MA: MIT Press.
Bayliss, A. P., Pellegrino, G. D., & Tipper, S. P. (2005). Sex differences in eye gaze and symbolic cueing of attention. The Quarterly Journal of Experimental
Psychology, 58(4), 631-650.
Benoit, D. (2004). Infant-parent attachment: Definition, types, antecedents, measurement and outcome. Paediatrics & child health, 9(8), 541.
Bimler, D. L., Kirkland, J., & Jameson, K. A. (2004). Quantifying variations in personal color spaces: Are there sex differences in color vision?. Color Research &
Application, 29(2), 128-134.
Birch, J. (2012). Worldwide prevalence of red-green color deficiency. JOSA A,29(3), 313-320.
Biro, S., Alink, L. R., van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (2014). Infants’ monitoring of social interactions: The effect of emotional cues.
Emotion, 14(2), 263-271.
Bjorklund, D. F. (2000). Children's thinking: Developmental function and individual differences. Wadsworth/Thomson Learning.
30 Bloom, P. (2009). Descartes' baby: How the science of child development explains what
makes us human. Basic Books.
Bloom, P. (2010, May 5). The moral life of babies. The new york times. Retrieved
from http://www.nytimes.com/2010/05/09/magazine/09babies-t.html?pagewanted=all. Bornstein, M. H. (1975). Qualities of color vision in infancy. Journal of Experimental Child
Psychology, 19(3), 401-419.
Bowlby, J. (1969). Attachment and loss: Vol. 1. Attachment. New York: Basic Books Cohen, L. B. (2004). Uses and misuses of habituation and related preference
paradigms. Infant and Child Development, 13(4), 349-352.
Csibra, G., Bıró, S., Koós, O., & Gergely, G. (2003). One-year-old infants use teleological representations of actions productively. Cognitive Science, 27(1), 111-133.
Csibra, G., & Gergely, G. (2007). ‘Obsessed with goals’: Functions and mechanisms of teleological interpretation of actions in humans. Acta Psychologica, 124(1), 60-78. Diamond, A. (1998). Understanding the A-not-B error: Working memory vs. reinforced
response, or active trace vs. latent trace. Developmental Science, 1 (2), 185-189. Franklin, A., Bevis, L., Ling, Y., & Hurlbert, A. (2010). Biological components of colour
preference in infancy. Developmental Science, 13(2), 346-354.
George, C. (1996). A representational perspective of child abuse and prevention: Internal working models of attachment and caregiving. Child Abuse & Neglect, 20(5), 411-424.
Geraci, A., & Surian, L. (2011). The developmental roots of fairness: Infants’ reactions to equal and unequal distributions of resources. Developmental science, 14(5), 1012-1020.
Gergely, G., Nádasdy, Z., Csibra, G., & Biro, S. (1995). Taking the intentional stance at 12 months of age. Cognition, 56(2), 165-193.
