Methodological Contributions: The Novel Paradigm

Get Complete Project Material File(s) Now! »

Chapter 3 Majority influence in infancy

This chapter is being revised to be resubmitted for publication


The findings of Chapter 2 leave a few open questions regarding preferences and the task context. When infants were given prior experience to the novel toys, an ingrained preference for the orange toy likely contributed to the findings in Experiment 1. While the findings are in line with other work on infants’ expectations of subjective preferences, of interest to the present research, is whether infants are sensitive to a majority’s preference. Assuming that majority preferences underlie day-to-day behaviour across different contexts, do infants attend to behaviours of the majority? Experiment 2 attempted to answer this question by omitting exposure to the novel toys. The findings indicated a marginal attentional bias for the majority toy; however, a looking time preference for the orange toy during test events did not rule out the possibility of a subjective preference for the orange toy. In light of the promising trends, it is possible that majorities guide attention in other contexts. Further, it is also likely that infants required additional cues to determine majority membership. While participants have full knowledge of the overall distribution of the majority in the paradigms used in previous work, the cues used to highlight the majority in Chapter 2 may have not been salient. The experiments in Chapter 3 address the concerns arising from the previous chapter; that is, whether contextual information and additional cues influence 19-month-old infants’ visual attention.
Imagine walking down the main street of an unfamiliar city while looking for a place to have dinner. You would prefer to eat at a Fusion restaurant and come across two suitable establishments. How do you choose between them? You could select one at random; or, check out an online/Internet-based resource to compare reviews provided by previous customers. Alternatively, you might just peek inside the window and choose the one with the most patrons. A relatively fail-safe strategy would be to rely on the wisdom of majorities to aid decision making. Indeed, we rely on majority information to guide our learning, behaviour, and decisions across a wide variety of contexts (Cialdini & Goldstein, 2004). In the present set of experiments, we extend the current literature on majority influence by examining the impact of majority information on infants’ behaviour.
There are certain advantages of a majority bias, especially while learning about others and the social environment. Behaviour of the majority is believed to have played a critical role in shaping the value and relevance that human beings place on information that is transmitted and shared (Boyd & Richerson, 2009). Thus, majorities influence people’s thoughts and behaviours across generations and over many centuries, promoting transmission of sociocultural knowledge. For instance, daily greetings differ across cultures based on what the majority of a specific culture adopts; some adopt a handshake, whilst others prefer bowing. Further, Henrich and McElreath (2003) posit that individuals defer to groups as an optimal strategy or heuristic to aid learning; access to majority information provides a consistent, valid, and reliable source of shared knowledge and practices within one’s environment.
Indeed, majorities have been shown to influence learning and behaviour across various species and domains of study. Several comparative studies demonstrate majority-biased learning in non-human primates and other animals (e.g., Battesti, Moreno, & Joly, 2015; Chou & Richerson, 1992; Haun, Rekers, & Tomasello, 2012; Pike & Laland, 2010). For instance, chimpanzees prefer to learn strategies to retrieve food from conspecifics shown to belong to a group (three in a group) rather than from one species member who used an alternative successful strategy. In humans, psychologists have long studied how majorities influence day-to-day behaviours and actions in adults (e.g., Sweeny & Whitney, 2014; for reviews, Cialdini & Goldstein, 2004; Levine & Tindale, 2015). For example, in a classic study, Asch (1956) demonstrated that adults disregard their own perceptions of a situation to conform to the opinions of a unanimous majority.
Haun, van Leeuwen, and Edelson (2013) categorized previous literature as demonstrating two ways in which majorities influence behaviour. The first is conformity, which relates to situations in which people forgo their own preferences, opinions, or behaviours for the group’s predominant strategy. Previous research has demonstrated that children conform to information provided by the majority across different tasks and contexts (e.g., Bishop & Beckman, 1971; Costanzo & Shaw, 1966; Iscoe, Williams, & Harvey, 1963; Walker & Andrade,). In their study, Walker and Andrade (1996) recreated the Asch paradigm to examine the developmental progression of influence. Their results indicated that while younger children were most likely to conform to majority opinions, conformity decreased through development. More recently, Haun and Tomasello (2011) demonstrated that the drive to conform is so powerful that preschool-aged children, like adults (e.g., Asch, 1956), relinquish their own perceptions of a situation to adhere to obviously incorrect perceptions of the majority.
The second categorization posited is majority-biased information transmission or the likelihood of naïve observers adopting behaviours displayed by a majority. In uncertain and novel situations, adults, and even toddlers, are more likely to be biased by information provided by a majority (e.g., Corriveau, Fusaro, & Harris, 2009; Corriveau & Harris, 2010; Corriveau, Kim, Song, & Harris, 2013; Haun et al., 2012; Sherif, 1937). For example, Corriveau and colleagues (2009) showed that pre-schoolers selectively endorse the opinions of a majority of adults in an unfamiliar object-labelling task, suggesting that majority bias is in place by three years of age. Further, 2-year-olds have been shown to adopt a majority strategy compared to another equally plausible strategy (Haun et al., 2012). Thus, very young children selectively attend, learn, and are biased towards information provided by a majority across different contexts.
Majority influences on our behaviour in adulthood and early childhood thus raises critical questions about the age at which individuals ‘learn to attend’ to and take heed of information provided by a majority. Indeed, evidence suggesting infants’ sensitivity to characteristics of social groups provides some reason to suspect that selective attention to groups is in place before their first birthday. For instance, 3-month-old infants selectively attend to members from their predominant racial group (Kelly et al., 2005). Kinzler, Dupoux, and Spelke (2007) demonstrated that 6-month-old infants look longer at people who had previously been shown to speak the same language and majority accent that they hear regularly. Further, 14-month-old infants selectively imitate an individual from their own linguistic community (Buttleman, Zmyj, Daum, & Carpenter, 2013; but see Howard, Henderson, Carrazza, & Woodward, 2014). The results of this previous work suggest that by their first birthday, infants are receptive to information based on familiar, socially constructed categories such as race and language.
As reviewed above, behavioural work with children indicates that pre-schoolers are biased by a majority for different reasons– informational-biased learning and conformity (Corriveau et al., 2009; Corriveau & Harris, 2010; Corriveau et al., 2013; Haun et al., 2012,2014; Haun & Tomasello, 2011). However, very little is known about the age at which these biases emerge. Do selective biases towards majorities exist in infancy? That is, when majority and individual information collide, which informational source would infants prioritize? Prioritizing majority information can be crucial when learning about people in one’s social environment and putting social information into perspective. From a naïve perspective, considering the behaviour of the majority infants may aid infants learn about sociocultural practices (Haun & Over, 2015).
The current studies address this gap in the literature by investigating whether majorities influence visual attention in infancy. Given the importance of visual attention in directing our focus and selectively responding to various aspects of the environment across varied learning contexts (Colombo, 2001; James, 1890), we opted to examine a majority informational bias at the foundational level – visual attention. In an environment, rife with ‘competing’ information and stimuli, previous research indicates that infants prioritize ostensive referential signals and other social cues over information such as object salience (Csibra, Biro, Koos, & Gergely, 2003; Gredebäck, Fikke, & Melinder, 2010; Hollich et al., 2000; Woodward, Sommerville, Gerson, Henderson, & Buresh, 2009; Wu & Kirkham, 2010; Wu, Tummeltshammer, Gliga, & Kirkham, 2014). These ostensive and social signals enable information processing in a manner that helps infants prioritize certain stimuli while ignoring others, which aids infants’ learning about objects and/or people in their social environment. However, it is unclear whether infants would prioritize these cues provided by a majority over those provided by an individual. This research answers this question by asking whether infants would show preferential attention towards a stimulus that had previously been examined by a majority over a stimulus that had only been examined by one individual.
We developed a novel eye-tracking paradigm to examine: 1) whether majority information influences infants’ visual attention in an ambiguous context, as literature indicates that, for both adults and young children, ambiguity increases the likelihood of attending to information provided by a majority (Bond & Smith, 1996; Corriveau et al., 2009; Haun, van Leeuwen, & Edelson, 2013; Sherif, 1937), 2) examine the developmental trajectory in the first two years of life by testing 12- and 19-month-old infants, and 3) identify possible factors that guide infants’ majority bias. 12- (Experiment 1) and 19-month-old (Experiment 2) infants were familiarized to a group of three actors (denoted by the members wearing the same colour t-shirt) completing a shared activity (i.e., stacking a tower of cups) in contrast with a fourth actor wearing a different coloured t-shirt and completing an action on her own (i.e., playing with a small car). Subsequently, infants were shown two video clips in which an actor looked in two boxes that had been placed on a table and selected one of the boxes. In one clip, one of the members from the majority always selected one of the boxes (e.g., green box) and in the other, the individual who was previously shown not to belong to the group, selected the other box (e.g., purple box). For the test trial, infants were shown a still image of the two boxes. The main question of interest was whether infants would be more likely to attend to the box that was previously favoured by the actors who had been shown to belong to a majority.

Experiment 1

Experiment 1 examined whether 12-month-old infants’ attention towards two similar boxes would be biased towards majority information. We selected this group as a starting point given previous evidence suggesting that infants as young as 6 months old have expectations of group members and show preferential looking towards members of their own linguistic ingroup (Buttleman et al., 2013; Kelly et al., 2005; Kinzler et al., 2007; Powell & Spelke, 2013). Further, by 12 months, infants respond to referential communicative gestures, and attribute and evaluate intentional behaviour of agents (Csibra & Gergely, 2009; Kuhlmeier, Wynn, & Bloom, 2003; Woodward, 2003).


Participants. Seventeen 12-month-olds (8 females; Mage = 12 months, 9 days; age range 11 months, 1 day to 13 months, 12 days) were recruited from a database of families who were interested in participating in developmental research in a large Australasian city. Parents identified their infant as belonging to one of the following ethnic groups: European (n = 13), Asian (n = 2), and Other/Mixed (n= 1). Data for one additional infant was removed from the final sample due to missing gaze data for the entire experiment. All infants received a toy; parents were given free parking during the visit and a $10 voucher for their participation.
Stimuli. Stimuli were presented on the Tobii T120 eye-tracker at a sampling rate of 120 Hz (Tobii Studio Version 3.2.2, Sweden). Stimuli included: two still images and nine video clips. One of the still images was a photo of the four actors that were shown throughout the experiment. Three people who we intended infants to perceive as belonging to a majority wore same coloured (e.g., blue) t-shirts and the fourth person, who we intended infants to perceive as not belonging to the majority wore a different coloured t-shirt (e.g., red). The second still image served as the test trial and depicted two similar, but different-coloured (green and purple) opaque boxes from the choice events. The video stimuli consisted of three majority familiarization events and two types of choice events.
Procedure. All infants participated in the following procedure. After completing the consent procedures, parents and infants were escorted to the eye-tracking room. Infants sat on their parent’s lap or in a highchair approximately 64 cm away from the Tobii T120 eye-tracker. After completing a five-point infant calibration procedure to calibrate the eye-tracker, the experimental procedure began.
Actor familiarization. To familiarize infants to the people that would appear in the ensuing videos, all infants were shown a picture of the four actors (Figure 1) for eight seconds.
Majority familiarization. To further introduce infants to the actors and more importantly, to emphasize the majority, infants were shown a video clip in which the three female actors who were wearing the same coloured shirt said “hi”, one of the group members said, “Look what we’re doing,” and then the members took turns building a tower with stacking cups. While completing the activity, the majority members looked and smiled at one another. After this video, infants were shown another video in which the female actor (i.e., the individual) who was wearing a different coloured shirt said, “Hi. Look what I’m doing” and played by herself with a toy car. During the clip, the actor had a pleasant demeanour. Infants were shown a third clip in which all four of the actors engaged in their respective activities as in the previous video (i.e., the majority built a tower together and the individual played with a car by herself), but were shown in the same frame. All three of the familiarization clips were of similar duration and the order of the first two familiarization clips (i.e., majority alone/individual alone) was counterbalanced across participants.
Choice event. Infants were shown a series of choice event video clips in which one of the actors (from Familiarization) was seated with her head down at a table on which there were two boxes on either side of the actor. The sequence of events was as follows: 1) after an attention-grabbing sound, the actor looked up, smiled, and said “hi”, 2) she then leaned over and peeked inside each of the two boxes, 3) looked and smiled at the infant, 4) then leaned over to the box that she looked into first, demonstrated positive regard (‘Ooh’), grasped the box, placed it in front of her and said, ‘Mmmm’, while still looking into the box (Figure 2). At this point, the clip froze and stayed on for another five seconds. Infants were shown two types of choice event videos: the majority choice event and the individual choice event. For the majority choice event, one of the individuals who had previously been shown to act with and belong to the majority completed the choice event. In the individual choice event, the actor who had previously acted alone and was not part of the majority completed the choice event. The choice event set up a context in which the majority members preferred one box and the individual preferred the other. A different actor from the majority completed each group choice event. Infants were shown each type of choice event in alternation three times resulting in infants being shown six choice events.
Test event. After the last choice event, infants were shown a still picture of the two boxes (Figure 3) for eight seconds (determined during pilot testing).
Data reduction. Raw data was processed using Tobii Studio’s I-VT fixation filters. For preliminary analyses, static areas of interest (AOIs) were created to test whether infants allocated similar levels of attention to the familiarization events (Majority familiarization: 919 px x 607 px; Individual familiarization: 541 px x 702 px; Joint familiarization: 1277 px x 642 px) and choice events (1163 px x 661 px). AOIs were created such that actors’ faces, shirt color, and the toys (familiarization) / boxes (choice events) were identifiable. Further, two same-sized (486 px x 409 px) AOIs were created for the test event stimulus (see Figure 3) to attain a measure of the duration of the test trial that infants spent looking towards each of the two boxes.

1.1 Majority Influence: A Background.
1.2 Why does Majority Rule? Factors that Motivate and Contribute to Majority Influence
1.3 Perspectives on Majority Influence
1.4 Developmental Perspectives and Evidence of Majority Influence
1.5 A Recap
1.6 The Present Research
1.7 Concluding Remarks
2.1. Introduction
2.2 Experiment 1
2.3. Method
2.4 Figure 1
2.5 Results
2.6 Table 1
2.7. Discussion
2.8 Experiment 2
2.9 Method
2.10 Results and Discussion
2.11 Figure 2
2.12 General Discussion
3.1 Prologue
3.2 Introduction
3.3 Experiment 1
3.4 Method
3.5 Figure 1
3.6 Figure 2
3.7 Figure 3
3.8 Results and Discussion
3.9 Table 1
3.10 Experiment 2
3.11 Method
3.12 Results and Discussion
3.13 Figure 4
3.14 General Discussion
4.1 Prologue
4.2 Introduction
4.3 Method
4.4 Figure 1
4.5 Figure 2
4.6 Figure 3
4.7 Results
4.8 Table 1
4.9 Figure 4
4.10 Discussion
5.1 Chapter Summaries and Findings
5.2 Empirical Contributions
5.3 Methodological Contributions: The Novel Paradigm
5.4 Theoretical Contributions
5.5 Open Questions and Future Avenues
5.6 Conclusion

Related Posts