(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Development of strategic social information seeking: Implications for cumulative culture

['Kirsten H. Blakey', 'Faculty Of Natural Sciences', 'Psychology', 'University Of Stirling', 'Stirling', 'United Kingdom', 'Eva Rafetseder', 'Mark Atkinson', 'Elizabeth Renner', 'Fía Cowan-Forsythe']

Date: 2021-09

Human learners are rarely the passive recipients of valuable social information. Rather, learners usually have to actively seek out information from a variety of potential others to determine who is in a position to provide useful information. Yet, the majority of developmental social learning paradigms do not address participants’ ability to seek out information for themselves. To investigate age-related changes in children’s ability to seek out appropriate social information, 3- to 8-year-olds (N = 218) were presented with a task requiring them to identify which of four possible demonstrators could provide critical information for unlocking a box. Appropriate information seeking improved significantly with age. The particularly high performance of 7- and 8-year-olds was consistent with the expectation that older children’s increased metacognitive understanding would allow them to identify appropriate information sources. Appropriate social information seeking may have been overlooked as a significant cognitive challenge involved in fully benefiting from others’ knowledge, potentially influencing understanding of the phylogenetic distribution of cumulative culture.

Introduction

Seeking out relevant information from appropriate social sources is ubiquitous in human adults. Human adults may demonstrate key differences in the way they seek, attend to, and use social information compared to children and non-human animals (henceforth animals). This propensity for identifying and gathering relevant social information has been proposed as one of a suite of cognitive mechanisms that may be required for distinctively human cumulative culture [1–3]. By ‘distinctively human’ cumulative culture we are referring to the accumulation of beneficial modifications to cultural traits over successive generations of learners, which results in increased functionality or efficiency [4–6]. Its rarity in animals and apparent importance in accounting for human evolutionary success has prompted interest regarding the emergence and development of cognitive mechanisms in human children that support cumulative culture [5,7,8]. It may be the lack, limited scope, or inflexibility of mechanisms such as information seeking that impedes development of human-like cumulative culture in animals despite them showing evidence of culture [9–11] and social learning abilities [12,13]. The current study aimed to address the gap in our understanding regarding age-related changes in children’s ability to seek out and use appropriate social information. Documenting the developmental trajectory of this capacity could provide insights into the cognitive demands involved and whether it is likely to be observed in animals.

In a bid to understand the discontinuity between humans’ and animals’ capacity for cumulative culture a number of sociocognitive mechanisms have been considered [4,7,14,15]. Some of these proposals focus on when and how social information is used. Social learning is generally considered to be adaptive when it is present within a population. However, models have shown that under most conditions it is the ability to flexibly switch from social to individual learning, when social learning proves unsatisfactory, that increases adaptability in a population [16–20]. Selective or flexible rules that influence individuals’ use of social information are known as social learning strategies (SLSs). These refer to heuristic biases or rules that dictate when, what, and from whom social information should be acquired [21,22], helping to filter out less useful aspects of available social information. The selective nature of SLSs such as ‘copy older individuals’ or ‘copy the majority’ makes them generally more effective than learning indiscriminately from others or individual learning. However, extensive evidence of SLSs in both young children and animals [22–26] suggests that selective social learning cannot account for the marked differences we see between humans and animals with regard to the capacity for cumulative culture.

One proposal that attempts to explain the distinctiveness of human cumulative culture outlines a dual-process view of social learning. This view, put forward by Heyes [27], suggests that there are distinct categories of SLSs each based on specific types of decision rules. According to the first half of this account, distinctively human cumulative culture could be attributed to the use of explicitly metacognitive SLSs [27,28]. These are rules that are consciously represented and (for verbal individuals) reportable, that reflect learners’ explicit awareness and understanding of the social learning strategies they are employing. Learners who employ an explicitly metacognitive SLS have been proposed to depend on theory of mind and advanced metacognitive capacities that enable human adults to flexibly identify, select, or disregard social information across varied contexts [2,27–29]. As such, these strategies afford learners the capacity to represent what information is required and to reason about where, or from whom, that information can be obtained. General SLSs such as ‘copy older individuals’ do not necessarily take into account the knowledge or experience of other individuals with regard to a particular task. By contrast, we propose that explicitly metacognitive SLSs afford learners the ability to flexibly reflect on others’ potential to provide valuable information by taking into consideration their knowledge, experience, or access to information in relation to a particular task. Thus, learning can be targeted towards the most appropriate source of information (e.g., when learning to use information technology one should learn from ‘digital natives’ [30]).

The second part of Heyes’s [27] dual-process account of social learning proposes that the majority of the behaviours that conform to SLSs in both humans and animals are based on decision rules that depend on general-purpose associative learning processes or biologically selected biases. These rules direct learning towards objects, agents, and events that are most likely to provide useful information across a broad range of contexts. However, unlike explicitly metacognitive SLSs, these implicit SLSs (as we will refer to them here) are not driven by a causal understanding of the potential value of social information. Although we refer to these strategies as implicit, we do not claim that learners employing them are necessarily devoid of insight regarding personal preferences that guide their social learning. It is likely that learners sometimes explicitly represent strategies related to salient yet superficial cues without appreciating why their strategy is successful. However, we suggest that such strategies, while they may be explicitly represented, are not explicitly metacognitive due to the absence of a causal understanding of informants’ potential to provide valuable information. Social learning biases in young children and animals are likely driven by such implicit, and relatively crude, heuristic decision rules [31]. Indeed, cases in which very young children [32–36] and animals [24,26,36,37] select knowledgeable others are likely the result of implicit SLSs (e.g., associative learning), rather than the reasoning-based strategies employed by adults. For example, model-based biases for older [25,32,38], higher-status [23,34], or even more reliable models [35] may be the result of repeated exposure to the successes of models with these characteristics, resulting in rule-like strategies.

While there is evidence of implicit SLSs, based on heuristic biases, in both young children and animals, in our view, there is as yet no solid evidence of explicitly metacognitive SLSs in either population. By contrast, adult humans are (with good reason) assumed to be able to use social information in an explicitly metacognitive manner (although this should not be taken to mean that they do not also use implicit SLSs). For example, it is routine for human adults to actively seek out models of social behaviour, using their understanding of others’ knowledge, access to information, or intentions relative to their own to strategically select and use the most relevant social information [39,40]. A critical question, then, is how children’s use of social information develops with age, particularly in relation to their explicit understanding of the value of information that can be provided by others, and how they might be able to benefit from it. Since we assume that this understanding is not present from infancy, but is certainly in place by adulthood, it follows that this transition must occur over the course of childhood. We may therefore be able to identify key stages during development when children begin to change how they respond to social information. This would provide insights into the cognitive capacities upon which such abilities depend, and therefore might also shed light on the reasons for the apparent absence of these abilities in animals.

Research into social learning has largely been restricted to investigating the circumstances under which social information is used, the efficacy of that use, and its role in cultural transmission [41–47]. Indeed, to date, developmental research into SLSs has focused on examining children’s responses to task-relevant social information (usually an effective solution) which is provided in advance of an opportunity to solve the same task [44,48–51]. However, while cultural transmission necessitates using information acquired from a social source, human learners are rarely passive recipients of valuable social information. Here, we propose that actively seeking out valuable information when faced with a particular problem to solve is more analogous to real world social learning scenarios, compared with being passively provided with relevant information.

Therefore, in the current study we were particularly interested in determining when children develop the ability to seek out social information based on an understanding of its value. This led us to consider selective information seeking paradigms that already exist in the literature. Specifically, we looked at the selective trust paradigm. This paradigm is commonly used to examine children’s preferences for information provided by models with conflicting social and/or epistemic characteristics, and is built on the premise that children learn from others’ testimony (see [52] for a full review).

These studies typically employ a ‘conflicting sources paradigm’ in which children first observe two informants who differ on social (e.g., gender or accent) and/or epistemic (e.g., accuracy or reliability) characteristics [35]. Following a familiarisation phase, children are faced with an unfamiliar scenario, for example, naming an unfamiliar object. In some tasks, children are required to select one of two potential informants to ‘seek’ information from (i.e., ‘ask’ questions) and/or required to make their selection following the informants’ claims about the name of the object (i.e., ‘endorse’ questions). The model selected by the child is considered to be the model whose claim they trust. With regards to the influence of social characteristics on selective trust, evidence suggests that in the absence of epistemic differences children ask and endorse informants who have positive social characteristics [53]. These selections are influenced by both the models’ and the learners’ own characteristics (e.g., age or gender [54]). In particular, social characteristics that signal a model’s similarity to the learner (i.e., ingroup membership) are consistently favoured by young children (e.g., preference for informant with a native accent [55]; preference for informant of the same gender [56]). Model-based biases such as these are unlikely to require explicit cognitive reasoning. Consistent with the selective preferences in children’s proclivity to copy and the SLS ‘copy successful individuals’, they are likely the result of implicit biases that promote learning from sources that are most likely to provide useful information across the broadest range of contexts. The selective trust literature reports that children as young as 3 years old are sensitive to informants’ social and epistemic characteristics [33,35,55,57]. Recent meta-analyses examined the relation between children’s age and their selective trust decisions [53]. Results indicated that children asked and endorsed more knowledgeable (accurate/reliable) informants when they differed on only epistemic characteristics. Specifically, 4-year-olds were more likely than 3-year-olds to endorse knowledgeable informants. When informants differed on epistemic and social characteristics simultaneously, 4- to 6-year-olds were more likely to endorse informants who were knowledgeable but had a negative social characteristic while 3-year-olds appeared to weigh both characteristics equally. Thus, from 4-years children appear to place greater value on epistemic characteristics.

Whilst these preferences may be linked to a developing explicit awareness of the potential value of social information, the design of these studies precludes this conclusion. The paradigm depends on participants being exposed to information about the accuracy or reliability of the two informants in the familiarisation phase, in order to establish the respective epistemic characteristics (e.g., knowledgeable/ignorant) of the conflicting sources. Thus, the literature on selective trust, like much of the literature on SLSs, depends on children making choices between models on the basis of characteristics for which they are likely to have a prior history of associations or a pre-existing bias (whether established as part of the experimental procedure, e.g., reputation for accuracy, or from the child’s own life experience, e.g., age [58]). The results can therefore be likewise attributed to implicit biases. However, the apparent transition to favouring epistemic characteristics (such as prior accuracy) over social ones (such as familiarity [59]) is perhaps suggestive of a developing insight into the value of others as sources of social information.

While it is useful to know how children use social information and who children prefer to learn from, we argue that this is not sufficient to determine the cognitive mechanisms that children are employing during social learning. Children’s ability and proclivity to seek out relevant social information has not yet been adequately addressed. That is, nothing in the SLSs or selective trust literature has examined children’s ability to select valuable social information on the basis of its relevance for solving a specific problem.

Explicitly metacognitive SLSs are proposed to be experience-dependent and learned through social interaction, so we would expect them to emerge relatively late in development [27]. Therefore, to help explain why cumulative culture appears to be restricted to humans we can look at whether age-related changes in children’s ability to seek out relevant social information coincide with advances in cognitive development. If we find particular ages at which children make significant advances in their appropriate social information seeking and these occur at a similar age to the development of particular cognitive capacities, then these capacities might be necessary prerequisites. Similarly to Heyes [27], Baldwin and Moses [1] proposed that motivations for initiating an appropriate search for social information likely rest on advanced metacognitive capacities. They emphasised that to seek social information effectively, the seeker should have awareness of what information is required and from whom it can be obtained. While behavioural tests of implicit metacognitive ability suggest that infants [60], young children [61], and animals [62,63] react to the state of ignorance, they do so without necessarily recognising a metacognitive awareness of that state. However, we know that children’s cognitive capacities continue to progress well beyond those of animals. In particular, as their cognition advances, children develop abilities such as evaluating their own knowledge state [64,65], understanding others’ mental states [66,67], and recognition that perceptual access to information facilitates knowledge formation [68–70]. Such abilities are thought to be cognitively demanding; thus, their requisite nature may preclude younger children’s (and by logical extension animals’) ability to seek out appropriate social information. Therefore, if we can identify when children develop the capacity to use explicitly metacognitive over implicit SLSs we may be able to use this to predict the likelihood of the capacity being available to animals. That is, if we find that this ability develops late in childhood then this would be consistent with the hypothesis that it is dependent on cognitive capacities that are not available to animals, and could help to explain the distinctiveness of human cumulative culture.

The aim of the current study was to investigate when children develop the ability to seek out social information using explicitly metacognitive SLSs. To explore the development of children’s appropriate information seeking we designed a task in which 3- to 8-year-old children were faced with a problem for which they could not use prior experience or knowledge. Rather, to solve the problem children had to reason about the information needed and who had the potential to provide that information. In certain circumstances, explicitly metacognitive SLSs dictate that social cues should be ignored, at least when cues are available which are more directly related to task relevant experience. This was the case for the task in the current study which required children to choose which of four possible demonstrators could provide critical information for unlocking a box. Specifically, children’s choice between social demonstrations was related to characteristics of the box each of the demonstrators had access to (relative to the box they themselves were faced with), rather than personal characteristics of the demonstrator. Attending to the social demonstration was critical to children’s successful use of the information. Thus, although the critical characteristics children had to attend to were not social, they functioned to determine what social information (in the form of the demonstrator’s attempt) would provide the details they needed to solve the problem. It should also be noted that while choices made by matching one’s own box with one of the demonstration options would appear to show appropriate information seeking, making such a choice without appreciating why it is useful is unlikely to result in successful use of the information. Understanding why a particular demonstration is useful involves some recognition of the value of the social information as it requires the child to understand that the demonstrator’s performance (rather than just the box) will be informative. The demonstrator’s success (or failure) in unlocking the box, was the key detail children required to unlock their own box.

Appropriate information seeking should be based on a reasoned understanding of the value, to oneself, of the social information. Heyes [27] argued that this kind of cognitive reasoning may not be available in young children or animals. In the absence of the ability to reason about the potential value of social information provided by others, we expected children to rely on less cognitively demanding implicit SLSs. In the current study we specifically looked at whether children’s information seeking might instead be influenced by heuristic model-based biases for superficial demonstrator characteristics, such as age or gender. Thus, the task was designed such that the use of model-based biases in the absence of cognitive reasoning would be inappropriate, leading to imperfect information seeking.

Overall, we expected to find an age-related transition from the use of heuristic model-based biases (implicit SLSs) to reasoning-based choices driven by the value of the information (explicitly metacognitive SLSs).

Specifically, we predicted that appropriate information seeking would improve with age due to our anticipation that advances in children’s explicit metacognition would enable them to identify appropriate sources of information based on an understanding of what information they required. We expected that younger children would struggle to employ explicitly metacognitive SLSs and instead rely on less cognitively demanding heuristic biases (implicit SLSs) related to superficial demonstrator characteristics such as age or gender. Any age-related changes in children’s appropriate information seeking could indicate use of different SLSs when approaching the task. Thus, we examined the developmental trajectory of appropriate social information seeking to provide insight into the emergence of cognitive reasoning as a mechanism required for distinctively human cumulative culture. Finding evidence of reasoning-based choices in older, but not younger, children would support the proposal [27] that explicitly metacognitive SLSs are experience-dependent, developing relatively late in children.

Examining whether children use social information appropriately (following appropriate information seeking) could further help to distinguish between learners with some level of metacognitive understanding, and those reliant on implicit rules for what, when, and whom to copy. That is, understanding the relevance of the acquired information would be expected to also result in more appropriate use of that information. By investigating social information use in conjunction with appropriate information seeking we can really begin to expose the cognitive mechanisms that underlie these processes. How children use particular types of information provided by others can tell us much about the nature of their social learning processes–most importantly whether they might be using reasoning-based strategies or relying upon adaptive heuristic biases. For instance, children who have sought out information based on an understanding of its value are more likely to be able to use it appropriately than children who selected it on the basis of an implicit SLS. In the current study, copying was not always the correct response, and success was sometimes dependent on making a different choice to the one made by the demonstrator. As with appropriate information seeking, we expected successful use of the social information to improve with age.

[END]

[1] Url: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0256605

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.