Children’s scale errors and object processing: Early evidence for cross-cultural differences

https://doi.org/10.1016/j.infbeh.2021.101631Get rights and content

Highlights

  • We examined the relationship between children’s scale errors and their categorization ability in Japan and UK.

  • UK children who showed greater local processing made more scale errors.

  • Japanese children, who overall showed greater global processing, showed no such relationship.

  • Suppression of scale errors in children may emerge not from attention to size per se, but from an integration of global and local information during object processing.

Abstract

Scale errors are observed when young children make mistakes by attempting to put their bodies into miniature versions of everyday objects. Such errors have been argued to arise from children’s insufficient integration of size into their object representations. The current study investigated whether Japanese and UK children’s (18–24 months old, N = 80) visual exploration in a categorization task related to their scale error production. UK children who showed greater local processing made more scale errors, whereas Japanese children, who overall showed greater global processing, showed no such relationship. These results raise the possibility that children’s suppression of scale errors emerges not from attention to size per se, but from a critical integration of global (i.e., size) and local (i.e., object features) information during object processing, and provide evidence that this mechanism differs cross-culturally.

Introduction

How infants and toddlers integrate feature information in object learning has attracted considerable attention in developmental psychology (e.g., Chen & Westermann, 2018; Cohen et al., 2002; Mareschal et al., 1999). Infants’ ability to individuate featural information such as shape, size, and color develops substantially in early infancy (Wilcox & Biondi, 2016). For example, by 4.5 months infants can reliably represent shape and size information, and by 11.5 months they can detect color information and use this information to individuate objects (Wilcox, 1999). However, early in development toddlers’ ability to integrate featural information with spatio-temporal information remains limited, presumably because young children have difficulty in maintaining representations of multiple object features (Bertenthal, 1996; Johnson & Mareschal, 2001). For example, infants within the first year of life are not able to represent featural and spatial information simultaneously, despite being able to process both types of information individually (Bremner et al., 2006; Oakes et al., 2006). Further, dissociations between perceived information and subsequent action are evident at these early stages of development; for example, 18- to 24-month-old children fail to integrate perceptual information with appropriate action responses (e.g., searching; Nardini et al., 2008).

The challenge of integrating perceptual information with the appropriate action representation is illustrated by an intriguing phenomenon known as the scale error (e.g., DeLoache et al., 2013; DeLoache et al., 2004; Rivière et al., 2020). When making a scale error, young children make mistakes by attempting to put their bodies into miniature versions of everyday objects (Arterberry et al. 2020). For example, children have been observed trying to get into a miniature sized car, just as they would do with a normal sized car. An example of a child performing such a scale error is depicted in Fig. 1. Children’s scale errors were first reported in laboratory settings, where children were exposed to miniature toys immediately after interacting with a appropriate-sized toy. In the first of these studies, DeLoache et al. (2004) demonstrated that approximately half of their sample of 18- to 30-month-old children performed at least one scale error. Subsequent studies have revealed that scale errors are also observed in the classroom (Rosengren, Carmichael et al., 2009) and in the home (Rosengren, Gutiérrez et al., 2009) without immediate prior exposure to corresponding real-sized objects. Importantly, then, scale errors are a real-world phenomenon that do not depend on a controlled lab environment for its elicitation (see also DeLoache et al., 2013), but whose prevalence is also marked by substantial individual differences between children (e.g., Rosengren et al., 2010).

Several mechanisms have been proposed as the potential cause of scale errors, most notably the immaturity of the brain network which governs object recognition/action planning (processed by the ventral stream) and online action control (processed by the dorsal stream; Glover 2004). Specifically, DeLoache et al. (2004) suggested that when children see a miniature replica of an object their mental representation of the corresponding real-size object is activated together with the motor routines associated with that real-size object. Typically, these motor routines are then inhibited, and an action plan based on the miniature category exemplar is executed instead, leading to size-appropriate behaviors such as pretending (as opposed to earnest attempts to carry out the action). However, in some cases this inhibition fails and a scale error is performed.

Several studies have explored the possibility that children’s failure to inhibit size-inappropriate motor routines may be associated with their insufficient encoding of size information when encountering miniature objects. For example, DeLoache and Uttal (2011) reported that some children in their study were not aware of the change in object size from normal to miniature. Related work has revealed that the occurrence of scale errors was negatively associated with children’s concept of size as measured using a parental questionnaire (Ishibashi & Moriguchi, 2017). In other work, using a looking time task designed to assess whether children could detect the appropriate or inappropriate object size for a particular tool, Grzyb et al. (2017) demonstrated that children who had exhibited scale errors were less likely to detect size changes than those who had not. Moreover, Grzyb et al. (2019) suggest that children’s lack of attention to object size may stem from their emerging shape bias: as children learn new words, they also learn that words tend to refer to category exemplars that share a shape, but that size is less relevant for category membership (Landau et al., 1988).

Taken together, these studies support accounts that assume that scale errors stem from an insufficient processing of size information when encountering objects with an atypical size (i.e., miniature replicas of larger real-world objects). Individual differences in the frequency of scale errors might on this account point to variation in the ability to integrate local (i.e, recognizing elements of object features; car door) and global features (i.e., recognize the whole object configuration; size or shape) in object processing.

Nevertheless, in studies using hierarchical patterns in which local features are arranged in global patterns (such as Navon figures in which e.g., a large T is made up of many small Es), infants as young as 3–4 months have been shown to be able to process both local and global information (Ghim & Eimas, 1988), with a processing advantage for global information (Frick et al., 2000), although there are considerable individual differences between infants (Guy et al., 2013). In addition, as attentional selection develops, children learn to inhibit irrelevant information, successfully focus their attention on relevant information (Krakowski et al., 2016; Krakowski et al., 2018), and flexibly switch from global to local target information (Zappullo et al., 2020).

However, although the ability to process both global and local elements individually emerges early on, the ability to integrate local with global information shows a more protracted development. This ability has, for example, been investigated in the context of infants’ formation of object categories. In a seminal study, Younger (1985) designed a set of eight line-drawn imaginary animals with four distinctive features (length of the neck and legs, thickness of the tail, distance between the ears) so that the features were correlated (e.g., a shorter legs would always co-occur with a thicker tail). If infants were able to detect these correlations they would form two categories based on correlated feature clusters. Younger (1985) found that 10-month-olds indeed formed two categories, indicating that they were able to encode the global feature relations across the stimuli. Nevertheless, follow-on studies have again found substantial individual differences between infants, with some infants categorizing on the basis of overall similarity and others on the basis of feature correlations (see Westermann & Mareschal, 2004), indicating that the ability to integrate local and global object information follows highly individual developmental trajectories.

Differences in global/local processing have not only been reported within homogeneous participant groups, but also on a larger scale between members of different cultures. Specifically, a substantial body of research has shown that members of Eastern cultures such as Japan and China show more holistic, global processing than members of Western cultures such as the US (Chua et al., 2005; Kelly et al., 2010; Masuda et al., 2014). For example, in tasks using Navon figures, East Asian adults showed a strong global advantage compared with Westerners in detecting target letters (McKone et al., 2010). Relatedly, several studies have shown that when viewing scenes containing local objects on a complex background, Westerners tend to look more at the local objects than East Asians (e.g., Chiu, 1972). Furthermore, when subsequently describing such scenes East Asians described surrounding information such as the color of water in an aquarium containing fish, whereas US participants’ descriptions related more directly to the objects, for example describing their motion (Masuda & Nisbett, 2001). Similar differences in processing have also been observed in abstract tasks. For example, in the Framed Line Task, in which participants are asked to judge the length of a line drawn inside a square, Japanese participants showed fewer errors when making judgements based on the length of the line relative to the background information (i.e., the size of the square), whereas US participants showed fewer errors when judging absolute length, irrespective of the square size (Kitayama et al., 2003; for a review, see Nisbett & Miyamoto, 2005).

Related results from the developmental literature suggest that children from Eastern and Western cultural backgrounds also show differences in processing. For example, like adults, Japanese children aged 6–13 years demonstrate heightened attention to context in the Framed Line Task compared with their US peers (Duffy et al., 2009). Differences have also been observed in children as young as 3, with Japanese children recognizing objects holistically whereas US children identified them locally (Kuwabara & Smith, 2016). In Kuwabara and Smith's (2016) study, US children relied on low-level features for object identification, whereas Japanese children showed more evidence of configural processing when the overall shape of an image was masked.

It is an open question whether scale error production is related to specific object processing strategies and thus, to cultural differences. In particular, although scale errors have been shown to be a robust phenomenon across cultures (in Western countries, e.g., Brownell et al., 2007; in Asian countries, e.g., Ishibashi & Moriguchi, 2017), the encoding processes by which children integrate local features of object properities (e.g., car door) into their global features (e.g., size, shape) may operate differently across different cultural backgrounds. On one hand, given that Western children pay more attention to local features, they may be better able to encode specific object properties. On the other hand, given that East Asians show stronger effects of an object’s surrounding environment on its encoding it is also possible that these children specifically encode global elements such as size more robustly than Western children. Irrespective of culture, we would expect children who show better object processing to be less likely to commit scale errors.

The current study explored these possibilities by examining the relationship between scale error production and object processing ability in 18- to 24-month-old Japanese and UK children. To index object processing, we tested infants in a categorization task. In a typical task, infants are presented with a series of pictures from one category during a familiarization phase, followed by a test phase in which a novel exemplar from the familiarized category is paired with an out-of-category exemplar. Infants’ longer looking at the out-of-category exemplar is taken as evidence that they have learned the category comprising the familiarization items and the within-category test item, and therefore show a novelty response to the out-of-category item. A rich body of work has shown that even young infants can succeed in such tasks and that they form categories on-line on the basis of the perceptual features of the presented items (e.g., Quinn et al., 1993; Younger & Cohen, 1986).

In the current study, we presented children with the categorization task used by Althaus and Westermann (2016). In this study, infants were familiarized with a novel category of morphed, cartoon animal stimuli, in which perceptual features (e.g., body morphology, posture; see Fig. 2) changed gradually along a continuum. Test stimuli consisted of three novel exemplars: a prototypical stimulus drawn from the averaged familiarization stimuli, a peripheral stimulus taken from the extreme end of the familiarization category, and a completely novel stimulus drawn from a novel category. Infants saw paired test stimuli across three types of test trial: novel/prototypical, novel/peripheral, and prototypical/peripheral. We took children’s preference for the more novel exemplar on each type of test trial as an index of their novelty preference and hence, their object processing.

To elicit scale errors, we presented Japanese and UK children with the scale error task employed by DeLoache et al. (2004), with the addition of an extra set of stimuli (shoes) for Japanese children to acknowledge cultural norms (Ishibashi & Moriguchi, 2017). We hypothesized (a) that due to different culture-specific object processing strategies we would find differences in the prevalence of scale errors between Japanese and UK children, and (b) that children who showed a stronger novelty preference in the categorization task, indicating more robust encoding of object features, should make fewer scale errors.

Section snippets

Participants

Participants were 40 typically-developing Japanese-learning children living in Japan (M = 19.73 months, SD = 1.63, 17 girls, range = 18.21–24.74 months) and 40 typically-developing English-learning children living in the UK (M = 19.75 months, SD = 1.86, 23 girls, range = 18.02–23.90 months). Sample sizes were determined a priori to reflect those in previous studies which were sufficient to detect scale errors (e.g., Ware et al., 2006).

From the UK sample, an additional nine children were

Results

All analyses were performed in R Studio (v. 3.4.2; R Studio Team, 2015); linear mixed effects models were conducted in lme4 (v. 1.1–15; Bates et al., 2014).

Discussion

The purpose of this study was to examine whether children’s scale errors were associated with their categorization ability, and if so, whether this relationship differed between Japanese and UK children. In particular, we reasoned that culture-specific differences in object processing may result in differences in category learning; specifically, we predicted that UK children’s local object processing would lead to better category learning. Further, we reasoned that because children who

Conclusion

This study offers a new perspective on the mechanisms underlying the intriguing phenomenon: children’s suppression of scale errors emerges not from attention to size per se, but from a critical integration of global and local information, and provides evidence that this mechanism may differ cross-culturally.

Declaration of Competing Interest

The authors report no declarations of interest.

Acknowledgements

The authors would like to thank Marina Bazhydai and Ayaka Kawasaki for coding. We are extremely grateful to all the children and parents who took part.

This work was supported by an ESRC Future Research Leaders fellowship to KT and the ESRC International Centre for Language and Communicative Development (GW, KT; LuCiD) [ES/L008955/1; ES/N01703X/1], and JSPS KAKENHI Grant Number JP18H05524(IU).

MI was affiliated with the Department of Psychology at Ochanomizu University until the submission of

References (58)

  • M. Kuwabara et al.

    Cultural differences in visual object recognition in 3-year-old children

    Journal of Experimental Child Psychology

    (2016)
  • B. Landau et al.

    The importance of shape in early lexical learning

    Cognitive Development

    (1988)
  • E. McKone et al.

    Asia has the global advantage: Race and visual attention

    Vision Research

    (2010)
  • R.E. Nisbett et al.

    The influence of culture: Holistic versus analytic perception

    Trends in Cognitive Sciences

    (2005)
  • K.S. Rosengren et al.

    Individual differences in children’s production of scale errors

    Infant Behavior and Development

    (2010)
  • T. Wilcox

    Object individuation: Infants’ use of shape, size, pattern, and color

    Cognition

    (1999)
  • M.E. Arterberry et al.

    Integration of thought and action continued: Scale errors and categorization in toddlers

    Infancy

    (2020)
  • D. Bates et al.

    Fitting linear mixed-effects models using lme4

    (2014)
  • B.I. Bertenthal

    Origins and early development of perception, action, and representation

    Annual Review of Psychology

    (1996)
  • C.A. Brownell et al.

    “So Big”: The development of body self-awareness in toddlers

    Child Development

    (2007)
  • K. Casler et al.

    Children’s scale errors with tools

    Developmental Psychology

    (2011)
  • Y.C. Chen et al.

    Different novelties revealed by infants’ pupillary responses

    Scientific Reports

    (2018)
  • L.H. Chiu

    A cross-cultural comparison of cognitive styles in Chinese and American children

    International Journal of Psychology

    (1972)
  • H.F. Chua et al.

    Cultural variation in eye movements during scene perception

    Proceedings of the National Academy of Sciences

    (2005)
  • J.S. DeLoache et al.

    Gulliver, Goliath and Goldilocks: Young children and scale errors

  • J.S. DeLoache et al.

    Scale errors offer evidence for a perception-action dissociation early in life

    Science

    (2004)
  • J.E. Frick et al.

    Temporal sequence of global-local processing in 3-month-old infants

    Infancy

    (2000)
  • H.R. Ghim et al.

    Global and local processing by 3- and 4-month-old infants

    Perception & Psychophysics

    (1988)
  • B.J. Grzyb et al.

    Children’s scale errors: A by-product of lexical development?

    Developmental Science

    (2019)
  • Cited by (2)

    • Object labeling activates young children's scale errors at an early stage of verb vocabulary growth

      2022, Journal of Experimental Child Psychology
      Citation Excerpt :

      For instance, young children’s immature inhibitory control may fail to suppress inappropriate motor plans (e.g., sitting on a miniature-sized chair), which leads to scale error production (DeLoache, LoBue, Vanderborght, & Chiong, 2013; Ishibashi & Moriguchi, 2021). Other explanations include that children might not be able to process object features of size properly (Grzyb, Cangelosi, Cattani, & Floccia, 2017; Ishibashi & Moriguchi, 2017; Ware, Uttal, Wetter, & DeLoache, 2006), understand their own body size precisely (Brownell, Zerwas, & Ramani, 2007), or integrate multiple features (e.g., local/global properties) that the object has (Ishibashi, Twomey, Westermann, & Uehara, 2021), which results in scale errors. Some researchers attribute these errors to a strong bias for a certain type of object features—for example, object shape (Grzyb, Cangelosi, et al., 2019) or object function (Casler, Eshleman, Greene, & Terziyan, 2011)—which induces children to execute object-specific actions irrespective of other associated features such as object size.

    View full text