Elsevier

Cognition

Volume 131, Issue 1, April 2014, Pages 44-59
Cognition

Intentions vs. resemblance: Understanding pictures in typical development and autism

https://doi.org/10.1016/j.cognition.2013.12.009Get rights and content

Highlights

  • Referential meaning of pictures can be determined by intentions or resemblance.

  • We examined these routes in children with autism and typically developing toddlers.

  • Typically developing toddlers link pictures to 3-D referents based on intentions.

  • Children with autism rely upon perceptual resemblance and are realists.

  • Thus we identify two distinct routes into the symbolic understanding of pictures.

Abstract

Research has debated whether children reflect on artists’ intentions when comprehending pictures, or instead derive meaning entirely from resemblance. We explore these hypotheses by comparing how typically developing toddlers and low-functioning children with autism (a population impaired in intentional reasoning) interpret abstract pictures. In Experiment 1, both groups mapped familiar object names onto abstract pictures, however, they related the same representations to different 3-D referents. Toddlers linked abstract pictures with intended referents they did not resemble, while children with autism mapped picture-referent relations based on resemblance. Experiment 2 showed that toddlers do not rely upon linguistic cues to determine intended referential relations. Experiment 3 confirmed that the responding of children with autism was not due to perseveration or associative word learning, and also provided independent evidence of their intention-reading difficulties. We argue that typically developing children derive meaning from the social-communicative intentions underlying pictures when resemblance is an inadequate cue to meaning. By contrast, children with autism do not reflect on artists’ intentions and simply relate pictures to whatever they happen to resemble.

Introduction

Symbolic understanding is at the core of human cognition (Deacon, 1997, Tomasello, 1999, Wittgenstein, 1953). Children must learn to master the symbol systems of their culture if they are to become functioning social beings (Callaghan, Moll, Rakoczy et al., 2011). In Western societies, pictorial symbols are universally used to teach children about the world, and, for many low-functioning children with autism (CWA), they provide an alternative means of communication (Frost & Bondy, 2002). In order to use pictures as symbols, children must understand that they are representations of independently existing objects. Typically-developing children (TDC) show awareness that pictures are symbols for real-world entities between 18 and 36 months (Callaghan, 2000, Callaghan, 2008, Callaghan et al., 2011, DeLoache and Burns, 1994, Ganea et al., 2008, Preissler and Carey, 2004, Suddendorf, 2003). However, comparatively little is known about picture comprehension in autism, and the question of how TDC and CWA map pictures to objects remains unanswered.

Some theorists contend that resemblance (i.e. similarity of perceptual features) defines picture-referent relations (e.g. Hopkins, 1995, Hopkins, 1998, Hyman, 2006, Peacocke, 1987), while others claim that a picture’s referent is determined by the intentions of the artist and that intention-monitoring skills are critical to picture comprehension (e.g. Bloom, 1996, Bloom and Markson, 1998, Preissler and Bloom, 2008, Taylor, 1998). The purpose of the present study was to establish which of these cues is most crucial to picture-object mapping in TDC and low-functioning CWA; do they always relate pictures to objects they resemble, or do they map pictures to objects they are intended to represent (irrespective of resemblance)? A novel way of teasing apart these hypotheses is to compare how these populations comprehend abstract pictures that relate to referents only by virtue of representational intent. Given that low-functioning CWA often have profound difficulty understanding the intentions of others (Baron-Cohen, 1995, D’Entremont and Yazbek, 2007; DSM-IV: American Psychiatric Association, 1994, Kanner, 1943), this comparison may yield the first evidence that TDC and CWA differ in their understanding of what fundamentally relates a picture to its referent. Mapping abstract pictures to objects based on resemblance, despite its inadequacy as a cue to intended referential meaning, would be consistent with naïve realism – a non-intentional theory of picture interpretation that privileges perceptual similarity and neglects external sources of meaning that are not immediately perceptible (e.g. the artist’s intentions, whether the picture was created accidentally, expectations of the viewer; Freeman, 1991, Freeman and Sanger, 1995).

Several studies have examined young children’s sensitivity to referential intentions when comprehending pictures. Bloom and Markson (1998) asked 3- and 4-year-olds to draw pairs of objects that closely resembled each other, such as a balloon and a lollipop. The pairs of pictures were virtually indistinguishable, and therefore could not be accurately matched to their original referents based on resemblance alone. When asked to name their drawings after a distractor task, both 3- and 4-year-olds correctly and consistently discriminated based on their original representational intentions. Even more remarkably, 2 and 3-year-old children have been shown to perform mentalistic reasoning when interpreting ambiguous pictures created by others. In one study, Gelman and Ebeling (1998) showed 2- and 3-year-olds a series of line drawings roughly shaped like familiar nameable objects (e.g. a kite). Some children were informed that the pictures had been created intentionally (e.g. someone painted a picture), while others were told that the pictures had been created accidentally (e.g. someone spilled some paint). Children were more likely to name the ambiguous pictures according to shape (thus regarding them as symbolic representations) when they believed the pictures were intentional creations. In another study, 2-year-olds watched an experimenter produce an ambiguous line drawing that looked equally like two unfamiliar objects (Preissler & Bloom, 2008). When asked to extend a novel label from the picture, the majority of children generalized the word to the object that the artist had been gazing at whilst drawing, suggesting that they perceived this object to be the picture’s intended referent. Together these studies indicate that, by 2-years of age, TDC consider intentional information when comprehending pictures created by themselves and by others.

Although TDC are capable of using intentional information to decipher ambiguous visual representations, it is undeniable that resemblance plays a vital role in children’s picture comprehension. Numerous studies have demonstrated that young children’s ability to map picture-referent relations is facilitated by high levels of iconicity – the extent that a picture resembles its referent (Callaghan, 2000, Ganea et al., 2008, Simcock and DeLoache, 2006). However, there is some evidence that children’s picture comprehension is governed primarily by resemblance, including when a picture’s appearance is in conflict with its creator’s intentions. When Browne and Woolley (2001) presented 4-year-olds, 7-year-olds, and adults with a puppet show in which the protagonist announced his intention to draw a bear, but actually produced a picture resembling a rabbit, all groups named the picture according to its appearance (e.g. a rabbit) rather than the artist’s stated intentions (e.g. a bear; also see Richert & Lillard, 2002).

The preceding results suggest that if a picture is sufficiently recognisable, resemblance rather than referential intent determines what it represents for both children and adults. However, as it is extremely irregular to encounter a drawing that is intended to represent X, but uniquely resembles Y, participants in these studies may have disregarded the artists’ intentions in an attempt to reconcile the conflicting cues. While it is doubtful that an artist would draw one thing whilst intending to represent something else, it is culturally acceptable to assign meaning to pictures that do not have a clearly recognisable referent (e.g. abstract art, infantile scribbles). Studying how children interpret abstract pictures can provide an innovative and ecologically valid method of assessing the relative importance of resemblance and representational intent to picture comprehension. Intentional theorists, such as Bloom and Markson (1998), claim that resemblance is beneficial because it provides a window to an artist’s intentions – “children might call a picture that looks like a bird “a bird” not merely because it looks like a bird, but because its appearance makes it likely that it was created with the intent to represent a bird” (Bloom & Markson, 1998, p. 203). Therefore, it may be that TDC derive meaning from resemblance only insofar as it provides an index of representation (Bloom and Markson, 1998, Gelman and Ebeling, 1998). If this theory holds, we would expect TDC to refrain from mapping shape-based picture-object relations when resemblance is a poor cue to a picture’s intended meaning.

To date, very few studies have examined children’s comprehension of abstract pictures, with Bloom and Markson (1998) being a notable exception. In their “Size Task”, 3- and 4-year-olds were shown pairs of differently-sized scribbles that had been ‘drawn’ by a child with a broken arm. For each pair of scribbles, the experimenter explained that the artist had attempted to draw two objects, such as an elephant (large) and a mouse (small). Crucially, the pictures looked nothing like the named objects, and could only be matched to their intended referents based on relative size. At test, children mapped labels for large and small objects to the abstract pictures based on relative size, which the authors interpreted as evidence for TDC inferring the artist’s representational intent. However, a more stringent test of intention reading in the domain of pictures would require children to map pictures to referent objects in the absence of resemblance.

Research investigating the development of pictorial understanding has shown that social-cognitive skills (e.g. intention reading and imitation) enable TDC to learn about pictures through interactions with symbolically-experienced adults (Callaghan and Rankin, 2002, Callaghan and Rochat, 2008, Callaghan et al., 2004, Rochat and Callaghan, 2005). However, many low-functioning CWA show deficits in the social-cognitive skills that underlie pictorial development (Baron-Cohen, 1989, Baron-Cohen, 1995; Charman et al., 1997; Griffin, 2002, Hobson, 2002, Mundy and Willoughby, 1996). If a nonverbal child with autism is unable to infer representational intentions from a social partner’s communicative actions involving pictures, they may fail to learn that pictures are symbolic. Instead, that child may treat pictures as signs, perhaps learning that viewing a certain picture precedes receiving a certain object and that particular pictures can be used to direct others’ behaviour in different ways, without understanding that pictures symbolize their referents. As a result, low-functioning CWA may not realise that information directed at pictures (e.g. verbal labels) actually relates to their real-world referents. Indeed, when the degree of resemblance between picture and referent is low, this population tends to form associative mappings between pictures and verbal labels (Preissler, 2008). However, when a picture is highly-iconic (e.g. a colour photograph), CWA are much more likely to treat it as a symbol (Hartley & Allen, in press). Thus, it appears that picture comprehension in low-functioning CWA may be contingent on a high-degree of perceptual resemblance.

If low-functioning CWA rely on resemblance to guide their mapping of picture-referent relations, we would expect to observe differences in the way these individuals comprehend abstract pictures. Whereas TDC might infer an artist’s intentions when iconicity is a poor cue to a picture’s referential meaning, the social-cognitive deficits that characterise autism (Baron-Cohen, 1989, Charman et al., 1997, Hobson, 2002, Mundy and Willoughby, 1996, Tomasello et al., 2005) may prevent low-functioning CWA from utilising this strategy. CWA also demonstrate a strong tendency to focus on localised perceptual elements when processing visual stimuli (Frith, 2003, Happé, 1999, Happé and Frith, 2006, Mottron et al., 2003, Mottron et al., 2006). Theoretically, an impaired ability to reason about others’ mental states coupled with a preference for fine-grained detail would make the comprehension of abstract pictures extremely difficult. Consequently, it may be that CWA derive meaning from abstract pictures based entirely on resemblance, despite its inadequacy as a referential cue.

As yet, no research has investigated how CWA comprehend abstract pictorial representations. However, Allen (2009) examined how CWA interpret ambiguous pictures that could not be matched to referents based on resemblance alone. Participants with a mean mental age of 4.5-years watched an adult turn towards one of two objects and produce a drawing. The representation resembled both objects on the table equally. When asked to label the picture, CWA named the object that was in the experimenter’s line of sight on just 25% of trials, whereas TDC made the same response on 75% of trials. These findings suggest that, unlike TDC, CWA do not use intentional information to decode the referents of ambiguous pictures drawn by others.

The objective of the present study was to establish whether children are reliant on resemblance when mapping picture-object relations. We approached this long-standing theoretical question from a novel angle: by examining how TDC and low-functioning CWA decipher abstract pictures that do not resemble their intended referents. While adults may call an arbitrary shape “an elephant” if they believe that its creator intended the image to represent an elephant, it is currently unknown whether children also believe that representational intent, rather than resemblance, determines the referents of non-iconic pictures. In Experiment 1, participants received a modified version of Bloom and Markson’s (1998) Size Task, which assessed whether they would map words and objects to abstract drawings based on intentional cues. Test trials comprised of a Picture Selection Stage (based on Bloom and Markson’s original procedure) followed by a novel Object Selection Stage. At the Picture Selection Stage, children were required to map object names to abstract pictures based on inferred referential intentions, in the absence of resemblance. Based on previous findings, we anticipated that TDC would reliably use intentional cues to guide their mapping of words to abstract pictures (Bloom and Markson, 1998, Gelman and Ebeling, 1998, Preissler and Bloom, 2008). By contrast, we expected that low-functioning CWA would have difficulty mapping correct word-picture relations in the absence of resemblance. At the Object Selection Stage, children were asked to link an abstract picture to a real 3-D referent. Participants selected from an ‘Intended Referent’ (an object that matched the artist’s representational intentions), a ‘Perceptual Referent’ (an object that resembled the picture) and a distractor. If, as Bloom and Markson (1998) suggest, 3-year-olds understand that resemblance is not necessary for a picture-referent relation to exist, we would expect TDC to select the object that corresponds with the artist’s intentions and ignore the closer perceptual match. Contrastingly, we predicted that resemblance would guide picture-object mapping in CWA, and that they would more frequently select the Perceptual Referent regardless of the artist’s intentions, making them naïve realists (Freeman, 1991, Freeman and Sanger, 1995).

Experiments 2 and 3 rule out alternative explanations for the response patterns of TDC and CWA on the Size Task and provide further evidence that these populations differ in their understanding of pictorial representations. Importantly, the results of these experiments will advance our understanding of symbolic development by elucidating the cues that fundamentally underpin picture-object relations for children with and without autism.

Section snippets

Participants

Fifteen CWA (15 male; M age: 9.7 years, range: 5.3–13.8 years) and fifteen TDC (10 males, 5 females; M age: 3.3 years, range: 2.5–5.3) were matched on receptive vocabulary as measured by the British Picture Vocabulary Scale (BPVS; Dunn, Dunn, Whetton, & Burley, 1997).

Abstract Picture-Selection

The number of trials (out of 4) in which participants selected the correct picture (e.g. large shape if the large object was requested) was calculated. TDC used relative size to identify the picture of the named object on 88% of trials (M = 3.53, SD = 0.92; see Fig. 3), significantly greater than chance, t(14) = 6.49, p < .001, d = 1.66. Surprisingly, CWA selected the correct picture on 68% of trials (M = 2.73, SD = 1.27; see Fig. 3), a rate that also significantly exceeded chance, t(14) = 2.22, p = .044, d = 

Discussion

Experiment 1 investigated whether TDC and low-functioning CWA derive meaning from abstract pictorial representations based on intentional information or perceptual resemblance. Our results indicate that TDC reliably assign verbal labels to abstract pictures based on intentional cues, which subsequently direct their mapping of picture-object relations. Surprisingly, CWA assigned labels to abstract representations based on intentional information, but then mapped picture-object relations based on

Participants

Participants were fourteen typically-developing 3-year-olds (6 males, 8 females; M age: 3.3 years, range: 3.0–3.6), with a mean BPVS receptive language age of 3.7-years (SD: 0.57). TDC in Experiment 2 did not differ from TDC in Experiment 1 on chronological age (t = 0.24, p = .81), or receptive vocabulary (t = −1.44, p = .16). Children were recruited from a day nursery in Kendal, United Kingdom.

Materials

Object stimuli were the large and small Intended Referents (3-D objects) used in Experiment 1, plus two

Results

Children were scored on the number of times they correctly matched the experimenter’s scribbles to their intended referents. For example, on an elephant–mouse trial, a child would score 2 if they correctly identified the large scribble as “an elephant” and the small scribble as “a mouse”. The maximum scores for familiar and unfamiliar objects were 8 and 2 respectively. Remarkably, in trials involving familiar objects, children mapped 96% of the experimenter’s scribbles to their intended object

Discussion

The results of Experiment 2 conclusively demonstrate that TDC are both able and willing to map picture-object relations based purely on intentional cues, in the complete absence of perceptual resemblance. It is apparent that, by 3-years of age, TDC understand that pictures are created with the intention of representing real-life referents, even if it is unclear what those referents actually are. Although the abstract pictures shared more perceptual features with familiar nameable shapes than

Participants

Participants were 17 low-functioning CWA (all male; M age: 9.7 years, range: 4.1–16.1 years) recruited from a specialist school in Preston, United Kingdom. None of these children participated in Experiment 1. All children received a diagnosis of autism from a qualified educational or clinical psychologist, using standardized instruments (i.e. Autism Diagnostic Observation Scale and Autism Diagnostic Interview – Revised; Lord et al., 1994, Lord et al., 2002) and expert judgment. Diagnosis was

Acknowledgements

We would like to thank the children, parents and staff at Hillside Specialist School, Preston (UK), Sandgate School, Kendal (UK), Ghyllside Primary School, Kendal (UK), and Burton Preschool, Burton-in-Kendal (UK).

References (65)

  • S. Baron-Cohen et al.

    Do children with autism use the speaker’s direction of gaze strategy to crack the code of language?

    Child Development

    (1997)
  • P. Bloom

    How children learn the meanings of words

    (2000)
  • P. Bloom et al.

    Intention and analogy in children’s naming of pictorial representations

    Psychological Science

    (1998)
  • C.A. Browne et al.

    Theory of mind in children’s naming of drawings

    Journal of Cognition and Development

    (2001)
  • T.C. Callaghan

    The origins and development of pictorial symbolic functioning

  • T.C. Callaghan et al.

    Early social cognition in three cultural contexts

    Monographs of the Society of Research in Child Development

    (2011)
  • T.C. Callaghan et al.

    Emergence of graphic symbol functioning and the question of domain specificity: A longitudinal study

    Child Development

    (2002)
  • T.C. Callaghan et al.

    Children’s understanding of artist-picture relations: Implications for their theories of pictures

  • T.C. Callaghan et al.

    Modeling referential actions in 6- to 18-month-old infants: A precursor to symbolic understanding

    Child Development

    (2004)
  • M. Carpenter et al.

    Understanding of others’ intentions in children with autism

    Journal of Autism and Developmental Disorders

    (2001)
  • T. Charman

    Theory of mind and the early diagnosis of autism

  • T. Charman et al.

    Infants with autism: An investigation of empathy, pretend play, joint attention, and imitation

    Developmental Psychology

    (1997)
  • B. D’Entremont et al.

    Imitation of intentional and accidental actions by children with autism

    Journal of Autism and Developmental Disorders

    (2007)
  • T. Deacon

    The symbolic species

    (1997)
  • L.M. Dunn et al.

    The British picture vocabulary scale

    (1997)
  • N.H. Freeman

    The theory of art that underpins children’s naïve realism

    Visual Arts Research

    (1991)
  • N.H. Freeman et al.

    Commonsense aesthetics of rural children

    Visual Arts Research

    (1995)
  • U. Frith

    Autism: Explaining the enigma

    (2003)
  • U. Frith et al.

    Language and communication in autistic disorders

    Philosophical Transactions of the Royal Society London B

    (1994)
  • L. Frost et al.

    PECS training manual

    (2002)
  • P.A. Ganea et al.

    Transfer between picture books and the real world by very young children

    Journal of Cognition and Development

    (2008)
  • S.A. Gelman et al.

    Shape and representational status in children’s early naming

    Cognition

    (1998)
  • Cited by (27)

    • Conveying symbolic relations: Children's ability to evaluate and create informative legends

      2020, Journal of Experimental Child Psychology
      Citation Excerpt :

      Based on this work, we know that by 3 or 4 years of age children consider the intentions of a symbol’s creator when interpreting a representation. For example, in Bloom and Markson (1998) and in Hartley and Allen (2014), experimenters presented participants with pictures that supposedly were drawn by a child with a broken arm. They explained that the artist had tried very hard to draw good pictures, but because of the artist’s broken arm, the pictures did not always look how the artist wanted.

    • From mind to picture: A systematic review on children's and adolescents’ understanding of the link between artists and pictures

      2020, Developmental Review
      Citation Excerpt :

      A total of six studies investigated whether children name pictures made by other people based on their appearance or the artist’s intention. Generally, it was found that children name pictures based on the artist’s intention when there is no conflict between appearance and intention, that is, when the image bore resemblance with what the artist meant to depict (Armitage & Allen, 2015; Browne & Woolley, 2001; Hartley & Allen, 2014). Furthermore, most of the studies suggested children favour intentionality over shape information when asked to label ambiguous drawings.

    • Is children's naming and drawing of pictures mediated by representational intentions? Evidence from typical development and autism

      2015, Cognitive Development
      Citation Excerpt :

      One possibility is that children simply label shape, without reflecting on factors that are external to the perceptible image (i.e. if an image is shaped like a cat, it is “a cat”; Browne & Woolley, 2001; Freeman, 1991; Freeman & Sanger, 1995). Alternatively, children might label pictures according to artists’ referential intentions (i.e. a picture is “a cat” only if it was created with the intention of representing a cat; Bloom & Markson, 1998; Gelman & Ebeling, 1998; Hartley & Allen, 2014). Our research has two primary aims: we investigate the influence of intention reading on picture interpretation in TD children aged 2–5 years, and also examine how children with intention monitoring difficulties derive meaning from pictures.

    View all citing articles on Scopus
    View full text