Brief Report
Eighteen-month-olds understand false beliefs in an unexpected-contents task

https://doi.org/10.1016/j.jecp.2013.10.002Get rights and content

Highlights

  • Eighteen-month-olds demonstrate false-belief understanding in an unexpected-contents task.

  • They decide how best to help others by taking into account others’ goals and beliefs.

  • Thus infants use their theory of mind flexibly in a variety of tasks.

Abstract

Recent studies suggest that infants understand that others can have false beliefs. However, most of these studies have used looking time measures, and the few that have used behavioral measures are all based on the change-of-location paradigm, leading to claims that infants might use behavioral rules instead of mental state understanding to pass these tests. We investigated infants’ false-belief reasoning using a different paradigm. In this unexpected-contents helping task, 18-month-olds were familiarized with boxes for blocks that contained blocks. When an experimenter subsequently reached for a box for blocks that now contained a spoon, infants based their choice of whether to give her a spoon or a block on her true or false belief about which object the block box contained. These results help to demonstrate the flexibility of infants’ false-belief understanding.

Introduction

“Theory of mind” is about predicting and explaining the actions of others by invoking their mental states such as goals and beliefs (Premack & Woodruff, 1978). For decades, it was thought that children first begin to understand that others can sometimes hold false beliefs at around 4 or 5 years of age (see Wellman, Cross, & Watson, 2001, for a review). This conclusion was based on studies using two different types of false-belief tests: (a) the change-of-location test, in which an object is moved in the absence of a story character (e.g., the “Sally–Anne” test; Baron-Cohen et al., 1985, Wimmer and Perner, 1983), and (b) the unexpected-contents test, in which a box does not contain what it is supposed to contain (the “Smarties” test; Hogrefe et al., 1986, Perner et al., 1987). However, several recent studies have claimed false-belief understanding in 1-year-olds or even younger infants (see Baillargeon, Scott, & He, 2010, for a review).

The primary reason for the difference between these two sets of findings might be the explicit versus implicit methods used; the classic tests passed by preschoolers are explicit verbal tests, whereas the tests passed by infants use implicit measures such as looking time (e.g., Kovács et al., 2010, Onishi and Baillargeon, 2005, Surian et al., 2007) and anticipatory looking (Clements and Perner, 1994, Southgate et al., 2007, Surian and Geraci, 2012; see Sodian, 2011, for a review). Researchers have also designed studies in which infants are asked to act based on their understanding of a protagonist’s beliefs. For example, Buttelmann, Carpenter, and Tomasello (2009) presented 16- and 18-month-olds with an experimenter who put a toy into one of two boxes and left the room. In his absence, an assistant “sneakily” transferred the toy to the other box and locked both boxes. When the experimenter returned and tried to open the first box, infants helped him by going to the box that now contained the toy—inferring that because the experimenter believed the toy to be in the first box, he wanted his toy. In contrast, in a true-belief condition, in which the experimenter watched the switch of the toy but nevertheless then tried to open the empty box, infants went to the empty box and showed him how to open it—inferring that, because the experimenter knew the current location of the toy, he wanted to open the empty box for some other reason. Thus, infants interpreted the experimenter’s goal differently depending on whether he believed his toy to be in the box he tried to open or not (see also Knudsen & Liszkowski, 2012a, and Southgate, Chevallier, & Csibra, 2010, for other behavioral tasks with 1-year-olds).

The literature on infants’ understanding of false belief, however, has proved to be deeply controversial. Some researchers have argued that, in many of the implicit tests, infants do not necessarily need to attribute beliefs to the protagonist but might instead solve the tasks by matching observed test events to associations built during the familiarization phase or by using behavioral rules rather than mentalistic understanding (Apperly and Butterfill, 2009, Perner and Roessler, 2012, Perner and Ruffman, 2005, Rakoczy, 2012, Ruffman and Perner, 2005). One possible rule infants could employ is that people look for objects where they last saw them. Explanations such as this apply primarily to change-of-location tasks. For this reason, it is important to test infants in a variety of tasks, scenarios, and paradigms (Perner, 2010). If infants are able to pass multiple different tasks, an explanation based on belief understanding gains increased plausibility.

There are a few studies that are not based on the change-of-location paradigm, and still 1-year-olds show differences between conditions. For example, 18-month-olds seem to attribute false beliefs about an object’s identity or properties (Scott and Baillargeon, 2009, Scott et al., 2010). Most relevant for the current study, Song and Baillargeon (2008) presented 14.5-month-olds with two objects: a stuffed skunk and a doll with blue pigtails. After familiarizing infants with the experimenter reaching for the doll, both objects were put into opaque boxes without the experimenter watching. Importantly, there was a tuft of blue hair protruding from the box containing the skunk. When the experimenter returned, infants looked significantly longer when she reached for the box containing the doll than when she reached for the box containing the skunk, demonstrating their expectation about the experimenter wanting the doll and being deceived by the tuft of blue hair. Although these studies do not use the change-of-location paradigm, they all use the same measure: looking time. Whereas looking time differences can clearly indicate, for example, a discrimination, an expectation, or a preference, behavioral measures go beyond this by showing that infants can translate their understanding into appropriate action. This is important because in real-world social interactions, children often need to actually respond to their partners (e.g., help or warn them about something; Buttelmann et al., 2009, Knudsen and Liszkowski, 2012a) based on the partners’ false belief rather than just understand that they have one.

So far, all behavioral studies of infants’ false-belief understanding have used change-of-location tests; thus, other types of tasks are needed. A further potential issue is that in most of the behavioral studies, there was a cue for infants that something unusual was going on, that is in the false-belief condition, an assistant transferred the object in a “sneaky” manner. This cue could have drawn infants’ attention to the correct answer (the transferred object) in the false-belief condition, whereas it was not present in the true-belief condition (note that in the one behavioral study that did not involve sneakiness, there was no corresponding true-belief condition; Knudsen & Liszkowski, 2012b). Sneakiness has been shown to boost children’s performance in verbal false-belief tests (Wellman et al., 2001), but this better performance does not necessarily reflect a better understanding of the others’ mental states (Sodian, Hülsken, & Thörmer, 1999). Therefore, converging evidence from studies using different paradigms, and from procedures not involving sneakiness, is needed to investigate infants’ flexibility in ascribing false beliefs to others.

In the current study, we presented 18-month-old infants with a new test modeled closely on the classic unexpected-contents (Smarties) test (Hogrefe et al., 1986, Perner et al., 1987) with no sneakiness. An assistant introduced infants to a series of boxes covered with pictures of children’s building blocks. Together with an experimenter, infants experienced that each of these “block boxes” contained a block. However, right before the test, one of the boxes (the target box) surprisingly did not contain a block but rather contained a spoon. What differed between conditions was whether the experimenter was present during that experience and therefore knew about the surprising contents (true-belief condition) or whether she was absent and so would believe that there was a block in this box too (false-belief condition). In both conditions, the experimenter subsequently reached for the target box unsuccessfully. The assistant invited infants to help and get what the experimenter wanted by giving them access to two test objects: a block (correct in the false-belief condition) and a spoon (correct in the true-belief condition). To interpret the experimenter’s goal correctly and help appropriately, infants must understand what the experimenter expected to find inside the box based on her belief about its contents. If they could do this, it would demonstrate their flexibility in attributing belief states to others and would rule out sneakiness as a possible explanation for their positive results in other behavioral studies.

Section snippets

Participants

Participants were 36 18-month-olds (mean age = 18 months; 9 days, range = 17;18–18;27, 17 girls and 19 boys). Half of the infants (n = 18) were assigned to each condition in a between-participants design. An additional 21 infants were tested but not included in the main analyses because of parental or experimenter (E) error (n = 4), because they gave one of the test objects to their parents (n = 4) or only touched or took an object for themselves (n = 7) (which made it unclear whether their choice was

Results

The object that infants gave to E differed significantly between conditions, χ2(1, 36) = 4.00, p = .046, ω = .33 (see Fig. 2). In the true-belief condition 66.7% of infants correctly gave E the spoon, whereas in the false-belief condition 66.7% of infants correctly gave E the block. Most infants (86%) responded after only A’s encouragement alone without any adults following them. Very similar results were found when excluding the 5 infants who were accompanied by adults; in the true-belief condition

Discussion

These results demonstrate false-belief understanding in infants using an unexpected-contents helping task. In the true-belief condition, in which the experimenter knew that there was a spoon in the target box, most infants gave her the spoon, whereas in the false-belief condition, in which the experimenter falsely believed that there was a block in the target box, most infants gave her the block (despite needing to set aside their own knowledge that the target box she was reaching for actually

Acknowledgments

We thank Franziska LeBoerk, Karen Kühn, Yvonne Naumann-Sparschuh, Jana Lenke, Irka Wetzig, Tatjana Krämer, Tabea Niche, and Elena Rossi for help with data collection and coding.

References (31)

  • J. Hogrefe et al.

    Ignorance versus false belief: A developmental lag in attribution of epistemic states

    Child Development

    (1986)
  • B. Knudsen et al.

    Eighteen- and 24-month-old infants correct others in anticipation of action mistakes

    Developmental Science

    (2012)
  • B. Knudsen et al.

    18-month-olds predict specific action mistakes through attribution of false belief, not ignorance, and intervene accordingly

    Infancy

    (2012)
  • Á.M. Kovács et al.

    The social sense: Susceptibility to others’ beliefs in human infants and adults

    Science

    (2010)
  • A. Leslie

    ToMM, ToBy, and Agency: Core architecture and domain specificity

  • Cited by (0)

    View full text