Elsevier

Cognition

Volume 217, December 2021, 104878
Cognition

Original Articles
Events structure information accessibility less in children than adults

https://doi.org/10.1016/j.cognition.2021.104878Get rights and content

Abstract

To manage the onslaught of continuously unfolding information in our complex environments, we adults are known to carve up our continuous experience into meaningful events, a process referred to as event segmentation. This segmentation directly shapes how our everyday experiences are construed: content experienced within an event is held mentally in an accessible state, which is then dropped after an event boundary. The greater accessibility of event-specific information has been shown to influence—at its most basic level—how information is processed and remembered. However, it is as yet unknown if accessibility is similarly influenced by event boundaries in children, who are still developing the working memory capacity and semantic knowledge thought to support event segmentation. Here, we tested seven- to nine-year-old children's and adults' recognition of objects experienced either within or across event boundaries of two cartoons. We found that children and adults were both more accurate and faster to correctly recognize objects that last occurred within events versus across event boundaries. We, however, additionally observed an interaction such that children's access to recent experience was less influenced by event boundaries than adults'. Thus, while the spontaneous segmentation of complex events emerges by middle childhood, event structure shapes the active contents of children's minds less reliably than adults'.

Introduction

One of the longest standing questions in psychology is how children make sense of the constant stream of information in our “blooming and buzzing”, ever-changing worlds (James, 1890). While discoveries of previously unimaginable learning systems (Saffran, Aslin, & Newport, 1996) and a greater understanding of children's input (Smith, Jayaraman, Clerkin, & Yu, 2018) have progressed us toward an answer, a pertinent and fertile research area in adults has been surprisingly underexplored in children – event segmentation theory (EST; Zacks, Speer, Swallow, Braver, & Reynolds, 2007). According to EST, adults carve experience into structured units by actively maintaining event models in working memory (Zacks et al., 2007; Zacks, Tversky, & Iyer, 2001).These experience-based models dynamically incorporate ongoing experiences with our rich semantic knowledge in real-time to predict how events will unfold. In turn, this segmentation influences how we construe experience in the moment and later remember its temporal aspects (DuBrow & Davachi, 2013; DuBrow & Davachi, 2014; DuBrow & Davachi, 2016; Ezzyat & Davachi, 2011; Ezzyat & Davachi, 2014; Heusser, Ezzyat, Shiff, & Davachi, 2018; Horner, Bisby, Wang, Bogus, & Burgess, 2016; Sols, DuBrow, Davachi, & Fuentemilla, 2017).

Despite EST's potential to illuminate how children make sense of and remember their experience, surprisingly little is known about whether and how event models structure experience across childhood. And, there is good reason to expect that children would maintain active event models very differently from adults. In particular, children's working memory capacities (Gathercole, Pickering, Ambridge, & Wearing, 2004) and semantic knowledge (Li et al., 2004) are much reduced relative to adults. They may, thus, be less likely to spontaneously generate event models to organize information during ongoing event perception. And when they do, their models may be both impoverished in their content and more short lived than adults'. If true, this means that, given the exact same experience as adults, children would organize this experience differently, make sense of it differently, and learn from it differently.

While little is understood about whether and how children spontaneously generate and maintain active event models, like adults, infants and children do appear to segment their ongoing experience. Indeed, infants have been shown to use input statistics to parse sequences of events (Roseberry, Richie, Hirsh-Pasek, Golinkoff, & Shipley, 2011; Stahl, Romberg, Roseberry, Golinkoff, & Hirsh-Pasek, 2014). After habituating to clips depicting different versions of the same event, infants looked longer at clips that contained pauses that interrupted an event than those that had pauses at event boundaries (Baldwin, Baird, Saylor, & Clark, 2001). Children also appear to segment their experience. For example, 3- and 4-year-olds took longer to advance a pictorial story at event boundaries (Meyer, Baldwin, & Sage, 2011). Also, kindergarten and primary school students reported boundaries when watching video clips, but did so less often and less consistently than high-schoolers and adults (Glebkin et al., 2019). Finally, 5- to 7-year-olds were found to place boundaries in locations similar to adults, but were more variable relative to both other children and adults (Zheng, Zacks, & Markson, 2020). Thus, both infants and children actively segment their ongoing experience, albeit more variably than adults.

These findings may suggest the possibility that children use event models to interpret their ongoing experience, despite their limited working memories and semantic knowledge, but this does not have to be the case. Consistent with developmental accounts of cognitive control (Blackwell & Munakata, 2014; Chatham, Frank, & Munakata, 2009; Munakata, Snyder, & Chatham, 2012), infants' and children's demonstrations of event segmentation could be achieved through a simpler, reactive, and data-driven mechanism, that is, through noting shifts in sensory information and even shifts in actor intentions or narrative structure after events have ended. This could lead them to draw a boundary—or indicate a break in event structure—without having to hold an event model in mind that, rich in semantic knowledge, predicts how events will unfold. Notably, while we know that adults spontaneously and proactively maintain event models (Carroll & Bever, 1976; Speer & Zacks, 2005; Swallow et al., 2011; Swallow, Zacks, & Abrams, 2009; Zwaan, 1996), they can also use perceptual, data-driven approaches to segment events (Cutting, Brunick, & Candan, 2012; Hard, Tversky, & Lang, 2006; Huff et al., 2017; Newtson, Engquist, & Bois, 1977), making this a likely possible mechanism for children. Importantly therefore, studies documenting that children have similar (but more variable) event segmentation behavior does not necessarily mean that children are as likely as adults to proactively keep event information active and more accessible in any given moment, leaving the nature of children's event models, and their tendency to use them, unclear.

Pertinent to this question, narrative comprehension research has shown that children can maintain situation models (Graesser, Singer, & Trabasso, 1994)—a type of event model that is not derived from live events or experiences (Radvansky & Zacks, 2011). One study demonstrated this by first showing children a physical model of a marketplace, then covering it while children performed a self-paced reading task while being interrupted periodically to answer questions. During these interruptions, children of all tested ages (9–16 years) more rapidly accessed information about marketplace locations that the protagonist would have passed by (but which were not directly referenced in the text) than those that were not relevant to the narrative (Barnes, Raghubar, Faulkner, & Denton, 2014). These findings suggest that children from 9-years on are capable of actively maintaining event models while reading. Relatedly, because interventions in the use of situation-models are found to enhance reading comprehension (Glenberg, Gutierrez, Levin, Japuntich, & Kaschak, 2004; Glenberg, 2008; Glenberg, Goldberg, & Zhu, 2009; Marley, Levin, & Glenberg, 2007), this burgeoning use of situation models may help children make sense of narratives.

However, do children also spontaneously maintain these models during live experience, as has been observed in adults (Swallow et al., 2009)? This is an important question because if the construction of event models is more costly for children (e.g., soaking up too much of their limited working memory resources), then their tendency to spontaneously generate them may be proportionate to their benefits. These benefits are more obvious when comprehending written narratives—which are only given life in our mind's eye—than in more real-to-life experience (like movie watching), where the already rich multimodal input provides more bottom-up support for comprehension.

The spontaneous use and updating of event models has been studied in adults by focusing on the accessibility of recently encountered information (Carroll & Bever, 1976; Speer & Zacks, 2005; Swallow et al., 2011, Swallow et al., 2009; Zwaan, 1996; also see Shin and DuBrow, 2021 for a review). In these clever designs, participants are asked questions about what was experienced seconds ago while manipulating whether an event boundary occurred in that intervening time. With this approach, participants have been shown to make more errors and respond more slowly to identify stimuli when an event boundary intervened, even when the time lapsed is the same. Notably, this is true when the probes (and lures) emphasize either perceptual (Heusser et al., 2018) or conceptual aspects of what has recently occurred (Swallow et al., 2009), and shows that information is more accessible from within an event, consistent with the active and dynamic use of event models during ongoing experience.

To understand whether children spontaneously generate and use event models, we likewise tested whether recently experienced information remains more accessible within an event versus across an event boundary. We chose to focus on children ages 7–9 years since this is a time during which both working memory (Gathercole et al., 2004) and semantic knowledge (Li et al., 2004) are developing at a fast pace but are still significantly reduced compared to adults. To measure the accessibility of information during spontaneous event segmentation, we asked children and adults to watch cartoons and interrupted them intermittently, asking them which of two objects they had recently encountered in the cartoon. Unbeknownst to them, half spanned event boundaries while half did not. To assess how event boundaries influenced the accessibility of this information, we focused on reaction time as our primary dependent variable and recorded response accuracy as a secondary measure. Given children's more limited working memory and their impoverished world knowledge, we predict that they will be less likely to generate and maintain event models, thus illuminating how information is managed and processed as children experience events.

Section snippets

Participants

A target sample size of 60 participants per age group (children and adults) was determined prior to data collection to achieve over 80% power in detecting age group differences with medium effect sizes (Cohen's d > 0.55), as there is no prior research investigating the development of this phenomenon. Notably, post hoc sensitivity analyses also confirmed that this sample size achieves over 80% power in detecting differences between event types within each age group, if these effect sizes fall in

Results

Object recognition accuracy was very high in children (Mean = 0.849, SD = 0.358) and adults (Mean = 0.899, SD = 0.301) across the two event types (within- vs. across-event boundary). Overall, 39 adults and 27 children were at ceiling (100% correct) for at least one of the two conditions. This ceiling effect and the coarse grain of accuracy as a binary variable make it a poor metric for answering the question of information accessibility. We, nonetheless, include it as a secondary dependent

Discussion

We found evidence that elementary school aged children do spontaneously generate mental models, even during passive movie viewing. Like adults, children's recognition was less accurate and slower for objects that occurred in the previous event as compared with the current event, even when the amount of time that elapsed since the most recent viewing of the object was matched. Of note, we also found that being asked about objects from the previous event slowed children's recognition less than

Author note

The authors thank Danielle Lim, Marlie Tandoc, Hannah Cho, Bharat Nadendla, and Gloria Rebello, for their assistance and support in running the study. This work was supported by the Natural Sciences and Engineering Research Council of Canada Discovery Grants (ASF:RGPIN-2016-05; KD:RGPIN/05582–201) and Canada Foundation for Innovation and Ontario Research Funds (ASF: 34947; KD:34479).

References (67)

  • L.B. Smith et al.

    The developing infant creates a curriculum for statistical learning

    Trends in Cognitive Sciences

    (2018)
  • T. Sonne et al.

    Occlusions at event boundaries during encoding have a negative effect on infant memory

    Consciousness and Cognition

    (2016)
  • N.K. Speer et al.

    Temporal changes as event boundaries: Processing and memory consequences of narrative time shifts

    Journal of Memory and Language

    (2005)
  • Y. Zheng et al.

    The development of event perception and memory

    Cognitive Development

    (2020)
  • D.A. Baldwin et al.

    Infants parse dynamic action

    Child Development

    (2001)
  • F.C. Bartlett

    Remembering: A study in experimental and social psychology

    (1932)
  • D. Bates et al.

    Fitting linear mixed-effects models using lme4

    Journal of Statistical Software

    (2015)
  • K.A. Blackwell et al.

    Costs and benefits linked to developments in cognitive control

    Developmental Science

    (2014)
  • G. Brod et al.

    Neural activation patterns during retrieval of schema-related memories: differences and commonalities between children and adults

    Developmental Science

    (2017)
  • G. Brod et al.

    The influence of prior knowledge on memory: A developmental cognitive neuroscience perspective

    Frontiers in Behavioral Neuroscience

    (2013)
  • J.M. Carroll et al.

    Segmentation in cinema perception

    Science

    (1976)
  • C.H. Chatham et al.

    Pupillometric and behavioral markers of a developmental shift in the temporal dynamics of cognitive control

    Proceedings of the National Academy of Sciences

    (2009)
  • D. Clewett et al.

    Transcending time in the brain: How event memories are constructed from experience

    Hippocampus

    (2019)
  • R. Core Team

    R: A language and environment for statistical computing

    (2020)
  • J.E. Cutting et al.

    Perceiving event dynamics and parsing Hollywood films

    Journal of Experimental Psychology: Human Perception and Performance

    (2012)
  • D. Davidson

    The role of schemata in Children’s memory

  • S. DuBrow et al.

    The influence of context boundaries on memory for the sequential order of events

    Journal of Experimental Psychology: General

    (2013)
  • S. DuBrow et al.

    Temporal memory is shaped by encoding stability and intervening item reactivation

    Journal of Neuroscience

    (2014)
  • Y. Ezzyat et al.

    What constitutes an episode in episodic memory?

    Psychological Science

    (2011)
  • R. Fivush

    Learning about school: The development of kindergartners’ school scripts

    Child Development

    (1984)
  • S.E. Gathercole et al.

    The structure of working memory from 4 to 15 years of age

    Developmental Psychology

    (2004)
  • M.A. Gernsbacher

    Language comprehension as structure building

    (1990)
  • S. Ghetti et al.

    The development of recollection and familiarity in childhood and adolescence: Evidence from the dual-process signal detection model

    Child Development

    (2008)
  • Cited by (0)

    1

    Katherine Duncan and Amy Finn made equal contributions to the present research.

    View full text